Dumbbell pin state real-time monitoring system and method based on magnetic induction signals

Abstract

The invention discloses a dumbbell pin state real-time monitoring system and method based on a magnetic induction signal, and the method comprises the steps: segmenting magnetic field signal time slice data received by a magnetic detection device in the operation process of a coal mining machine, segmenting time slice data corresponding to each dumbbell pin, carrying out the sample marking of the segmented dumbbell pin time slice data, and carrying out the real-time monitoring of the dumbbell pin state through the previous historical data. Constructing a data set containing a large number of normal and abnormal samples; constructing a state recognition model of the dumbbell pin by using a long short-term memory (LSTM) network, and training the constructed state recognition model according to the constructed data set; deploying the trained state recognition model to an upper computer to realize real-time monitoring; magnetic induction intensity signals collected when the coal mining machine advances are transmitted to an upper computer in real time, the state of each dumbbell pin is rapidly judged through a model, once the dumbbell pin is detected to be missing, an alarm is triggered immediately, the specific position is recorded, and timely maintenance is facilitated.

Description

Technical Field

[0001] The invention relates to the technical field of scraper conveyors, and in particular to a system and method for real-time monitoring of dumbbell pin states based on magnetic induction signals. Background Art

[0002] In underground coal mine operations, the scraper conveyor is one of the key equipment, and the dumbbell pin, as a connecting component of the middle trough, directly affects the normal operation of the scraper conveyor. Once the dumbbell pin fails, such as breaking or falling off, it may cause equipment shutdown, and in serious cases, it may cause a safety accident. Traditional detection methods are mostly manual regular inspections, which are inefficient and cannot be monitored in real time. Especially for the dumbbell pins on the coal wall side, the coal blocks mined by the coal mining machine have accumulated here for a long time, resulting in the dumbbell pins not being buried. It is extremely difficult, time-consuming and labor-intensive to use manual detection. Therefore, efficient and convenient detection of dumbbell pins has always been the focus of research on scraper conveyor equipment.

[0003] At present, the main methods for detecting dumbbell pin failure are: ① Detection based on radio frequency identification (RFID), installing RFID tags on dumbbell pins, and monitoring dumbbell pins in real time through RFID readers and writers underground. RFID readers and writers can output RFID tag information and induced current strength signals to the host computer to determine whether the scraper has dumbbell pins that are detached or broken. ② Detection based on ultrasonic waves, that is, high-frequency sound waves are emitted through ultrasonic probes. When the sound waves propagate in the dumbbell pin material and encounter cracks or fractures, they will be reflected. The reflected waves are received and analyzed to detect whether there are cracks or fractures inside the dumbbell pins.

[0004] The dumbbell pin detection method based on RFID needs to overcome the signal penetration problem and the durability problem of the tag in the harsh environment underground, especially when the dumbbell pin on the coal wall side is buried by coal; the dumbbell pin detection method based on ultrasound requires special detection equipment and technicians, and the detection efficiency is low, especially in the harsh environment underground, which is inconvenient to operate and increases safety risks. Summary of the invention

[0005] In order to solve the technical problems existing in the above-mentioned technology, it is necessary to provide a real-time monitoring method of the dumbbell pin state based on magnetic induction signals.

[0006] A method for real-time monitoring of dumbbell pin status based on magnetic induction signals comprises the following steps:

[0007] Step S1: Segment the time slice data of the magnetic field signal received by the magnetic detection device during the operation of the coal mining machine to obtain the time slice data corresponding to each dumbbell pin;

[0008] Step S2: labeling samples of the segmented dumbbell pin time slice data to distinguish between normal state and missing state;

[0009] Step S3: construct a data set containing a large number of normal and abnormal samples through past historical data;

[0010] Step S4: construct a state recognition model of the dumbbell pin using a long short-term memory network (LSTM), and train the constructed state recognition model according to the constructed data set; deploy the trained state recognition model to the host computer to achieve real-time monitoring;

[0011] Step S5: The magnetic induction intensity signal collected when the coal mining machine is moving is transmitted to the host computer in real time, and the state of each dumbbell pin is quickly determined through the state recognition model. Once the dumbbell pin is detected to be missing, an alarm is immediately triggered and the specific location is recorded.

[0012] Preferably, in step S3, the constructed data set includes a training set and a test set, the training set is used for model training, and the test set is used to verify model performance.

[0013] Preferably, in step S4, after training is completed, the accuracy, recall and F1 score of the model are evaluated on the test set to ensure that the model has good generalization ability and robustness.

[0014] Preferably, in step S1, the time slice data of each dumbbell pin is obtained by:

[0015] S11: Obtain the time point when the front end of the nth dumbbell pin contacts the sensor The time point when the rear end of the nth dumbbell pin touches the sensor

[0016] S12: Set the sampling value set of the magnetic field signal time slice data s(t) when the nth dumbbell pin completely passes through the monitoring area of ​​the magnetic detection device as S n ; Get the time interval Δt between two consecutive sampling points;

[0017] S12: Construct the time slice data S corresponding to the nth dumbbell pin n ;

[0018] S13: Time slice data S n Make optimizations.

[0019] Preferably, the time point at which the front end of the nth dumbbell pin contacts the sensor Obtained through the following methods,

[0020]

[0021] Where V 采 is the travel speed of the coal mining machine, L哑 is the length of the dumbbell pin, L d is the distance between the front end position of the n-1th dumbbell pin and the front end position of the nth dumbbell pin, f s is the sampling frequency of the magnetic detection device.

[0022] Preferably, the time point at which the rear end of the nth dumbbell pin contacts the sensor Obtained through the following methods,

[0023]

[0024] Where V 采 is the travel speed of the coal mining machine, L 哑 is the length of the dumbbell pin, L d is the distance between the front end position of the n-1th dumbbell pin and the front end position of the nth dumbbell pin, f s is the sampling frequency of the magnetic detection device.

[0025] Preferably, the time slice data S corresponding to the nth dumbbell pin n Obtained through the following methods,

[0026] Suppose the sampling value set of the magnetic field signal time slice data s(t) when the nth dumbbell pin completely passes through the monitoring area of ​​the magnetic detection device is S n ;

[0027] Calculate the time interval Δt between two consecutive sampling points, Where f s is the sampling frequency of the given signal;

[0028] Then the time slice signal S corresponding to the nth dumbbell pin is n It can be expressed as arrive The discrete sampling value sequence of s(t) in the time interval, that is:

[0029]

[0030] In the formula, k represents the sampling point number, corresponding to arrive The sampling point number between , Z represents an integer set, and kΔt represents the sampling time point.

[0031] Preferably, considering in practical applications, due to and It will not fall exactly on the sampling time point, and usually the closest and The sampling time point is taken as the time slice signal S n The starting and ending time points are:

[0032]

[0033] In the formula, Represents a floor function that ensures that the sampling points selected are integer multiples of the closest time point.

[0034] Preferably, in step S2, the obtained several dumbbell pin time slice signals are converted into an input three-dimensional array format suitable for LSTM processing, corresponding to the number of samples, the time step of each sample, and the eigenvalue (magnetic induction intensity signal) of each time step; the output is one-dimensional data (i.e., 0 and 1), 1 represents a normal state, and 0 represents a missing state.

[0035] It is also necessary to provide a real-time monitoring system for the state of a dumbbell pin based on magnetic induction signals.

[0036] A real-time monitoring system for dumbbell pin status based on magnetic induction signals, comprising a magnetic detection device and a control device;

[0037] The magnetic detection device is installed on the shearer sliding shoe on the coal wall side and is at the same level as the dumbbell pin. The magnetic detection device receives signals from the external magnetic field at a fixed frequency during the movement of the shearer.

[0038] The control device forms a communication connection with the magnetic detection device, and the control device stores a computer program, which can execute the above-mentioned real-time monitoring method of the dumbbell pin state based on the magnetic induction signal when running, thereby monitoring the dumbbell pin state in real time.

[0039] Compared with the prior art, the present invention provides a real-time monitoring system and method for dumbbell pin status based on magnetic induction signals. The magnetic field signal time slice data received by the magnetic detection device during the operation of the coal mining machine is segmented to obtain the time slice data corresponding to each dumbbell pin. The segmented dumbbell pin time slice data is sampled and the normal state 1 and the missing state 0 are distinguished. Through the past historical data, a data set containing a large number of normal and abnormal samples is constructed for subsequent model training and verification. The state recognition model of the dumbbell pin is constructed using a long short-term memory network (LSTM). The labeled data set is divided into a training set and a test set. The training set is used for model training, and the test set is used to verify the model performance. After the training is completed, the accuracy, recall rate and F1 score of the model are evaluated on the test set to ensure that the model has good generalization ability and robustness. The trained model is deployed to the host computer to realize real-time monitoring. The magnetic induction intensity signal collected when the coal mining machine is moving is transmitted to the host computer in real time. The state of each dumbbell pin is quickly judged by the model. Once the dumbbell pin is detected to be missing, an alarm is immediately triggered and the specific location is recorded to facilitate timely maintenance. The present invention solves the practical problems of durable equipment and unmanned continuous and efficient detection under harsh conditions underground. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0041] Figure 1 It is a schematic diagram of the process flow of the dumbbell pin status monitoring system of the present invention.

[0042] Figure 2 This is a schematic diagram of installing the magnetic detection device of the present invention on the sliding shoe of a coal mining machine.

[0043] In the figure: magnetic detection device 01, dumbbell pin 02. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Please see Figure 1 In one embodiment, the present invention provides a method for real-time monitoring of dumbbell pin status based on magnetic induction signals, comprising the following steps:

[0046] Step S1: Segment the time slice data of the magnetic field signal received by the magnetic detection device during the operation of the coal mining machine to obtain the time slice data corresponding to each dumbbell pin;

[0047] Step S2: labeling samples of the segmented dumbbell pin time slice data to distinguish between normal state and missing state;

[0048] Step S3: construct a data set containing a large number of normal and abnormal samples through past historical data;

[0049] Step S4: construct a state recognition model of the dumbbell pin using a long short-term memory network (LSTM), and train the constructed state recognition model according to the constructed data set; deploy the trained state recognition model to the host computer to achieve real-time monitoring;

[0050] Step S5: The magnetic induction intensity signal collected when the coal mining machine is moving is transmitted to the host computer in real time. The state of each dumbbell pin is quickly determined through the state recognition model. Once the dumbbell pin is detected to be missing, an alarm is immediately triggered and the specific location is recorded.

[0051] To explain more specifically, in step S3, the constructed data set includes a training set and a test set, the training set is used for model training, and the test set is used to verify the model performance.

[0052] To be more specific, in step S4, after training is completed, the accuracy, recall and F1 score of the model are evaluated on the test set to ensure that the model has good generalization ability and robustness.

[0053] To explain more specifically, in step S1, the time slice data of each dumbbell pin is obtained in the following manner:

[0054] S11: Obtain the time point when the front end of the nth dumbbell pin contacts the sensor The time point when the rear end of the nth dumbbell pin touches the sensor

[0055] S12: Set the sampling value set of the magnetic field signal time slice data s(t) when the nth dumbbell pin completely passes through the monitoring area of ​​the magnetic detection device as S n ; Get the time interval Δt between two consecutive sampling points;

[0056] S12: Construct the time slice data S corresponding to the nth dumbbell pin n ;

[0057] S13: Time slice data S n Make optimizations.

[0058] Of course, the time point when the front end of the nth dumbbell pin touches the sensor Obtained through the following methods,

[0059]

[0060] Where V 采 is the travel speed of the coal mining machine, L 哑 is the length of the dumbbell pin, L d is the distance between the front end position of the n-1th dumbbell pin and the front end position of the nth dumbbell pin, f s is the sampling frequency of the magnetic detection device.

[0061] Of course, the time point when the rear end of the nth dumbbell pin touches the sensor Obtained through the following methods,

[0062]

[0063] Where V 采 is the travel speed of the coal mining machine, L 哑 is the length of the dumbbell pin, L d is the distance between the front end position of the n-1th dumbbell pin and the front end position of the nth dumbbell pin, fs is the sampling frequency of the magnetic detection device.

[0064] Among them, the time slice data S corresponding to the nth dumbbell pin n Obtained through the following methods,

[0065] Suppose the sampling value set of the magnetic field signal time slice data s(t) when the nth dumbbell pin completely passes through the monitoring area of ​​the magnetic detection device is S n ;

[0066] Calculate the time interval Δt between two consecutive sampling points, Where f s is the sampling frequency of the given signal;

[0067] Then the time slice signal S corresponding to the nth dumbbell pin is n It can be expressed as arrive The discrete sampling value sequence of s(t) in the time interval, that is:

[0068]

[0069] In the formula, k represents the sampling point number, corresponding to arrive The sampling point number between , Z represents an integer set, and kΔt represents the sampling time point.

[0070] Considering the practical application, and It will not fall exactly on the sampling time point, and usually the closest and The sampling time point is taken as the time slice signal S n The starting and ending time points are:

[0071]

[0072] In the formula, Represents a floor function that ensures that the sampling points selected are integer multiples of the closest time point.

[0073] Considering that the actual operation and working conditions of the coal mining machine may affect the stability of the speed, thus affecting the time slice signal S n The calculation of arrive The discrete sampling value sequences of s(t) in the time interval have different lengths. The pad_sequences method in keras is used to fill the sequences with 0 to the same length, and then the Masking layer is used in the model to ignore these padding values. In this way, even if the sequence lengths are different, the model can correctly process the real part of each sequence.

[0074] To explain more specifically, in step S2, the obtained dumbbell pin time slice signals are converted into an input three-dimensional array format suitable for LSTM processing, corresponding to the number of samples, the time step of each sample, and the eigenvalue (magnetic induction intensity signal) of each time step; the output is one-dimensional data (i.e., 0 and 1), 1 represents a normal state, and 0 represents a missing state.

[0075] Please see Figure 2 , In one embodiment, the present invention provides a real-time monitoring system for dumbbell pin status based on magnetic induction signals, including a magnetic detection device, a control device;

[0076] The magnetic detection device is installed on the shearer sliding shoe on the side of the coal wall and is at the same horizontal plane as the dumbbell pin. The magnetic detection device receives signals from the external magnetic field at a fixed frequency during the movement of the coal shearer. Since the middle groove and dumbbell pin of the scraper conveyor are installed in a fixed pattern, when the coal shearer runs on the scraper conveyor at a fixed speed, the magnetic field signals received by the magnetic detection device will show a certain pattern. When a dumbbell pin is missing, the magnetic field signal detected by the magnetic detection device passing through this place will change significantly. This relationship is used to detect the missing dumbbell pin.

[0077] The control device forms a communication connection with the magnetic detection device, and the control device stores a computer program, which can execute the real-time monitoring method of the dumbbell pin state based on the magnetic induction signal when running, so as to monitor the dumbbell pin state in real time.

[0078] What is disclosed above is only a preferred embodiment of the present invention, which certainly cannot be used to limit the scope of rights of the present invention. A person skilled in the art can understand that all or part of the processes of the above embodiments and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A method for real-time monitoring of dumbbell pin status based on magnetic induction signals, characterized in that: The following steps are included: Step S1: Segment the time slice data of the magnetic field signal received by the magnetic detection device during the operation of the coal mining machine to obtain the time slice data corresponding to each dumbbell pin; Step S2: labeling samples of the segmented dumbbell pin time slice data to distinguish between normal state and missing state; Step S3: construct a data set containing a large number of normal and abnormal samples through past historical data; Step S4: construct a state recognition model of the dumbbell pin using a long short-term memory network (LSTM), and train the constructed state recognition model according to the constructed data set; deploy the trained state recognition model to the host computer to achieve real-time monitoring; Step S5: The magnetic induction intensity signal collected when the coal mining machine is moving is transmitted to the host computer in real time, and the state of each dumbbell pin is quickly determined through the state recognition model. Once the dumbbell pin is detected to be missing, an alarm is immediately triggered and the specific location is recorded.

2. The method for real-time monitoring of dumbbell pin status based on magnetic induction signals according to claim 1 is characterized in that: In step S3, the constructed data set includes a training set and a test set, the training set is used for model training, and the test set is used to verify the model performance.

3. The method for real-time monitoring of dumbbell pin status based on magnetic induction signals according to claim 2 is characterized in that: In step S4, after training is completed, the accuracy, recall and F1 score of the model are evaluated on the test set to ensure that the model has good generalization ability and robustness.

4. The method for real-time monitoring of dumbbell pin status based on magnetic induction signals according to claim 1 is characterized in that: In step S1, the time slice data of each dumbbell pin is obtained in the following way, S11: Obtain the time point when the front end of the nth dumbbell pin contacts the sensor The time point when the rear end of the nth dumbbell pin touches the sensor S12: Set the sampling value set of the magnetic field signal time slice data s(t) when the nth dumbbell pin completely passes through the monitoring area of ​​the magnetic detection device as S n ; Get the time interval Δt between two consecutive sampling points; S12: Construct the time slice data S corresponding to the nth dumbbell pin n ; S13: Time slice data S n Make optimizations.

5. The method for real-time monitoring of dumbbell pin status based on magnetic induction signals according to claim 4 is characterized in that: The time point when the front end of the nth dumbbell pin touches the sensor Obtained through the following methods, Where V 采 is the travel speed of the coal mining machine, L 哑 is the length of the dumbbell pin, L d is the distance between the front end position of the n-1th dumbbell pin and the front end position of the nth dumbbell pin, f s is the sampling frequency of the magnetic detection device.

6. The method for real-time monitoring of dumbbell pin status based on magnetic induction signals according to claim 5 is characterized in that: The time point when the rear end of the nth dumbbell pin touches the sensor Obtained through the following methods, Where V 采 is the travel speed of the coal mining machine, L 哑 is the length of the dumbbell pin, L d is the distance between the front end position of the n-1th dumbbell pin and the front end position of the nth dumbbell pin, f s is the sampling frequency of the magnetic detection device.

7. The method for real-time monitoring of dumbbell pin status based on magnetic induction signals according to claim 4 is characterized in that: The time slice data S corresponding to the nth dumbbell pin n Obtained through the following methods, Suppose the sampling value set of the magnetic field signal time slice data s(t) when the nth dumbbell pin completely passes through the monitoring area of ​​the magnetic detection device is S n ; Calculate the time interval Δt between two consecutive sampling points, Where f s is the sampling frequency of the given signal; Then the time slice signal S corresponding to the nth dumbbell pin is n It can be expressed as arrive The discrete sampling value sequence of s(t) in the time interval, that is: In the formula, k represents the sampling point number, corresponding to arrive The sampling point number between , Z represents an integer set, and kΔt represents the sampling time point.

8. The method for real-time monitoring of dumbbell pin status based on magnetic induction signals according to claim 7 is characterized in that: Considering the practical application, and It will not fall exactly on the sampling time point, and usually the closest and The sampling time point is taken as the time slice signal S n The starting and ending time points are: In the formula, Represents a floor function that ensures that the sampling points selected are integer multiples of the closest time point.

9. The method for real-time monitoring of dumbbell pin status based on magnetic induction signals according to claim 1, characterized in that: In step S2, the obtained dumbbell pin time slice signals are converted into an input three-dimensional array format suitable for LSTM processing, corresponding to the number of samples, the time step of each sample, and the eigenvalue (magnetic induction intensity signal) of each time step; the output is one-dimensional data (i.e., 0 and 1), 1 represents the normal state, and 0 represents the missing state.

10. A real-time monitoring system for dumbbell pin status based on magnetic induction signals, characterized in that: Including magnetic detection device and control equipment; The magnetic detection device is installed on the shearer sliding shoe on the coal wall side and is at the same level as the dumbbell pin. The magnetic detection device receives signals from the external magnetic field at a fixed frequency during the movement of the shearer. The control device forms a communication connection with the magnetic detection device, and the control device stores a computer program, which, when running, can execute the real-time monitoring method of the dumbbell pin state based on the magnetic induction signal described in any one of claims 1-9, thereby monitoring the dumbbell pin state in real time.

Images