CN110553644B - A precise positioning system and method for mining electric shovel - Google Patents

Abstract

The invention provides a mining electric shovel accurate positioning system and method, which are characterized by comprising an electric shovel accurate positioning running track measurement terminal, a data analysis and management system and a user management terminal, wherein the electric shovel accurate positioning running track measurement terminal enters an enterprise internal network through a WIFI wireless workstation, is connected with the data analysis and management system and the user management terminal, adopts a deep learning calculation method to eliminate swinging of an electric shovel loading, researches actual running tracks, calculates and describes the running tracks through continuous training of a large amount of acquired real-time data, realizes real-time accurate positioning of the electric shovel, and realizes the measurement of the operation direction and the running distance according to shifts; the obtained real-time accurate data is transmitted to a background production management center for storage in real time, and a reliable basis is provided for electric shovel operation assessment.

Description

A precise positioning system and method for mining electric shovel

technical field

The invention relates to a positioning combination technology utilizing inertial navigation and Beidou satellites, in particular to a method for precise positioning and running track measurement of electric shovels for open-pit mining.

Background technique

The inertial navigation system is a navigation attitude calculation system composed of inertial sensor accelerometer, gyroscope and magnetic sensor as the basic measurement components. It integrates computer, control, mechanics, mathematics, optics, electromechanical and other disciplines, and connects various disciplines together to form a modern technology with high-tech content. The inertial navigation system is based on the theoretical basis of the inertial principle. It not only does not rely on any information from the outside world, but also does not radiate energy to the outside world. No matter what kind of propagation medium space or changes in climate conditions, it can independently complete three-axis orientation and positioning only by its own navigation characteristics. Other navigation systems also include celestial navigation, radio navigation and satellite navigation, etc. The performance of these navigation systems is superior to that of inertial navigation systems in some respects. However, the unique advantages of inertial navigation systems, which are concealed, autonomous, and capable of real-time output information, as well as large information output and the ability to obtain complete motion information of moving objects, cannot be surpassed by other navigation systems. Therefore, the inertial navigation system has been widely used as an important device of the navigation system. The main sensitive elements of inertial sensors include accelerometers and gyroscopes. The accelerometer is used to measure the linear acceleration of the object, the amplitude of the small vibration of the object, etc.; the gyroscope is used to measure the changes of various angles when the moving object rotates or the change of the inclination angle of the object.

In the production process of open-pit mines, the electric shovel is an important tool for mining and loading, and its working direction and actual running track are the important foundations of the mining process. The position of the electric shovel carries the explosive material information. In the past, the GPS satellite positioning technology used by the electric shovel generally had an error of about 5-15 meters, but the actual working distance of the actual production and mining in each shift was generally about 20 meters. Usually, the electric shovel rotates left and right to load the production vehicles, and it also needs to move forward and backward to dig materials. Therefore, pure GPS cannot meet the requirements of precise positioning and running track measurement and assessment in actual production.

At the same time, due to the vibration and interference of the electric shovel, some detection systems cannot operate on the electric shovel. How to combine the inertial navigation system and GPS into the electric shovel to realize the precise positioning of the electric shovel has become the key to solving the problem.

Contents of the invention

The purpose of the present invention is to provide a precise positioning system and method for mining electric shovels, with dust release, vibration release and fanless industrial computer as the core, and through inertial and Beidou satellite combined positioning technology to realize electric shovel precise positioning and running track measurement technology, to achieve the purpose of precise mining and intelligent ore distribution, facilitate production management, and improve production efficiency.

To achieve the above object, the present invention is achieved through the following technical solutions:

A mine electric shovel precise positioning system according to the present invention is characterized in that it includes a precise positioning electric shovel running trajectory measurement terminal, a data analysis and management system and a user management terminal,

The electric shovel precise positioning operation trajectory measurement terminal includes an industrial computer, a power supply module, a keyboard and mouse, a display module, a Beidou GPS positioning module, a WIFI communication module and an inertial navigation positioning module, and the described power supply module, keyboard and mouse, display module, Beidou GPS positioning module and inertial navigation positioning module are all electrically connected to the industrial control computer;

The data analysis management system includes a production server, an intranet, a graphic terminal and a data storage module;

Described user management terminal comprises computer 1, computer 2 and computer n;

The electric shovel accurately locates the running trajectory measurement terminal and connects with the internal network in the data analysis management system through the WIFI communication module and the WIFI wireless workstation. Meanwhile, the internal network is also connected with the production server and all computers of the user management terminal.

The inertial navigation positioning module is an MPU6050 sensor installed on a forklift, embedded with a three-axis digital gyroscope.

The Beidou GPS positioning module is used to receive satellite signals, and the inertial navigation and positioning module is used to more accurately locate the position information of the electric shovel. The position information of the electric shovel includes the longitude and latitude of the electric shovel, the direction of the trajectory of the electric shovel, the grade of the bucket, and the operation plan of the electric shovel; the industrial control computer is used to calculate the position information of the electric shovel according to the satellite signals received by the Beidou GPS positioning module and the dual-mode positioning module of the inertial navigation and positioning module.

A method for precise positioning of electric shovels used in mines according to the present invention is characterized in that it comprises the following steps:

(1) Initialization: Initialize the Beidou GPS positioning module and inertial navigation positioning module;

(2) Inertial navigation calibration: calibrate the accelerometer, gyroscope, magnetic field and altitude to 0 for the inertial navigation module under static conditions, so as to ensure the error analysis of the measurement accuracy of the inertial navigation module and prevent the inertial navigation module from drifting too much;

(3) Data acquisition: The industrial control computer performs inertial navigation data acquisition and Beidou satellite signal acquisition every 10 milliseconds. The industrial control computer receives the data of the Beidou GPS positioning module and inertial navigation module, and transmits the collected data to the deep learning training machine for model training;

(4) Data processing: the data of the actual operation of the electric shovel is collected at a period of 10 milliseconds, and digitally filtered, and one data is generated per second for data processing;

(5) Model establishment: the deep learning convolutional neural network and the regression model of the time recursive neural network are used. The convolutional neural network is used to extract the eigenvalues of the input signal, and the time recursive neural network is used to perform time series analysis on the extracted eigenvalues. The loss function formula of the model is used to obtain the training model parameters;

(6) The industrial control computer performs data collection and signal processing according to the training model parameters, calculates the position and working direction of the electric shovel every 10 minutes through the loss function formula of the model, and uploads the longitude and latitude data to the production server through the wireless network;

(7) The production server stores the position and working direction for 10 minutes in the data storage, and marks the position on the X, Y plane coordinate map, and displays the actual position according to the distance of 100 meters, and produces a chart of working distance and working direction for each electric shovel produced in each shift;

(8) Each manager inquires the current location of each electric shovel through the company's internal network computer, the moving distance and position of the electric shovel every day and every shift, including the specific operation direction, starting and final location information.

4. A kind of mining electric shovel precise positioning method according to claim 3, is characterized in that, the loss function formula of described model:

where y _i is the actual distance and angle identified in the training data, is the distance and angle predicted by the model, and the model obtains the minimum value of the deviation by iterating the function of formula (1).

The advantages of the present invention are as follows:

1. The present invention uses inertial and Beidou satellite combined positioning technology to realize the precise positioning of electric shovels and the measurement technology of running trajectory, so as to realize the purpose of precise mining and intelligent mineral distribution;

The inertial navigation module calculates the azimuth: east, south, west, north, northeast, northwest, southeast and southwest 8 azimuths, the distance along a certain direction, using the inertial navigation as the main calculation basis, improves the position accuracy of the shovel.

2. Using the multi-sensor fusion correction parameter method of Beidou and inertial navigation system, the advantages of Kalman filter in positioning and navigation technology are fully utilized, providing accurate positioning parameters, and providing a more powerful theoretical basis for electric shovel positioning.

3. According to the actual production, the intelligent positioning and navigation system of the electric shovel is designed, and the walking distance of the electric shovel in each shift is verified through the Beidou GPS positioning system to ensure the accuracy of the positioning system. Through the precise positioning of the electric shovel, intelligent ore distribution can be realized and the mining production effect can be improved.

The present invention adopts the combined technology based on inertial navigation and Beidou satellites, collects a large amount of real-time data produced by electric shovels, uses the hybrid dynamic algorithm based on SVM to analyze and predict, establishes a deep learning data analysis and processing model, and combines with the current on-site total station measurement technology to meet the needs of "precision mining and intelligent ore distribution", and provide convenient and fast performance methods and information acquisition modes for future intelligent mines.

Description of drawings

Fig. 1 is a system composition block diagram of a mine electric shovel precise positioning system and method.

Fig. 2 is a block diagram of general steps of a mine electric shovel precise positioning system and method;

Fig. 3 is a structure diagram of a deep learning model of a precise positioning system and method for a mining electric shovel.

Fig. 4 is a schematic diagram of an active positioning technology of a precise positioning system and method for a mining electric shovel.

Fig. 5 is a schematic diagram of electric shovel positioning of a mining electric shovel precise positioning system and method.

Fig. 6 is a schematic diagram of the electric shovel running trajectory of a mining electric shovel precise positioning system and method.

Detailed ways

The technical content of the present invention will be further described in detail below in conjunction with the accompanying drawings.

As shown in Figure 1, a mine electric shovel precise positioning system of the present invention is characterized in that it includes a precise positioning electric shovel running track measurement terminal, a data analysis and management system and a user management terminal,

The electric shovel precise positioning operation trajectory measurement terminal includes an industrial computer, a power supply module, a keyboard and mouse, a display module, a Beidou GPS positioning module, a WIFI communication module and an inertial navigation positioning module, and the described power supply module, keyboard and mouse, display module, Beidou GPS positioning module and inertial navigation positioning module are all electrically connected to the industrial control computer;

The data analysis management system includes a production server, an intranet, a graphic terminal and a data storage module;

The electric shovel accurately locates the running trajectory measurement terminal and connects with the internal network in the data analysis management system through the WIFI communication module and the WIFI wireless workstation. Meanwhile, the internal network is also connected with the production server and all computers of the user management terminal.

The inertial navigation positioning module is an MPU6050 sensor installed on a forklift, embedded with a three-axis digital gyroscope.

The Beidou GPS positioning module is used to receive satellite signals, and the inertial navigation and positioning module is used to more accurately locate the position information of the electric shovel. The position information of the electric shovel includes the longitude and latitude of the electric shovel, the direction of the trajectory of the electric shovel, the grade of the bucket, and the operation plan of the electric shovel; the industrial control computer is used to calculate the position information of the electric shovel according to the satellite signals received by the Beidou GPS positioning module and the dual-mode positioning module of the inertial navigation and positioning module.

Preferably, according to the present invention, the mine electric shovel precise positioning system directly deduces the moving distance and direction of the forklift through the data collection of the inertial navigation sensor (accelerometer, gyroscope, etc.) installed on the forklift and the Beidou GPS positioning module and the corresponding relationship between these data and the forklift position data, so that when the position data is not available, the inertial navigation data can still be used to measure and predict the actual operating position and trajectory of the forklift. A convolutional neural network is used. The regression model of the time recurrent neural network is used to extract the characteristics of the input signal. Value, the time recursive neural network is used to analyze the time series of the extracted feature values, and the output of the multi-layer convolutional network layer is used as the input of the time recursive layer to improve the training speed of the model. The system uses a structure of 4 convolutional layers plus double time recursive layers. Each convolutional layer is attached with a pooling layer (4x2). The final model is a learning model of 11 layers of convolutional neural layer plus time recursive layer. Through the combined calculation of GPS and inertial navigation, the calculation method of deep learning is used to eliminate the swing of the electric shovel loading. The actual running trajectory is continuously trained through a large amount of real-time data collection, and the running trajectory is calculated and described to achieve precise positioning and precise mineral allocation.

The inertial navigation module uses the MPU6050 sensor, embedded with a three-axis digital gyroscope.

The Beidou GPS positioning module is used to receive satellite signals, and the inertial navigation and positioning module is used to more accurately locate the position information of the electric shovel. The position information of the electric shovel includes the longitude and latitude of the electric shovel, the direction of the trajectory of the electric shovel, the grade of the bucket, and the operation plan of the electric shovel; the industrial control computer is used to calculate the position information of the electric shovel according to the satellite signals received by the Beidou GPS positioning module and the dual-mode positioning module of the inertial navigation and positioning module.

The data storage is connected with the production server for data backup and storage.

As shown in Figure 2, a method for precise positioning of a mining electric shovel according to the present invention is characterized in that it comprises the following steps:

(1) Initialization: Initialize the Beidou GPS positioning module and inertial navigation positioning module;

(2) Inertial navigation calibration: calibrate the accelerometer, gyroscope, magnetic field and altitude to 0 for the inertial navigation module under static conditions, so as to ensure the error analysis of the measurement accuracy of the inertial navigation module and prevent the inertial navigation module from drifting too much;

(3) Data collection: The industrial control computer performs inertial navigation data collection and Beidou satellite signal collection every 10 milliseconds. The industrial control computer receives the data of the Beidou GPS positioning module and inertial navigation module, and transmits the collected data to the deep learning training machine for model training; data collection includes collection time, inertial navigation ax=x-axis acceleration, ay: y-axis acceleration, az: z-axis acceleration, pitch: x rotation angle, roll: y-axis rotation angle, yaw: z-axis rotation angle, wx: x-axis angular velocity, wy: y-axis angular velocity , wz: z-axis angular velocity, gv: earth speed and Beidou satellite longitude and latitude data;

(4) Data processing: the data of the actual operation of the electric shovel is collected at a period of 10 milliseconds, and digitally filtered, and one data is generated per second for data processing;

7:30 in the morning and 18:30 in the evening are calculated as the starting point of the static state at 7:30 in the morning and 18:30 in the evening, and the longitude and latitude of these two time points are used as the origin of the inertial navigation system. Then, the moving distance generated by the inertial navigation movement every 1 minute is converted into longitude and latitude, and then the latitude and longitude changes generated by the Beidou satellite in 1 minute are compared.

(5) Model establishment: as shown in Figure 3, the regression model of deep learning convolutional neural network collocation time recurrent neural network is adopted, the convolutional neural network is used to extract the eigenvalues of the input signal, the time recursive neural network is used to perform time series analysis on the extracted eigenvalues, and the training model parameters are obtained by using the loss function formula of the model;

We use the output of the multi-layer convolutional network layer as the input of the time recursive layer to improve the training speed of the model. The model training optimizer is Adam. It uses a structure of 4 convolutional layers plus a double time recursive layer. Each convolutional layer is accompanied by a pooling layer (4x2). The final model is a learning model of an 11-layer convolutional neural layer plus a time recurrent layer.

The key parameters of the model are as follows:

Number of convolutional layer convolution kernels: 32/64/128/64

Convolutional layer convolution kernel size: 8

Convolutional layer convolution kernel moving step: 8

Pooling layer movement step size: 2

Pooling Layer Pool Size: 4

Number of temporally recurrent layer neurons: 64.

The loss function formula of the described model:

where yi is the actual distance and angle identified in the training data, is the distance and angle predicted by the model, and the model achieves the purpose of training model parameters by iteratively seeking the minimum value of the function of formula (1).

After training, 80,000 pieces of training data were actually accumulated to train the model, and the error of the loss function was reduced to within 1.0 meters of distance and within 30 degrees of angle.

Preferably, according to the present invention, the mine electric shovel precise positioning system directly deduces the moving distance and direction of the forklift through the data collection of the inertial navigation sensor (accelerometer, gyroscope, etc.) installed on the forklift and the Beidou GPS positioning module and the corresponding relationship between these data and the forklift position data, so that when the position data is not available, the inertial navigation data can still be used to measure and predict the actual operating position and trajectory of the forklift. A convolutional neural network is used. The regression model of the time recurrent neural network is used to extract the characteristics of the input signal. Value, the time recursive neural network is used to analyze the time series of the extracted feature values, and the output of the multi-layer convolutional network layer is used as the input of the time recursive layer to improve the training speed of the model. The system uses a structure of 4 convolutional layers plus double time recursive layers. Each convolutional layer is attached with a pooling layer (4x2). The final model is a learning model of 11 layers of convolutional neural layer plus time recursive layer. Through the combined calculation of GPS and inertial navigation, the calculation method of deep learning is used to eliminate the swing of the electric shovel loading. The actual running trajectory is continuously trained through a large amount of real-time data collection, and the running trajectory is calculated and described to achieve precise positioning and precise mineral allocation.

(6) The industrial control computer performs data collection and signal processing according to the training model parameters, calculates the position and working direction of the electric shovel every 10 minutes through the loss function formula of the model, and uploads the longitude and latitude data to the production server through the wireless network;

(7) The production server stores the position and working direction for 10 minutes in the data storage, and marks the position on the X, Y plane coordinate map, and displays the actual position according to the distance of 100 meters, and produces a chart of working distance and working direction for each electric shovel produced in each shift;

Preferably according to the present invention, in step (2), solve based on the motion parameter of MEMS sensor object and analyze its relevant accuracy error, concrete steps include:

a. Study the characteristics and applicable conditions of low-precision MEMS sensors, improve the measurement accuracy of low-precision MEMS sensors, and reduce errors;

b. Establish an error model, and use the model as a basis to conduct error calibration and compensation experiments to improve the accuracy of the sensor, hoping to use the sensor to accurately measure the attitude and position information of the moving body;

Before the inertial navigation module is used, the module needs to be calibrated to ensure the main accurate measurement of the sensor. The calibration of the inertial navigation module includes gyroscope calibration, magnetic field calibration and zero altitude. Gyroscope calibration is used to remove bias from gyroscope measurements. When the module is stationary, if the angular velocity figure is not near 0°/s, the gyroscope needs to be recalibrated. After operating through the gyroscope calibration software, wait for the data to be read out to stabilize, indicating that the calibration is completed. Then the configuration software saves the zero offset data to the internal FLASH of the module for power-off storage. Thereafter, at rest, the output of the gyroscope will return to around 0°/s.

c. Calibrate and compensate the acceleration sensor, reduce the output error of the acceleration sensor, improve the accuracy of the sensor, and make the output data of the sensor more accurate and reliable;

d. Read the longitude and latitude of the Beidou GPS positioning module in the sensor, and provide calibration parameters;

Each shift reads the Beidou GPS positioning data once, performs corrections, forms model correction error parameters, and continuously calibrates. The error calculated each time is an important part of the calculation.

Preferably according to the present invention, in step (4), based on the electric shovel positioning principle of the inertial navigation module, the specific steps include:

a. Active positioning (Active positioning) principle, see Figure 5, it determines the position of the electric shovel by collecting and calculating the displacement and angle of the electric shovel movement. Its positioning principle is very simple: if the initial coordinates (x ₀ , y ₀ , θ ₀ ) of the mobile carrier are known, the coordinates (x ₁ , y ₁ , θ ₁ ) after the target is calculated using the measured movement displacement and angle. After continuous iterative calculations, the trajectory of the moving target is finally obtained.

Due to its simple and reliable positioning principle and low cost, active positioning is mainly affected by the design of its own positioning system and the accuracy of positioning components, and is less affected by external environmental factors. Therefore, it is especially suitable for positioning needs in small-scale, low-cost and complex environments, and has high application prospects and research value.

b. The principle of active positioning of the electric shovel is shown in Figure 6. It is different from passive positioning. It does not rely on the assistance of external positioning equipment, but only collects and calculates the displacement and angle of the electric shovel’s own motion to determine the position of the electric shovel.

The initial coordinates (x0, y0) of the positioning target are known, if the measured displacement

S1 and angle Q1, the coordinates (x1, y1) after the target can be calculated by using formula (2.2).

After continuous iterative calculation, the moving trajectory and position coordinates of the positioning target are obtained.

c. According to the above moving track and position coordinates, draw the running track of the electric shovel.

Due to the vibration and interference of the electric shovel, as well as the interference of the inertial navigation module itself, environmental interference, vibration interference and electromagnetic interference, etc., it is difficult to locate the direction of the electric shovel. It needs to be eliminated by software filtering to ensure that the distance measurement of the electric shovel can reach 3%.

(8) Each manager inquires the current location of each electric shovel through the company's internal network computer, the moving distance and position of the electric shovel every day and every shift, including the specific operation direction, starting and final location information.

Although the specific implementation of the present invention has been described above in conjunction with the accompanying drawings, it does not limit the protection scope of the present invention. Those skilled in the art should understand that on the basis of the technical solution of the present invention, various modifications or deformations that those skilled in the art can make without creative labor are still within the protection scope of the present invention.

Claims (4)

1. A method for precise positioning of electric shovels for mines, characterized in that, a precise positioning system for electric shovels for mining is adopted, the system includes a precise positioning operation track measurement terminal for electric shovels, a data analysis and management system and a user management terminal, and its precise positioning method for electric shovels for mines comprises the following steps: (1) Initialization: Initialize the Beidou GPS positioning module and inertial navigation positioning module; (2) Inertial navigation calibration: calibrate the accelerometer, gyroscope, magnetic field and altitude to 0 for the inertial navigation and positioning module under static conditions to ensure the error analysis of the measurement accuracy of the inertial navigation module and prevent the inertial navigation and positioning module from drifting too much; (3) Data acquisition: The industrial control computer performs inertial navigation data acquisition and Beidou satellite signal acquisition every 10 milliseconds. The industrial control computer receives the data from the Beidou GPS positioning module and inertial navigation positioning module, and transmits the collected data to the deep learning training machine for model training; (4) Data processing: The actual operation data of the electric shovel is collected at a period of 10 milliseconds, and digital filtering is performed, and one data is generated per second for data processing; (5) Model establishment: the regression model of deep learning convolutional neural network and time recursive neural network is used. The convolutional neural network is used to extract the eigenvalues of the input signal, and the time recursive neural network is used to perform time series analysis on the extracted eigenvalues. The loss function formula of the model is used to obtain the training model parameters; (6) The industrial control computer performs data collection and signal processing according to the actual situation on site, calculates the position and working direction of the electric shovel every 10 minutes through the loss function formula of the model, and uploads the latitude and longitude data to the production server through the wireless network; (7) The production server stores the 10-minute location and working direction in the data storage, and marks the location on the X, Y plane coordinate map, and displays the actual location according to the distance of 100 meters, and produces a working distance and working direction chart for each electric shovel produced in each shift; (8) Each manager inquires the current location of each electric shovel through the company's internal network computer, the moving distance and position of the electric shovel every day and every shift, including the specific operation direction, starting and final location information; The electric shovel precise positioning operation trajectory measurement terminal includes an industrial control computer, a power supply module, a keyboard and mouse, a display module, a Beidou GPS positioning module, a WIFI communication module and an inertial navigation positioning module, and the described power supply module, keyboard and mouse, display module, Beidou GPS positioning module and inertial navigation positioning module are all electrically connected to the industrial control computer; The data and analysis management system includes a production server, an intranet, a graphic terminal and a data storage module; Described user management terminal comprises computer 1, computer 2 and computer n; The electric shovel accurately locates the running trajectory measurement terminal and connects with the internal network in the data analysis management system through the WIFI communication module and the WIFI wireless workstation. Meanwhile, the internal network is also connected with the production server and all computers of the user management terminal. 2. A kind of mine electric shovel precise positioning method according to claim 1, is characterized in that, the loss function formula of described model: (1) Wherein is the actual distance and angle identified in the training data, is the distance and angle predicted by the model, and the model obtains the minimum value of the deviation by iterating the function of formula (1). 3. The method for precise positioning of electric mining shovels according to claim 1, wherein the inertial navigation positioning module is an MPU6050 sensor installed on a forklift, embedded with a three-axis digital gyroscope. 4. The precise positioning method for mining electric shovels according to claim 1, wherein the Beidou GPS positioning module is used to receive satellite signals, and the inertial navigation positioning module is used to more precisely locate the position information of the electric shovel, the position information of the electric shovel includes the longitude and latitude of the electric shovel, the direction of the trajectory of the electric shovel, the grade of the bucket, and the operation plan of the electric shovel;