Method and device for controlling coal collection support based on neural network

Through the neural network-based control method, the scaffold control model is trained using historical and working environment data, the problem of low control efficiency of coal acquisition scaffolds in the existing technology is solved, and more efficient and accurate scaffold control is achieved, adapting to complex collection environments and improving Production safety.

CN118938756APending Publication Date: 2024-11-12TIANJIN HUANING ELECTRONICS
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411141359.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing coal collection bracket control method is inefficient and cannot adapt to the complex coal collection working surface environment.

Method used

A neural network-based control method is adopted to generate a training data set by obtaining historical environment and state data, train the scaffold control model, and use the working environment data for real-time control and model updates to improve the efficiency and accuracy of scaffold control.

Benefits of technology

It improves the control efficiency and accuracy of the coal collection bracket, can better adapt to the complex collection environment, reduces the dependence of manual operations, and ensures safe and efficient production of the working face.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118938756A_ABST
    Figure CN118938756A_ABST
Patent Text Reader

Abstract

The invention provides a control method and device of a coal collection support based on a neural network, and the method comprises the steps: obtaining historical data, the historical data comprises historical environment data and historical state data, and the historical environment data and the historical state data have a corresponding relation; and taking the historical environment data and the corresponding historical state data in the historical data as training data, and generating a training data set. And training the initial network model based on the training data set to obtain a support control model. And acquiring working environment data. And taking the working environment data as input of the bracket control model, and obtaining control data through the bracket control model. And controlling the coal collection bracket based on the control data. The working environment data and the corresponding control data serve as real-time training data, and parameters of the support control model are updated through incremental training. According to the method provided by the invention, accurate and efficient control on the coal collection bracket can be realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of coal mine automation, and in particular to a control method for a coal collection support based on a neural network. Background Art

[0002] With the continuous development of science and technology, a variety of intelligent equipment has been applied to all aspects of people's work and life. Coal is one of the indispensable sources of energy for people's production and life, and coal collection is a very important part of the entire coal industry. Coal collection has a working face, which refers to the first production site of coal. The working face of coal collection has the characteristics of small working space, many mechanical equipment, poor visual environment and high temperature. It is the key area in coal mining work.

[0003] The coal collection support is an indispensable equipment in the working face. The coal collection support refers to the equipment that supports the top and bottom plates in the coal collection working face to maintain the safety of the working space. The coal collection support can be a hydraulic support, for example.

[0004] The control of the coal collecting support by the related technology mainly relies on fixed control logic or manual control. The existing control method of the coal collecting support cannot adapt to various scenarios in the complex coal collecting working face, and the control efficiency of the coal collecting support is low.

[0005] Therefore, how to improve the control efficiency of the coal collection support has become a technical problem that technical personnel in this field need to solve urgently. Summary of the invention

[0006] Based on the above problems, the present application provides a control method and device for a coal collection support based on a neural network to solve the problem of low control efficiency of the coal collection support in the related technology. The method comprises the following steps:

[0007] Acquire historical data, wherein the historical data includes historical environmental data and historical status data, wherein the historical environmental data and the historical status data have a corresponding relationship, wherein the historical environmental data includes historical environmental temperature and historical environmental humidity, and wherein the historical status data includes historical displacement, historical pressure, and historical angle of the coal collection support;

[0008] Using the historical environment data and the corresponding historical status data in the historical data as training data to generate a training data set, wherein one piece of the training data includes one piece of the historical environment data and the corresponding historical status data;

[0009] Training the initial network model based on the training data set to obtain a support control model;

[0010] Acquiring working environment data, wherein the working environment data includes working environment temperature and working environment humidity;

[0011] Using the working environment data as input of the support control model, and obtaining control data through the support control model;

[0012] controlling the coal collecting support based on the control data;

[0013] The working environment data and the corresponding control data are used as real-time training data, and the parameters of the bracket control model are updated through incremental training. The bracket control model with updated parameters is used to predict the next working environment data to obtain the corresponding control data.

[0014] In a possible implementation, the training of the initial network model based on the training data set to obtain the support control model includes:

[0015] Performing feature extraction on the training data in the training data set to obtain a first feature extraction result corresponding to the historical environment data and a second feature extraction result corresponding to the historical state data;

[0016] Using the initial network model to predict the first feature extraction result to obtain a first prediction result;

[0017] Based on the difference between the first prediction result and the corresponding second feature extraction result, the parameters of the initial network model are adjusted until a training cutoff condition is reached to obtain a stent control model.

[0018] In a possible implementation, adjusting the parameters of the initial network model based on the difference between the first prediction result and the corresponding second feature extraction result until a training cutoff condition is reached to obtain the stent control model includes:

[0019] Using the first prediction result to obtain predicted stent stability, and using the second feature extraction result to obtain actual stent stability;

[0020] The parameters of the initial network model are adjusted based on the difference between the first prediction result and the corresponding second feature extraction result, and the training is stopped when the difference between the predicted stent stability and the actual stent stability reaches a preset threshold to obtain a stent control model.

[0021] In a possible implementation, adjusting the parameters of the initial network model based on the difference between the first prediction result and the corresponding second feature extraction result until a training cutoff condition is reached to obtain the stent control model includes:

[0022] Using the first prediction result and the corresponding second feature extraction result as inputs of a back propagation algorithm, and obtaining a weight coefficient through the back propagation algorithm;

[0023] Based on the difference between the weight coefficient, the first prediction result and the corresponding second feature extraction result, the parameters of the initial network model are adjusted until a training cutoff condition is reached to obtain a stent control model.

[0024] In a possible implementation, the method further includes:

[0025] Obtaining the parameter accuracy of the support control model;

[0026] The parameter accuracy of the support control model is quantified from high-bit accuracy to low-bit accuracy.

[0027] In a possible implementation, the method further includes:

[0028] Obtaining the changing trend of environmental parameters through the historical environmental data;

[0029] Using the historical state data, obtaining a historical pressure change rate and a historical displacement velocity;

[0030] Taking the historical environmental data whose difference from the environmental parameter change trend is less than the environmental parameter threshold in the historical environmental data as the historical environmental data in the training data;

[0031] The historical state data whose difference with the historical pressure change rate and the historical displacement speed is less than the state threshold value is used as the historical state data in the training data.

[0032] In a possible implementation, the acquiring historical data includes:

[0033] Get the historical data to be optimized;

[0034] If there are missing data in the historical data to be optimized, the historical data to be optimized are optimized by an interpolation method or a mean filling method to obtain historical data.

[0035] The present application also provides a control device for a coal collection support based on a neural network, the device comprising:

[0036] A first acquisition module is used to acquire historical data, wherein the historical data includes historical environmental data and historical status data, wherein the historical environmental data and the historical status data have a corresponding relationship, wherein the historical environmental data includes historical environmental temperature and historical environmental humidity, and wherein the historical status data includes historical displacement, historical pressure, and historical angle of the coal collection support;

[0037] A training data set generation module, used to use the historical environment data and the corresponding historical status data in the historical data as training data to generate a training data set, wherein one piece of training data includes one piece of historical environment data and the corresponding historical status data;

[0038] A training module, used to train the initial network model based on the training data set to obtain a support control model;

[0039] A second acquisition module is used to acquire working environment data, wherein the working environment data includes working environment temperature and working environment humidity;

[0040] A prediction module, used for taking the working environment data as input of the support control model, and obtaining control data through the support control model;

[0041] A control module, used for controlling the coal collection support based on the control data;

[0042] The optimization module is used to use the working environment data and the corresponding control data as real-time training data, and update the parameters of the bracket control model through incremental training. The bracket control model with updated parameters is used to predict the next working environment data to obtain the corresponding control data.

[0043] The present application also provides an electronic device, the electronic device comprising a processor and a memory:

[0044] The memory is used to store a computer program and transmit the computer program to the processor;

[0045] The processor is used to execute the steps of the control method of the coal collection support based on the neural network according to the instructions in the computer program.

[0046] The present application also provides a computer-readable storage medium, characterized in that the computer-readable storage medium is used to store a computer program, and when the computer program is executed by an electronic device, the steps of the control method of the above-mentioned neural network-based coal collection support are implemented.

[0047] Compared with the prior art, this application has the following beneficial effects:

[0048] The method provided in the present application first constructs a model for controlling a coal collection support. By acquiring historical data, and using the corresponding historical environment data and the corresponding historical status data in the historical data as training data, a training data set is generated. Then, the initial network model is trained based on the training data set to obtain a support control model. Since the historical data in the coal collection process is used when training the support control model, the support control model finally trained can better adapt to the complex coal collection environment. When controlling the coal collection support, the working environment data is first obtained. Then, the working environment data is used as the input of the support control model, and the control data is obtained through the support control model. The coal collection support is controlled based on the control data. In the method provided in the present application, in the coal collection working face, the support control model obtained by the above training can be combined with the support control model, and the support control model can efficiently obtain the control data of the coal collection support according to the working environment data, which improves the control efficiency of the coal collection support compared with the prior art. The method provided in the present application can also use the working environment data and the corresponding control data as real-time training data, and update the parameters of the support control model through incremental training. In this way, the control accuracy of the coal collection support is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0050] Figure 1 A flowchart of a control method for a coal collection support based on a neural network provided in this application;

[0051] Figure 2 A schematic structural diagram of a control device for a coal collection support based on a neural network provided in this application. DETAILED DESCRIPTION

[0052] As described above, in the process of coal mining, the control of the support of the coal collection working face is a complex and critical link. In the relevant technology, the support control mainly relies on fixed control logic and manual experience, which is difficult to adapt to the complex and changing working environment, and easily causes problems such as irregular support operation, substandard action, low mining efficiency and energy waste.

[0053] After research, it was found that in a complex and changeable coal collection environment, fixed control logic is often difficult to be competent, and the control of the coal collection bracket mainly depends on manual experience. The staff will adopt the control logic of the coal collection bracket corresponding to different coal collection environments. This is very dependent on the experience of the staff, and the control efficiency of the coal collection bracket is very low. In order to improve the control efficiency of the coal collection bracket, this application combines artificial intelligence with the control of the coal collection bracket.

[0054] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0055] It is understandable that the method provided in the present application can be applied to a processing device, which is a processing device that can train the initial network model, such as a terminal device or a server that can train the initial network model. The method provided in the present application can be executed independently by a terminal device or a server, or it can be applied to a network scenario in which a terminal device and a server communicate, and is executed by the cooperation of a terminal device and a server. Among them, the terminal device can be a computer, a mobile phone and other devices. The server can be understood as an application server or a Web server. In actual deployment, the server can be a stand-alone server or a cluster server.

[0056] Figure 1 A flowchart of a control method for a coal collection support based on a neural network provided in the present application, the method comprising the following steps:

[0057] S101: Acquire historical data.

[0058] The processing device can obtain historical data in a variety of different ways. For example, the processing device can obtain historical data in a database. Historical data includes historical environmental data and historical status data. Historical environmental data can be data used to represent the environment of the coal collection working face, and the historical environmental data can include, for example, historical environmental temperature and historical environmental humidity. Historical status data can be the state of the coal collection support, for example, it can be data such as the historical displacement, historical pressure and historical angle of the coal collection support. There is a corresponding relationship between the historical environmental data and the historical status data, and the historical status data is the status data of the coal collection support in the environment of the corresponding coal collection working face.

[0059] It is understandable that in some cases, historical data may have a certain impact on model training. For example, data with large errors in historical data may affect the accuracy of the trained support control model. For another example, missing data may affect the speed of training the support control model. In order to reduce the impact of historical data on model training, the processing device can perform certain processing on the historical data.

[0060] In a possible implementation, the processing device can obtain the trend of environmental parameter changes through historical environmental data, and obtain the historical pressure change rate and historical displacement speed by using historical state data. Then, the historical environmental data in the historical environmental data whose difference with the trend of environmental parameter changes is less than the environmental parameter threshold is used as the historical environmental data in the training data. If the historical environmental data includes the historical environmental temperature, the processing device can obtain the trend of the historical environmental temperature change according to the historical environmental temperature, such as the temperature continues to rise steadily, with an increase of about 1°C, and the environmental parameter threshold can be, for example, 0.5°C. If in the historical environmental data, the temperature corresponding to the first time is 30°C, and the temperature of the second time after the first time is 38°C, the difference between the two is 8°C, the increase is about 1°C, and the acceptable difference is between 0.5°C-1.5°C, therefore, the processing device can delete the historical environmental data corresponding to the second time. If in the historical environmental data, the temperature corresponding to the first time is 30°C, and the temperature of the second time after the first time is 30.7°C, therefore, the processing device can use the historical environmental data of the second time as the historical environmental data in the training data.

[0061] The processing device may also use the historical state data in which the difference between the historical pressure change rate and the historical displacement speed is less than the state threshold as the historical state data in the training data.

[0062] In the above method, the processing device can analyze and process the historical data, thereby eliminating the data that has an impact on the model training, and improving the accuracy of the model training data and the bracket control model obtained by training.

[0063] In another possible implementation, the processing device may first obtain the historical data to be optimized, and obtain the historical data by optimizing the historical data to be optimized. The optimization process may be to supplement the historical data. If there is missing data in the historical data to be optimized, the processing device may optimize the historical data to be optimized by interpolation or mean filling to obtain the historical data.

[0064] After obtaining the historical data, the processing device can also normalize the historical data. The processing device can use Z-score normalization or Min-Max normalization to perform normalization. The normalization operation can make the data get a unified scale to facilitate subsequent model training.

[0065] S102: Using historical environment data and corresponding historical status data in the historical data as training data to generate a training data set.

[0066] The processing device uses the historical environment data and the corresponding historical status data in the historical data as training data to generate a training data set, where one piece of training data includes one piece of historical environment data and the corresponding historical status data.

[0067] S103: Training the initial network model based on the training data set to obtain a support control model.

[0068] The processing device trains the initial network model based on the training data set to obtain the bracket control model. The initial network model can be selected according to the actual usage scenario. In one possible implementation, the initial network model can choose to use a multi-layer perceptron and a convolutional neural network structure. The initial network model can have an input layer, a hidden layer, and an output layer. The number of neurons in the input layer can be the same as the number of features corresponding to the training data set. The hidden layer is used to process and convert the input data so that it can be used by the output layer. In one possible implementation, the hidden layer can use a RelU activation function. The number of neurons in the output layer can be equal to the number of control parameters that need to be predicted. The activation function of the output layer can be selected according to the actual application scenario. For example, a linear activation function can be used for regression tasks, and a Softmax function can be used for classification tasks.

[0069] When training the initial network model, specifically, the processing device can extract features from the training data in the training data set to obtain a first feature extraction result corresponding to the historical environment data and a second feature extraction result corresponding to the historical state data. Then, the initial network model is used to predict the first feature extraction result to obtain a first prediction result. Finally, based on the difference between the first prediction result and the corresponding second feature extraction result, the parameters of the initial network model are adjusted until the training cutoff condition is reached to obtain the bracket control model.

[0070] In one possible implementation, the processing device can obtain a loss function based on the difference between the first prediction result and the corresponding second feature extraction result, and use the loss function to adjust the parameters of the initial network model. Different loss functions can be selected in different application scenarios. For example, for regression tasks, the mean square error can be selected as the loss function; for classification tasks, the cross entropy can be selected as the loss function. When training the model, an optimization algorithm can be used in combination, such as the Adam optimization algorithm for optimization. The optimization algorithm makes the trained model have good adaptability and the training process has a faster convergence speed.

[0071] In the process of training the model, the training cutoff condition can be set by different dimensions. For example, the processing device can determine the training cutoff condition in the dimension of the stability of the stent. The processing device can use the first prediction result to obtain the predicted stent stability, and use the second feature extraction result to obtain the actual stent stability. Afterwards, the parameters of the initial network model are adjusted based on the difference between the first prediction result and the corresponding second feature extraction result, until the difference between the predicted stent stability and the actual stent stability reaches a preset threshold, and the training is stopped to obtain the stent control model.

[0072] In the process of model training, in addition to the forward gradient (stent stability), the reverse gradient (model weight coefficient) can also be used in combination. The processing device can use the first prediction result and the corresponding second feature extraction result as inputs of the back propagation algorithm, and obtain the weight coefficient through the back propagation algorithm. Based on the difference between the weight coefficient, the first prediction result and the corresponding second feature extraction result, the parameters of the initial network model are adjusted until the training cutoff condition is reached to obtain the stent control model.

[0073] In the above method, a dual gradient is designed, namely the forward gradient (stent stability) and the reverse gradient (weight coefficient of the model). On the one hand, the forward gradient makes the stability of the stent an indicator, and on the other hand, the reverse gradient adjusts the weight coefficient. The dual gradient design ensures that the best balance is found between stent stability and energy consumption optimization.

[0074] After obtaining the support control model, the processing device can also perform a quantization operation on the support control model. The processing device can obtain the parameter accuracy of the support control model, and then quantize the parameter accuracy of the support control model from high-bit accuracy to low-bit accuracy. For example, if the parameter accuracy of the support control model is a 32-bit floating point number, the processing device can quantize the high-bit precision 32-bit floating point number into a low-bit precision 8-bit integer.

[0075] The above method can significantly reduce the storage and computing overhead of the model and reduce costs through quantization operations.

[0076] S104: Acquire working environment data.

[0077] The processing equipment obtains the working environment data. The working environment data can be obtained by continuously collecting the working surface through sensors or monitoring equipment. The working environment data may include the working environment temperature and the working environment humidity.

[0078] After acquiring the working environment data, the processing device can standardize and normalize the working environment data so that the working environment data and the historical data have the same scale, thereby ensuring the consistency of the working environment data and the historical data.

[0079] S105: Using the working environment data as input of the support control model, and obtaining control data through the support control model.

[0080] The processing equipment can input the acquired working environment data into the support control model, and the support control model performs prediction to obtain control data, which can be used to control the coal collection support.

[0081] S106: Control the coal collection support based on the control data.

[0082] The processing device controls the coal gathering support based on the control data.

[0083] S107: Using the working environment data and the corresponding control data as real-time training data, and updating the parameters of the support control model through incremental training.

[0084] The processing device can use the working environment data and the corresponding control data as real-time training data, and update the parameters of the support control model through incremental training. The support control model with updated parameters is used to predict the next working environment data to obtain the corresponding control data.

[0085] The above method can input real-time data into the trained support control model for real-time prediction and analysis. The model parameters can be updated in real time using real-time data. Under the premise of ensuring the stability of the model, the model can be gradually optimized to adapt to the dynamically changing working environment.

[0086] The method provided in the present application first constructs a model for controlling a coal collection support. By acquiring historical data, and using the corresponding historical environment data and the corresponding historical status data in the historical data as training data, a training data set is generated. Then, the initial network model is trained based on the training data set to obtain a support control model. Since the historical data in the coal collection process is used when training the support control model, the support control model finally trained can better adapt to the complex coal collection environment. When controlling the coal collection support, the working environment data is first obtained. Then, the working environment data is used as the input of the support control model, and the control data is obtained through the support control model. The coal collection support is controlled based on the control data. In the method provided in the present application, in the coal collection working face, the support control model obtained by the above training can be combined with the support control model, and the support control model can efficiently obtain the control data of the coal collection support according to the working environment data, which improves the control efficiency of the coal collection support compared with the prior art. The method provided in the present application can also use the working environment data and the corresponding control data as real-time training data, and update the parameters of the support control model through incremental training. In this way, the control accuracy of the coal collection support is improved.

[0087] In the actual application process, first of all, the processing equipment can use the working environment data and the support control model to evaluate the status of the current working face support in real time, including stability, displacement and environmental changes, which helps to complete the safe production of the working face.

[0088] Secondly, the processing device can obtain the optimal control data under the current environment based on the support control model, and generate the optimal control strategy based on the control data. The control strategy may include parameters such as the pressure and flow of the hydraulic system and the action time limit of the support. The solution provided in this application can take into account the historical state and current environmental conditions of the support, dynamically adjust the control strategy, and ensure that the support can maintain the optimal state under various working conditions.

[0089] Thirdly, since the method provided in the present application controls the bracket through a bracket control model, no manual operation is required and the displacement and angle of the bracket can be precisely controlled to ensure that the bracket always remains in the optimal position and achieve the flatness of the working surface.

[0090] The present application also provides a Figure 2 The structure diagram of the control device of the coal collection support based on the neural network is shown, and the control device 200 of the coal collection support based on the neural network includes:

[0091] The first acquisition module 201 is used to acquire historical data, wherein the historical data includes historical environmental data and historical status data, wherein the historical environmental data and the historical status data have a corresponding relationship, wherein the historical environmental data includes historical environmental temperature and historical environmental humidity, and wherein the historical status data includes historical displacement, historical pressure and historical angle of the coal collection support;

[0092] A training data set generation module 202, configured to use the historical environment data and the corresponding historical status data in the historical data as training data to generate a training data set, wherein one piece of the training data includes one piece of the historical environment data and the corresponding historical status data;

[0093] A training module 203 is used to train the initial network model based on the training data set to obtain a support control model;

[0094] The second acquisition module 204 is used to acquire working environment data, where the working environment data includes working environment temperature and working environment humidity;

[0095] A prediction module 205, configured to use the working environment data as an input of the support control model, and obtain control data through the support control model;

[0096] A control module 206, configured to control the coal collection support based on the control data;

[0097] The optimization module 207 is used to use the working environment data and the corresponding control data as real-time training data, and update the parameters of the bracket control model through incremental training. The bracket control model with updated parameters is used to predict the next working environment data to obtain the corresponding control data.

[0098] In a possible implementation, the training module is specifically used to:

[0099] Performing feature extraction on the training data in the training data set to obtain a first feature extraction result corresponding to the historical environment data and a second feature extraction result corresponding to the historical state data;

[0100] Using the initial network model to predict the first feature extraction result to obtain a first prediction result;

[0101] Based on the difference between the first prediction result and the corresponding second feature extraction result, the parameters of the initial network model are adjusted until a training cutoff condition is reached to obtain a stent control model.

[0102] In a possible implementation, the optimization module is specifically used to:

[0103] Using the first prediction result to obtain predicted stent stability, and using the second feature extraction result to obtain actual stent stability;

[0104] The parameters of the initial network model are adjusted based on the difference between the first prediction result and the corresponding second feature extraction result, and the training is stopped when the difference between the predicted stent stability and the actual stent stability reaches a preset threshold to obtain a stent control model.

[0105] In a possible implementation, the optimization module is specifically used to:

[0106] Using the first prediction result and the corresponding second feature extraction result as inputs of a back propagation algorithm, and obtaining a weight coefficient through the back propagation algorithm;

[0107] Based on the difference between the weight coefficient, the first prediction result and the corresponding second feature extraction result, the parameters of the initial network model are adjusted until a training cutoff condition is reached to obtain a stent control model.

[0108] In a possible implementation manner, the device further includes a quantization module, and the quantization module is specifically configured to:

[0109] Obtaining the parameter accuracy of the support control model;

[0110] The parameter accuracy of the support control model is quantified from high-bit accuracy to low-bit accuracy.

[0111] In a possible implementation manner, the device further includes a data processing module, and the data processing module is specifically configured to:

[0112] Obtaining the changing trend of environmental parameters through the historical environmental data;

[0113] Using the historical state data, obtaining a historical pressure change rate and a historical displacement velocity;

[0114] Taking the historical environmental data whose difference from the environmental parameter change trend is less than the environmental parameter threshold in the historical environmental data as the historical environmental data in the training data;

[0115] The historical state data whose difference with the historical pressure change rate and the historical displacement speed is less than the state threshold value is used as the historical state data in the training data.

[0116] In a possible implementation manner, the first acquisition module is specifically configured to:

[0117] Get the historical data to be optimized;

[0118] If there are missing data in the historical data to be optimized, the historical data to be optimized are optimized by an interpolation method or a mean filling method to obtain historical data.

[0119] It can be seen from the above device that the above device first constructs a model for controlling the coal collection support. By acquiring historical data, and using the corresponding historical environment data and the corresponding historical status data in the historical data as training data, a training data set is generated. Then, the initial network model is trained based on the training data set to obtain a support control model. Since the historical data in the coal collection process is used when training the support control model, the support control model finally trained can better adapt to the complex coal collection environment. When controlling the coal collection support, the working environment data is first obtained. Then, the working environment data is used as the input of the support control model, and the control data is obtained through the support control model. The coal collection support is controlled based on the control data. In the method provided in the present application, in the coal collection working face, the support control model obtained by the above training can efficiently obtain the control data of the coal collection support according to the working environment data, which improves the control efficiency of the coal collection support compared with the prior art. The method provided in the present application can also use the working environment data and the corresponding control data as real-time training data, and update the parameters of the support control model through incremental training. In this way, the control accuracy of the coal collection support is improved.

[0120] An embodiment of the present application also provides a control device for a coal collection support based on a neural network, wherein the device includes a memory and a processor, the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device performs the steps of the control method for a coal collection support based on a neural network as described in any embodiment of the present application.

[0121] In practical applications, the computer-readable storage medium may be any combination of one or more computer-readable media, which may be a computer-readable signal medium or a computer-readable storage medium.

[0122] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0123] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0124] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0125] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0126] It should be noted that each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is merely schematic, in which the unit described as a separate component may or may not be physically separated, and the component prompted as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0127] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A control method for a coal collection support based on a neural network, characterized in that: include: Acquire historical data, wherein the historical data includes historical environmental data and historical status data, wherein the historical environmental data and the historical status data have a corresponding relationship, wherein the historical environmental data includes historical environmental temperature and historical environmental humidity, and wherein the historical status data includes historical displacement, historical pressure, and historical angle of the coal collection support; Using the historical environment data and the corresponding historical status data in the historical data as training data to generate a training data set, wherein one piece of the training data includes one piece of the historical environment data and the corresponding historical status data; Training the initial network model based on the training data set to obtain a support control model; Acquiring working environment data, wherein the working environment data includes working environment temperature and working environment humidity; Using the working environment data as input of the support control model, and obtaining control data through the support control model; controlling the coal collecting support based on the control data; The working environment data and the corresponding control data are used as real-time training data, and the parameters of the bracket control model are updated through incremental training. The bracket control model with updated parameters is used to predict the next working environment data to obtain the corresponding control data.

2. The method according to claim 1, characterized in that The training of the initial network model based on the training data set to obtain the support control model includes: Performing feature extraction on the training data in the training data set to obtain a first feature extraction result corresponding to the historical environment data and a second feature extraction result corresponding to the historical state data; Using the initial network model to predict the first feature extraction result to obtain a first prediction result; Based on the difference between the first prediction result and the corresponding second feature extraction result, the parameters of the initial network model are adjusted until a training cutoff condition is reached to obtain a stent control model.

3. The method according to claim 2, characterized in that The adjusting the parameters of the initial network model based on the difference between the first prediction result and the corresponding second feature extraction result until the training cutoff condition is reached to obtain the stent control model includes: Using the first prediction result to obtain predicted stent stability, and using the second feature extraction result to obtain actual stent stability; The parameters of the initial network model are adjusted based on the difference between the first prediction result and the corresponding second feature extraction result, and the training is stopped when the difference between the predicted stent stability and the actual stent stability reaches a preset threshold to obtain a stent control model.

4. The method according to claim 1, characterized in that: The adjusting the parameters of the initial network model based on the difference between the first prediction result and the corresponding second feature extraction result until the training cutoff condition is reached to obtain the stent control model includes: Using the first prediction result and the corresponding second feature extraction result as inputs of a back propagation algorithm, and obtaining a weight coefficient through the back propagation algorithm; Based on the difference between the weight coefficient, the first prediction result and the corresponding second feature extraction result, the parameters of the initial network model are adjusted until a training cutoff condition is reached to obtain a stent control model.

5. The method according to claim 1, characterized in that The method further comprises: Obtaining the parameter accuracy of the support control model; The parameter accuracy of the support control model is quantified from high-bit accuracy to low-bit accuracy.

6. The method according to claim 1, characterized in that The method further comprises: Obtaining the changing trend of environmental parameters through the historical environmental data; Using the historical state data, obtaining a historical pressure change rate and a historical displacement velocity; Taking the historical environmental data whose difference from the environmental parameter change trend is less than the environmental parameter threshold in the historical environmental data as the historical environmental data in the training data; The historical state data whose difference with the historical pressure change rate and the historical displacement speed is less than the state threshold value is used as the historical state data in the training data.

7. The method according to claim 1, characterized in that The acquisition of historical data includes: Get the historical data to be optimized; If there are missing data in the historical data to be optimized, the historical data to be optimized are optimized by an interpolation method or a mean filling method to obtain historical data.

8. A control device for a coal collection support based on a neural network, characterized in that: include: A first acquisition module is used to acquire historical data, wherein the historical data includes historical environmental data and historical status data, wherein the historical environmental data and the historical status data have a corresponding relationship, wherein the historical environmental data includes historical environmental temperature and historical environmental humidity, and wherein the historical status data includes historical displacement, historical pressure, and historical angle of the coal collection support; A training data set generation module, used to use the historical environment data and the corresponding historical status data in the historical data as training data to generate a training data set, wherein one piece of training data includes one piece of historical environment data and the corresponding historical status data; A training module, used to train the initial network model based on the training data set to obtain a support control model; A second acquisition module is used to acquire working environment data, wherein the working environment data includes working environment temperature and working environment humidity; A prediction module, used for taking the working environment data as input of the support control model, and obtaining control data through the support control model; A control module, used for controlling the coal collection support based on the control data; The optimization module is used to use the working environment data and the corresponding control data as real-time training data, and update the parameters of the bracket control model through incremental training. The bracket control model with updated parameters is used to predict the next working environment data to obtain the corresponding control data.

9. An electronic device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store the computer program; The processor is used to execute the computer program to implement the control method of the coal collection support based on the neural network as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the control method of the coal collection support based on a neural network as described in any one of claims 1 to 7 is implemented.

Citation Information

Cited By

  • Coal mining automatic optimization control method based on reinforcement learning

    CN121832308A