CN120102997A - A method and system for rapid detection of transient faults of power devices in a DC charging pile - Google Patents

Abstract

The invention discloses a method and a system for rapidly detecting transient faults of a power device of a direct current charging pile, comprising the steps of collecting a first characteristic of the direct current charging pile; the method comprises the steps of collecting first characteristics through a charging pile fault identification model, processing and analyzing the collected first characteristics, detecting changes of each charging stage and analyzing change rules of the first characteristics, monitoring abnormal fluctuation of charging current and charging voltage in a charging process based on the change rules, classifying fault types, analyzing the abnormal fluctuation through an abnormal detection model, and determining the fault types in the charging process. The invention can rapidly detect possible faults or abnormal states of the power device so as to intelligently sense faults in the charging process.

Description

Method and system for rapidly detecting transient faults of direct-current charging pile power device

Technical Field

The invention relates to the technical field of power detection, in particular to a method and a system for rapidly detecting transient faults of a power device of a direct current charging pile.

Background

Extensive research has been conducted in the technical field of detection of charging equipment in foreign countries, and many enterprises are devoted to research and develop charging pile detection products and have achieved remarkable results. For example, the company of Fluke (Fluke) in the united states is one of the leaders in the measurement field, and 6658A ac/dc charging pile detection devices have been successfully developed. The device integrates a universal meter and various testing instruments, writes an automatic detection program, can measure maximum voltage to 1000V, maximum current to 250A and measurement accuracy to 0.05 level. The device is suitable for authentication detection mechanisms such as power companies, metering homes and the like, and can be used for efficiently completing the detection of the electrical performance of the charging pile.

When optimizing the early warning rules, the primary challenge is comprehensive consideration of battery state changes in multiple scenarios. The diversity of charging scenarios results in variations in battery charging SOC being affected by factors such as charge rate, temperature, current, etc., which may be quite different in different scenarios. A prediction model of charging scene classification is established by a machine learning method through large-scale data collection and analysis so as to more accurately predict potential charging safety hazards.

Another challenge is to address the factor identification challenge of large amounts of unlabeled fault information. Failure factors that lack clear knowledge result in failure to definitively categorize or annotate the failure information. To apply feature engineering, cluster analysis and expert knowledge, a model is built by adopting a supervised learning method and a semi-supervised learning method, fault factors are identified and classified to form a fault feature library of the charging facility, and potential fault factors are analyzed and mined based on classification trees and the like to realize mining of potential fault feature factors of the charging pile.

The accuracy of predicting the potential safety hazard of charging is also an important issue. The change of the charging SOC of the battery is closely related to the potential safety hazard of charging, but the dynamics and complexity of the charging process pose challenges for accurate prediction. A real-time data acquisition technology is required to be adopted, a dynamic prediction model is established by combining a machine learning algorithm and model prediction, and abnormal behaviors in the charging process are classified and predicted, so that the recognition and early warning capability is improved.

Disclosure of Invention

This section is intended to outline some aspects of embodiments of the application and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section as well as in the description of the application and in the title of the application, which may not be used to limit the scope of the application.

The present invention has been made in view of the above-described problems occurring in the prior art.

Therefore, the invention provides a method and a system for rapidly detecting transient faults of a direct-current charging pile power device, which solve the problem of classifying and predicting abnormal behaviors in the charging process.

In order to solve the technical problems, the invention provides the following technical scheme:

in a first aspect, the invention provides a method for rapidly detecting transient faults of a direct current charging pile power device, which comprises the following steps:

collecting a first characteristic of the direct current charging pile;

Processing and analyzing the collected first characteristics through a charging pile fault identification model, detecting the change of each charging stage and analyzing the change rule;

Based on a change rule, monitoring abnormal fluctuation of charging current and charging voltage in the charging process, and classifying fault types;

and analyzing the abnormal fluctuation through an abnormal detection model, and determining the fault type in the charging process.

As a preferable scheme of the rapid detection method for transient faults of the direct current charging pile power device, the invention comprises the following steps:

the first characteristic includes a charging current and a charging voltage of the direct current charging post, and charge state information of the charging vehicle.

As a preferable scheme of the rapid detection method for transient faults of the direct current charging pile power device, the invention comprises the following steps:

the charging pile fault identification model comprises the following steps:

Taking the random forest, the K nearest neighbor and the extreme gradient lifting as sub-classifiers;

each sub-classifier calculates the probability of each type of fault occurring on the input first characteristics;

And carrying out weighted summation on the probability of various faults of the first feature calculated by each classifier, and determining the fault type with the highest summation result as the discrimination result of the recognition model.

As a preferable scheme of the rapid detection method for transient faults of the direct current charging pile power device, the invention comprises the following steps:

the charging stage comprises a starting stage, a constant current charging stage and a charging completion stage;

The fault types are classified into direct current bus faults, inter-charging electrode faults, direct current side outlet faults of the converter, load side faults and power supply side faults.

As a preferable scheme of the rapid detection method for transient faults of the direct current charging pile power device, the invention comprises the following steps:

The method for detecting the change of each charging stage and analyzing the change rule comprises the following steps:

If abnormal fluctuation of charging power occurs during the starting phase, the direct current bus fault is indicated;

if abnormal fluctuation of charging power occurs during the starting phase, the power supply side fault is indicated;

if during the start-up phase, a build-up stagnation of charging power occurs, indicating that a load side fault exists;

If the charging power is rapidly and greatly fluctuated and reduced and is lower than a preset safety threshold value when entering a constant current charging stage, the abnormal DC side outlet of the converter is indicated;

if the charging power frequently fluctuates and cannot reach the expected minimum point in the charging completion stage, it indicates that there is an inter-charging-electrode abnormality.

As a preferable scheme of the rapid detection method for transient faults of the direct current charging pile power device, the invention comprises the following steps:

The anomaly detection model includes modeling and training current and voltage data using an unsupervised learning algorithm.

As a preferable scheme of the rapid detection method for transient faults of the direct current charging pile power device, the invention comprises the following steps:

The analysis of the abnormal fluctuation through the abnormal detection model comprises the steps of inputting current and voltage data acquired in real time into the constructed abnormal detection model, identifying data points which do not accord with a normal mode, and marking the data points as abnormal.

In a second aspect, the present invention provides a system for rapidly detecting transient faults of a power device of a direct current charging pile, including:

the acquisition module is used for acquiring the first characteristics of the direct current charging pile;

The detection module is used for processing and analyzing the collected first characteristics through the charging pile fault identification model, detecting the change of each charging stage and analyzing the change rule;

the classification module is used for monitoring abnormal fluctuation of charging current and charging voltage in the charging process based on a change rule and classifying fault types;

and the analysis module is used for analyzing the abnormal fluctuation through an abnormal detection model and determining the fault type in the charging process.

In a third aspect, the present invention provides a computing device comprising:

A memory for storing a program;

and the processor is used for executing the computer executable instructions, and the computer executable instructions realize the steps of the method for quickly detecting the transient faults of the direct current charging pile power device when being executed by the processor.

In a fourth aspect, the invention provides a computer readable storage medium, which comprises the steps of realizing the method for rapidly detecting transient faults of the power device of the direct current charging pile when the program is executed by a processor.

The method has the beneficial effects that the charging characteristics of the direct current charging pile are comprehensively and deeply analyzed by collecting the charging information in real time and utilizing the charging pile fault identification model and the abnormality detection model, an operator is helped to quickly detect the fault type in the charging process, the operation management of the charging pile is optimized, the operation efficiency and the service quality are improved, meanwhile, the layout of the charging pile can be optimized through the analysis of data in the charging process, the charging experience of a user is improved, the operation cost is reduced, and scientific and effective decision references are provided for planning and policy making of a charging network of a future electric vehicle.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. Wherein:

fig. 1 is a basic flow diagram of a method for quickly detecting transient faults of a power device of a direct current charging pile according to an embodiment of the present invention;

fig. 2 is a schematic technical circuit diagram of a method for quickly detecting transient faults of a power device of a direct current charging pile according to an embodiment of the present invention;

Fig. 3 is a charging power differential graph of a method for rapidly detecting transient faults of a power device of a direct current charging pile according to an embodiment of the present invention;

Fig. 4 is a schematic diagram of a soft classification-based charging pile fault recognition model of a method for rapidly detecting transient faults of a power device of a direct current charging pile according to an embodiment of the present invention;

Fig. 5 is a schematic diagram of a frame of a dc charging pile fault recognition model of a method for fast detecting a transient fault of a dc charging pile power device according to an embodiment of the present invention;

fig. 6 is a schematic diagram of normal probability distribution and cumulative distribution of a normal sample of a method for fast detecting transient faults of a power device of a direct current charging pile according to an embodiment of the present invention;

fig. 7 is a schematic diagram of failure probability cumulative distribution of a normal sample of a method for rapidly detecting transient failure of a power device of a direct current charging pile according to an embodiment of the present invention.

Detailed Description

So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.

Further, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic can be included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.

While the embodiments of the present invention have been illustrated and described in detail in the drawings, the cross-sectional view of the device structure is not to scale in the general sense for ease of illustration, and the drawings are merely exemplary and should not be construed as limiting the scope of the invention. In addition, the three-dimensional dimensions of length, width and depth should be included in actual fabrication.

Also in the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "upper, lower, inner and outer", etc. are based on the orientation or positional relationship shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

The terms "mounted, connected, and coupled" as used herein, unless otherwise specifically indicated and defined, shall be construed broadly and include, for example, fixed, removable, or integral, as well as mechanical, electrical, or direct, as well as indirect via intermediaries, or communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.

Example 1

Referring to fig. 1, for one embodiment of the present invention, a method for quickly detecting transient faults of a power device of a dc charging pile is provided, including:

s1, collecting a first characteristic of a direct current charging pile;

s2, processing and analyzing the collected first characteristics through a charging pile fault identification model, detecting the change of each charging stage and analyzing the change rule;

s3, based on a change rule, monitoring abnormal fluctuation of charging current and charging voltage in the charging process, and classifying fault types;

and S4, analyzing the abnormal fluctuation through an abnormal detection model, and determining the fault type in the charging process.

In the embodiment of the application, the current and the voltage of the direct current charging pile are monitored in real time, and the data mining and the pattern recognition technology are applied, so that the possible power device faults or abnormal states are rapidly detected by adopting an unsupervised learning method, and the faults in the charging process are intelligently sensed. Meanwhile, the battery state of health and the battery thermal coupling model are considered, the fault data of the charging pile and the battery failure mode are analyzed by utilizing an optimized machine learning algorithm, and the accuracy and the reliability of various early warning methods are evaluated.

Example 2

Referring to fig. 2-5, for an embodiment of the present invention, a method for quickly detecting a transient fault of a power device of a dc charging pile is provided based on the previous embodiment, including:

In the embodiment of the application, the charging facility is high-power output equipment, and the power device of the charging facility can be out of control under the conditions of aging and the like, so that potential safety hazards are caused. Transient fault prediction and fast sensing based on big data methods are critical to improve charging safety. By monitoring the current and the voltage of the direct-current charging pile in real time and applying the data mining and pattern recognition technology, an unsupervised learning method is adopted to rapidly detect possible faults or abnormal states of the power device so as to intelligently sense the faults in the charging process. Meanwhile, the battery state of health and the battery thermal coupling model are considered, the fault data of the charging pile and the battery failure mode are analyzed by utilizing an optimized machine learning algorithm, and the accuracy and the reliability of various early warning methods are evaluated. A technical circuit diagram of the fast detection of transient faults of the DC charging pile power device for early warning in the charging process is shown in fig. 2.

In the embodiment of the application, as shown in fig. 2, the intelligent sensing of the fault of the charging pile is specifically as follows:

According to the structure of the charging pile, the charging pile faults can be classified into a plurality of categories such as alternating current side faults, converter faults, direct current network faults, battery faults and the like. Specifically, the five faults represent a dc bus fault, a charging pole abnormality, an inverter dc side outlet abnormality, a load side fault, and a power supply side fault, respectively. In general, the probability of the cable line of the charging pile to fail is lower, and the direct current bus is more likely to fail. In addition, the malfunction of the dc converter may be timely shut off by the protection action of the internal switching element. As for the failure of the battery pack, the conditions of short circuit, overload, aging, breakdown, etc. of the battery are mainly included.

1) Abnormal charge curve analysis

The real-time charging power can be calculated through the charging current and voltage information monitored by the charging pile terminal in real time so as to reflect the electric quantity output condition of the charging pile in unit time. Meanwhile, the electric quantity input condition of the electric automobile can be known by combining real-time SOC information from the automobile end. In the normal charging process, the charging of the direct current charging pile can be divided into three different stages, each stage has specific charging characteristics, namely, the SOC value of the corresponding vehicle end in stage 1 is in the range of 0% -90%, the charging power of the pile end is kept at a relatively high level and slowly increases, the SOC value of the corresponding vehicle end in stage 2 is generally in the range of 90% -95%, the charging power of the pile end is quickly reduced from the highest value to the lower value, the SOC value of the corresponding vehicle end in stage 3 is close to 100%, and the charging power of the pile end is always kept at the lowest level. For comparison with normal charging laws, fig. 3 shows a partial fault type charging power differential graph. During the fault charging process, the power difference value will drop significantly in advance, and a large fluctuation occurs around the value 0. Such a trend of variation provides a powerful clue to identify and monitor the type of fault of the charging pile.

2) Construction of fault identification model of charging pile

Referring to the schematic diagram of the soft classification-based charging pile fault identification model shown in fig. 4, a direct current charging pile fault identification model is established by means of an integration method of Random Forest (RF), K Nearest Neighbor (KNN) and extreme gradient lifting (XGBoost). In this model, RF, KNN, and XGBoost act as sub-classifiers, each of which computes the probability of each type of fault occurring for the input samples. The integrated model performs weighted summation on the various fault probabilities of the samples calculated by the classifiers, so that the fault type with the highest summation result is determined as the discrimination result of the identification model. The integration method combines the advantages of different sub-classifiers and improves the accuracy and the robustness of the fault identification model

In the data acquisition stage, the direct current charging pile only outputs current and voltage values of A phase, and the current and voltage values of B phase and C phase are always zero. And taking the A-phase current, the A-phase voltage and the SOC record data of the direct current charging pile in each charging process as independent sample information. For each sample, the maximum value of the current and voltage sequences is extracted and a differential sequence of powers is calculated, and these values are then combined with the SOC sequence of the corresponding charging process to form the charging characteristics of the sample. And meanwhile, taking basic information such as rated power, maximum output voltage and the like of the direct current charging pile as input characteristics of a fault identification model. To build the fault recognition model, a kernel based on "rf+knn+ XGBoost" integration is selected. The model is capable of outputting probability values for each type of fault occurrence. In the whole, a frame schematic diagram of the direct-current charging pile fault identification model is shown in fig. 5, key information such as current, voltage and SOC is fully considered, and a reliable basis is provided for accurately identifying faults.

In the embodiment of the application, as shown in fig. 2, the charge early warning research and evaluation index are researched, specifically as follows:

The early warning threshold is closely connected with the accuracy and the early warning time of the early warning, namely the early warning accuracy of the model is improved along with the increase of the early warning threshold, and the reliability of the early warning is correspondingly improved, however, the early warning time is shortened along with the increase of the early warning threshold, and the time of actually informing in advance is reduced, so that the practical value of the early warning is reduced. Because different types of faults have respective characteristics, independent pre-warning thresholds should be set for the different fault types. To set these thresholds, statistical methods are employed. And fitting various fault probability values obtained by calculating a model of the normal sample of the non-fault charging pile to obtain a probability density function, and finally calculating a corresponding early warning threshold value.

In the embodiment of the application, as shown in fig. 2, in the rapid fault detection and intelligent sensing module for charging current and voltage, by analyzing charging characteristics, possible power device faults or abnormal states are inferred, so that rapid fault detection and intelligent sensing for charging current and voltage are realized. The charging AC/DC information of the DC charging pile is monitored in real time, and the method comprises the following steps:

1) Power device fault or abnormal state analysis

The charging pile is used as a key electric facility of the electric automobile, the working environment of the charging pile is bad, and various power device faults or abnormal states can be encountered in daily operation. These conditions may include:

(a) Overload. Long-term high-load operation, such as high-power charging demands exceeding the design load capacity of the charging pile, may cause thermal damage to the circuit board, overload of electronic components, and even damage to critical components.

(B) And (5) overheating. Prolonged operation or poor heat dissipation in high temperature environments may cause internal components to overheat, causing electronic component failure and even cable or connector damage.

(C) The circuit is short-circuited or open-circuited. Short circuits or open circuits of the circuit elements inside the charging post may cause abnormal operation of the charging post, such as interruption of charging or reduction of charging rate.

(D) The voltage is abnormal. Abnormal voltage fluctuations may suggest that internal circuit problems, such as power management unit failure, may affect charging efficiency or safety.

(E) The current is abnormal. Abnormal current output may represent a current sensor fault or power supply problem, which may affect charging efficiency or safety.

(F) Communication failure. A communication failure of the charging stake with an external system may result in data transmission errors or the inability to remotely monitor and manage the charging stake.

(G) The controller fails. Failure of the controller unit or related circuitry may result in inefficient charging or an inability to properly complete the charge.

According to the structure of the charging pile, the charging pile faults are divided into alternating current side faults, converter faults, direct current network faults, battery faults and the like. As shown in the figure, five faults respectively represent a dc bus fault, a charging pole abnormality, an inverter dc side outlet abnormality, a load side fault, and a power supply side fault. Generally, the fault probability of a cable line of the charging pile is low, the direct current bus is more prone to fault, the fault of the direct current converter can be timely removed by the protection action of an internal switching element, and the faults of the battery pack are mainly short circuit, overload, aging, breakdown and the like of the battery.

2) Current-voltage monitoring and charging profile analysis

And (3) collecting real-time data, and monitoring the current and the voltage of the direct-current charging pile in real time by using a sensor. The current sensor and the voltage sensor are embedded in the charging pile or on a charging circuit, capture charging AC/DC information and convert analog signals into digital signals.

And (3) data analysis and feature extraction, and processing and analyzing the acquired current and voltage data. Charging characteristics, such as charging phase, stability, etc., are extracted using data processing techniques, including filtering, fourier transforms, etc. And detecting the change of the charging stage, such as a starting stage, a constant current charging stage and a charging completion stage, and analyzing the change rule.

Abnormality detection and fault diagnosis, an abnormality detection model is established, and abnormal fluctuation of current and voltage or unexpected modes in the charging process are monitored. By comparison with the normal mode, a possible power device failure or abnormal state is identified. Machine learning algorithms, such as support vector machines, neural networks, etc., are employed to rapidly detect and diagnose anomalies in charging current and voltage.

Intelligent sensing and quick detection, and a model is built to sense and analyze charging current and voltage in real time by using an intelligent sensing technology. The rapid detection of current and voltage anomalies is achieved using a rapid detection algorithm, such as threshold-based anomaly detection.

The method combines real-time data acquisition and analysis, abnormal detection and establishment of a fault diagnosis model, can realize monitoring of charging AC/DC information of the DC charging pile and analysis of charging characteristics, and can infer possible faults or abnormal states of the power device, thereby realizing rapid detection and intelligent sensing of charging current and voltage.

3) Unsupervised learning algorithm application

Unsupervised learning is a branch of machine learning whose goal is to discover patterns and structures from unlabeled data. The main principle is based on statistical properties of the data, independent of labels or known results. Based on this, unsupervised learning can be used to infer power device faults or abnormal conditions and to label them. The operation flow comprises the following steps:

(a) Data collection and preprocessing. First, data is collected and preprocessed. This involves flushing the data, filling in missing values, processing outliers, and performing normalization and normalization operations to ensure data quality and consistency for algorithmic processing.

(B) Algorithms and models are selected. Next, an appropriate unsupervised learning algorithm or model is selected based on the task and data characteristics. For example, a K-means clustering algorithm may be selected for the clustering task, and a Principal Component Analysis (PCA) algorithm may be used for the dimension reduction task.

(C) And (5) model training. The data is trained using the selected algorithm. The algorithm attempts to find patterns or structures in the data, fits according to the characteristics of the data, and creates a corresponding model.

(D) Evaluation and analysis. The performance of the model is evaluated and the patterns or clusters generated are interpreted to obtain potential information behind the data. The evaluation may include an internal index (e.g., profile coefficient) and an external index (e.g., cluster accuracy), and the interpretation involves analysis and understanding of the cluster result or feature dimension reduction.

Unsupervised learning is widely used in a variety of fields, and its cluster analysis can group data into similar groups, which is suitable for market segments and social network analysis. Anomaly detection can discover anomalies in data such as network security and financial fraud detection. At the same time, it performs equally well in terms of dimension reduction and feature extraction for data compression and extraction of features most relevant, such as image and speech recognition. Through association rule mining, it helps discover associations in data, such as shopping cart analysis and recommendation systems. Unsupervised learning is a powerful tool for processing large amounts of unlabeled data, discovering potential information and knowledge, and provides a powerful means for deep knowledge of data structures and patterns.

4) Power device fault or abnormal state inference

4.1 Method for diagnosing fault or abnormal state of charging pile power device

Real-time monitoring and data analysis. And the sensor is used for monitoring data such as current, voltage and the like of the charging pile in real time. Irregular patterns or anomalies may be found by data analysis techniques such as anomaly detection and pattern recognition. Abnormality detection is performed by using time sequence analysis, a machine learning algorithm, etc., and possible fault signals are identified.

Fault codes and alarms. The charging stake is typically designed with a fault code or alarm system. When the system detects an abnormal condition, a corresponding code or alarm is generated to remind the user of the problem and further check or process the problem.

And (5) remote diagnosis. The data and status of the charging stake can be accessed remotely through a remote monitoring system. This approach allows remote diagnostics and anomaly analysis to discover and resolve potential problems in time, reducing the impact of faults on the user.

Professional detection and maintenance. Periodic professional equipment detection and maintenance is critical to preventing failure. Periodic equipment inspection, maintenance and maintenance are carried out by professional maintenance personnel for periodic detection and fault removal, so that potential problems can be prevented and solved, and the reliability and continuous operability of the charging facility are improved.

By comprehensively using the method, the fault or abnormal state of the charging pile power device can be timely found and diagnosed, and the safe and stable operation of the equipment is ensured.

And 4.2, deducing a power device fault or abnormal state based on an unsupervised learning algorithm, and realizing rapid detection and intelligent sensing of charging current and voltage. The inference process based on the unsupervised learning algorithm is as follows:

a) And (5) data acquisition and preprocessing. Real-time charging current and voltage data are collected and pre-processed, including outlier removal, normalization data, etc., to ensure data quality.

B) And (5) extracting characteristics. Useful features are extracted from the current and voltage data. Frequency domain features (e.g., frequency, spectrum), time domain features (e.g., waveform, periodicity), etc. may be utilized to characterize the current and voltage.

C) And (5) constructing an abnormality detection model. Current and voltage data are modeled and trained using an unsupervised learning algorithm, such as density-based anomaly detection (e.g., DBSCAN), distance-based anomaly detection (e.g., LOF, KNN), or cluster-based methods (e.g., K-means), among others.

D) Anomaly detection and intelligent sensing. In the real-time monitoring process, current and voltage data acquired in real time are input into a constructed abnormality detection model. The model will identify data points that do not conform to the normal mode and identify as abnormal. These outliers may indicate potential problems such as equipment failure, circuit problems, etc.

E) Fast response and processing. Upon detection of an anomaly, the system may automatically trigger an alarm or notify the relevant personnel to quickly respond and take appropriate action. Such as interrupting charging, safely stopping device operation, notifying maintenance personnel, etc.

F) Model optimization and feedback. According to the actual running condition and the abnormal feedback, the abnormal detection model is continuously optimized and updated so as to improve the detection accuracy and reliability.

The abnormal detection model constructed by the unsupervised learning algorithm can rapidly detect and intelligently sense the abnormal conditions of the charging current and the charging voltage, and provides support for real-time monitoring and fault diagnosis, so that the safe operation of the charging equipment is ensured.

In the embodiment of the application, as shown in fig. 2, in the charging early warning rule making and optimizing module, the accuracy and reliability of different early warning methods are evaluated by making and optimizing the charging early warning rule, and according to the charging pile and the battery thermal coupling model, the battery charging SOC change is considered, and three steps of battery safety and state monitoring, battery charging physical model construction and charging early warning rule making and optimizing are carried out. The method comprises the following steps:

1) Charging pile and battery thermal coupling model research

In the development of charging piles and battery technology, it is critical to build a thermal coupling model to achieve effective charge management and battery protection. The model relates to various considerations, from the measurement of thermal physical parameters to the establishment of a thermal conduction model to the development and optimization of a thermal coupling algorithm. The key steps of this process will be studied below and the key considerations for modeling in this field will be introduced.

(A) And (5) measuring a thermophysical parameter. First, the parameters are obtained, and the thermal physical parameters of the charging pile and the battery, such as thermal conductivity, specific heat capacity, etc., need to be obtained. These parameters can be obtained through laboratory testing or literature investigation. Then, the experiment verifies that the accuracy and the reliability of the parameters used in the experiment verification model are ensured. For example, thermal conductivity may be obtained by thermal-container experiments or thermal-conduction experiments.

(B) And (5) installing a temperature sensor. First, sensor location selection and placement, mounting temperature sensors on the charging post and battery, and selecting appropriate locations to accurately monitor temperature changes. Optimizing the sensor arrangement may maximize capture of changes in temperature distribution. Then, accurate data acquisition is performed, and the acquired temperature data needs to have high accuracy and high sampling rate in order to establish an accurate thermal coupling model.

(C) And (5) establishing a heat conduction model. First, a model is selected, a heat conduction model can be established by using a heat conduction equation or finite element analysis, etc., and the heat exchange process and the heat coupling effect between the charging pile and the battery are considered. And then, parameter adjustment, namely, adjusting the model parameters to improve the accuracy and fidelity of the model through comparing the model with actual data.

(D) Thermal coupling algorithm. Firstly, integrating models, combining thermal physical models of a battery and a charging pile, developing a thermal coupling algorithm, and simulating temperature change in a charging process. Then, thermal management strategies are developed, and algorithms are used to perform thermal management control to optimize charging efficiency, extend battery life, and predict and address potential thermal issues.

(E) Verification and optimization. Firstly, experimental data verification is carried out, the established thermal coupling model is compared with actual data for verification, and parameters and assumptions in the model are corrected to improve the accuracy and reliability of the model. And then, carrying out iterative optimization, continuously optimizing the model, and ensuring that the model is consistent with the actual situation and has predictability by using actual feedback and continuously collected data.

2) Battery SOH estimation and SOC estimation

(A) The state of health estimation (SOH) of a lithium ion battery is an important index for evaluating the service life and performance degradation degree of the battery, and when the battery is charged, the state of health of the battery is considered, and the change of the charging SOC of the battery is fused, so that the faults of a charging pile and the failure condition of the battery can be reduced. The SOH is typically estimated using several methods:

a) Capacity fade analysis. The number of charge and discharge cycles and capacity loss of the battery were recorded by the periodic charge and discharge test. By tracking the difference between the rated capacity and the actual capacity of the battery, the capacity fade rate can be estimated to infer SOH.

B) And (5) testing internal resistance. The internal resistance has the greatest effect on the battery performance. Internal resistance testing is a commonly used evaluation method, which can measure internal resistance through techniques such as alternating current impedance spectroscopy (AC IMPEDANCE) and the like, and evaluate the health status of the battery according to the internal resistance.

C) Temperature and cycle number monitoring. The operating temperature and cycle times of the battery are tracked. High temperatures and excessive charge and discharge cycles can lead to battery aging. Recording these parameters helps to assess the state of health of the battery.

D) And (5) a prediction model. The historical data and the monitored parameters are utilized to predict the state of health of the battery based on a physical model or data driven approach. These models may include machine learning and statistical methods such as neural networks, regression analysis, and the like.

E) And (5) comprehensive evaluation. And combining various monitoring parameters with the model, and adopting a comprehensive evaluation method, such as weighted average or multi-index fusion, to obtain the evaluation of the overall health state of the battery.

(B) The state of charge (SOC) estimation of the lithium ion battery is an important component for realizing a Battery Management System (BMS), and the SOC of the battery is estimated in real time in the charging process, so that the charging strategy is timely adjusted, the safety of the battery is better ensured, and the service life of the battery is prolonged. The following are some common methods of estimating SOC:

a) Voltage method. SOC is estimated by voltage measurement. This is one of the most commonly used methods, which uses the relationship between battery voltage and SOC during charge and discharge. But is affected by factors such as charge-discharge rate, temperature, capacity fading, etc., the estimation accuracy is to be improved.

B) Ampere-hour method. The SOC is estimated based on the charge-discharge amount accumulation of the battery. The method uses the product of current and time to track the charge and discharge of the battery, and estimates the SOC by calculation. Over time, errors in this approach may accumulate.

C) Filtering and kalman filtering. These methods use filters to process measured data of current, voltage, etc. to improve estimation accuracy of SOC. The Kalman filtering method is particularly suitable for a dynamic system, and can consider priori information and noise to improve estimation accuracy.

D) Equivalent circuit model method. Based on an equivalent circuit model of the battery, parameters such as internal resistance of the battery, voltage of the battery, open circuit voltage and the like are adopted to estimate the SOC by combining current and voltage measurement data. This approach is generally accurate but requires appropriate model parameters.

E) Deep learning and machine learning methods based on comprehensive methods. The SOC is predicted and estimated using deep learning and machine learning models in combination with multiple data sources and sensor measurements. These methods can utilize large amounts of data to improve accuracy and take into account more complex battery behavior.

3) Battery charging physical model construction

The construction of a battery charging physical model based on a machine learning algorithm is an important research direction, and aims to improve the safety in the battery charging process and early warn potential problems. Through collection, conversion and analysis of a large amount of charging data, a reliable physical model is constructed, and the prediction capability of the physical model is utilized to provide early warning and control for the battery charging process, so that the safety and efficiency of battery charging are improved.

(A) And (5) data collection. Data including parameters such as charging voltage, current, temperature and the like in the battery charging process are collected. The data should cover different battery types, operating conditions and charging modes to ensure that the model is representative and generalizing.

(B) Feature extraction and selection. Based on the collected data, feature extraction and selection is performed to convert the raw data into a feature set acceptable to the model. This includes frequency domain or time domain feature extraction and related feature selection methods.

(C) Model selection and training. An appropriate data-driven method, such as a neural network, support vector machine, regression model, etc., is selected to construct the model. The model is trained using the prepared dataset.

(D) Model evaluation and optimization. The model is evaluated, and its performance is evaluated using cross-validation or a method that retains a validation set, such as k-fold cross-validation. And optimizing according to the prediction accuracy, generalization capability and stability of the model. It may be necessary to adjust model parameters or try different model architectures.

(E) Model verification and application. And finally, verifying the optimized model, and checking the performance of the model on new data. If the model meets the expectations and has good prediction capability, the model can be deployed into practical application for physical model construction and prediction of the battery charging process.

4) Charging early warning rule formulation and optimization

(A) Charge-related data collection and analysis

The safety and performance of the charging pile ensure compliance with compliance standards and regulations. When formulating charging early warning rules, it is important to understand and comply with relevant charging standards and regulations. The collection and analysis of data are the basis, and relate to the collection of real-time parameters of a charging pile and a battery, and meanwhile, the typical behavior patterns and abnormal characteristics of the battery in different charging states are mastered through historical data analysis. In addition, considering the influence of the charging standard (regulation), sufficient charging standard research and compliance requirement analysis are performed to ensure that the charging process meets the safety and regulation requirements. And the standard and data analysis are comprehensively considered, so that more reliable and safe charging early warning rules can be formulated, and the safety and compliance of the charging process are ensured.

(B) Charging early warning rule formulation

When the early warning rule is determined, the early warning rule under different battery states needs to be considered. Indicators of SOC (state of charge), SOH (state of health), etc. may be used to set state evaluation rules, such as triggering an early warning when SOC falls below a certain threshold or SOH drops sharply. In addition, it is also important to develop rules for monitoring abnormal values, which cover abnormal temperature, current, voltage, etc. conditions to identify abnormal charge states. In addition, safety factors should be considered when setting the charging process rule. The conditions of too fast charging rate, too high temperature in the charging process, frequent disconnection of the charging pile and the like can have potential risks, and corresponding rules are formulated to warn and avoid the occurrence of potential dangerous conditions. The setting of the rules needs to comprehensively consider the actual conditions under different scenes and reasonably adjust according to experience and expertise so as to ensure the safety, stability and compliance of the charging process.

(C) Charging early warning rule optimization

The charging early warning rule can be optimized and adjusted by combining expert knowledge and experience rules through intelligent search algorithms such as genetic algorithm and a rule engine. First, a large amount of data generated during charging is collected and utilized as input to an optimization algorithm. Through intelligent algorithms such as genetic algorithms, an optimization model can be established according to a large amount of data and expert knowledge, and then the existing early warning rules can be evaluated and adjusted by using the models.

The genetic algorithm can be used for searching an optimal solution, and when the early warning rule is adjusted, the parameters of the early warning rule can be generated and modified by simulating the biological evolution process, and the parameters and the threshold of the rule are optimized so as to improve the performance and the accuracy of the rule. The rules engine may then help manage and execute these rules.

Expert knowledge and empirical rules refer to rules set based on expert understanding of the charging process and battery behavior. These rules may be empirical, based on industry standards, or obtained from an actual production environment. By combining the rules with the intelligent algorithm, the flexibility and adaptability of the early warning rules can be better improved, so that the early warning rules can be better adapted to different charging scenes and conditions.

5) Early warning research on fault and battery failure mode of charging pile

In the background of rapid development of the current electric vehicle technology, faults of a charging pile and battery failure have important influence on the service performance of the electric vehicle. Analyzing the fault data of the charging pile and the failure mode of the battery, and evaluating the accuracy and reliability of different early warning methods becomes a crucial task. In this process, a number of key steps are involved:

(a) Data collection and cleaning. Historical data of the charging pile and the battery are collected, wherein the historical data comprise parameters such as current, voltage, temperature, charging rate and the like in the charging process, and running state records of the charging pile. When cleaning data, care should be taken to handle outliers and missing data to ensure data quality.

(B) And (5) fault mode analysis. And exploring fault modes existing in the historical data by utilizing a data analysis technology such as a statistical method, cluster analysis, an anomaly detection algorithm and the like. By mining the data, different types of failure modes and possible failure features are identified to gain insight into possible problems.

(C) And (5) evaluating an early warning method. And selecting a proper early warning method to deal with the identified fault mode. This may involve the use of a rules engine, a machine learning model (e.g., decision tree, neural network), anomaly detection algorithms, and the like. The performance of different methods is evaluated, and the accuracy and the practicability of the method in fault detection and prediction are examined.

(D) Model verification and verification. The selected early warning method is applied to the real data, and the predicted result is compared with the actually occurring fault condition. Such verification helps evaluate the accuracy and reliability of the predictive model, discover the limitations of the model and make corrections.

(E) Improvement and optimization. And adjusting and optimizing the early warning model based on the verification result. It may be desirable to adjust model parameters, introduce new features, improve data processing methods, etc., to increase prediction accuracy and sensitivity to fault detection.

(F) And continuously monitoring and updating. And continuously monitoring the performance of the early warning method along with the change of the service conditions of the charging pile and the battery. The model and the data set are updated regularly, and the early warning system is continuously improved to adapt to the change of actual conditions.

The embodiment also provides a system for rapidly detecting transient faults of the direct current charging pile power device, which comprises:

the acquisition module is used for acquiring the first characteristics of the direct current charging pile;

The detection module is used for processing and analyzing the collected first characteristics through the charging pile fault identification model, detecting the change of each charging stage and analyzing the change rule;

the classification module is used for monitoring abnormal fluctuation of charging current and charging voltage in the charging process based on a change rule and classifying fault types;

and the analysis module is used for analyzing the abnormal fluctuation through an abnormal detection model and determining the fault type in the charging process.

Still further, still include:

A memory for storing a program;

And the processor is used for loading the program to execute the method for rapidly detecting the transient faults of the direct current charging pile power device.

The embodiment also provides a computer readable storage medium which stores a program, and when the program is executed by a processor, the method for quickly detecting the transient faults of the power device of the direct current charging pile is realized.

The storage medium provided in this embodiment belongs to the same inventive concept as the method for quickly detecting transient faults of the dc charging pile power device provided in the foregoing embodiment, and technical details not described in detail in this embodiment can be seen in the foregoing embodiment, and this embodiment has the same beneficial effects as the foregoing embodiment.

From the above description of embodiments, it will be clear to a person skilled in the art that the present invention may be implemented by means of software and necessary general purpose hardware, but of course also by means of hardware, although in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as a floppy disk, a read only Memory (ReadOnly, memory, ROM), a random access Memory (Random Access Memory, RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method of the embodiments of the present invention.

Example 3

Referring to fig. 6-7 and tables 1-2, a method for quickly detecting transient faults of a power device of a direct current charging pile is provided for one embodiment of the invention, and in order to verify the beneficial effects, comparison results of various schemes are provided.

In order to evaluate the performance of the direct current charging pile fault identification model, accuracy (Precision), recall (Recall), and F1-Score were used as evaluation indexes. The larger the values of these three indices, the better the model performance. Considering that the model involves the recognition of a plurality of failure tags, macro Average and Micro Average values of respective evaluation indexes are calculated as comprehensive evaluations of the failure recognition model performance. The numerical average reflects the average performance of various faults on the same type of index, and the weighted average comprehensively considers the duty ratio information of various samples, so that the performance of the model on the whole data set is comprehensively reflected.

Table 1 below summarizes the number of 12 charging pile faults, the data comprising historical information of over 6000 dc charging piles, including real-time acquisition, fault reporting, and equipment information, including real-time voltage, real-time current, and real-time SOC data. The real-time power is obtained through calculation based on the real-time current and the voltage, and then the power is subjected to first-order differential processing so as to analyze the charging power of the direct-current charging pile. According to the statistical result, more than 40 fault types are involved in the data, and through the selection of experienced staff, the fault types with higher fault frequency and relative importance are focused, so that the type and the occurrence frequency of the fault of the charging pile can be known in depth.

Table 1 fault type statistics table

The information such as maximum current, maximum voltage, power difference sequence, SOC sequence, rated power and maximum allowable output voltage of each charging process of the direct current charging pile is used as an input characteristic of one sample. Samples located in the fault start and end time periods are marked by fault information. In the fault time period, if there is no relevant charging record, that is, the charging pile stops providing the charging service due to the fault, the charging process sample closest to the fault starting time is marked as a fault sample. The training set and the testing set are divided according to the proportion of 7:3, a five-fold cross validation method is adopted for training, and model parameters are optimized through grid parameters. Under the same data set, the integrated model proposed in this embodiment is compared with KNN, XGBoost, RF, and the training and testing effects of each model are shown in table 2 below:

table 2 model performance comparison table

The result shows that compared with a fault identification model of a single algorithm, the integrated model has higher accuracy and more excellent performance, so that the effectiveness of the model in the aspect of fault diagnosis of the charging pile is verified.

In addition, the integrated model is used for predicting the charging pile samples which do not fail, and the model presents probability distribution and probability accumulation distribution of the normal samples which do not fail, and is specifically shown in a normal probability distribution and accumulation distribution schematic diagram of the normal samples in fig. 6. In this case, 90% of the normal samples are predicted by the model to have a probability value of no occurrence of a fault exceeding 0.8, which indicates that the model has high accuracy in recognition.

In addition, in the prediction of the model, the cumulative probability distribution of occurrence of various faults in the normal sample is shown in the cumulative probability distribution diagram of faults in the normal sample of fig. 7. It was observed that the cumulative distribution of the types of faults approaches 100% at the locations where the model predictive probability values were low, further confirming the accuracy of the model.

It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.

Claims (10)

1. The method for rapidly detecting the transient faults of the direct current charging pile power device is characterized by comprising the following steps of:

collecting a first characteristic of the direct current charging pile;

Processing and analyzing the collected first characteristics through a charging pile fault identification model, detecting the change of each charging stage and analyzing the change rule;

Based on a change rule, monitoring abnormal fluctuation of charging current and charging voltage in the charging process, and classifying fault types;

and analyzing the abnormal fluctuation through an abnormal detection model, and determining the fault type in the charging process.

2. The method for rapidly detecting transient faults of a direct current charging pile power device according to claim 1, wherein the first characteristic comprises charging current and charging voltage of the direct current charging pile and charging state information of a charging vehicle.

3. The method for rapidly detecting transient faults of a direct current charging pile power device according to claim 1 or 2, wherein the charging pile fault identification model comprises the following steps:

Taking the random forest, the K nearest neighbor and the extreme gradient lifting as sub-classifiers;

each sub-classifier calculates the probability of each type of fault occurring on the input first characteristics;

And carrying out weighted summation on the probability of various faults of the first feature calculated by each classifier, and determining the fault type with the highest summation result as the discrimination result of the recognition model.

4. The method for rapidly detecting transient faults of a direct current charging pile power device according to claim 3, wherein the charging stage comprises a starting stage, a constant current charging stage and a charging completion stage;

The fault types are classified into direct current bus faults, inter-charging electrode faults, direct current side outlet faults of the converter, load side faults and power supply side faults.

5. The method for rapidly detecting transient faults of a direct current charging pile power device according to claim 4, which is characterized by detecting changes of each charging stage and analyzing change rules thereof, comprises the following steps:

If abnormal fluctuation of charging power occurs during the starting phase, the direct current bus fault is indicated;

if abnormal fluctuation of charging power occurs during the starting phase, the power supply side fault is indicated;

if during the start-up phase, a build-up stagnation of charging power occurs, indicating that a load side fault exists;

If the charging power is rapidly and greatly fluctuated and reduced and is lower than a preset safety threshold value when entering a constant current charging stage, the abnormal DC side outlet of the converter is indicated;

if the charging power frequently fluctuates and cannot reach the expected minimum point in the charging completion stage, it indicates that there is an inter-charging-electrode abnormality.

6. The method for rapidly detecting transient faults of a direct current charging pile power device according to claim 5, wherein the anomaly detection model comprises modeling and training current and voltage data by using an unsupervised learning algorithm.

7. The method for rapidly detecting transient faults of a direct current charging pile power device according to claim 6, which is characterized in that the analysis of abnormal fluctuation through an abnormal detection model comprises the steps of inputting current and voltage data acquired in real time into a constructed abnormal detection model, identifying data points which do not accord with a normal mode, and marking the data points as abnormal.

8. A system based on the rapid detection method for transient faults of the direct current charging pile power device according to claim 1, which is characterized in that:

the acquisition module is used for acquiring the first characteristics of the direct current charging pile;

The detection module is used for processing and analyzing the collected first characteristics through the charging pile fault identification model, detecting the change of each charging stage and analyzing the change rule;

the classification module is used for monitoring abnormal fluctuation of charging current and charging voltage in the charging process based on a change rule and classifying fault types;

and the analysis module is used for analyzing the abnormal fluctuation through an abnormal detection model and determining the fault type in the charging process.

9. An electronic device, comprising:

A memory for storing a program;

a processor for loading the program to perform the steps of the method for fast detecting transient faults of a dc charging pile power device according to any of claims 1-7.

10. A computer readable storage medium storing a program, wherein the program, when executed by a processor, implements the steps of the method for fast detecting a transient fault of a dc charging pile power device according to any one of claims 1-7.