CN114398770B - A life prediction method for liquid particle counting sensors - Google Patents
A life prediction method for liquid particle counting sensors Download PDFInfo
- Publication number
- CN114398770B CN114398770B CN202111643921.6A CN202111643921A CN114398770B CN 114398770 B CN114398770 B CN 114398770B CN 202111643921 A CN202111643921 A CN 202111643921A CN 114398770 B CN114398770 B CN 114398770B
- Authority
- CN
- China
- Prior art keywords
- degradation
- working state
- particle counting
- counting sensor
- liquid particle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
- G01N15/10—Investigating individual particles
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
- G01N15/10—Investigating individual particles
- G01N2015/1026—Recognising analyser failures, e.g. bubbles; Quality control for particle analysers
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Analytical Chemistry (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- Dispersion Chemistry (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Testing Resistance To Weather, Investigating Materials By Mechanical Methods (AREA)
Abstract
本发明公开的一种用于液体颗粒计数传感器的寿命预测方法，属于液体颗粒计数传感器寿命预测领域。本发明以传感器内狭缝面积的形变比作为衡量指标，开展非工作状态和工作状态下的加速退化试验并收集指标的退化数据；根据试验所得数据和试验包含的使用、环境条件，建立液体颗粒计数传感器在两阶段下退化率与影响因子之间的函数关系；建立基于Wiener随机过程的两阶段退化模型，利用极大似然估计法进行参数估计，获得参数的全局最优解；根据液体颗粒计数传感器的常规贮存温度和工作额定流量条件，外推实际情况下非工作和工作两状态下的常规退化速率，进一步根据传感器内部狭缝面积形变的失效阈值，实现对液体颗粒计数传感器寿命与可靠性的预测。
The present invention discloses a life prediction method for a liquid particle counting sensor, which belongs to the field of life prediction of liquid particle counting sensors. The present invention uses the deformation ratio of the slit area in the sensor as a measurement index, conducts accelerated degradation tests in non-working and working states, and collects degradation data of the index; establishes a functional relationship between the degradation rate and the influencing factors of the liquid particle counting sensor in two stages based on the test data and the use and environmental conditions included in the test; establishes a two-stage degradation model based on the Wiener random process, uses the maximum likelihood estimation method to estimate parameters, and obtains the global optimal solution of the parameters; extrapolates the conventional degradation rate in the non-working and working states under actual conditions based on the conventional storage temperature and working rated flow conditions of the liquid particle counting sensor, and further realizes the prediction of the life and reliability of the liquid particle counting sensor based on the failure threshold of the deformation of the slit area inside the sensor.
Description
Technical Field
The invention relates to a life prediction method for a liquid particle counting sensor, in particular to a two-stage degradation modeling life prediction method based on a Wiener random process, which is suitable for accurately predicting the life of the liquid particle counting sensor and belongs to the field of life prediction of the liquid particle counting sensor.
Background
In a plurality of industries such as aviation, aerospace, ships, power plants and the like, the pollution degree detection of oil has high requirements. According to historical data, 60% -70% of hydraulic system faults are caused by pollution of oil, and the cleanliness of the oil is related to the safety and reliability of the whole system. At present, a liquid particle counter is a main testing device for detecting oil pollution, as shown in fig. 1, and is used for testing the diameter size and the size distribution of solid impurities in oil, and the weak and key components of the liquid particle counter are liquid particle counting sensors.
If the performance of the particle counting sensor fails or degenerates, the control command and the indication deviate, and the particle counter and even the whole hydraulic system eventually fail catastrophically, so that it is important to accurately evaluate the reliability of the particle counting sensor, and therefore, the particle counting sensor is replaced at proper time to avoid serious failure of the system, so that a degeneration modeling and service life method for attaching the actual use condition and environmental factors of the liquid particle counting sensor needs to be provided.
Existing studies on degradation models generally assume that the degradation process is controlled by a single process, and conventional single-stage degradation models are only suitable for the case that the degradation process is generally smooth and has small fluctuation, but many products have degradation characteristics which are obvious multi-stage characteristics, and conventional single-stage degradation modeling methods are not suitable. During the full life cycle of the liquid particle count sensor, its degradation process is composed of both the natural degradation process when not in operation and the wear process when in operation. During the storage period of the liquid particle counting sensor in the non-working state, natural degradation phenomenon occurs, wherein the natural degradation is represented by the reduction of the strength of the material, and finally the function of the sensor is disabled. When the liquid particle counting sensor is in a working state and oil to be measured passes through the sensor, solid particles in the oil can damage an internal slit of the sensor, the gradual accumulation of the damage can finally lead to deformation of a structure around the slit, and the deformation is increased to exceed a limit value, so that the sensor fails. Because the degradation processes of the liquid particle counting sensor in the non-working state and the working state have great difference, the whole degradation process of the liquid particle counting sensor is obviously two-stage degradation, and the service life of the liquid particle counting sensor is predicted by using the traditional single-stage degradation modeling method, so that the prediction result deviation is larger.
Meanwhile, the degradation model based on the random process is more suitable for describing the degradation process, and the degradation model introduced with the random process has good fitting capacity and statistical properties, and becomes a mainstream method of related researches. Degradation models based on wiener processes, gamma processes, inverse gaussian distributions are the most widely used degradation modeling methods. In consideration of the failure mechanism of the liquid particle counting sensor and certain randomness of external environmental conditions, a degradation model based on a Wiener random process is established, and the diffusion parameters contained in the Wiener random process are utilized to describe the fluctuation effect of the particle counting sensor caused by factors such as environment, manufacturing materials and the like in two states.
Currently, research on liquid particle count sensors is focused on improving structural processes to increase lifetime, but little research is done on liquid particle count sensor lifetime prediction. The existing method for carrying out degradation modeling and life prediction on various sensors is also focused on establishing a single-stage degradation model, a random process is not introduced into the degradation model, and the failure mechanism of the sensor and the randomness existing in external environmental conditions are ignored. The existing method is not attached to the actual situation of the whole life cycle of the liquid particle counting sensor, and the accurate prediction of the life of the liquid particle counting sensor cannot be realized.
Disclosure of Invention
The invention discloses a life prediction method for a liquid particle counting sensor, which is characterized in that degradation data are obtained by carrying out a grouping Constant Stress Acceleration Degradation Test (CSADT), a degradation model is established by respectively considering the characteristics of the degradation process of the sensor in a non-working state and a working state and influence factors thereof, the deformation ratio of the slit area in the sensor is selected as an index parameter for measuring the degradation degree, a functional relation between the degradation rate and the influence factors is established, meanwhile, a random process is introduced into the model to describe the fluctuation effect caused by uncertain factors and the difference of the degradation process among individuals, a two-stage degradation model is established based on the Wiener random process, the degradation influence factors of the non-working state and the working state of the sensor are comprehensively considered, the actual situation of the degradation process of the whole life cycle of the liquid particle counting sensor is more attached, and the efficient and accurate life prediction of the liquid particle counting sensor is realized.
The invention aims at realizing the following technical scheme:
The invention discloses a life prediction method for a liquid particle counting sensor, which takes the deformation ratio of the area of a slit in the sensor as an index parameter for measuring the degradation degree, develops an accelerated degradation test under a non-working state and a working state and collects degradation data of the index, establishes a functional relation between the degradation rate of the liquid particle counting sensor under two stages and an influence factor according to the obtained data and the use and environmental conditions contained in the test, further establishes a two-stage degradation model based on a Wiener random process, carries out parameter estimation by using a maximum likelihood estimation method to obtain a global optimal solution of the parameter, extrapolates the conventional degradation rate under the non-working and working states according to the conventional storage temperature and the working rated flow condition of the liquid particle counting sensor, and further realizes the prediction of the life and the reliability of the liquid particle counting sensor according to the failure threshold of the deformation of the area of the slit in the sensor.
The invention discloses a life prediction method for a liquid particle counting sensor, which comprises the following steps of:
Step one, by analyzing the characteristics of degradation process and the influence factors thereof in the non-working state and the working state, selecting the deformation ratio of the area of the slit in the sensor as an index parameter for measuring the degradation degree, respectively carrying out accelerated degradation tests in the non-working state and the working state, and collecting degradation data of the influence degree of the degradation influence factors in each state on the deformation ratio of the area of the slit.
The implementation method of the first step is as follows:
In order to quantify the degradation degree of the liquid particle counting sensor, the area deformation ratio X (t) of a slit is used as a degradation characteristic parameter by measuring the area of the slit in the sensor, and the degradation characteristic parameter X (t) is used as an index for representing the degradation state of the liquid particle counting sensor:
Where S (t) represents the area of the slit inside the sensor at time t and S 0 represents the initial area of the slit.
The liquid particle counting sensor is naturally degraded in the storage environment under the non-working state, and the phenomenon that the slit area is deformed under the storage environment of the liquid particle counting sensor is shown. The higher the storage temperature, the faster the intensity of the liquid particle count sensor slit material decreases, accelerating the change in the liquid particle count sensor slit area deformation ratio X (t).
On the other hand, the main degradation form of the liquid particle counting sensor in the working state is expressed as abrasion, when the detected oil passes through the sensor, solid particles in the oil can damage slits in the liquid particle counting sensor, the temperature and the flow are main factors influencing the abrasion rate of the slits in the sensor in the working state, namely the degradation process of the liquid particle counting sensor is influenced by the temperature and the flow of the detected oil, and on the premise that other conditions of the detected oil are unchanged, the higher the temperature and the higher the flow of the detected oil are, the faster the slit abrasion rate is, and the change of the slit area deformation ratio X (t) of the liquid particle counting sensor is accelerated.
And respectively carrying out accelerated degradation tests of the liquid particle counting sensor in a non-working state and a working state, and collecting degradation data of the degradation influence factors on the degree of the influence of the slit area deformation ratio. And in the non-working state, the degradation influencing factors are storage environment temperature, and in the working state, the degradation influencing factors comprise measured oil temperature and flow.
The controllability and operability of the environmental factors, as well as the sensitivity of the degradation rate to the environmental factors, are considered when performing accelerated degradation tests. Preferably, the single stress accelerated degradation test is performed in a non-working state and a working state, the working state uses the measured oil flow as the acceleration stress, the non-working state uses the storage environment temperature as the acceleration stress, the liquid particle counting sensor samples are grouped, and the grouping constant accelerated degradation test is performed (CSADT). And after the degradation test in the non-working state, collecting degradation data of the degradation influence factor storage environment temperature on the extent of the influence of the slit area deformation ratio, and after the degradation test in the working state, collecting degradation data of the degradation influence factor flow on the extent of the influence of the slit area deformation ratio.
As a further preferable mode, a plurality of particle counting sensors with consistent states are taken as samples, the temperature of the group I of tests is set to be T 1＜T2＜…＜Tl in the non-working state and is larger than the normal storage temperature T 0, the flow of the group I of tests is set to be v 1＜v2＜…＜vl in the working state and is larger than the rated flow v 0 of oil. In the implementation process of the accelerated degradation test, the area of the slit in the sensor of each group of test samples is collected at equal time intervals, and the accelerated degradation test data of each group of test samples, namely the area deformation ratio of the slit in the sensor, is obtained through calculation according to a formula (1).
And step two, dividing the degradation process into two degradation stages in the non-working state and the working state based on the characteristics of the degradation process and the influence factors thereof in the non-working state and the working state which are analyzed in the step one, simultaneously introducing random process description to the fluctuation effect caused by uncertain factors and the difference of the degradation process among individuals, and establishing a functional relation between the degradation rate and the influence factors of the liquid particle counting sensor in the two stages according to the degradation data of the influence degree of the degradation influence factors collected in the step one on the slit area deformation ratio, wherein the functional relation is a degradation rate functional model in the two states.
In consideration of controllability and operability of environmental factors and sensitivity of degradation rate to environmental factors when the accelerated degradation test is performed, it is preferable that the degradation test data obtained by the single stress acceleration test in the non-working and working states in the first step is corresponding.
The implementation method of the second step is as follows:
And 2.1, establishing a degradation rate function model in a non-working state.
Arrhenius acceleration model describes degradation rateRelationship with temperature T:
Wherein M is degradation amount of a certain characteristic value, r Arrhenius represents degradation rate at temperature T, k b is Boltzmann constant, C 0 is undetermined parameter, ea is reaction activation energy, and the same failure mode of the same product is constant by taking eV as a unit.
The liquid particle counting sensor generates a natural degradation process under the storage condition of a non-working state, a degradation rate function model under the non-working state is established based on the form of the Arrhenius acceleration model, the storage temperature is used as single acceleration stress under the non-working state, other influencing factors are combined into one influencing factor, and the degradation rate r 1 under the non-working state is expressed as:
Wherein T represents the storage environment temperature in a non-working state, and A and B are parameters to be estimated.
And 2.2, establishing a degradation rate function model in a working state.
Abrasion is an irreversible damage process of the surface of a material, and the abrasion loss Archard model is as follows:
Where r m denotes the wear rate, W denotes the wear amount, t denotes the wear time, k denotes the wear coefficient under certain conditions, P, v denotes the pressure on the friction surface and the relative linear velocity, and m and n denote the undetermined correlation coefficients.
The degradation of the liquid particle counting sensor in the working state is mainly expressed as abrasion, a degradation rate function model in the working state is established based on the form of the Arcard model, the flow of the detected oil passing through the slit is single acceleration stress, other influencing factors are combined into one influencing factor, and the degradation rate r 2 in the working state is expressed as:
r2＝C·(υ)D (3)
wherein, v represents the liquid flow under the working state, C, D is the parameter to be estimated.
And thirdly, accumulating the natural aging degradation effect in a non-working state and the degradation effect caused by abrasion in a working state based on an accumulated damage theory, and establishing a two-stage degradation model of the liquid particle counting sensor based on a Wiener random process according to the functional relation between the degradation rate and the influence factor of the liquid particle counting sensor under the two stages established in the second step.
The implementation method of the third step is as follows:
The two-stage degradation process processed based on the accumulated damage theory has different degradation rates, so that the two-stage Wiener process in the two-stage degradation model of the liquid particle counting sensor has different drift coefficients, and the drift coefficient term is represented by a step two degradation rate function. The diffusion parameters of the Wiener process represent the differences among individuals caused by materials and process levels, are used for describing the fluctuation characteristics of the whole degradation process, and can be regarded as being approximate.
Wherein r 1 represents the natural degradation rate in the non-working state, r 2 represents the degradation rate of wear in the working state, σ represents the fluctuation effect of the degradation process, and B (t) is the standard Brownian motion. The initial degradation amount X (0) =0, and X (J) is defined as the accumulated degradation amount corresponding to the operating state accumulation time t J as the initial degradation amount in the non-operating stage.
According to the functional relation between the degradation rate and the influence factor of the liquid particle counting sensor under the two stages, which is established in the second step, establishing a two-stage degradation model of the liquid particle counting sensor based on a Wiener random process is expressed as:
And step four, carrying out parameter estimation on the two-stage degradation model of the liquid particle counting sensor established in the step three, and carrying out parameter estimation by utilizing a maximum likelihood estimation method to obtain a global optimal solution of the two-stage degradation model parameters of the liquid particle counting sensor, wherein the parameters comprise all parameters in a degradation rate functional relation and diffusion parameters in a random process.
Carrying out parameter estimation on a two-stage degradation model of the liquid particle counting sensor by using a maximum likelihood estimation method, and obtaining a global optimal solution of model parameters in one step, wherein the implementation method in the fourth step is as follows:
The total sample size input in the degradation test is n, the test times of the area deformation ratio in the test are m times, the test times in the first stage, namely the accumulated non-working stage in the non-working state, are J times, the test times in the second stage, namely the accumulated working stage in the working state, are (m-J) times, and the test data obtained in the degradation test are expressed as:
The amount of degradation of sample i between the (j-1) th and the j-th tests is noted as Δx (t ij-ti(j-1))＝ΔXij, available based on the independent delta nature of the Wiener process:
ΔXij～N(r1·Δtj,σ2·Δtj)1≤j≤J
ΔXij～N(r2·Δtj,σ2·Δtj)J+1≤j≤m
In the whole life cycle of the particle counting sensor, when the ratio of the non-working state accumulation time to the working state accumulation time is known, parameter estimation is carried out by utilizing a maximum likelihood method in stages according to degradation data obtained in two stages, and likelihood functions are expressed as follows;
Maximizing the likelihood function L 1(·),L2 (,) to obtain the estimated values of the parameters A, B, C, D and sigma;
And fifthly, according to the parameters of the two-stage degradation model of the liquid particle counting sensor estimated in the step four, extrapolating the degradation rate under the non-working and working states of the actual situation, and according to the failure threshold value of the area deformation ratio of the slit inside the sensor, deducing a failure distribution function to realize the accurate prediction of the reliability and the service life of the liquid particle counting sensor.
The fifth implementation method comprises the following steps:
And extrapolating the degradation rate r s` of the particle counting sensor in the actual condition under the non-working condition according to the normal temperature condition during storage, and extrapolating the degradation rate r s2 of the particle counting sensor in the actual condition according to the rated flow of the tested oil.
The area deformation ratio of the degradation state of the particle counting sensor exceeding the failure threshold value W for the first time is regarded as the internal slit deformation is out of tolerance, the sensor is caused to be failed, and the corresponding time T w is regarded as the service life of the particle counting sensor:
Tw＝inf{t≥tJ:X(t)≥W}
R(t)=P(Tw≥t≥tJ)＝P(t≥tJ:X(t)≤W) (7)
based on the statistical characteristics of the two-stage Wiener random process, the time when the failure threshold value is exceeded for the first time obeys the two-stage inverse Gaussian distribution, and the derived failure distribution function is as follows:
Wherein, the An estimated degradation amount representing the turning moment at which the particle count sensor accumulates over the entire operating time, that is,
Predicting the lifetime of the liquid particle count sensor according to the failure distribution F (t), wherein the median lifetime t (0.5) of the particle count sensor is R (t) =1-F (t) =0.5.
The beneficial effects are that:
1. the invention discloses a life prediction method for a liquid particle counting sensor, which considers degradation influence factors in a non-working state and a working state, and respectively establishes a functional relation between degradation rate and storage temperature in the non-working state and a functional relation between degradation rate and rated flow of tested oil in the working state. By establishing a reasonable functional relation between the degradation rate and the actual storage environment and working conditions, the degradation rate under the extrapolation actual condition is more accurate, and the purpose of accurately predicting the service life of the particle counting sensor is achieved.
2. According to the life prediction method for the liquid particle counting sensor, differences between degradation processes of the particle counting sensor in a non-working state and a working state are considered, a non-working and working two-stage degradation model is established based on an accumulated damage theory, and compared with a single-stage degradation model, the life prediction method is closer to an actual degradation rule of the sensor, and accuracy of life prediction results is effectively improved.
3. According to the life prediction method for the liquid particle counting sensor, disclosed by the invention, the grouped constant acceleration degradation test is utilized to obtain degradation data, so that a large amount of degradation data can be obtained in a short time, and the efficiency of the test and life prediction is improved.
4. The life prediction method for the liquid particle counting sensor provides an assessment method for predicting the life of the particle counting sensor, avoids larger deviation when the pollution level test is carried out on oil, and provides guarantee for the safety of the whole hydraulic system.
Drawings
FIG. 1 is a schematic diagram of particle count sensor operation;
FIG. 2 is a schematic diagram of a general flow chart of a lifetime prediction method for a liquid particle count sensor in accordance with the present disclosure;
FIG. 3 is a simulation result of a non-operational phase temperature accelerated degradation test in an example of the present invention;
FIG. 4 is a simulation result of a flow acceleration degradation test in the working phase of the example of the invention;
FIG. 5 simulation results of a two-stage degradation process in the conventional case of the example of the present invention;
FIG. 6 is a graph comparing reliability predictions in an example of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
In order to verify the effectiveness and feasibility of the method, an accelerated degradation simulation test under the non-working state and the working state is firstly carried out. Further, according to the accelerated degradation simulation test data, a two-stage degradation model is established and a reliability prediction curve is obtained by using the life prediction method of the liquid particle counting sensor. And finally, comparing and analyzing test results, on one hand, verifying that the method is feasible by comparing with a K-M curve under the actual condition, and on the other hand, verifying that the method can improve the life prediction precision by comparing with a reliability prediction curve based on a single-stage degradation model.
The general flow of the method for predicting the lifetime of a liquid particle count sensor disclosed in this embodiment is shown in fig. 2, and the embodiment includes the following steps:
and 1, developing a grouping acceleration degradation simulation test in a non-working state and a working state, and collecting degradation data of the degradation influence factors in each state on the degree of influence of the slit area deformation ratio.
(1) Accelerated degradation simulation test in non-working state
The temperature is used as single acceleration stress in the non-working state, temperature acceleration test simulation in the non-working state is carried out, the acceleration tests are divided into 3 groups according to temperature gradients, the set temperatures are 50 ℃, 60 ℃ and 70 ℃, each group of samples are 4, the simulation test duration is 36 time units, other parameter settings in the degradation model and the degradation rate acceleration model are shown in table 1, degradation data obtained by the non-working state simulation test are collected, and 3 groups of randomly generated degradation data are shown in fig. 3.
TABLE 1 simulation parameter set for non-operational temperature acceleration test
| A | B | Sigma1 |
| 2000 | -3000 | 0.1 |
(2) Accelerated degradation simulation test for working state
In the working state, the oil flow is used as single acceleration stress, flow acceleration test simulation under the working state is carried out, the acceleration tests are divided into 3 groups according to flow gradients, the acceleration flow is set to be 60mL/min, 70mL/min and 80mL/min respectively, each group of samples is set to be 4, the simulation test duration is set to be 36 units, other parameters in the degradation model and the degradation rate acceleration model are set as shown in table 2, degradation data obtained by the working state simulation test are collected, and 3 groups of degradation data are randomly generated as shown in fig. 4.
TABLE 2 simulation parameter settings for flow acceleration test for operating conditions
| C | D | Sigma2 |
| 0.1 | 0.5 | 0.1 |
And 2, establishing a degradation rate function model of the liquid particle counting sensor in a non-working state and a working state.
(1) Establishing a degradation rate function model in a non-working state
Establishing a degradation rate function model under a non-working state, wherein the storage temperature is used as a single acceleration stress under the non-working state, other influencing factors are combined into one influencing factor, and the degradation rate r 1 under the non-working state is expressed as:
Wherein T represents the storage environment temperature in a non-working state, and A and B are parameters to be estimated.
(2) Establishing a degradation rate function model under the working state
Establishing a degradation rate function model under a working state, wherein the flow of the oil to be measured passing through the slit is single acceleration stress, other influencing factors are combined into one influencing factor, and the degradation rate r 2 under the working state is expressed as:
r2＝C·(υ)D (2)
wherein, v represents the liquid flow under the working state, C, D is the parameter to be estimated.
And 3, establishing a non-working and working two-stage degradation model based on a Wiener random process.
According to the functional relation between the degradation rate and the influence factor of the liquid particle counting sensor under the two stages, which is established in the step 2, establishing a two-stage degradation model of the liquid particle counting sensor based on a Wiener random process:
Wherein sigma represents the fluctuation effect of the degradation process, B (t) is standard Brownian motion, an initial degradation amount X (0) =0 is defined, and X (J) is the accumulated degradation amount corresponding to the accumulated time t J of the working state and is used as the initial degradation amount of the non-working stage.
And 4, carrying out parameter estimation by using a maximum likelihood estimation method to obtain a global optimal solution of the two-stage degradation model parameters of the liquid particle counting sensor.
Parameter estimation is carried out on the degradation model and the degradation rate acceleration model by using a maximum likelihood estimation method, and a maximum likelihood function is maximized according to simulation test data to obtain a parameter estimation result, wherein the parameter estimation result is shown in Table 3:
TABLE 3 parameter estimation results
| A | B | C | D | Sigma |
| 1871 | -2976 | 0.1020 | 0.4947 | 0.0966 |
And 5, extrapolating degradation rates of the actual condition under the non-working and working states, and predicting the reliability and the service life of the liquid particle counting sensor.
According to the conventional storage temperature and the oil nominal flow condition, the non-working degradation rate r s1, the working degradation rate r s2 and the Brownian motion diffusion parameter sigma s are calculated, when the proportion of the working state accumulation time to the non-working state accumulation time is 1/2, the simulation non-working state accumulation time is 200 units, the working state accumulation time is 100 units, and two-stage degradation data of 10 samples are generated, and the result is shown in fig. 5.
Calculating a failure distribution function:
Wherein, the An estimated degradation amount representing the turning moment at which the particle count sensor accumulates over the entire operating time, that is,
The area deformation ratio failure threshold value is set to be W=40 per mill in the test, the reliability prediction result is further obtained according to the failure distribution function, and a reliability result curve within 0-300 units of time is drawn, so that a two-stage reliability prediction curve is shown in fig. 6.
And 6, verifying the effectiveness and the accuracy of the life prediction method for the liquid particle counting sensor.
(1) Comparing the result of a lifetime prediction method for a liquid particle count sensor disclosed in this example with a K-M curve, and verifying the feasibility of the method
According to the time that the simulated degradation data actually exceeds the failure threshold value for the first time, a K-M comparison curve is drawn, the K-M comparison curve is shown in table 4 from small to large, and the K-M comparison curve is further drawn, and the result is shown as the K-M curve in fig. 6.
Table 4 sample life results
| 245 | 245 | 247 | 248 | 249 |
| 251 | 253 | 255 | 256 | 263 |
(2) Establishing a single-stage degradation model to obtain a single-stage reliability prediction result, comparing the single-stage reliability prediction result with the result of a life prediction method for a liquid particle counting sensor disclosed by the patent, and verifying the accuracy of the method
The method comprises the steps of establishing a single-stage degradation model, wherein the degradation process of the particle counting sensor is stable, the degradation rate is regarded as the whole process to be constant, the degradation rate is represented by r 0, and the single-stage degradation model of the particle counting sensor is as follows:
X(t)=X(0)+r0·t+σB(t) (5)
and carrying out parameter estimation of the single-stage degradation model according to the generated two-stage degradation data, further obtaining a single-stage reliability prediction result based on the single-stage degradation modeling method, and drawing a result curve, wherein the result is shown in the single-stage reliability prediction curve in fig. 6.
The comparative analysis of 3 curves in fig. 6 can prove that the reliability evaluation curve obtained based on the two-stage degradation modeling method disclosed by the patent is very close to the K-M curve, and the feasibility and the effectiveness of the method are verified. Meanwhile, compared with a reliability evaluation result obtained by the single-stage degradation model, the reliability evaluation method is obviously closer to a K-M curve, and the reliability evaluation method has higher accuracy compared with a single-stage degradation modeling and life prediction method.
While the foregoing detailed description has described the objects, aspects and advantages of the invention in further detail, it should be understood that the foregoing description is only illustrative of the invention, and is intended to cover various modifications, equivalents, alternatives, and improvements within the spirit and scope of the present invention.
Claims (8)
1. A life prediction method for a liquid particle count sensor is characterized by comprising the following steps,
The method comprises the steps of firstly, respectively carrying out accelerated degradation tests in a non-working state and a working state by analyzing characteristics of degradation processes and influence factors thereof in the non-working state and the working state, selecting the deformation ratio of the area of a slit in a sensor as index parameters for measuring the degradation degree, and collecting degradation data of the degradation influence factors in the non-working state and the working state on the influence degree of the deformation ratio of the area of the slit, wherein the degradation influence factors are storage environment temperature in the non-working state and comprise the temperature and flow of oil to be measured in the working state;
Dividing the degradation process into two degradation stages in the non-working state and the working state based on the characteristics of the degradation process and the influence factors thereof in the non-working state and the working state which are analyzed in the first step, introducing random process description to the fluctuation effect caused by uncertain factors and the difference of the degradation process among individuals, and establishing a functional relation between the degradation rate and the influence factors of the liquid particle counting sensor in the two stages according to the degradation data of the influence degree of the degradation influence factors collected in the first step on the slit area deformation ratio, wherein the functional relation is a degradation rate functional model in the two states;
Accumulating natural aging degradation effect in a non-working state and degradation effect caused by abrasion in a working state based on accumulated damage theory, and establishing a two-stage degradation model of the particle counting sensor based on a Wiener random process according to the functional relation between the degradation rate and the influence factor of the liquid particle counting sensor established in the second step;
step four, carrying out parameter estimation on the two-stage degradation model of the particle counting sensor established in the step three, and carrying out parameter estimation by utilizing a maximum likelihood estimation method to obtain a global optimal solution of parameters of the two-stage degradation model of the particle counting sensor, wherein the parameters comprise various parameters in a degradation rate function relation and diffusion parameters in a random process;
and fifthly, according to the parameters of the two-stage degradation model of the particle counting sensor estimated in the step four, extrapolating the conventional degradation rate under the non-working and working states of the actual situation, and according to the failure threshold value of the area deformation ratio of the slit inside the sensor, deducing a failure distribution function to realize the accurate prediction of the reliability and the service life of the liquid particle counting sensor.
2. A life prediction method for a liquid particle count sensor according to claim 1 wherein step one is accomplished by,
In order to quantify the degradation degree of the liquid particle counting sensor, the area deformation ratio X (t) of a slit is used as a degradation characteristic parameter by measuring the area of the slit in the sensor, and the degradation characteristic parameter X (t) is used as an index for representing the degradation state of the liquid particle counting sensor:
wherein S (t) represents the area of the slit inside the sensor at the moment t, and S 0 represents the initial area of the slit;
the liquid particle counting sensor is naturally degraded in a storage environment under a non-working state, and the deformation of the slit area of the liquid particle counting sensor is shown in the storage environment;
when the detected oil passes through the sensor, solid particles in the oil can damage slits in the liquid particle counting sensor, namely the degradation process of the liquid particle counting sensor is influenced by the temperature and the flow of the detected oil, and under the premise that other conditions of the detected oil are unchanged, the higher the temperature and the higher the flow of the detected oil are, the faster the slit abrasion speed is, and the change of the slit area deformation ratio X (t) of the liquid particle counting sensor is accelerated;
And respectively developing accelerated degradation tests of the liquid particle counting sensor in a non-working state and a working state, and collecting degradation data of the degradation influence factors on the degree of influence of the area deformation ratio of the slit, wherein the degradation influence factors are storage environment temperature in the non-working state, and the degradation influence factors comprise measured oil temperature and flow in the working state.
3. The method for predicting the service life of the liquid particle counting sensor according to claim 2, wherein controllability and operability of environmental factors and sensitivity of degradation rate to the environmental factors are considered when an accelerated degradation test is implemented, single stress accelerated degradation tests are respectively implemented in a non-working state and a working state, the measured oil flow is taken as acceleration stress in the working state, the storage environment temperature is taken as acceleration stress in the non-working state, liquid particle counting sensor samples are grouped, a grouping constant acceleration degradation test (CSADT) is implemented, degradation data of the degree of influence of the storage environment temperature of the degradation influence factors on the slit area deformation ratio is collected after the degradation test in the non-working state, and degradation data of the degree of influence of the oil flow of the degradation influence factors on the slit area deformation ratio is collected after the degradation test in the working state.
4. A life prediction method for a liquid particle counting sensor according to claim 3, wherein a plurality of particle counting sensors with consistent states are taken as samples, the temperature of a group of tests is set as T 1＜T2＜…＜Tl in an acceleration test under a non-working state and is larger than a normal storage temperature T 0, the flow rate of the group of tests is set as v 1＜v2＜…＜vl in an acceleration degradation test under a working state and is larger than the rated oil flow rate v 0, slit areas in the sensors of the test samples of each group are collected according to equal time intervals in the implementation process of the acceleration degradation test, and the acceleration degradation test data of the test samples of each group, namely the slit area deformation ratio in the sensor, is obtained according to a formula (1).
5. A life prediction method for a liquid particle count sensor according to claim 1, 2, 3 or 4, wherein the controllability and operability of environmental factors and sensitivity of degradation rate to environmental factors in performing accelerated degradation test are considered, and the degradation test data obtained by performing single stress acceleration test in the non-operating and operating states respectively in step one are corresponding to the degradation test data obtained by performing single stress acceleration test in the non-operating and operating states,
Step 2.1, establishing a degradation rate function model in a non-working state;
Arrhenius acceleration model describes degradation rate Relationship with temperature T:
Wherein M is degradation amount of a certain characteristic value, r Arrhenius represents degradation rate at temperature T, k b is Boltzmann constant, C 0 is undetermined parameter, ea is reaction activation energy, eV is taken as a unit, and the same failure mode of the same type of product is constant;
The particle counting sensor generates a natural degradation process under the storage condition of a non-working state, a degradation rate function model under the non-working state is established based on the form of the Arrhenius acceleration model, the storage temperature is used as single acceleration stress under the non-working state, other influencing factors are combined into one influencing factor, and the degradation rate r 1 under the non-working state is expressed as:
Wherein T represents the storage environment temperature in a non-working state, and A and B are parameters to be estimated;
Step 2.2, establishing a degradation rate function model in a working state;
Abrasion is an irreversible damage process of the surface of a material, and the abrasion loss Archard model is as follows:
Wherein r m represents the wear speed, W represents the wear amount, t represents the wear time, k represents the wear coefficient under certain working conditions, P, v represents the pressure on the friction surface and the relative linear speed, and m and n represent the undetermined correlation coefficients;
The degradation of the particle counting sensor in the working state is mainly expressed as abrasion, a degradation rate function model in the working state is established based on the form of the Arcard model, the flow of the measured oil passing through the slit is single acceleration stress, other influencing factors are combined into one influencing factor, and the degradation rate r 2 in the working state is expressed as:
r2＝C·(υ)D (3)
wherein, v represents the liquid flow under the working state, C, D is the parameter to be estimated.
6. A life prediction method for a liquid particle count sensor according to claim 5, wherein step three is implemented by,
The two-stage degradation process processed based on the accumulated damage theory has different degradation rates, so that the two-stage Wiener process in the two-stage degradation model of the particle counting sensor has different drift coefficients, and a drift coefficient term is represented by a step two degradation rate function;
Wherein, r 1 represents the natural degradation rate in the non-working state, r 2 represents the degradation rate of abrasion in the working state, sigma represents the fluctuation effect of the degradation process, B (t) is standard Brownian motion, and the initial degradation amount X (0) =0 is defined, X (J) is the accumulated degradation amount corresponding to the working state accumulation time t J and is taken as the initial degradation amount in the non-working stage;
according to the functional relation between the degradation rate and the influence factor of the liquid particle counting sensor under the two stages, which is established in the second step, establishing a two-stage degradation model of the particle counting sensor based on a Wiener random process is expressed as:
。
7. A life prediction method for a liquid particle count sensor according to claim 5, wherein the two-stage degradation model of the particle count sensor is subjected to parameter estimation by using maximum likelihood estimation, a global optimal solution of model parameters is obtained in one step, and the four implementation methods are as follows,
The total sample size input in the degradation test is n, the test times of the area deformation ratio in the test are m times, the test times in the first stage, namely the accumulated non-working stage in the non-working state, are J times, the test times in the second stage, namely the accumulated working stage in the working state, are (m-J) times, and the test data obtained in the degradation test are expressed as:
The amount of degradation of sample i between the (j-1) th and the j-th tests is noted as Δx (t ij-ti(j-1))＝ΔXij, available based on the independent delta nature of the Wiener process:
ΔXij～N(r1·Δtj,σ2·Δtj)1≤j≤J
ΔXij～N(r2·Δtj,σ2·Δtj)J+1≤j≤m
In the whole life cycle of the particle counting sensor, when the ratio of the non-working state accumulation time to the working state accumulation time is known, parameter estimation is carried out by utilizing a maximum likelihood method in stages according to degradation data obtained in two stages, and likelihood functions are expressed as follows;
maximizing a likelihood function L 1(·),L2 (DEG) to obtain estimated values of parameters A, B, C, D and sigma;
8. A life prediction method for a liquid particle count sensor according to claim 7 wherein step five is accomplished by,
According to normal temperature condition when actually storing, extrapolating degradation rate r s` of the particle counting sensor under the non-working state, according to the rated flow rate of tested oil under the conventional condition, extrapolating degradation rate r s2 of the working state of the particle counting sensor;
the area deformation ratio of the degradation state of the particle counting sensor exceeding the failure threshold value W for the first time is regarded as the internal slit deformation is out of tolerance, the sensor is caused to be failed, and the corresponding time T w is regarded as the service life of the particle counting sensor:
Tw＝inf{t≥tJ:X(t)≥W}
R(t)=P(Tw≥t≥tJ)＝P(t≥tJ:X(t)≤W) (7)
based on the statistical characteristics of the two-stage Wiener random process, the time when the failure threshold value is exceeded for the first time obeys the two-stage inverse Gaussian distribution, and the derived failure distribution function is as follows:
Wherein, the An estimated degradation amount representing the turning moment at which the particle count sensor accumulates over the entire operating time, that is,
Predicting the lifetime of the liquid particle count sensor according to the failure distribution F (t), wherein the median lifetime t (0.5) of the particle count sensor is R (t) =1-F (t) =0.5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111643921.6A CN114398770B (en) | 2021-12-30 | 2021-12-30 | A life prediction method for liquid particle counting sensors |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111643921.6A CN114398770B (en) | 2021-12-30 | 2021-12-30 | A life prediction method for liquid particle counting sensors |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114398770A CN114398770A (en) | 2022-04-26 |
| CN114398770B true CN114398770B (en) | 2025-07-11 |
Family
ID=81229822
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111643921.6A Active CN114398770B (en) | 2021-12-30 | 2021-12-30 | A life prediction method for liquid particle counting sensors |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114398770B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117574681B (en) * | 2024-01-11 | 2024-04-26 | 北京市计量检测科学研究院 | Sensor service life prediction method based on oil smoke gas interference |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2604118A1 (en) * | 2007-11-01 | 2008-04-23 | Ashok Ak Koul | A system and method for real-time prognostics analysis and residual life assessment of machine components |
| CN102436515A (en) * | 2010-08-27 | 2012-05-02 | 通用电气公司 | Method and system for estimating remaining life of a turbine airfoil |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111046564B (en) * | 2019-12-16 | 2024-04-05 | 重庆大学 | Residual life prediction method for two-stage degraded product |
| CN111274737A (en) * | 2020-02-25 | 2020-06-12 | 山东大学 | Method and system for predicting remaining service life of mechanical equipment |
-
2021
- 2021-12-30 CN CN202111643921.6A patent/CN114398770B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2604118A1 (en) * | 2007-11-01 | 2008-04-23 | Ashok Ak Koul | A system and method for real-time prognostics analysis and residual life assessment of machine components |
| CN102436515A (en) * | 2010-08-27 | 2012-05-02 | 通用电气公司 | Method and system for estimating remaining life of a turbine airfoil |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114398770A (en) | 2022-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110779745B (en) | Heat exchanger early fault diagnosis method based on BP neural network | |
| US20230316051A1 (en) | Pre-alarming method for rotary stall of compressors based on temporal dilated convolutional neural network | |
| CN107436963B (en) | O-shaped rubber sealing ring service life prediction method based on Copula function multi-degradation | |
| CN111353482A (en) | LSTM-based fatigue factor recessive anomaly detection and fault diagnosis method | |
| CN107423495B (en) | A Method of Degradation Modeling and Life Prediction under Dynamic Conditions | |
| CN107515965B (en) | A kind of acceleration degeneration modelling evaluation method based on uncertain course | |
| CN107766628A (en) | A kind of dynamic Degradation Reliability appraisal procedure based on life information fusion | |
| CN106647718A (en) | Non-linear industrial process fault detection method based on Bayes kernel slow feature analysis | |
| CN103884359B (en) | A kind of satellite gyroscope component fault diagnosis method based on pivot analysis algorithm | |
| US8560279B2 (en) | Method of determining the influence of a variable in a phenomenon | |
| CN106886213A (en) | A kind of batch process fault detection method based on core similarity Support Vector data description | |
| CN105930629B (en) | An Online Fault Diagnosis Method Based on Massive Operating Data | |
| CN103278343B (en) | A kind of recognition methods of key structural fatigue component | |
| CN107702905A (en) | A kind of rubber ring Q-percentile life Forecasting Methodology based on Weibull distribution | |
| CN106203637A (en) | A kind of mobile system health evaluating method based on mahalanobis distance and health index | |
| CN102830618A (en) | Quality-fluctuation prediction method for multi-procedure processing process | |
| CN102607641A (en) | Cluster anomaly detection method of combustion gas turbine | |
| CN102706563A (en) | Detection method for neighbor abnormities of gas turbine | |
| CN114398770B (en) | A life prediction method for liquid particle counting sensors | |
| CN116910677A (en) | Industrial instrument fault diagnosis method and system | |
| CN112949735A (en) | Liquid hazardous chemical substance volatile concentration abnormity discovery method based on outlier data mining | |
| CN108830006B (en) | Linear-nonlinear industrial process fault detection method based on linear evaluation factor | |
| CN108492026B (en) | Soft measurement method based on integrated orthogonal component optimization regression analysis | |
| Shanling et al. | Real-time rubber quality model based on CNN-LSTM deep learning theory | |
| Skaf et al. | A simple state-based prognostic model for filter clogging |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |