Conveyor belt deviation prediction and warning system and prediction and warning method based on BP neural network algorithm

Abstract

The present invention discloses a system and a method for predicting and warning conveyor belt deviation based on BP neural network algorithm, including a laser acquisition device for monitoring the distance between the laser point of the roller and the edge of the conveyor belt; a speed detector for monitoring the speed of the transmission belt and the speed of the roller; a power supply control device for controlling the switch of the power supply; a communication device 1, as a terminal, functions to transmit the data processed by the speed detector and the single-chip microcomputer to the communication device 2; the communication device 2 plays the role of a hub, uploading the signal in the communication device 1 to the host computer and the monitor; and also includes a conveyor belt tension detection device for collecting the tension of the conveyor belt. The system and method of the present invention are simple, easy to use, highly feasible, highly automated, safe and reliable, and can realize the prediction of the lateral deviation of the conveyor belt of the belt conveyor.

Description

Technical Field

[0001] The invention relates to a system and a method for predicting and warning conveyor belt deviation based on a BP neural network algorithm. Background Art

[0002] Belt conveyors, also known as rubber belt conveyors, are widely used in various industries such as home appliances, electronics, electrical appliances, machinery, tobacco, injection molding, post and telecommunications, printing, food, etc. for the assembly, testing, debugging, packaging and transportation of objects.

[0003] Belt conveyor is the most ideal high-efficiency continuous transportation equipment for coal mines. Compared with other transportation equipment (such as locomotives), it has the advantages of long transportation distance, large transportation capacity, continuous transportation, etc., and it is reliable in operation and easy to realize automation and centralized control. Especially for high-yield and high-efficiency mines, belt conveyor has become the key equipment of coal mining mechatronics technology and equipment.

[0004] Belt conveyors are widely used in metallurgy, coal, transportation, hydropower, chemical industry and other departments because of their large conveying capacity, simple structure, convenient maintenance, low cost, strong versatility, etc. At the same time, belt conveyors are also used in building materials, electricity, light industry, food, ports, ships and other departments.

[0005] Belt conveyor deviation is the most common fault of belt conveyor, which not only affects the progress of the project, but also affects the service life of the conveyor, and in serious cases, causes casualties. Therefore, real-time detection of conveyor belt lateral deviation and alarm is of great significance for the complete and stable operation of belt conveyor.

[0006] At present, although conveyor belt lateral deviation detection devices have been developed at home and abroad, their accuracy and application conditions are limited to a certain extent. The existing online detection methods include visual methods, which use linear array CCD cameras to collect conveyor belt running images, obtain conveyor belt binary images after binarization, and then project the binary images into one-dimensional fault feature functions, and then extract the deviation angle and offset from the one-dimensional fault feature functions for deviation fault identification. This method can collect conveyor belt images in real time, but the deviation identification accuracy is greatly affected by the high powder layer environment. Based on photoelectric detection, the system has strong anti-interference ability, high accuracy, and fast response, but the system is relatively complex and cannot accurately know the deviation distance. Therefore, it will be of great practical value to explore a high-precision real-time online belt conveyor belt lateral deviation prediction method that is not affected by operating conditions and is not affected by operating conditions, and apply it to the conveyor belt lateral deviation prediction system. Summary of the invention

[0007] The purpose of the present invention is to provide a system and a prediction and warning method for conveyor belt deviation prediction and warning based on BP neural network algorithm which is easy to use, highly feasible, highly automated, safe and reliable.

[0008] The technical solution of the present invention is:

[0009] A system for predicting and warning conveyor belt deviation based on BP neural network algorithm, characterized by: including a laser acquisition device for monitoring the distance between the laser point of the roller and the edge of the conveyor belt; a speed detector for monitoring the speed of the transmission belt and the speed of the roller; a power supply control device for controlling the switch of the power supply; a communication device 1, as a terminal, functions to transmit the data processed by the speed detector and the single-chip microcomputer to a communication device 2; the communication device 2 plays the role of a hub, uploading the signal in the communication device 1 to the upper computer and the monitor; and also includes a conveyor belt tension detection device for collecting the tension of the conveyor belt;

[0010] The power control module and the alarm module are used to issue an alarm or disconnect the power supply according to the conveyor belt deviation predicted by the BP algorithm;

[0011] During the operation of the system, the signal acquisition and processing module continuously collects and processes the conveyor belt offset signals detected by the laser acquisition device and the photoelectric encoder, and then transmits them wirelessly to the host computer. After prediction by the BP neural network algorithm, it can decide whether to start the alarm and disconnect the power judgment program, thereby obtaining the current level and subsequent alarms and disconnecting the power signal; the power control module and the alarm module are started or shut down according to the host computer signal.

[0012] A prediction and early warning method using the system for predicting and early warning of conveyor belt deviation based on BP neural network algorithm is characterized by comprising the following steps:

[0013] Step 1: Initialization, the current frame of the laser scanner is k = 0, the acquisition time is t = 0, the roller speed vt = 0, the belt speed is vd = 0, the current cycle n = 0, and the number of laser scanning points per frame section m = 0;

[0014] Step 2: Start the belt speed detection vm and roller speed vn detection, as well as the conveyor belt tension f detection. The data acquisition time of the laser scanner is T, which is the time of one frame of the laser scanner;

[0015] Conveyor belt speed: The speed sensor measures the roller angular velocity W1, roller radius: R1, so the conveyor belt speed vd:

[0016] vd=W1*R1 (1)

[0017] Find the average value vm of the conveyor belt speed measured within time T:

[0018]

[0019] The roller speed uses a speed sensor to measure the roller's angular velocity W2 and roller radius R2, so the roller linear velocity vt is:

[0020] vt = W2·R2 (3)

[0021] Find the average value vn of the idler speed measured within T:

[0022]

[0023] Obtain the relative speed Vk of the conveyor belt with respect to the idler:

[0024]

[0025] Step 3: Start the laser scanner and scan within time T, record the data angles m and distances s of the data points on the contour line. The laser scanning points are A1, A2....., A n , B1, B2……, B n The angles C1, C2……, Cn, measure and record the distances between the left drum edge line and the right belt edge line as: D1, D2, ……, D n ; At the same time, A1, A2....., A n , B1, B2……, B n can be measured by the laser scanner, and the included angle C between the single scan points Ai and Bi is measured i ;

[0026] When |D1 - D n | < L mm, calculate the lateral deviation distance:

[0027] The size of the single - measurement lateral deviation distance D i can be obtained by the cosine theorem as

[0028]

[0029] When |D1 - D4| < L mm, compare the size of D i with the normal offset value D of the conveyor belt for the lateral deviation speed:

[0030]

[0031] Lateral deviation acceleration:

[0032] a = d(vx) / dt (8)

[0033] Step 4: For the deviation process, establish a deviation model with the conveyor belt speed, idler speed, conveyor belt tension, lateral deviation acceleration, and relative speed of the conveyor belt with respect to the idler as inputs and the corresponding conveyor belt deviation amount as the output. Collect the data with |D1 - D4| < L in advance P times (P is large enough), and establish a neural network model in MATLAB for training with the collected data;

[0034] Step 5: Remove the invalid or erroneous measurement parts of the original data; process the data to reduce the error: for data with missing values ​​not exceeding 10%, directly remove the data with more than 10%;

[0035] Step 6: After data deletion, data completion is performed: missing or erroneous data are completed according to the regression replacement method;

[0036] Step 7: Input the normalized training sample data set into the Matlab workspace in the form of vectors, so as to establish the BP neural network for this experiment. The training accuracy is high and can meet the needs of deviation prediction.

[0037] Step 8: After the BP neural network prediction model completes the training of the training set, the test set data is used to predict the deviation amount to test the prediction ability of the network model;

[0038] Step 9: The input variable data are the measured conveyor belt speed, roller speed, conveyor belt tension, lateral deviation acceleration, and conveyor belt relative roller speed. The five comprehensive variables are normalized and input into the network prediction model to obtain the output variable, i.e., the lateral deviation amount. The deviation amount is predicted through the output variable.

[0039] Step 10: When the predicted deviation is greater than L mm, start the alarm system, disconnect the power supply of the belt conveyor, and start the braking device to stop the movement of the conveyor belt in time; if the predicted deviation is less than L mm, continue the prediction.

[0040] The system and method of the present invention are simple, easy to use, highly feasible, highly automated, safe and reliable, and can realize the prediction of lateral deviation of the conveyor belt of a belt conveyor. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0042] Figure 1 It is a block diagram of the system composition of the prediction and early warning of the present invention.

[0043] Figure 2 It is a schematic diagram of the system structure of the present invention.

[0044] Figure 3 This is the schematic diagram for measuring the deviation distance of the conveyor belt.

[0045] A is the edge of the roller, B is the edge of the conveyor belt, and the laser scanner is located above it.

[0046] Figure 4 Detailed view of measuring the edge of the conveyor belt roller.

[0047] The laser scanning points are A1, A2, ...A n ,B1,B2...B n The angles C1, C2, ..., Cn are used to measure and record the distances between the roller edge line on the left side of the conveyor belt and the belt edge line on the right side: D1, D2, D3, D4. At the same time, A1, A2, ..., A n and B1 B2...B n And measured A i and B i The angle c between i °.

[0048] Figure 5 Schematic diagram of the process steps of the method of the present invention. DETAILED DESCRIPTION

[0049] A system for predicting and warning conveyor belt deviation based on BP neural network algorithm, including a laser acquisition device for monitoring the distance between the laser point of the roller and the edge of the conveyor belt; a speed detector for monitoring the speed of the transmission belt and the speed of the roller; a power supply control device for controlling the switch of the power supply; a communication device 1, as a terminal, is used to transmit the data processed by the speed detector and the single-chip microcomputer to a communication device 2; the communication device 2 plays the role of a hub, uploading the signal in the communication device 1 to the upper computer and the monitor; and also includes a conveyor belt tension detection device for collecting the tension of the conveyor belt;

[0050] The power control module and the alarm module are used to issue an alarm or disconnect the power supply according to the conveyor belt deviation predicted by the BP algorithm;

[0051] During the operation of the system, the signal acquisition and processing module continuously collects and processes the conveyor belt offset signals detected by the laser acquisition device and the photoelectric encoder, and then transmits them wirelessly to the host computer. After prediction by the BP neural network algorithm, it can decide whether to start the alarm and disconnect the power judgment program, thereby obtaining the current level and subsequent alarms and disconnecting the power signal; the power control module and the alarm module are started or shut down according to the host computer signal.

[0052] A prediction and early warning method using the system for predicting and early warning of conveyor belt deviation based on BP neural network algorithm comprises the following steps:

[0053] Step 1: Initialization, the current frame of the laser scanner is k = 0, the acquisition time is t = 0, the roller speed vt = 0, the belt speed is vd = 0, the current cycle n = 0, and the number of laser scanning points per frame section m = 0;

[0054] Step 2: Start the belt speed detection vm and roller speed vn detection, as well as the conveyor belt tension f detection. The data acquisition time of the laser scanner is T;

[0055] The speed of the conveyor belt: Measure the angular velocity W1 of the roller with a speed sensor. The radius of the roller is R1. Therefore, the conveyor belt speed vd is:

[0056] vd = W1 * R1 (1)

[0057] Find the average value vm of the measured conveyor belt speed within time T:

[0058]

[0059] The idler speed Measure the angular velocity W2 of the idler with a speed sensor. The radius of the idler is R2. Therefore, the linear speed vt of the idler is:

[0060] vt = W2 · R2 (3)

[0061] Find the average value vn of the measured idler speed within T:

[0062]

[0063] Obtain the speed vk of the conveyor belt relative to the idler:

[0064]

[0065] Step 3: Start the laser scanner, scan within time T, and record the data angles m and distances s of the data points on the contour line. The laser scan points are A1, A2....., A n , B1, B2……, B n The angles C1, C2……, Cn. Measure and record the distances between the left roller edge line and the right belt edge line as: D1, D2, ……, D n ; At the same time, A1, A2……, A n , B1, B2……, B n can be measured by the laser scanner, and the included angle C between the single scan points Ai and Bi is measured i ;

[0066] When |D1 - D n | < L mm, calculate the lateral deviation distance:

[0067] The size of the single measurement lateral deviation distance D i can be obtained by the cosine theorem as

[0068]

[0069] When |D1 - D4| < L mm, compare the size of D i with the normal offset value D of the conveyor belt for the lateral deviation speed:

[0070]

[0071] Lateral deviation acceleration:

[0072] a = d(vx) / dt (8)

[0073] Step 4: For the deviation process, establish a deviation model with the conveyor belt speed, idler speed, conveyor belt tension, lateral deviation acceleration, and relative speed of the conveyor belt to the idler as inputs and the corresponding conveyor belt deviation amount as the output; collect the data of |D1 - D4| < L in advance for P times (P is large enough), and establish a neural network model in MATLAB with the collected data for training;

[0074] Step 5: Remove the invalid or incorrectly measured parts from the original data; for data processing to reduce errors: for data with missing values not exceeding 10%, and directly remove data with more than 10% missing values;

[0075] Step 6: After data deletion, perform data completion: complete the missing or incorrect data according to the regression replacement method;

[0076] Step 7: Input the normalized training sample data set into the Matlab workspace in vector form to establish the BP neural network for this experiment, with relatively high training accuracy, which can meet the needs of deviation prediction;

[0077] Step 8: After the BP neural network prediction model completes the training of the training set, use the test set data to predict the deviation amount to test the prediction ability of the network model;

[0078] Step 9: The input variable data are the measured conveyor belt speed, idler speed, conveyor belt tension, lateral deviation acceleration, and relative speed of the conveyor belt to the idler. After normalizing the 5 input variables, input them into the network prediction model, and the output variable, that is, the lateral deviation amount, can be obtained. The deviation amount is predicted through the output variable;

[0079] Step 10: When the predicted deviation amount is greater than L mm, start the alarm system, disconnect the power supply of the belt conveyor, and start the braking device to stop the movement of the conveyor belt in time; if the predicted deviation amount is less than L mm, continue the prediction.

Claims

1. A prediction and early warning method for a conveyor belt deviation prediction and early warning system based on a BP neural network algorithm, characterized by: The system for predicting and warning the deviation of the conveyor belt based on the BP neural network algorithm includes a laser acquisition device for monitoring the distance between the roller and the laser point at the edge of the conveyor belt; a speed detector for monitoring the rotation speed of the conveyor belt and the speed of the idler; a power control device for controlling the switch of the power supply; a communication device 1, as a terminal, for transmitting the data obtained by the speed detector and the single-chip microcomputer processing to the communication device 2; a communication device 2, acting as a hub, for uploading the signal in the communication device 1 to the upper computer and the monitor; and also includes a conveyor belt tension detection device for collecting the tension of the conveyor belt. A power control module and an alarm module for issuing an alarm or disconnecting the power supply according to the deviation amount of the conveyor belt predicted by the BP algorithm. During the operation of the system, the signal acquisition and processing module continuously collects and processes the conveyor belt offset signals detected by the laser acquisition device and the photoelectric encoder, and then wirelessly transmits them to the upper computer. After prediction by the BP neural network algorithm, it can decide whether to start the alarm and power-off judgment program, so as to obtain the current-level and subsequent alarm and power-off signals; the power control module and the alarm module are started or closed according to the upper computer signal. The prediction and warning method includes the following steps: Step 1: Initialization, the current frame of the laser scanner is k = 0, the acquisition time is t = 0, the idler speed vt = 0, the belt speed is vd = 0, the current cycle n = 0, and the number of laser scan points per frame section m = 0. Step 2: Start the detection of the belt speed vm and the idler speed vn, and the detection of the conveyor belt tension f. The data acquisition time of the laser scanner is T. Conveyor belt speed: Measure the angular velocity W1 of the roller and the roller radius R1 with a speed sensor, so the conveyor belt speed vd: vd = W1 * R1 (1) Find the average value vm of the measured conveyor belt speed within the time T: The idler speed is measured by a speed sensor to measure the angular velocity W2 of the idler and the idler radius R2, so the linear speed vt of the idler: vt = W2 * R2 (3) Find the average value vn of the measured idler speed within T: Obtain the relative speed vk of the conveyor belt with respect to the idler: Step 3: Start the laser scanner, scan within time T, and record the angle m and distance s of the data points on the contour line. The laser scanning points are A1, A2, A n , B1, B2..., B n , angles C1, C2, ..., Cn, the distances between the left drum edge and the right belt edge are measured and recorded as: D1, D2, ..., D n ; At the same time, A1, A2, A n ,B1,B2……,B n , and the angle C between the single scanning points Ai and Bi is measured i ; When |D1 - D n| < L mm, calculate the lateral deviation distance: The single-measurement lateral deviation distance D can be obtained by the cosine theorem. i The size is When |D1 - D4| < L mm, compare the size of D i with the normal offset value D of the conveyor belt. Lateral deviation speed: Lateral deviation acceleration: a = d(vx) / dt (8) Step 4: For the deviation process, establish a deviation model with the conveyor belt speed, idler speed, conveyor belt tension, lateral deviation acceleration, and relative speed of the conveyor belt with respect to the idler as inputs and the corresponding conveyor belt deviation amount as the output; collect the data of |D1 - D4| < Lmm in advance for P times, and establish a neural network model in MATLAB for the collected data for training. Step 5: Remove the invalid or incorrectly measured parts in the original data; data processing to reduce errors: for data with missing values not exceeding 10%, data exceeding 10% is directly removed. Step 6: After data deletion, perform data filling: fill in the missing or incorrect data according to the regression replacement method. Step 7: Input the normalized training sample data set into the Matlab workspace in vector form, so as to establish the BP neural network for this experiment, with a relatively high training accuracy, which can meet the needs of deviation prediction. Step 8: After the BP neural network prediction model completes the training of the training set, the test set data is used to predict the deviation amount to test the prediction ability of the network model; Step 9: The input variable data are the measured conveyor belt speed, roller speed, conveyor belt tension, lateral deviation acceleration, and conveyor belt relative roller speed. The five comprehensive variables are normalized and input into the network prediction model to obtain the output variable, i.e., the lateral deviation amount. The deviation amount is predicted through the output variable. Step 10: When the predicted deviation is greater than L mm, start the alarm system, disconnect the power supply of the belt conveyor, and start the braking device to stop the movement of the conveyor belt in time; if the predicted deviation is less than L mm, continue the prediction.

Images