Method and system for generating cutting curve of coal mining machine, electronic equipment and storage medium

Abstract

The invention provides a coal mining machine cutting curve generation method and system, electronic equipment and a storage medium, and relates to the technical field of coal mining machines. The method comprises the following steps: acquiring operation data of at least two nearest coal mining machine operations, generating a corresponding behavior map node according to the operation data of each coal mining machine operation, and acquiring working condition migration data between adjacent coal mining machine operations in the at least two nearest coal mining machine operations, according to the working condition migration data, generating a corresponding behavior map edge, based on the behavior map node and the behavior map edge, obtaining a behavior map of at least two nearest coal mining machine operations, inputting the behavior map into a cutting curve generation model, and obtaining a cutting curve of the next coal mining machine operation output by the cutting curve generation model; the cutting curve generation model is obtained based on the sample behavior map of the operation data of the at least two sample coal cutter operations and the sample cutting curve of the next coal cutter operation, and the accuracy of the generated coal cutter cutting curve can be improved.

Description

Technical Field

[0001] This invention relates to the field of coal mining machine technology, and in particular to a method, system, electronic device, and storage medium for generating cutting curves for a coal mining machine. Background Technology

[0002] Intelligent construction of coal mines is a key direction for the development of my country's coal mining industry. In this process, the coal mining machine cutting curve, as a key control command guiding the automatic cutting of the coal mining machine, is of great significance for improving coal mining efficiency and the overall level of intelligence in coal mines.

[0003] However, the underground environment of coal mines is complex, and the coal-rock interface is difficult to identify, making it hard to directly perceive the actual boundary curve between the coal seam and the surrounding rock. Currently, the conventional approach mainly relies on the inertial navigation equipment of the coal mining machine to calculate the cutting curve, but this method has two limitations: first, it requires high sensor accuracy; second, it fails to fully consider the real-time changes in the attitude of the coal mining machine during the cutting process. These factors result in inaccurate cutting curve outputs, hindering the development of planned cutting technology towards a higher level of intelligence. Summary of the Invention

[0004] This invention provides a method, system, electronic device, and storage medium for generating coal mining machine cutting curves, in order to solve the reliance on inertial navigation equipment of coal mining machines in the prior art and improve the accuracy of the generated coal mining machine cutting curves.

[0005] This invention provides a method for generating a coal mining machine cutting curve, comprising: Obtain the operation data of the two most recent coal mining machine operations, and generate corresponding behavior graph nodes based on the operation data of each coal mining machine operation; Obtain the working condition migration data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations, and generate corresponding behavioral graph edges based on the working condition migration data; Based on the behavior graph nodes and the behavior graph edges, obtain the behavior graph of the most recent at least two coal mining machine operations; The behavior map is input into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model; wherein, the cutting curve generation model is trained based on the sample behavior map of the operation data of at least two sample coal mining machine operations and the sample cutting curve of the next coal mining machine operation.

[0006] According to a method for generating a coal mining machine cutting curve provided by the present invention, the step of inputting the behavior map into a cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model includes: Based on the spatial feature extraction layer of the truncation curve generation model, spatial semantic feature extraction processing is performed on the behavior graph to obtain the coal mining machine operation spatial embedding vector sequence of the behavior graph; Based on the time feature extraction layer of the cut curve generation model, the time semantic feature extraction process is performed on the coal mining machine operation embedding vector sequence of the behavior graph to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph. Based on the fully connected layer of the cut curve generation model, the spatiotemporal embedding vector sequence of the coal mining machine operation in the behavior graph is mapped to obtain the cut curve of the next coal mining machine operation.

[0007] According to the method for generating a coal mining machine cutting curve provided by the present invention, before performing spatial semantic feature extraction processing on the behavior map based on the spatial feature extraction layer of the cutting curve generation model to obtain the coal mining machine operation space embedding vector sequence of the behavior map, the method further includes: Obtain the time interval term of the behavior graph edge, and calculate the time decay factor based on the time interval term; The attention weights of the spatial feature extraction layer of the truncated curve generation model are set based on the time decay factor.

[0008] According to the present invention, a method for generating a cutting curve of a coal mining machine is provided, wherein the spatial feature extraction layer of the cutting curve generation model includes at least two graph embedding layers; Before performing spatial semantic feature extraction processing on the behavior graph based on the spatial feature extraction layer of the truncation curve generation model to obtain the coal mining machine operation space embedding vector sequence of the behavior graph, the method further includes: The static characteristics of each coal mining machine operation are determined based on the parameters of the coal mining machine and the coal mining environment. The bias vector of each graph embedding layer of the spatial feature extraction layer of the truncation curve generation model is set based on the static features; or, the conditional embedding vector of the output layer of the spatial feature extraction layer of the truncation curve generation model is set based on the static features.

[0009] According to a method for generating a coal mining machine cutting curve provided by the present invention, the step of extracting temporal semantic features from the coal mining machine operation embedding vector sequence of the behavior graph based on the temporal feature extraction layer of the cutting curve generation model to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph includes: The static characteristics of each coal mining machine operation are determined based on the parameters of the coal mining machine and the coal mining environment, so as to determine the initial hidden state of the time feature extraction layer of the cutting curve generation model based on the static characteristics. The coal mining machine operation embedding vector sequence of the behavior graph is arranged in chronological order to obtain the chronological embedding vector sequence of the behavior graph; Based on the initial hidden state of the temporal feature extraction layer of the truncation curve generation model, the hidden state of each behavior graph node in the temporal embedding vector sequence is determined to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph.

[0010] According to the method for generating a coal mining machine cutting curve provided by the present invention, the number of operation data of the at least two sample coal mining machines is greater than the number of operation data of the at least two sample coal mining machines.

[0011] According to the present invention, a method for generating a cutting curve for a coal mining machine is provided. The loss function of the cutting curve generation model is determined based on the deviation between the predicted cutting curve output by the original model of the cutting curve generation model and the sample cutting curve, the gradient penalty of the operation data of the sample coal mining machine, and the deviation between the predicted cutting curve and the coal-rock boundary. The coal-rock boundary is obtained after the operation of the at least two sample coal mining machines.

[0012] The present invention also provides a coal mining machine cutting curve generation system, comprising: The behavior graph node acquisition module is used to acquire the operation data of the most recent at least two coal mining machine operations, so as to generate corresponding behavior graph nodes based on the operation data of each coal mining machine operation; The behavior graph edge acquisition module is used to acquire the working condition migration data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations, so as to generate corresponding behavior graph edges based on the working condition migration data. The behavior graph acquisition module is used to acquire the behavior graph of at least the two most recent coal mining machine operations based on the behavior graph nodes and the behavior graph edges. The cutting curve generation module is used to input the behavior map into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model; wherein, the cutting curve generation model is trained based on the sample behavior map of the operation data of at least two sample coal mining machine operations and the sample cutting curve of the next coal mining machine operation.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the coal mining machine cutting curve generation method as described above.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the coal mining machine cutting curve generation method as described above.

[0015] The present invention provides a method, system, electronic device, and storage medium for generating coal mining machine cutting curves. It generates corresponding behavior graph nodes by acquiring operational data from at least the two most recent coal mining machine operations, and generates corresponding behavior graph edges by acquiring condition transition data between adjacent coal mining machine operations from the two most recent operations. Based on the behavior graph nodes and edges, it obtains the behavior graph of the two most recent coal mining machine operations, inputs the behavior graph into a cutting curve generation model, and obtains the cutting curve for the next coal mining machine operation output by the model. This allows the generation of the cutting curve for the next coal mining machine operation based on the historical operational data of the coal mining machine, providing direct input to an automatic cutting control system. This solves the reliance on the inertial navigation equipment of the coal mining machine in the prior art and improves the accuracy of the generated coal mining machine cutting curves. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the method for generating the cutting curve of a coal mining machine provided by the present invention.

[0018] Figure 2 This is a behavioral graph diagram of the coal mining machine cutting curve generation method provided by the present invention.

[0019] Figure 3 This is a schematic diagram of the coal mining machine cutting process parameters in the coal mining machine cutting curve generation method provided by the present invention.

[0020] Figure 4 This is a schematic diagram of the structure of the coal mining machine cutting curve generation system provided by the present invention.

[0021] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] The following is combined with Figures 1 to 5 The present invention describes a method, system, electronic device, and storage medium for generating coal cutting curves for a coal mining machine.

[0024] Figure 1 This is a flowchart illustrating the method for generating the cutting curve of a coal mining machine provided by the present invention, as shown below. Figure 1 As shown, the method includes the following steps.

[0025] Step 101: Obtain the operation data of the most recent at least two coal mining machine operations, and generate corresponding behavior graph nodes based on the operation data of each coal mining machine operation.

[0026] The operation data for each cutting coal mining machine can also be referred to as the operation data for each cutting cycle coal mining machine.

[0027] Coal mining machine operation, also known as coal mining machine behavior or coal mining machine operation event, refers to a series of orderly actions or instructions triggered by the control system or operators to cause the coal mining machine or its components to transition from one defined state to another in order to complete a specific coal mining task.

[0028] Operational data for coal mining machines refers to a multi-dimensional data set used to describe, characterize, or drive a complete operation process. It can be understood that operational data for coal mining machines is equivalent to a snapshot of the machine's behavior with spatiotemporal markers, capable of describing the machine's actions, preceding and subsequent states, and environmental context.

[0029] A coal mining machine can also be simply referred to as a coal mining machine. In some embodiments, the operation data of the coal mining machine may include multiple feature dimensions such as operation category data, initial values ​​of operation parameters, target values ​​of operation parameters, the position of the coal mining machine corresponding to the operation action, and time data of the operation action, as well as the heights of the front drum before and after the operation and the heights of the rear drum before and after the operation. The operation category data may be an operation category code, the initial values ​​of operation parameters may be initial values ​​of the action parameters, the target values ​​of operation parameters may be target values ​​of the action parameters, the position of the coal mining machine corresponding to the operation action may be the position of the coal mining machine at the time of operation, and the time data of the operation action may include the absolute timestamp of the operation action and the duration of the operation.

[0030] It should be noted that obtaining operational data from at least the two most recent coal mining machine operations can be achieved through various feasible technical means. For example, it can be extracted from the historical command logs of the working face electro-hydraulic control system, or read from the operating data storage of the coal mining machine's main controller. Those skilled in the art can choose the appropriate acquisition method based on the actual system's data architecture and interface type; this embodiment does not impose specific limitations in this regard.

[0031] For example, the data sources for some operational data of the coal mining machine are shown in the table below: Data types Source Encoder position Coal mining machine body displacement encoder Rocker arm angle / speed Front and rear rocker arm angle sensors Drum speed Roller drive unit sensor Traction speed and direction Electro-hydraulic control system operation record / PLC status status Operation command timestamp Console operation log / event trigger time It is understandable that by generating corresponding behavioral graph nodes based on the operational data of each coal mining machine operation, an independent coal mining machine operation event can be represented through the behavioral graph nodes.

[0032] For example, behavioral graph nodes can be simply referred to as nodes, and the node feature dimensions are shown in the table below: Node feature dimensions illustrate Operation category code (01-front rocker arm movement, 02-rear rocker arm movement, 03-drum speed, 04-traction speed, 05-traction direction) Initial values ​​of motion parameters (initial angle of rocker arm, initial speed, rotational speed) Target values ​​for motion parameters (target angle, target velocity) Coal mining machine position (units) during operation. Absolute timestamp of the operation Operation duration (unit) Height before and after front roller operation The height of the rear roller before and after operation Step 102: Obtain the working condition migration data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations, so as to generate corresponding behavior graph edges based on the working condition migration data.

[0033] Among them, the working condition transition data, also known as the dynamic status of the coal mining machine, is used to quantitatively describe and associate the transition process and contextual relationship between two adjacent coal mining machine operations.

[0034] In some embodiments, the condition migration data between adjacent coal mining machine operations includes the time interval between adjacent coal mining machine operations, spatial position displacement, environmental feedback data, and sequence markers.

[0035] Environmental feedback data refers to the environmental feedback caused by the operation results of the coal mining machine. Sequence flags are used to determine whether adjacent coal mining machine operations belong to the same logical operation sequence.

[0036] It is understandable that the working condition migration data between adjacent coal mining machine operations generates corresponding behavioral graph edges, which can represent the transfer relationship between two adjacent coal mining machine operations, that is, the behavioral transfer and environmental feedback between the two.

[0037] For example, the edges of the behavior graph can be simply referred to as edges, and the edge feature dimensions are shown in the table below: Edge feature dimension illustrate Time interval between two coal mining machine operations Spatial position displacement Environmental feedback resulting from the operation: the range of drum position adjustment, changes in equipment stability (whether there is side spalling, large chunks of coal falling). Do they belong to the same logical operation sequence (a group of consecutive rocker arm movements)? Step 103: Obtain the behavior graph of the most recent at least two coal mining machine operations based on the behavior graph nodes and the behavior graph edges.

[0038] It should be noted that behavior graph nodes can be linked with behavior graph edges to construct structures such as... Figure 2 The graph shown is a behavioral graph where nodes represent coal mining machine operations and edges represent migration relationships between adjacent coal mining machine operations. G(t) .

[0039] Understandably, compared to the sequential operation data of coal mining machine operations, the behavioral map in this embodiment can not only reflect the order of coal mining machine operations, but also the evolution path such as changes in geological conditions and equipment status from the previous coal mining machine operation to the next coal mining machine operation.

[0040] Step 104: Input the behavior map into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model.

[0041] The cutting curve generation model is trained based on the sample behavior map of the operation data of at least two sample coal mining machine operations and the sample cutting curve of the next coal mining machine operation.

[0042] The cutting curve of a coal mining machine is used to characterize the spatiotemporal state and motion planning of the machine over time in the next cut or the next cutting cycle. One cut refers to a complete coal mining cycle.

[0043] It should be noted that the cutting curve of the coal mining machine can be used to determine a set of control commands generated by controlling the coal mining machine to execute a cut, which can be used as the direct input of the automatic cutting control system.

[0044] The at least two sample coal mining machine operations and the next coal mining machine operation used for model training can be a certain number of consecutive historical coal mining machine operations. The specific value of the number can be set according to the actual working conditions, and this embodiment does not limit it.

[0045] For example, at least two sample coal mining machine operations used for model training can be the first 30 of 31 consecutive historical coal mining machine operations, and the next coal mining machine operation can be the 31st historical coal mining machine operation.

[0046] For example, the operational data of a coal mining machine can be the cutting curve of the coal mining machine operation. That is, the cutting curve generation model can predict the cutting curve of the next coal mining machine operation based on the behavioral map corresponding to the cutting curves of historical coal mining machine operations.

[0047] The sample cut-off curve for the next coal mining machine operation can be the cut-off curve corresponding to the actual operation data of the last coal mining machine operation in the continuous historical coal mining machine operations used for model training.

[0048] A feasible training scheme for the truncation curve generation model may include: Input the sample behavior maps of at least two sample coal mining machine operations into the original model to obtain the predicted cut-off curve for the next coal mining machine operation output by the original model. Then, calculate the loss function value based on the predicted cut-off curve and the sample cut-off curve of the next coal mining machine operation. Finally, update the model parameters of the original model based on the loss function value. Iterate through the above input and calculation processes until the loss function converges or the preset number of iterations is reached to obtain the cut-off curve generation model. The preset number of iterations can be set as needed and is not specifically limited here.

[0049] Understandably, compared to training a model using sequential operational data from coal mining machines, the behavioral graph of coal mining machine operations in this embodiment can explicitly and structurally display the transition patterns and state change patterns between coal mining machine operations to the original model. Therefore, training the model using the behavioral graph eliminates the need for the original model to implicitly learn these patterns from the data, reducing the risk of overfitting or getting trapped in local optima. Furthermore, the behavioral graph serves as an interpretable framework for understanding the model's decision-making process, making the operational logic more transparent.

[0050] Furthermore, the behavioral graph clearly reveals the transition patterns and state change patterns between coal mining machine operations in a structured form, which can reduce trial-and-error training during model training and significantly reduce the number of iterations and computational resource consumption required for model training.

[0051] In some embodiments, the model can be updated online based on the operation data of the coal mining machine over a recent period to achieve a closed loop of data-driven, behavior-aware, and cut curve output, thereby improving the adaptability, interpretability, and deployment practicality of the cut curve generation model.

[0052] The coal mining machine cutting curve generation method provided in this invention generates corresponding behavior graph nodes by acquiring operation data from at least the two most recent coal mining machine operations, and generates corresponding behavior graph edges by acquiring condition migration data between adjacent coal mining machine operations in the two most recent coal mining machine operations. Based on the behavior graph nodes and behavior graph edges, the behavior graph of the two most recent coal mining machine operations is obtained. The behavior graph is then input into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model. This method can obtain the cutting curve of the next coal mining machine operation based on the operation data of the historical operation behavior of the coal mining machine, and can be used as a direct input for the automatic cutting control system. This solves the dependence on the inertial navigation equipment of the coal mining machine in the prior art and improves the accuracy of the generated coal mining machine cutting curve.

[0053] Meanwhile, the behavior graph of the coal mining machine cutting curve generation method provided in this embodiment of the invention can show the sequence of coal mining machine operations and the working condition transition between adjacent coal mining machine operations. Based on this, according to the behavior graph, the cutting curve generation model can actively adapt to the changes in specific working conditions such as coal seam hardness and interbedded rock distribution while following the process sequence, and generate a cutting curve for the next coal mining machine operation that is more in line with the actual mining conditions. This further enhances the planning effect of the cutting curve driven by the coal mining machine operation behavior and improves mining efficiency and coal quality.

[0054] Based on the above embodiments, the step of inputting the behavior map into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model includes: Based on the spatial feature extraction layer of the truncation curve generation model, spatial semantic feature extraction processing is performed on the behavior graph to obtain the coal mining machine operation spatial embedding vector sequence of the behavior graph; Based on the time feature extraction layer of the cut curve generation model, the time semantic feature extraction process is performed on the coal mining machine operation embedding vector sequence of the behavior graph to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph. Based on the fully connected layer of the cut curve generation model, the spatiotemporal embedding vector sequence of the coal mining machine operation in the behavior graph is mapped to obtain the cut curve of the next coal mining machine operation.

[0055] It should be noted that the spatial feature extraction layer can also be called a graph convolutional layer, graph neural network layer, etc. It can perform graph convolution calculation on the behavior graph through the spatial feature extraction layer to fuse the node features and edge features of the behavior graph to aggregate neighbor node information, realize spatial semantic feature extraction, and finally obtain the coal mining machine operation space embedding vector of each behavior graph node through a learnable adjacency aggregation method. Based on the coal mining machine operation space embedding vector of each behavior graph node, the coal mining machine operation space embedding vector sequence of the behavior graph is obtained.

[0056] Among them, the coal mining machine operation space embedding vector of each behavior graph node can also be called the operation state embedding vector of each behavior graph node, which contains the spatial semantic features of the behavior graph node obtained by graph structure perception.

[0057] In some embodiments, a spatial feature extraction layer of the truncated curve generation model can be obtained by constructing a graph neural network structure based on GraphSAGE.

[0058] It should be noted that the time feature extraction layer can also be called the temporal feature extraction layer, dependency feature extraction layer, etc. It can model the long-term and short-term dependencies in the evolution process of coal mining machine operation by embedding vectors based on the operation state of the behavioral graph nodes, so as to obtain temporal semantic features. Combined with the spatial embedding vector sequence of coal mining machine operation and the corresponding temporal semantic features, the spatiotemporal embedding vector sequence of coal mining machine operation in the behavioral graph is obtained.

[0059] Among them, the coal mining machine operation space embedding vector of the behavior graph node refers to the vectorized representation of the semantics and state of the coal mining machine operation in the graph structure space, which is calculated by the spatial feature extraction layer on the behavior graph.

[0060] The spatiotemporal embedding vector of the coal mining machine operation of the behavior graph node refers to the joint vector representation of the unified coded operation behavior spatial structure and temporal evolution law obtained after time-dependent modeling through the time feature extraction layer on the basis of the coal mining machine operation space embedding vector sequence.

[0061] It should be noted that the fully connected layer can also be called the fully connected mapping layer, etc. The spatiotemporal embedding vector sequence of the coal mining machine operation in the behavior graph can also be called the hidden state sequence of the coal mining machine operation in the behavior graph. The spatiotemporal embedding vector sequence of the coal mining machine operation in the behavior graph can be mapped to the target parameter prediction result through the fully connected layer.

[0062] The target parameter prediction results may include the next coal cutter cutting trajectory sequence, the traction speed at the current stage, and the coal cutter traction direction. The next coal cutter cutting trajectory sequence may be the drum height trajectory.

[0063] In some embodiments, the fully connected layer can be a multidimensional parallel output structure, thereby supporting joint modeling in multi-task prediction scenarios and improving the overall fitting accuracy of the cutting curve for the next coal mining machine operation.

[0064] In some embodiments, after obtaining the target parameter prediction results of the next coal mining machine operation through the fully connected layer, the cutting process parameters of the next coal mining machine can be generated based on the target parameter prediction results, and the cutting process parameters of the next coal mining machine can be output in the form of a time series to form the cutting curve of the next coal mining machine operation.

[0065] Understandably, by constructing a behavioral map of coal mining machine operation behavior and structurally displaying the correlation between coal mining machine operation behavior, dynamic state, and working environment information, joint modeling through spatial feature extraction layers and temporal feature extraction layers yields a joint vector representation of the spatial structure and temporal evolution of unified coding of coal mining machine operation behavior. This approach can provide structural dependencies between adjacent coal mining machine operations and suppress predictive abrupt changes caused by data noise or discontinuities, thus maintaining high fitting accuracy to the real coal and rock boundary even under conditions of missing data or perceptual ambiguity. Furthermore, it avoids reliance on traditional laser / ultrasonic sensors.

[0066] Based on any of the above embodiments, before performing spatial semantic feature extraction processing on the behavior graph based on the spatial feature extraction layer of the truncation curve generation model to obtain the coal mining machine operation space embedding vector sequence of the behavior graph, the method further includes: Obtain the time interval term of the behavior graph edge, and calculate the time decay factor based on the time interval term; The attention weights of the spatial feature extraction layer of the truncated curve generation model are set based on the time decay factor.

[0067] For example, the time interval between two coal mining machine operations can be determined based on the timestamps of the preceding and following coal mining machine operations in the coal mining machine operations corresponding to two coal mining machine operation behaviors connected by the behavior graph edge in the behavior graph, thus obtaining the time interval term of the behavior graph edge between the corresponding two behavior graph nodes. For example, the time interval term can be 10.5s.

[0068] It should be noted that the time interval term can be mapped to a weight between 0 and 1, and the time decay factor can be obtained based on this weight. This ensures that the smaller the time interval term of the behavior graph edge, the greater the importance of the information of the behavior graph node that is earlier in time to the behavior graph node that is later in time.

[0069] Understandably, setting the attention weights of the spatial feature extraction layer of the cut curve generation model based on the time decay factor can introduce a time-aware mechanism. When the spatial feature extraction layer aggregates neighbor node information, it can use the time decay factor to weight the neighbor node information, making the cut curve generation model more expressive of short-term highly correlated coal mining machine operations, thus providing a foundation for improving the accuracy and intelligence level of automatic cut control in the future.

[0070] Based on any of the above embodiments, the spatial feature extraction layer of the truncation curve generation model includes at least two graph embedding layers.

[0071] The graph embedding layer is a hierarchical structure based on the spatial feature extraction layer framework.

[0072] It should be noted that the graph embedding layer can also be called a convolutional layer or a stacked layer. The spatial feature extraction layer of the truncated curve generation model includes multiple graph embedding layers. Each graph embedding layer introduces a time decay factor as an attention weight during the aggregation process, and the output of the previous graph embedding layer is the input of the next graph embedding layer. In this way, through multi-layer stacking, nodes can fuse information from multi-hop neighbor nodes.

[0073] In some embodiments, before performing spatial semantic feature extraction processing on the behavior graph based on the spatial feature extraction layer of the truncation curve generation model to obtain the coal mining machine operation space embedding vector sequence of the behavior graph, the method further includes: The static characteristics of each coal mining machine operation are determined based on the parameters of the coal mining machine and the coal mining environment. Based on the static features, set the bias vector of each graph embedding layer of the spatial feature extraction layer of the truncation curve generation model.

[0074] Static features, also known as static data, refer to the set of physical and structural parameters whose values ​​remain constant or are considered constant within the target time window. The specific value of the target time window can be set according to the actual working conditions; this embodiment does not impose any limitations on it.

[0075] Static features are used to characterize the inherent constraints of the working environment and the equipment itself, and do not change dynamically with the time progression of the coal mining machine's operating behavior sequence.

[0076] It should be noted that before the cutting curve of the next coal mining machine operation is output through the cutting curve generation model, the static features have been determined and used as known inputs.

[0077] There are many ways to determine the static characteristics of each coal mining machine operation based on the parameters of the coal mining machine and the mining environment. For example, the parameters of the coal mining machine and the mining environment can be obtained through the geological exploration report of the working face or the mining design documents to determine the static characteristics of the coal mining machine operation. This embodiment does not limit this method.

[0078] In some embodiments, static features may include multiple feature dimensions such as mining height, coal thickness, machine face height, and maximum cutting height.

[0079] For example, the feature dimensions of static features are shown in the table below: Feature name type illustrate High H Continuous Current working face mining height Coal Thickness Tc Continuous Coal seam thickness Aircraft surface height hm Continuous Distance between coal mining machine chassis and machine track ground Maximum cutting height Continuous The maximum lifting angle of the coal mining machine rocker arm corresponds to the height. It should be noted that, in the process of recalculating the embedding vector of each behavioral graph node in the behavioral graph of the spatial feature extraction layer of the truncated curve generation model, after obtaining an intermediate result based on the neighbor aggregation result and the embedding vector obtained from the previous layer, the updated embedding vector of each behavioral graph node in the behavioral graph of the current graph embedding layer is obtained based on the bias vector of the graph embedding layer of the spatial feature extraction layer of the truncated curve generation model and the intermediate result.

[0080] It is understandable that by setting the bias vector of each graph embedding layer of the spatial feature extraction layer of the cut curve generation model based on static features, each graph embedding layer of the spatial feature extraction layer of the cut curve generation model can recalculate the embedding vector of each behavioral graph node in the behavioral graph under the direct constraint of static features. This realizes the global constraint of static features on the spatial semantic feature extraction processing of the spatial feature extraction layer, enabling the trained model to specialize to the working condition of the fully mechanized mining face corresponding to the current static features, and improving the accuracy of the cut curve of the next coal mining machine operation output by the trained model.

[0081] In other embodiments, before performing spatial semantic feature extraction processing on the behavior graph based on the spatial feature extraction layer of the truncation curve generation model to obtain the coal mining machine operation space embedding vector sequence of the behavior graph, the method further includes: The static characteristics of each coal mining machine operation are determined based on the parameters of the coal mining machine and the coal mining environment. Based on the static features, set the conditional embedding vector of the output layer of the spatial feature extraction layer of the truncation curve generation model.

[0082] It should be noted that after the dynamic spatial embedding representation of each behavior graph node in the behavior graph is obtained by iteratively calculating the embedding vector of each behavior graph node in the behavior graph through multiple graph embedding layers of the spatial feature extraction layer of the truncated curve generation model, the coal mining machine operation spatial embedding vector of each behavior graph node in the current graph embedding layer is obtained based on the conditional embedding vector of the output layer of the spatial feature extraction layer of the truncated curve generation model and the dynamic spatial embedding representation.

[0083] Understandably, by setting the conditional embedding vector of each graph embedding layer in the spatial feature extraction layer of the cut curve generation model based on static features, the final output stage after obtaining the dynamic spatial embedding representation of each behavioral graph node in the behavioral graph through multiple graph embedding layers of the spatial feature extraction layer of the cut curve generation model can be fused with static features to obtain the coal mining machine operation spatial embedding vector of each behavioral graph node in the updated behavioral graph of the current graph embedding layer. This allows the coal mining machine operation spatial embedding vector to simultaneously contain the dynamic evolution information of the behavioral graph and the invariant static feature constraints, enabling the trained model to specialize to the fully mechanized mining face condition corresponding to the current static features, thereby improving the accuracy of the cut curve of the next coal mining machine operation output by the trained model.

[0084] Based on any of the above embodiments, the temporal feature extraction layer based on the truncation curve generation model performs temporal semantic feature extraction processing on the coal mining machine operation embedding vector sequence of the behavior graph to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph, including: The static characteristics of each coal mining machine operation are determined based on the parameters of the coal mining machine and the coal mining environment, so as to determine the initial hidden state of the time feature extraction layer of the cutting curve generation model based on the static characteristics. The coal mining machine operation embedding vector sequence of the behavior graph is arranged in chronological order to obtain the chronological embedding vector sequence of the behavior graph; Based on the initial hidden state of the temporal feature extraction layer of the truncation curve generation model, the hidden state of each behavior graph node in the temporal embedding vector sequence is determined to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph.

[0085] The working principle and technical effect of determining the static characteristics of each coal mining machine operation based on the parameters of the coal mining machine and the coal mining environment in this embodiment are basically the same as the aforementioned working principle and technical effect of determining the static characteristics of each coal mining machine operation based on the parameters of the coal mining machine and the coal mining environment, and will not be repeated here.

[0086] It should be noted that the temporal feature extraction layer of the cut curve generation model can be constructed based on the Long Short-Term Memory (LSTM) network.

[0087] The main difference between the coal mining machine operation embedding vector sequence and the time-series embedding vector sequence is that the order of the coal mining machine operation embedding vectors may be different.

[0088] For example, the corresponding vector can be obtained based on static features. Then, the static feature vector is mapped to the initial hidden state of the LSTM through a transformation function. As shown in the following formula: The temporal feature extraction layer of the truncated curve generation model can be input into the temporal input sequence. Under the guidance of the initial hidden state, the hidden state of each time step is obtained, forming a hidden state sequence.

[0089] Understandably, the initial hidden state of the time feature extraction layer of the model for generating the cutting curve based on the static features of the coal mining machine operation can achieve pre-emptive and in-depth guidance of the dynamic time evolution process. From the starting point of the dynamic time evolution process, it is under the constraint of the actual working conditions of the current fully mechanized mining face, so that the trained model can be specialized to the working conditions of the fully mechanized mining face corresponding to the current static features, thereby improving the accuracy of the cutting curve of the next coal mining machine operation output by the trained model.

[0090] Furthermore, both the spatial feature extraction layer and the temporal feature extraction layer of the truncation curve generation model apply static features. This dual-path fusion mechanism based on static features can ensure the consistency of static features in the spatial semantic feature extraction process and the temporal semantic feature extraction process.

[0091] Based on any of the above embodiments, the number of operation data of the at least two sample coal mining machines is greater than the number of operation data of the at least two coal mining machines.

[0092] In some embodiments, the number of operation data for at least two sample coal mining machine operations during the model training phase can be 30, and the number of operation data for at least two coal mining machine operations during the model application phase can be 10.

[0093] Among them, the number of operation data of at least two sample coal mining machine operations in the model training phase can be determined by analyzing the cutting process and judging the number of historical coal mining machine operation operation data that have a significant time correlation with the cutting trajectory of the next coal mining machine operation.

[0094] Specifically, based on variables such as drum height changes, traction speed, and coal-rock contact state, it can be determined whether there is a clear temporal evolution pattern between the historical coal mining machine operation data and the cutting trajectory of the next coal mining machine operation.

[0095] The amount of operational data for at least two coal mining machine operations during the model application phase can be determined based on actual working conditions.

[0096] Understandably, training the model with operational data from at least two sample coal mining machine operations allows the original model to fully learn and capture the temporal evolution of coal mining machine operations. The trained model can then predict the cutting curve of the next coal mining machine operation using the operational data from at least two coal mining machine operations. This effectively predicts the cutting curve of the next coal mining machine operation while reducing the dependence on the amount of real-time coal mining machine operation data, thus ensuring the timeliness of the output under actual working conditions.

[0097] Based on any of the above embodiments, the loss function of the cut-off curve generation model is determined based on the deviation between the predicted cut-off curve output by the original model of the cut-off curve generation model and the sample cut-off curve, the gradient penalty of the operation data of the sample coal mining machine, and the deviation between the predicted cut-off curve and the coal-rock boundary. The coal-rock boundary is obtained after the operation of the at least two sample coal mining machines.

[0098] For example, the deviation between the predicted cutoff curve output by the original model of the cutoff curve generation model and the sample cutoff curve can be measured by the following formula. : in, It is the total time step of at least two sample coal mining machine operations. It is the predicted cutoff curve output by the original model at time step t. It is the sample cutoff curve at time step t.

[0099] The gradient penalty for the operation data of the sample coal mining machine can be determined using the following formula. : in, It is the predicted cutoff curve output by the original model at time step t.

[0100] Understandably, the gradient penalty of the operating data of the sample coal mining machine can suppress drastic fluctuations in control parameters over a continuous period of time, thereby enhancing the physical feasibility of the operating trajectory.

[0101] The deviation between the predicted cutoff curve and the coal-rock boundary can be determined using the following formula. : in, It is the cut trajectory height of the predicted cut curve output by the original model at time step t. It is the coal-rock boundary height at time step t.

[0102] In some embodiments, the deviation between the predicted cut-off curve output by the original model of the cut-off curve generation model and the sample cut-off curve, the gradient penalty of the operation data of the sample coal mining machine, and the deviation between the predicted cut-off curve and the coal-rock boundary can be weighted to determine the loss function of the cut-off curve generation model.

[0103] For example, the loss function of the cut curve generation model can be expressed by the following formula. : in, It is a weighted coefficient representing the deviation between the predicted cutoff curve output by the original model of the cutoff curve generation model and the sample cutoff curve. These are the weighting coefficients of the gradient penalty for the operational data of the sample coal mining machine. It is the weighting coefficient for predicting the deviation between the cut-off curve and the coal-rock boundary.

[0104] It is understandable that the aforementioned weight coefficients are adjustable and can be adjusted according to the needs of the training scenario.

[0105] To illustrate the functionality of the coal mining machine cutting curve generation method provided in this implementation, a specific example is given below.

[0106] First, the original model of the cut-off curve generation model can be determined, and the loss function used for model training can be determined based on the deviation between the predicted cut-off curve output by the original model and the sample cut-off curve, the gradient penalty of the operation data of the sample coal mining machine, and the deviation between the predicted cut-off curve and the coal-rock boundary. Then, the sample behavior map of the operation data of 30 sample coal mining machines is input into the original model to obtain the predicted cut-off curve of the next coal mining machine operation output by the original model. Next, based on the predicted cut-off curve and the sample cut-off curve of the next coal mining machine operation, the loss function value is calculated. Finally, the model parameters of the original model are updated based on the loss function value. This input and calculation process is iteratively executed until the loss function converges or the preset number of iterations is reached, resulting in the cut-off curve generation model. The preset number of iterations can be set as needed and is not specifically limited here. The parameter settings for the original model are shown in the table below: Model parameters Setting value enter Static data: mining height, coal thickness, maximum cutting height, machine face height; Dynamic data: behavioral patterns of coal mining machine operation. Output Spatiotemporal state sequence of the next coal mining machine operation Initial learning rate 0.01 (cosine annealing) Optimizer Adam Epoch 1000 Batch_Size 32 Number of training rounds 100 epochs When applying the cut curve generation model, you can input the operation data of 10 historical coal mining machine operations to obtain the spatiotemporal state sequence of the next coal mining machine operation output by the cut curve generation model: Based on the spatiotemporal state sequence of the next coal mining machine operation, a sequence such as... can be generated. Figure 3 The cutting process parameters of the next coal cutter shown are output in time series form to form a complete cutting curve of the next coal cutting machine operation, which serves as the input of the automatic cutting control system.

[0107] The cutting process parameters for the next coal cutter may specifically include the following: Coal machine position: Predict the traction path of the next cut and output the position of the coal machine on the working face at each moment according to spatial coordinates; Front drum cutting height: indicates the cutting point height of the front drum of the coal mining machine, fitted to the coal-rock boundary curve; Rear drum cutting height: indicates the cutting point height of the rear drum of the coal mining machine, and supports independent prediction; Drum speed: Outputs the drum speed curve required for the corresponding cutting intensity; Traction speed: Output travel speed matched to geological conditions, balancing cutting efficiency and equipment load; Traction direction: Determine the current direction of movement of the coal mining machine (left traction / right traction) to ensure the continuity and safety of operations.

[0108] The coal mining machine cutting curve generation method provided in this invention can avoid dependence on traditional laser / ultrasound sensors. By constructing a behavior map of coal mining machine operation behavior, it can structurally display the correlation between coal mining machine operation behavior, dynamic state of coal mining machine and working environment information. Through joint modeling of spatial feature extraction layer and temporal feature extraction layer, it can obtain a joint vector representation of the spatial structure and temporal evolution law of unified coding of coal mining machine operation behavior. It can provide structural dependence between adjacent coal mining machine operations and suppress prediction mutations caused by data noise or discontinuity, thereby maintaining high fitting accuracy to the real coal and rock boundary under conditions of missing data or perceptual ambiguity.

[0109] The coal mining machine cutting curve generation system provided by the present invention is described below. The coal mining machine cutting curve generation system described below can be referred to in correspondence with the coal mining machine cutting curve generation method described above.

[0110] Figure 4 This is a schematic diagram of the structure of the coal mining machine cutting curve generation system provided by the present invention, as shown below. Figure 4 As shown, the device includes: The behavior graph node acquisition module 410 is used to acquire the operation data of the most recent at least two coal mining machine operations, so as to generate corresponding behavior graph nodes based on the operation data of each coal mining machine operation. The behavior graph edge acquisition module 420 is used to acquire the working condition migration data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations, so as to generate corresponding behavior graph edges based on the working condition migration data. The behavior graph acquisition module 430 is used to acquire the behavior graph of at least the two most recent coal mining machine operations based on the behavior graph nodes and the behavior graph edges. The cutting curve generation module 440 is used to input the behavior map into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model; wherein, the cutting curve generation model is trained based on the sample behavior map of the operation data of at least two sample coal mining machine operations and the sample cutting curve of the next coal mining machine operation.

[0111] Based on any of the above embodiments, the truncation curve generation module 440 is specifically used to perform spatial semantic feature extraction processing on the behavior map based on the spatial feature extraction layer of the truncation curve generation model, so as to obtain the coal mining machine operation spatial embedding vector sequence of the behavior map. Based on the time feature extraction layer of the cut curve generation model, the time semantic feature extraction process is performed on the coal mining machine operation embedding vector sequence of the behavior graph to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph. Based on the fully connected layer of the cut curve generation model, the spatiotemporal embedding vector sequence of the coal mining machine operation in the behavior graph is mapped to obtain the cut curve of the next coal mining machine operation.

[0112] Based on any of the above embodiments, the system further includes an attention weight setting module, used to obtain the time interval term of the behavior graph edge, so as to calculate the time decay factor based on the time interval term; The attention weights of the spatial feature extraction layer of the truncated curve generation model are set based on the time decay factor.

[0113] Based on any of the above embodiments, the spatial feature extraction layer of the truncation curve generation model includes at least two graph embedding layers; The system further includes a bias vector setting module, used to determine the static features of each coal mining machine operation based on the coal mining machine parameters and the coal mining environment; and to set the bias vector of each graph embedding layer of the spatial feature extraction layer of the truncation curve generation model based on the static features; and / or, The conditional embedding vector setting module is used to determine the static characteristics of each coal mining machine operation based on the parameters of the coal mining machine and the coal mining environment. Based on the static features, set the conditional embedding vector of the output layer of the spatial feature extraction layer of the truncation curve generation model.

[0114] Based on any of the above embodiments, the cut curve generation module 440 is specifically used for: The static characteristics of each coal mining machine operation are determined based on the parameters of the coal mining machine and the coal mining environment, so as to determine the initial hidden state of the time feature extraction layer of the cutting curve generation model based on the static characteristics. The coal mining machine operation embedding vector sequence of the behavior graph is arranged in chronological order to obtain the chronological embedding vector sequence of the behavior graph; Based on the initial hidden state of the temporal feature extraction layer of the truncation curve generation model, the hidden state of each behavior graph node in the temporal embedding vector sequence is determined to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph.

[0115] Based on any of the above embodiments, the number of operation data of the at least two sample coal mining machines is greater than the number of operation data of the at least two coal mining machines.

[0116] Based on any of the above embodiments, the loss function of the cut-off curve generation model is determined based on the deviation between the predicted cut-off curve output by the original model of the cut-off curve generation model and the sample cut-off curve, the gradient penalty of the operation data of the sample coal mining machine, and the deviation between the predicted cut-off curve and the coal-rock boundary. The coal-rock boundary is obtained after the operation of the at least two sample coal mining machines.

[0117] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5 As shown, the electronic device may include: a processor 510, a communications interface 520, a memory 530, and a communications bus 540, wherein the processor 510, the communications interface 520, and the memory 530 communicate with each other through the communications bus 540. The processor 510 can call logic instructions in the memory 530 to execute a method for generating a coal mining machine cutting curve. This method includes: acquiring operation data from the most recent at least two coal mining machine operations; generating corresponding behavior graph nodes based on the operation data of each coal mining machine operation; acquiring working condition transition data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations; generating corresponding behavior graph edges based on the working condition transition data; acquiring the behavior graph of the most recent at least two coal mining machine operations based on the behavior graph nodes and the behavior graph edges; inputting the behavior graph into a cutting curve generation model; and obtaining the cutting curve of the next coal mining machine operation output by the cutting curve generation model. The cutting curve generation model is trained based on sample behavior graphs of operation data from at least two sample coal mining machine operations and sample cutting curves of the next coal mining machine operation.

[0118] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0119] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the coal mining machine cutting curve generation method provided by the above methods. The method includes: acquiring operation data of the most recent at least two coal mining machine operations to generate corresponding behavior graph nodes based on the operation data of each coal mining machine operation; acquiring working condition migration data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations to generate corresponding behavior graph edges based on the working condition migration data; acquiring the behavior graph of the most recent at least two coal mining machine operations based on the behavior graph nodes and the behavior graph edges; inputting the behavior graph into a cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model; wherein, the cutting curve generation model is trained based on the sample behavior graph of the operation data of at least two sample coal mining machine operations and the sample cutting curve of the next coal mining machine operation.

[0120] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for generating a coal mining machine cutting curve provided by the above methods. This method includes: acquiring operational data from the most recent at least two coal mining machine operations to generate corresponding behavioral graph nodes based on the operational data of each coal mining machine operation; acquiring condition migration data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations to generate corresponding behavioral graph edges based on the condition migration data; acquiring a behavioral graph of the most recent at least two coal mining machine operations based on the behavioral graph nodes and the behavioral graph edges; inputting the behavioral graph into a cutting curve generation model to obtain a cutting curve for the next coal mining machine operation output by the cutting curve generation model; wherein the cutting curve generation model is trained based on sample behavioral graphs of operational data from at least two sample coal mining machine operations and sample cutting curves of the next coal mining machine operation.

[0121] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0122] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating a coal mining machine cutting curve, characterized in that, include: Obtain the operation data of the two most recent coal mining machine operations, and generate corresponding behavior graph nodes based on the operation data of each coal mining machine operation; Obtain the working condition migration data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations, and generate corresponding behavioral graph edges based on the working condition migration data; Based on the behavior graph nodes and the behavior graph edges, obtain the behavior graph of the most recent at least two coal mining machine operations; The behavior map is input into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model; wherein, the cutting curve generation model is trained based on the sample behavior map of the operation data of at least two sample coal mining machine operations and the sample cutting curve of the next coal mining machine operation.

2. The method for generating the cutting curve of a coal mining machine according to claim 1, characterized in that, The step of inputting the behavior map into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model includes: Based on the spatial feature extraction layer of the truncation curve generation model, spatial semantic feature extraction processing is performed on the behavior graph to obtain the coal mining machine operation spatial embedding vector sequence of the behavior graph; Based on the time feature extraction layer of the cut curve generation model, the time semantic feature extraction process is performed on the coal mining machine operation embedding vector sequence of the behavior graph to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph. Based on the fully connected layer of the cut curve generation model, the spatiotemporal embedding vector sequence of the coal mining machine operation in the behavior graph is mapped to obtain the cut curve of the next coal mining machine operation.

3. The method for generating the cutting curve of a coal mining machine according to claim 2, characterized in that, Before performing spatial semantic feature extraction processing on the behavior graph based on the spatial feature extraction layer of the truncation curve generation model to obtain the coal mining machine operation space embedding vector sequence of the behavior graph, the method further includes: Obtain the time interval term of the behavior graph edge, and calculate the time decay factor based on the time interval term; The attention weights of the spatial feature extraction layer of the truncated curve generation model are set based on the time decay factor.

4. The method for generating the cutting curve of a coal mining machine according to claim 2, characterized in that, The spatial feature extraction layer of the truncation curve generation model includes at least two graph embedding layers; Before performing spatial semantic feature extraction processing on the behavior graph based on the spatial feature extraction layer of the truncation curve generation model to obtain the coal mining machine operation space embedding vector sequence of the behavior graph, the method further includes: The static characteristics of each coal mining machine operation are determined based on the parameters of the coal mining machine and the coal mining environment. The bias vector of each graph embedding layer of the spatial feature extraction layer of the truncation curve generation model is set based on the static features; or, the conditional embedding vector of the output layer of the spatial feature extraction layer of the truncation curve generation model is set based on the static features.

5. The method for generating the cutting curve of a coal mining machine according to claim 2, characterized in that, The temporal feature extraction layer based on the truncation curve generation model performs temporal semantic feature extraction processing on the coal mining machine operation embedding vector sequence of the behavior graph to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph, including: The static characteristics of each coal mining machine operation are determined based on the parameters of the coal mining machine and the coal mining environment, so as to determine the initial hidden state of the time feature extraction layer of the cutting curve generation model based on the static characteristics. The coal mining machine operation embedding vector sequence of the behavior graph is arranged in chronological order to obtain the chronological embedding vector sequence of the behavior graph; Based on the initial hidden state of the temporal feature extraction layer of the truncation curve generation model, the hidden state of each behavior graph node in the temporal embedding vector sequence is determined to obtain the spatiotemporal embedding vector sequence of the coal mining machine operation of the behavior graph.

6. The method for generating the cutting curve of a coal mining machine according to claim 1, characterized in that, The number of operation data of the at least two sample coal mining machines is greater than the number of operation data of the at least two sample coal mining machines.

7. The method for generating the cutting curve of a coal mining machine according to claim 1, characterized in that, The loss function of the cut-off curve generation model is determined based on the deviation between the predicted cut-off curve output by the original model of the cut-off curve generation model and the sample cut-off curve, the gradient penalty of the operation data of the sample coal mining machine, and the deviation between the predicted cut-off curve and the coal-rock boundary. The coal-rock boundary is obtained after the operation of the at least two sample coal mining machines.

8. A coal mining machine cutting curve generation system, characterized in that, include: The behavior graph node acquisition module is used to acquire the operation data of the most recent at least two coal mining machine operations, so as to generate corresponding behavior graph nodes based on the operation data of each coal mining machine operation; The behavior graph edge acquisition module is used to acquire the working condition migration data between adjacent coal mining machine operations in the most recent at least two coal mining machine operations, so as to generate corresponding behavior graph edges based on the working condition migration data. The behavior graph acquisition module is used to acquire the behavior graph of at least the two most recent coal mining machine operations based on the behavior graph nodes and the behavior graph edges. The cutting curve generation module is used to input the behavior map into the cutting curve generation model to obtain the cutting curve of the next coal mining machine operation output by the cutting curve generation model; wherein, the cutting curve generation model is trained based on the sample behavior map of the operation data of at least two sample coal mining machine operations and the sample cutting curve of the next coal mining machine operation.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the method for generating the cutting curve of a coal mining machine as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for generating the cutting curve of a coal mining machine as described in any one of claims 1 to 7.

Images