Autonomous Learning and Intelligent Decision-Making Methods for Control Behavior of Hydraulic Support Clusters

Abstract

This invention relates to an autonomous learning and intelligent decision-making method for controlling the behavior of hydraulic support clusters, belonging to the field of intelligent coal mine technology. It includes: mathematically representing the movement of hydraulic supports within the hydraulic support cluster; acquiring and determining a reference hydraulic support movement step matrix based on coal mining machine position data, hydraulic support movement control data, and hydraulic support stroke data; obtaining a target hydraulic support movement reference matrix from the reference hydraulic support movement step matrix; inputting the target hydraulic support movement reference matrix into a pre-trained intelligent decision-making model for movement adjustment, which outputs the movement distance for the target hydraulic support to perform the adjustment action; and controlling the target hydraulic support to perform the adjustment action based on the movement distance. This invention saves labor costs, improves the efficiency of hydraulic support movement adjustment, and has a high degree of intelligence.

Description

Technical Field

[0001] This invention relates to the field of intelligent coal mining technology, and in particular to an autonomous learning and intelligent decision-making method for the control behavior of hydraulic support clusters. Background Technology

[0002] The movement of hydraulic supports following the coal mining machine is a key control function in coal mining production. According to the coal mining process requirements, each hydraulic support in the hydraulic support cluster should move sequentially after the coal mining machine cuts the coal to provide timely support to the surrounding rock and ensure sufficient space in the mining area. Currently, hydraulic support control systems can achieve automatic follow-the-machine operation through automatic programming. However, in actual field applications, due to uncertainties such as geological conditions and equipment status, the execution effect of controlling the movement according to the stroke target value determined by the current automatic follow-the-machine control system cannot perfectly achieve the straight alignment requirements of the hydraulic support cluster in normal production, thus affecting the overall straightness requirements of the working face. This is the main reason why the current automatic follow-the-machine control system is difficult to operate routinely. In fact, in actual production, after the hydraulic supports complete their initial follow-the-machine movement (i.e., the initial movement), it is necessary to control the movement again to adjust the straight alignment of the hydraulic support cluster until the production requirements are met.

[0003] Currently, the initial movement of hydraulic supports at the coal mining face is automatically controlled by the hydraulic support automatic following control system. Subsequent adjustments are made manually based on on-site observation of the straight alignment of the hydraulic support cluster. In other words, the adjustment distance for the subsequent adjustments is determined manually based on the on-site conditions of the coal mining face. This not only requires high labor costs but also results in low adjustment efficiency and a low level of intelligence. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an autonomous learning and intelligent decision-making method for the control behavior of hydraulic support clusters. The technical solution of this invention is as follows:

[0005] An autonomous learning and intelligent decision-making method for controlling the behavior of a hydraulic support cluster, comprising:

[0006] S1, Mathematicalize the hydraulic support shifting action in the hydraulic support cluster to construct the initial hydraulic support shifting step distance matrix. The shifting action performed by the hydraulic support during each coal cutting process of the coal mining machine includes basic actions and adjustment actions. The initial hydraulic support shifting step distance matrix includes the shifting distance of each hydraulic support during each coal cutting process of the coal mining machine, including the basic actions and adjustment actions performed by each hydraulic support.

[0007] S2, acquire the position data of the coal mining machine, the hydraulic support moving action control data and the hydraulic support stroke data, and determine the value of each element in the initial hydraulic support moving step matrix based on the acquired coal mining machine position data, hydraulic support moving action control data and hydraulic support stroke data, and obtain the reference hydraulic support moving step matrix;

[0008] S3, obtain the target hydraulic support's shifting reference matrix from the reference hydraulic support shifting step matrix, wherein the target hydraulic support is the hydraulic support in the hydraulic support cluster whose shifting distance needs to be determined for adjustment.

[0009] S4, input the target hydraulic support's shifting reference matrix into the pre-trained shifting adjustment intelligent decision model, and the shifting adjustment intelligent decision model outputs the shifting distance of the target hydraulic support to perform the adjustment action;

[0010] S5, when the target hydraulic support performs the moving action, control the target hydraulic support to perform the adjustment action according to the moving distance of the target hydraulic support to perform the adjustment action.

[0011] Optionally, S1 includes:

[0012] S11, determine that the hydraulic support cluster of the coal mining face includes n hydraulic supports, and the coal mining machine advances m cutters during coal mining;

[0013] S12, based on the moving actions performed by each hydraulic support in the hydraulic support cluster, including basic action a and adjustment action b, construct the initial hydraulic support moving step matrix, represented as:

[0014]

[0015] in, It refers to the distance the support shifts during the basic action a of the i-th hydraulic support when the coal mining machine cuts the j-th coal (1≤i≤n); It refers to the shifting distance (1≤j≤m) formed by the adjustment action b of the i-th hydraulic support when the coal mining machine is cutting the j-th coal.

[0016] Optionally, in step S2, the initial hydraulic support shifting step matrix is ​​determined based on the acquired coal mining machine position data, hydraulic support shifting action data, and hydraulic support stroke data. and When the value is , it includes:

[0017] S21, Determine the time sequence set T of events where the coal mining machine passes the hydraulic support based on the coal mining machine's position data. i , represented as:

[0018]

[0019] Among them, Ti Each element in the text represents a moment in the process of the coal mining machine cutting m pieces of coal at the coal mining face, when the coal mining machine passes the i-th hydraulic support m times.

[0020] S22, Determine the sequence set t of hydraulic support moving action events based on the hydraulic support moving action control data. i , is represented as;

[0021]

[0022] Where t represents the moment when the hydraulic support moving action event occurs; t i Each element in the text represents the moment when the i-th hydraulic support moves o times during the process of cutting m-cut coal at the coal mining face;

[0023] S23, when to There are a total of c+1 time points in the state and Between, that is, satisfying When the time is right, it is determined that the kth to k+cth moving action of the i-th hydraulic support occurs at the j-th cutter.

[0024] S24, define Δl[t,t+d] as the change in the hydraulic support stroke data over time d after time t. and The specific values ​​are as follows:

[0025] If c = 0, then

[0026] If c = 1, then

[0027] If c≥2, then

[0028] Optionally, when the target hydraulic support is the i-th hydraulic support, step S3, when obtaining the shifting reference matrix of the i-th hydraulic support from the shifting step matrix of the reference hydraulic support, includes:

[0029] Take the moving distance values ​​within the range [ix, i+x] and the advancing range [jy, j] from the reference hydraulic support moving step matrix to form a vector dataset of (2x+1)×(2y+1). This vector dataset is the moving reference matrix of the i-th hydraulic support, expressed as:

[0030]

[0031] Optionally, before step S4, the method further includes: training a smart decision-making model for moving and adjusting the frame, and the step of training the smart decision-making model for moving and adjusting the frame includes:

[0032] S41, Creating Training Samples: Obtain historical coal mining machine position data, historical hydraulic support shifting action control data, and historical hydraulic support stroke data during the historical coal cutting process of the coal mining machine after cutting m cutters. Construct a training hydraulic support shifting step matrix based on these data. For hydraulic support i, during the coal mining machine's cutting of m cutters, determine the shifting distance of its adjustment action b. For target estimation, the moving distance values ​​within the range of [ix, i+x] frame number and [jy, j] advance range are taken from the training hydraulic support moving step distance matrix to form a vector dataset of (2x+1)×(2y+1). This vector dataset of (2x+1)×(2y+1) is used as the input of the intelligent decision model for moving adjustment.

[0033] S42, Labeling: Determine the minimum accuracy of the hydraulic support shifting distance as δ, divide the output of the intelligent decision-making model for support shifting into B output categories, then the labels of the intelligent decision-making model for support shifting satisfy:

[0034]

[0035] In the formula, This represents the floor function; the displacement distances corresponding to the B output categories are {[0,2δ),[2δ,4δ),…,[2(B-1)δ,l} max ]}, l max This refers to the maximum stroke value of the hydraulic support during each moving action;

[0036] Define the training hydraulic support shifting step matrix If the label falls within any output category range, then the label is that output category;

[0037] S43 uses training samples and labels to train the intelligent decision-making model for adjusting the transfer frame through machine learning.

[0038] Optionally, before determining the values ​​of each element in the initial hydraulic support shifting step matrix based on the acquired coal mining machine position data, hydraulic support shifting action control data, and hydraulic support stroke data, step S2 further includes:

[0039] Fluctuation and missing value processing are performed on the hydraulic support stroke data.

[0040] All of the above-mentioned optional technical solutions can be combined arbitrarily, and the present invention will not provide a detailed description of the structure after each combination.

[0041] By means of the above solution, the beneficial effects of the present invention are as follows:

[0042] By acquiring coal mining machine position data, hydraulic support movement control data, and hydraulic support stroke data, and determining the values ​​of each element in the initial hydraulic support movement step matrix based on these data, a reference hydraulic support movement step matrix is ​​obtained. For any target hydraulic support in the hydraulic support cluster that requires a movement distance for adjustment, the movement reference matrix determined based on the reference hydraulic support movement step matrix is ​​input into a pre-trained intelligent decision-making model for movement adjustment. The intelligent decision-making model outputs the movement distance for the target hydraulic support to perform the adjustment action, which is then used as the movement distance for the adjustment action. This provides a method for intelligently learning a movement adjustment intelligent decision-making model based on massive data from the coal mining face. This method intelligently decides the movement distance for hydraulic supports to perform adjustment actions, which not only saves labor costs but also improves the efficiency of hydraulic support movement adjustment, demonstrating a high degree of intelligence.

[0043] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0044] Figure 1 This is a flowchart of an embodiment of the present invention, which describes an autonomous learning and intelligent decision-making method for the control behavior of a hydraulic support cluster.

[0045] Figure 2 This is a schematic diagram showing the fluctuation of stroke data of a hydraulic support obtained in an embodiment of the present invention.

[0046] Figure 3 This is a schematic diagram of the composition structure of the intelligent decision-making model for frame adjustment in an embodiment of the present invention.

[0047] Figure 4 This is an overall flowchart of the autonomous learning and intelligent decision-making method for hydraulic support cluster control behavior provided in the embodiments of the present invention. Detailed Implementation

[0048] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0049] The autonomous learning and intelligent decision-making method for hydraulic support cluster control behavior provided in this invention can be implemented using any electronic device with computing capabilities, such as a PC, mobile terminal, or server. The massive amounts of data from the coal mining face, including hydraulic support stroke data, hydraulic support moving action control data, and coal mining machine position data, contain information such as the hydraulic support cluster arrangement conditions and manual operation trajectories. The autonomous learning and intelligent decision-making method for hydraulic support cluster control behavior provided in this invention utilizes appropriate machine learning methods to autonomously learn on-site manual operation methods from massive amounts of data, constructing an intelligent decision-making model for support moving and adjustment, and achieving a fully adaptive control mode for the entire process of the hydraulic support's initial basic action followed by subsequent adjustment actions. Figure 1 As shown, the autonomous learning and intelligent decision-making method for hydraulic support cluster control behavior provided in this embodiment of the invention includes the following steps S1 to S6:

[0050] S1, the hydraulic support shifting action in the hydraulic support cluster is mathematically quantified to construct the initial hydraulic support shifting step distance matrix. The shifting action performed by the hydraulic support during each coal cutting process of the coal mining machine includes basic actions and adjustment actions. The initial hydraulic support shifting step distance matrix includes the shifting distance of each hydraulic support during each coal cutting process of the coal mining machine, including the basic actions and adjustment actions performed by each hydraulic support.

[0051] Specifically, assuming there are n hydraulic supports in the coal mining face, and the coal mining machine advances m cutters, the hydraulic supports complete two types of moving actions during each coal cutting process: basic actions (denoted as a) and adjustment actions (denoted as b). Basic action a refers to the normal moving control completed by the hydraulic support according to the machine-following process requirements. This is generally the initial moving control with a full stroke, and is often implemented in the field by the hydraulic support's automatic following control system through a preset automated control strategy. Adjustment action b refers to the control strategy for adjusting the moving action of the hydraulic support after completing the basic actions, based on on-site conditions such as straightness. This is generally the moving control with a small stroke, and is currently often implemented in the field by manual judgment and operation. Let l be the moving distance along the working face advancement direction caused by each moving action of the hydraulic support, with a value range of [0 ≤ l ≤ l]. max ], l max This refers to the maximum stroke value of each hydraulic support movement. In this embodiment of the invention, based on the composition of the hydraulic support movement in a coal mining face hydraulic support cluster, the hydraulic support movement in the cluster is mathematically quantified to construct an initial hydraulic support movement step matrix.

[0052] Based on the above, S1 includes: S11, determining that the hydraulic support cluster of the coal mining face includes n hydraulic supports, and the coal mining machine advances m cutters during coal mining; here, m cutters refers to the mth cut of coal that has been cut by the coal mining machine during the coal cutting process;

[0053] S12, based on the moving actions performed by each hydraulic support in the hydraulic support cluster, including basic action a and adjustment action b, construct the initial hydraulic support moving step matrix, represented as:

[0054]

[0055] in, It refers to the distance the support shifts during the basic action a of the i-th hydraulic support when the coal mining machine cuts the j-th coal (1≤i≤n); This refers to the shifting distance (1≤j≤m) caused by the adjustment action b of the i-th hydraulic support during the cutting of the j-th coal cut by the coal mining machine; it is worth noting that or A value of 0 indicates that the action did not occur. Under normal circumstances, the basic action... Generally greater than 0 and close to l max Value, Adjustment Action The value is typically 0 (no adjustment action is performed) or a small value (adjustment action is performed).

[0056] S2, acquire the position data of the coal mining machine, the hydraulic support moving action control data, and the hydraulic support stroke data, and determine the value of each element in the initial hydraulic support moving step matrix based on the acquired coal mining machine position data, hydraulic support moving action control data, and hydraulic support stroke data, to obtain the reference hydraulic support moving step matrix.

[0057] Specifically, various types of sensors are installed on the coal mining face. After collecting data, these sensors send it to electronic equipment, which can then retrieve data from the corresponding sensor based on the required data type. For example, the position data of the coal mining machine can be obtained from the data collected by the positioning sensor installed on the coal mining machine; the stroke data of the hydraulic support can be obtained from the data collected by the hydraulic support stroke sensor installed on the hydraulic support. The hydraulic support moving action control data reflects whether the hydraulic support has moved. In this embodiment of the invention, this data is taken as the time when the hydraulic support moving action event occurs.

[0058] In one optional embodiment, due to issues such as data fluctuations and data loss during the acquisition, transmission, and storage of hydraulic support stroke data, the accuracy of decision-making results can be affected. Therefore, this embodiment of the invention can further process the hydraulic support stroke data for fluctuation and missing values ​​after acquisition. The same applies to the coal mining machine position data and the hydraulic support moving action control data.

[0059] Specifically, to address the issue of data fluctuations, this embodiment employs filtering techniques to smooth the data, reducing the impact of short-term fluctuations and resulting in more consistent data. Regarding the issue of data loss, this embodiment uses previous values ​​to fill in the gaps, making the data more complete.

[0060] like Figure 2 As shown, this illustrates the fluctuations in the hydraulic support stroke data of hydraulic support number 66. If the hydraulic support does not move, but the hydraulic support stroke data shows repeated small-range changes, then the hydraulic support stroke data fluctuation is determined, and the hydraulic support stroke data values ​​during this period are [v1, v2, ..., v...]. z v1 represents the first point at the start of the fluctuation, z represents the total number of points, and the average value of the hydraulic support stroke data during the fluctuation is denoted as v. mean , will [v mean v mean , ..., v mean The data replaces the original fluctuation data. The mathematical formula can be expressed as:

[0061]

[0062] If data loss occurs, such as when the stroke data of hydraulic support i is empty for a certain period of time, then the value before the stroke data loss is used to fill the gap. Let v be the value of the hydraulic support stroke data before it is lost. p The value during the missing period is recorded as None. This can be expressed mathematically as:

[0063] v p ,None,None,None,None→v p v p v p v p v p .

[0064] By employing the aforementioned methods for handling fluctuation and missing values, not only is the integrity and accuracy of the hydraulic support stroke data ensured, but a high-quality data foundation is also provided for subsequent model updates.

[0065] In one specific embodiment, step S2 determines the initial hydraulic support shifting step matrix based on the acquired coal mining machine position data, hydraulic support shifting action data, and hydraulic support stroke data. and When the value is , it includes:

[0066] S21, Determine the time sequence set T of events where the coal mining machine passes the hydraulic support based on the coal mining machine's position data. i , represented as:

[0067]

[0068] Among them, T i Each element in the diagram represents a moment in the process of the coal mining machine cutting m pieces of coal at the coal face, specifically the time when the coal mining machine passes the i-th hydraulic support m times.

[0069] S22, Determine the sequence set t of hydraulic support moving action events based on the hydraulic support moving action control data. i , is represented as;

[0070]

[0071] Where t represents the moment when the hydraulic support moving action event occurs; t i Each element in the diagram represents a moment when the i-th hydraulic support moves o times during the process of cutting m-cut coal at the coal mining face.

[0072] S23, when to There are a total of c+1 time points in the state and Between, that is, satisfying When the time is right, it is determined that the kth to k+cth moving action of the i-th hydraulic support occurs at the j-th cutter.

[0073] S24, define Δl[t,t+d] as the change (decrease) in the hydraulic support stroke data over a time interval d after time t. and The specific values ​​are as follows:

[0074] If c = 0, then

[0075] If c = 1, then

[0076] If c≥2, then

[0077] Through the above mathematical process, massive amounts of data collected on-site at the working face, such as hydraulic support stroke data, hydraulic support moving action control data, and coal mining machine position data, can be transformed into a mathematical expression of the spatiotemporal evolution of the hydraulic support moving process throughout the entire coal mining face. This provides a data foundation for the autonomous learning and intelligent decision-making of the hydraulic support cluster control behavior in the next step.

[0078] S3, obtain the target hydraulic support's shifting reference matrix from the reference hydraulic support shifting step matrix, wherein the target hydraulic support is the hydraulic support in the hydraulic support cluster whose shifting distance needs to be determined for adjustment.

[0079] The frame shifting reference matrix is ​​determined by the frame shifting action between the first preset value hydraulic supports on the left and right sides of the target hydraulic support and the second preset value hydraulic supports before the current cutter of the coal mining machine, in the frame shifting step matrix of the reference hydraulic support. The specific values ​​of the first preset value and the second preset value can be determined as needed, and this embodiment of the invention does not limit them.

[0080] In one specific embodiment, when the target hydraulic support is the i-th hydraulic support, when S3 obtains the moving reference matrix of the i-th hydraulic support from the moving step matrix of the reference hydraulic support, it takes the moving distance values ​​within the range of [ix, i+x] and the advancing range of [jy, j] from the moving step matrix of the reference hydraulic support to form a vector dataset of (2x+1)×(2y+1). The vector dataset is the moving reference matrix of the i-th hydraulic support, expressed as:

[0081]

[0082] It should be noted that if the target hydraulic support is located at the end position, leaving no hydraulic support on one side (e.g., hydraulic support number 1), and there is no hydraulic support on its left side, then the shifting reference matrix is ​​supplemented with the standard value on the left side according to the number of missing hydraulic supports. That is, the position of basic action 'a' is supplemented with 'l'. max The value is adjusted by filling the position of action b with a 0 value, which means that no adjustment action has occurred.

[0083] S4. Input the target hydraulic support's shifting reference matrix into the pre-trained intelligent decision-making model for shifting and adjusting the hydraulic support. The intelligent decision-making model for shifting and adjusting the hydraulic support will output the shifting distance for the target hydraulic support to perform the adjustment action.

[0084] The intelligent decision-making model for support adjustment takes the support adjustment reference matrix of the target hydraulic support as input and outputs 0 or a positive value. An output of 0 indicates that no adjustment is needed, while a positive output is one of several categorical values, each corresponding to a support adjustment distance.

[0085] It should be noted that step S4 includes: training the intelligent decision-making model for moving and adjusting the frame. Specifically, the steps for training the intelligent decision-making model for moving and adjusting the frame include:

[0086] S41, Creating Training Samples: Obtain historical coal mining machine position data, historical hydraulic support shifting action control data, and historical hydraulic support stroke data during the historical coal cutting process of the coal mining machine after cutting m cutters. Construct a training hydraulic support shifting step matrix based on these data. For hydraulic support i, during the coal mining machine's cutting of m cutters, determine the shifting distance of its adjustment action b. For target estimation, the moving distance values ​​within the range of frame number [ix, i+x] and the advancing range [jy, j] are taken from the training hydraulic support moving step distance matrix to form a vector dataset of (2x+1)×(2y+1). This vector dataset of (2x+1)×(2y+1) is used as the input of the intelligent decision model for moving adjustment.

[0087] Through the above mathematical process, historical hydraulic support stroke data, historical coal mining machine position data, and historical hydraulic support moving action control data collected on-site at the working face can be transformed into a mathematical expression of the spatiotemporal evolution of the hydraulic support moving process in the entire coal mining working face, providing a complete and usable sample space for the intelligent decision-making model for support moving adjustment.

[0088] The method for obtaining historical coal mining machine position data, historical hydraulic support movement control data, and historical hydraulic support stroke data, and constructing a training hydraulic support movement step matrix based on these data, is similar to the principle in embodiment S2 above, and will not be repeated here.

[0089] Through the above S41, multiple (2x+1)×(2y+1) vector datasets can be obtained. These vector datasets are used as sample datasets, that is, the input of the intelligent decision-making model is adjusted.

[0090] It should be noted that if, during the training sample preparation process, the hydraulic supports are located at the end position, causing the number of hydraulic supports on one side to not meet the sampling conditions (e.g., hydraulic support number 1 has no hydraulic supports on its left side), then the standard value should be supplemented on its left side according to the missing number of hydraulic supports, i.e., the position of basic action 'a' should be supplemented with 'l'. max The value is adjusted by adding a 0 value to the position of action b, indicating that no adjustment action has occurred. This yields the vector data for i=1 as follows. Similarly, if there is no support on the right, the vector data is formed by adding the corresponding values.

[0091]

[0092] S42, Labeling: Determine the minimum accuracy of the hydraulic support shifting distance as δ, divide the output of the intelligent decision-making model for support shifting into B output categories, then the labels of the intelligent decision-making model for support shifting satisfy:

[0093]

[0094] In the formula, This represents the floor function; the displacement distances corresponding to the B output categories are {[0,2δ),[2δ,4δ),…,[2(B-1)δ,l} max ]}, l maxThis refers to the maximum stroke value of each movement of the hydraulic support; δ is determined based on the minimum stroke value of the hydraulic support during movement.

[0095] Define the training hydraulic support shifting step matrix If a value falls within any output category range, then the label is that output category.

[0096] Based on the aforementioned training hydraulic support shifting step matrix, multiple labels for the intelligent decision-making model for support shifting adjustment can be obtained, forming a label database.

[0097] S43 uses training samples and labels to train the intelligent decision-making model for adjusting the transfer frame through machine learning.

[0098] Specifically, when training the intelligent decision-making model for frame relocation adjustment using training samples and labels, the training samples and their corresponding labels are first divided into a training set and a test set. Then, after assigning initial values ​​to the intelligent decision-making model for frame relocation adjustment, it undergoes autonomous learning and continuous iteration using the training set, and is validated using the test set. Once the test is passed, the trained intelligent decision-making model for frame relocation adjustment is obtained.

[0099] Since the training matrix of hydraulic support shifting steps contains the spatiotemporal evolution trajectory of basic and adjustment action control strategies, especially since the adjustment action is controlled and executed after manual on-site observation, it contains a large amount of manual operation experience. Therefore, this embodiment of the invention uses machine learning to train the intelligent decision-making model for shifting adjustment by using the shifting distance of a certain hydraulic support adjustment action as the output, and using the shifting distances of a certain number of adjacent hydraulic supports on the left and right (spatial arrangement characteristics) and the shifting distances of the previous few cuts (time cumulative characteristics) as the input to the intelligent decision-making model for shifting adjustment.

[0100] Regarding the specific type of intelligent decision-making model for relocation and adjustment, the embodiments of the present invention do not impose specific limitations. For example, the intelligent decision-making model for relocation and adjustment may be KNN, SVM, decision tree, or random forest, or it may be a combination of several models.

[0101] As coal mining continues, the amount of data from the working face will increase dramatically. This means that the number of training samples and labels generated by training the hydraulic support shifting step matrix will also increase, allowing for the learning of more and more manual operational experience. By analyzing the characteristics of the training samples, which exhibit both time-series and spatial features, this invention combines convolutional networks, long short-term memory neural networks, and attention mechanisms to construct an intelligent decision-making model for support shifting adjustments. Figure 3The diagram shown illustrates the structural composition of the intelligent decision-making model for frame relocation and adjustment in this embodiment of the invention. Through continuous iterative updates, the intelligent decision-making model for frame relocation and adjustment can autonomously learn more on-site manual operation methods. Currently, through autonomous learning from massive amounts of data on the working face, the intelligent decision-making model for frame relocation and adjustment achieves an accuracy rate of 83.84%, meeting the standards for industrial field applications.

[0102] Specifically, such as Figure 3 As shown, taking a (7*13) vector dataset as the input and three-class classification intelligent decision-making model parameters for relocation and adjustment as an example, the specific architecture diagram of the intelligent decision-making model for relocation and adjustment is as follows:

[0103] The network parameters for each layer are as follows: The input is a (7*13) vector dataset, which is converted to 1D before passing through the first convolutional layer. The parameters of the first convolutional layer are: type "Conv2d", 1 input channel, 32 output channels, kernel size (2, 3), padding: 1. The parameters of the first batch normalization layer are: type "BatchNorm2d", 32 features. The first activation function is ReLU. The first random dropout layer has a probability of 0.5. The parameters of the second convolutional layer are: type "Conv2d", 32 input channels, 64 output channels, kernel size (2, 3), padding: 1. The parameters of the second batch normalization layer are: type "BatchNorm2d", 64 features. The second activation function is ReLU. The second random dropout layer has a probability of 0.3. The flattening layer converts the multidimensional data into one dimension. Parameters of the bidirectional LSTM layer: Type "LSTM", input size 64, hidden layer size 100, number of layers 2; batch priority: True, bidirectional: True. First fully connected layer: Type "Linear", input size: 200, output size: 50. Third batch normalization layer: Type "BatchNorm1d", number of features: 50. Attention layer: Composed as a linear layer to convert to 25 dimensions, applying the Tanh activation function, then passing through another linear layer to convert back to 1 dimension for weight calculation. Softmax is used to normalize the weights, and then the input is summed with weights. Second fully connected layer: Type "Linear", input size: 50, output size: 3.

[0104] S5, when the target hydraulic support performs the moving action, control the target hydraulic support to perform the adjustment action according to the moving distance of the target hydraulic support to perform the adjustment action.

[0105] Specifically, when controlling the target hydraulic support to perform adjustment actions based on the shifting distance of the target hydraulic support, the target support can be adjusted manually based on the shifting distance of the target hydraulic support, or the target hydraulic support can be controlled to perform adjustment actions by an automated following system.

[0106] The method provided in this invention, through the processing of massive amounts of data from the working face, enables the autonomous learning and intelligent decision-making method for the control behavior of the hydraulic support cluster to intelligently determine how the hydraulic support cluster should be moved and adjusted again after the initial control of the hydraulic support cluster, thus helping to adjust the straight alignment of the hydraulic support cluster until it meets production requirements.

[0107] In summary, the autonomous learning and intelligent decision-making method for hydraulic support cluster control behavior provided in this invention preprocesses massive amounts of on-site data from the working face, such as historical coal mining machine position data, historical hydraulic support moving action control data, and historical hydraulic support stroke data (processing for fluctuation values ​​and missing values), mathematically transforms the hydraulic support moving action, and creates a training hydraulic support moving step matrix. Then, based on the training hydraulic support moving step matrix, a vector dataset of size (2x+1)×(2y+1) is extracted as the sample input to the intelligent decision-making model for support adjustment, and the moving distance of adjustment action b is used as the output of the intelligent decision-making model for support adjustment. By analyzing the characteristics of the dataset, which simultaneously possesses time-series and spatial features, a convolutional network, a long short-term memory neural network, and an attention mechanism are combined to construct the intelligent decision-making model for support adjustment. The intelligent decision-making model for support adjustment is then autonomously learned and continuously iterated based on the training hydraulic support moving step matrix. During subsequent coal mining operations, once the reference hydraulic support shifting step matrix is ​​obtained, the intelligent decision-making model for support adjustment can output the control behavior of the hydraulic support cluster, thereby achieving intelligent decision-making and accurately classifying the vector dataset to achieve intelligent decision-making effects for the hydraulic support cluster control behavior. This autonomous learning and intelligent decision-making method for hydraulic support cluster control behavior can help underground workers adjust the shifting step of the hydraulic supports in a timely manner, thereby adjusting the straightness of the hydraulic support cluster, which is beneficial to improving the efficiency of coal mining and the overall support effect. Figure 4 The flowchart illustrates the overall process of the autonomous learning and intelligent decision-making method for the control behavior of hydraulic support clusters provided in this embodiment of the invention.

[0108] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for autonomous learning and intelligent decision-making of hydraulic support cluster control behavior, characterized in that, include: S1, Mathematicalize the hydraulic support shifting action in the hydraulic support cluster to construct the initial hydraulic support shifting step distance matrix. The shifting action performed by the hydraulic support during each coal cutting process of the coal mining machine includes basic actions and adjustment actions. The initial hydraulic support shifting step distance matrix includes the shifting distance of each hydraulic support during each coal cutting process of the coal mining machine, including the basic actions and adjustment actions performed by each hydraulic support. Wherein, S1 includes: S11, determining that the hydraulic support cluster at the coal mining face includes... n A hydraulic support is used to propel the coal mining machine during coal mining. m knife; S12, based on the basic movements performed by each hydraulic support in the hydraulic support cluster. a and adjusting movements b Construct the initial hydraulic support shifting step matrix, which is represented as: ； in, It refers to the first i The hydraulic support is in the coal cutting machine. j Basic actions when cutting coal a The resulting shifting distance (1≤ i ≤ n ); It refers to the first i The hydraulic support is in the coal cutting machine. j Adjusting the action when cutting coal b The resulting shifting distance (1≤ j ≤ m ); S2, acquire the position data of the coal mining machine, the hydraulic support moving action control data and the hydraulic support stroke data, and determine the value of each element in the initial hydraulic support moving step matrix based on the acquired coal mining machine position data, hydraulic support moving action control data and hydraulic support stroke data, and obtain the reference hydraulic support moving step matrix; S3, obtain the target hydraulic support's shifting reference matrix from the reference hydraulic support shifting step matrix, wherein the target hydraulic support is the hydraulic support in the hydraulic support cluster whose shifting distance needs to be determined for adjustment. Among them, when the target hydraulic support is the first i When the hydraulic support is in position number 1, S3 obtains the first hydraulic support from the reference hydraulic support shifting step matrix. i When referring to the moving reference matrix of the hydraulic support, the following is included: Take from the reference hydraulic support shifting step matrix Within the range of number The moving distance value within the propulsion range forms The vector dataset, which is the first... i The reference matrix for the movement of hydraulic support No. 1 is represented as follows: ； S4, input the target hydraulic support's shifting reference matrix into the pre-trained shifting adjustment intelligent decision model, and the shifting adjustment intelligent decision model outputs the shifting distance of the target hydraulic support to perform the adjustment action; S5, when the target hydraulic support performs the moving action, control the target hydraulic support to perform the adjustment action according to the moving distance of the target hydraulic support to perform the adjustment action.

2. The autonomous learning and intelligent decision-making method for controlling the cluster behavior of hydraulic supports according to claim 1, characterized in that, S2 determines the initial hydraulic support shifting step matrix based on the acquired coal mining machine position data, hydraulic support shifting action data, and hydraulic support stroke data. and When the value is , it includes: S21, Determine the time sequence set of events when the coal mining machine passes the hydraulic support based on the coal mining machine's position data. , is represented as: ； in, The elements in the text represent the coal cutting machine at the coal mining face. m During the coal cutting process, the coal mining machine passes through the first... i No. 1 hydraulic support m At each moment of the next period; S22, Determine the sequence set of hydraulic support moving action events based on the hydraulic support moving action control data. , is represented as; ； in, t Indicates the moment when the hydraulic support relocation event occurs; The elements in the text represent the cutting of the coal mining face. m During the coal cutting process, the first i The hydraulic support shifting action occurred. o At each moment of the sequence; S23, when to total c +1 time is in and Between, that is, satisfying When, then determine the first i The hydraulic support No. 1 k Next to k+c The second frame-shifting action occurred during the... j knife; S24, Definition for t After a moment d The change in the hydraulic support stroke data over a period of time, then and The specific values ​​are as follows: like c =0, then , ; like c =1, then , ; like c ≥2, then , .

3. The autonomous learning and intelligent decision-making method for controlling the cluster behavior of hydraulic supports according to claim 1, characterized in that, Before step S4, the method further includes: training the intelligent decision-making model for moving and adjusting the frame, and the step of training the intelligent decision-making model for moving and adjusting the frame includes: S41, Creating training samples: Obtaining historical coal cutting data from the coal mining machine. m The system collects historical coal mining machine position data, historical hydraulic support movement control data, and historical hydraulic support stroke data. Based on these data, a training hydraulic support movement step matrix is ​​constructed. For the hydraulic support... i Command the coal mining machine to cut m During the coal cutting process, its adjustment actions b The distance of the moving frame For target estimation, the step distance matrix of the training hydraulic support is taken as follows: Within the range of frame numbers The moving distance value within the propulsion range forms The vector dataset, The vector dataset is used as input to the intelligent decision-making model for frame adjustment; S42, Label creation: Determine the minimum accuracy of the hydraulic support shifting distance as δ, and divide the output of the intelligent decision-making model for support shifting adjustment into... B If there are 1 output category, then the labels of the smart decision-making model for relocation and adjustment satisfy the following: ； In the formula, ⌊⌋ represents the floor function; B The transfer distances corresponding to the output categories are respectively , This refers to the maximum stroke value of the hydraulic support during each moving action; Define the training hydraulic support shifting step matrix If the label falls within any output category range, then the label is that output category; S43 uses training samples and labels to train a machine learning-based intelligent decision-making model for adjusting the transfer mechanism.

4. The autonomous learning and intelligent decision-making method for controlling the cluster behavior of hydraulic supports according to claim 1, characterized in that, Before determining the values ​​of each element in the initial hydraulic support shifting step matrix based on the acquired coal mining machine position data, hydraulic support shifting action control data, and hydraulic support stroke data, step S2 also includes: Fluctuation and missing value processing are performed on the hydraulic support stroke data.

Images