An optimization method for digital design of excavators

Abstract

The present invention discloses an optimization method for digital design of an excavator, and specifically relates to the technical field of excavator design. The key points of the technical solution are: obtaining feature data of the excavator, constructing a three-dimensional model of the excavator based on digital twins, and creating a simulated working environment for the excavator; using a multi-layer perceptron to perform spatial fusion processing on the feature data to obtain a feature data set; using the feature data in the feature data set to construct a feature undirected graph, inputting the feature undirected graph into a graph neural network for processing in combination with a multi-head attention mechanism, and obtaining importance scores of the nodes of the feature undirected graph, wherein the feature data are used as nodes of the undirected graph, and the association relationship between the feature data generates the edges of the undirected graph; constructing an initial excavator relationship model based on a deep neural network, inputting the feature data and the importance scores into the initial excavator relationship model for training, and obtaining the excavator relationship model that is ultimately used to generate the excavator structure association coefficient after training.

Description

Technical Field

[0001] The invention relates to the technical field of excavator design, and in particular to an optimization method for digital design of an excavator. Background Art

[0002] In recent years, driven by the continuous development of science and technology, more advanced technologies have been applied to large mining excavators for specific tasks in the design and manufacturing process, and the overall technical level of excavators has been greatly improved. The new theories, methods and technologies adopted in many aspects such as overall machine size, structural parameters, structural design, mechanism design, performance parameters and electrical design have played a role in promoting the reliability and performance of excavator products to varying degrees.

[0003] In the field of excavator digitalization, the key technology research of excavator digital design mainly focuses on the random, vibration, and impact loads of the excavator under complex natural working conditions, as well as the use requirements of high reliability and long life. In the existing excavator design process, the mapping mechanism between dynamic performance and design parameters is relatively vague, and the correlation between the optimal dynamic performance and design parameters has not been established, which affects the design of the excavator structure.

[0004] Therefore, the present invention aims to provide an optimization method for digital design of an excavator to solve the above-mentioned related problems. Summary of the invention

[0005] The technical problem to be solved by the present invention is that in the existing excavator design process, the mapping mechanism between dynamic performance and design parameters is relatively vague, and the correlation relationship between the optimal dynamic performance and the design parameters has not been established, which affects the design of the excavator structure. The purpose is to provide an optimization method for the digital design of an excavator. In the present invention, by acquiring the feature data of the excavator, a multi-layer perceptron is used to perform spatial fusion processing on the feature data, and a feature undirected graph is constructed based on the feature data. The feature undirected graph is input into a graph neural network and processed in combination with a multi-head attention mechanism to obtain the importance scores of the feature undirected graph nodes. Finally, a neural network model is used to combine the feature data and the importance scores to construct an excavator relationship model for generating the excavator structure correlation coefficient, so as to solve the problem that the correlation relationship between the optimal dynamic performance and the design parameters has not been established in the prior art, thereby affecting the design of the excavator structure.

[0006] The present invention is achieved through the following technical solutions:

[0007] An optimization method for digital design of an excavator, the method comprising:

[0008] Obtain the characteristic data of the excavator, build a 3D model of the excavator based on digital twins, and create a simulated working environment for the excavator;

[0009] Use a multi-layer perceptron to perform spatial fusion processing on the feature data to obtain a feature data set;

[0010] The feature data in the feature data set is used to construct a feature undirected graph, which is then input into the graph neural network and processed in combination with the multi-head attention mechanism to obtain the importance scores of the nodes in the feature undirected graph. The feature data is used as the nodes of the undirected graph, and the associations between the feature data generate the edges of the undirected graph.

[0011] An initial excavator relationship model is constructed based on a deep neural network, and feature data and importance scores are input into the initial excavator relationship model for training. After training, an excavator relationship model is finally obtained for generating an excavator structure correlation coefficient;

[0012] Inputting the feature data to be processed into the excavator relationship model to output the excavator structure correlation coefficient of the correlation relationship between the feature data and the excavator structure components, and using the excavator structure correlation coefficient to correct the excavator three-dimensional model;

[0013] Run the excavator 3D model in the excavator simulation working environment, and return various performance indicators and reward values ​​of the excavator according to the operation status of the excavator 3D model;

[0014] The excavator's three-dimensional model is corrected using the excavator's various performance indicators and reward values, and then the corrected excavator's three-dimensional model is input into the excavator's relationship model to output the excavator's structural correlation coefficient. The optimal excavator three-dimensional model is obtained through repeated iterative corrections until the reward value is stable.

[0015] Furthermore, a multi-layer perceptron is used to perform spatial fusion processing on the feature data to obtain a feature data set, specifically:

[0016] Setting category weights for feature data of the excavator based on preset feature categories, extracting Pearson correlation coefficients between feature data, marking feature data whose Pearson correlation coefficients are greater than a preset coefficient threshold, obtaining a plurality of marked feature data, and setting association weights for the marked feature data according to the respective Pearson correlation coefficients of the plurality of marked feature data;

[0017] Based on the category weight and association weight, the feature data is spatially fused using a multi-layer perceptron to obtain a feature data set.

[0018] Furthermore, the feature data in the feature data set is used to construct a feature undirected graph, which is then input into the graph neural network and processed in combination with the multi-head attention mechanism to obtain the importance scores of the nodes in the undirected graph. The feature data is used as the nodes of the undirected graph, and the associations between the nodes generate the edges of the undirected graph, specifically:

[0019] An undirected graph is constructed according to the feature data set, each feature data in the feature data set is used as a node of the undirected graph, the correlation relationship between each node is obtained according to the Pearson correlation coefficient, and the edge of the undirected graph is generated based on the correlation relationship between each node, so as to obtain a feature undirected graph associated with each feature data;

[0020] Obtain the maximum mutual information coefficient between different nodes of the feature undirected graph, and construct the adjacency matrix of the feature undirected graph according to the maximum mutual information coefficient;

[0021] The adjacency matrix of the feature undirected graph is input into the graph neural network, which outputs neighbor node vectors of multiple neighbor nodes and obtains the multi-head attention coefficients of multiple neighbor nodes in the adjacency matrix based on the multi-head attention mechanism;

[0022] The neighbor node vectors of multiple neighbor nodes are multiplied by the corresponding multi-head attention coefficients and then averaged to obtain the importance score of the feature undirected graph node.

[0023] Furthermore, an initial excavator relationship model is constructed based on a deep neural network, and the feature data and importance scores are input into the initial excavator relationship model for training. After training, the excavator relationship model used to generate the excavator structure correlation coefficient is obtained, which is specifically:

[0024] Obtain an autoencoder network, use the autoencoder network as the input layer of a deep neural network model, and obtain an initial excavator relationship model;

[0025] The particle swarm algorithm is used to optimize the network structure of the initial excavator relationship model and obtain the initial excavator relationship model with the optimal structure.

[0026] The feature data and importance scores are input into the initial excavator relationship model for training, and after training, the excavator relationship model finally used to generate the excavator structure association coefficient is obtained.

[0027] Furthermore, the particle swarm algorithm is used to optimize the network structure of the initial excavator relationship model to obtain the initial excavator relationship model with the optimal structure, which is specifically:

[0028] Initialize the particle swarm algorithm parameters, particle position and velocity information, use the output accuracy of the initial excavator relationship model as the fitness, and obtain the individual extreme value and global extreme value of the particle;

[0029] Based on the simulated annealing algorithm, it is repeated to optimize, and the particle position and velocity are updated by using the simulated annealing neighborhood search;

[0030] When the preset termination condition is met, the optimization is stopped, the optimal structural parameters of the autoencoder network are obtained according to the optimal particle position, and the initial excavator relationship model of the optimal structure is constructed based on the optimal structural parameters.

[0031] Furthermore, the preset feature categories include excavator working environment, power performance and working performance.

[0032] Furthermore, the simulated working environment of the excavator includes farmland, mountains, cities and rivers.

[0033] The present invention also provides an optimization system for digital design of an excavator, the system comprising:

[0034] The feature data acquisition module is used to acquire the feature data of the excavator, build a three-dimensional model of the excavator based on the digital twin, and create a simulated working environment for the excavator;

[0035] A feature data set construction module is used to perform spatial fusion processing on feature data using a multi-layer perceptron to obtain a feature data set;

[0036] An undirected graph construction module is used to construct a feature undirected graph using the feature data in the feature data set, and input the feature undirected graph into the graph neural network for processing in combination with the multi-head attention mechanism to obtain the importance scores of the nodes in the feature undirected graph, where the feature data are used as nodes of the undirected graph, and the associations between the feature data generate the edges of the undirected graph;

[0037] A model training module is used to construct an initial excavator relationship model based on a deep neural network, input feature data and importance scores into the initial excavator relationship model for training, and obtain the excavator relationship model that is ultimately used to generate the excavator structure correlation coefficient after training;

[0038] A structure correction module is used to input the feature data to be processed into the excavator relationship model to output the excavator structure correlation coefficient of the correlation relationship between the feature data and the excavator structural components, and to correct the excavator three-dimensional model using the excavator structure correlation coefficient;

[0039] The simulation operation module is used to run the excavator 3D model in the excavator simulation working environment and return various performance indicators and reward values ​​of the excavator according to the operation status of the excavator 3D model;

[0040] The structure correction module is also used to correct the excavator three-dimensional model using the excavator's various performance indicators and reward values, and then input the corrected excavator three-dimensional model into the excavator relationship model to output the excavator structure correlation coefficient. After repeated iterative corrections until the reward value is stable, the optimal excavator three-dimensional model is obtained.

[0041] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above-mentioned methods when executing the computer program.

[0042] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any one of the methods described above are implemented.

[0043] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0044] In the present invention, the feature data of the excavator is obtained, a multi-layer perceptron is used to perform spatial fusion processing on the feature data, and a feature undirected graph is constructed based on the feature data. The feature undirected graph is input into a graph neural network and processed in combination with a multi-head attention mechanism to obtain the importance scores of the nodes of the feature undirected graph. Finally, a neural network model is used to combine the feature data and the importance scores to construct an excavator relationship model for generating the excavator structure correlation coefficient, thereby solving the related problem that the correlation relationship between the optimal dynamic performance and the design parameters has not been established in the prior art, thereby affecting the design of the excavator structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0046] Figure 1 A method flow chart of an optimization method for digital design of an excavator in this embodiment;

[0047] Figure 2 This is a method flow chart of step S3 in an optimization method for digital design of an excavator in this embodiment;

[0048] Figure 3 This is a method flow chart of step S4 in an optimization method for digital design of an excavator in this embodiment;

[0049] Figure 4A schematic diagram of a module of an optimization system for digital design of an excavator in this embodiment;

[0050] Figure 5 A schematic diagram of the structure of a computer device. DETAILED DESCRIPTION

[0051] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0052] In the present disclosure, unless otherwise specified, the use of the terms "first", "second", etc. to describe various elements is not intended to limit the positional relationship, timing relationship, or importance relationship of these elements, and such terms are only used to distinguish one element from another element. In some examples, the first element and the second element may refer to the same instance of the element, and in some cases, based on the description of the context, they may also refer to different instances.

[0053] The terms used in the description of various examples in this disclosure are only for the purpose of describing specific examples and are not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the element can be one or more. In addition, the term "and / or" used in this disclosure covers any one of the listed items and all possible combinations.

[0054] Example 1

[0055] An optimization method for digital design of an excavator, the method comprising:

[0056] S1: Acquire characteristic data of the excavator, build a three-dimensional model of the excavator based on the digital twin, and create a simulated working environment for the excavator. In this embodiment, the simulated working environment for the excavator includes farmland, mountain, city, and river;

[0057] It should be noted that in this embodiment, the basic modules in the simulink model library of the engineering simulation software are called according to the structural parts of the excavator to construct the three-dimensional model of the excavator, ensure the correct scale, texture and material of the model, the three-dimensional spatial position of each component and the coordinated motion relationship information between each component, and realize the synchronization of the virtual three-dimensional model movement. The constructed three-dimensional model of the excavator is encapsulated to automatically generate PLC code, and the constructed three-dimensional model of the excavator is simulated by the solver to perform error troubleshooting. The parameter configuration includes setting the solver step size, setting the system target file generated by the PLC code, setting the annotation option and setting the report format, setting the solver to a fixed step size, and setting the basic sampling period of the excavator three-dimensional model to be consistent with the interrupt period of the target controller hardware chip; the generated PLC code is placed in the selected compilation environment for integration to generate an executable program file; the excavation action is planned according to the specific shape of the excavation pile; to control the relevant units of the PLC, the movement specifies the travel of the stepper motor rotary encoder in the planning drive unit, and can control the relevant parts in the action implementation unit to perform excavation operations according to the specified excavation trajectory;

[0058] At the same time, the characteristic data of the excavator includes characteristic parameters of different types of excavators, and the characteristic data are classified and organized according to the excavator's working environment, power performance and working performance; to create an excavator simulation working environment, aerial photography modeling can be used in combination with manual modeling to perform 3D rendering and establishment of the excavator's working scene.

[0059] It should be noted that dynamics and working performance mainly include tonnage size, engine power, bucket capacity, movement speed, climbing ability, digging angle, hydraulic and electrical parameters, etc.

[0060] Specifically, in this embodiment, by constructing a three-dimensional model of the excavator based on the digital twin, it is convenient to display the corrected and optimized changes of the excavator structure in a realistic, intuitive and real-time manner. At the same time, the correction process can be monitored so as to optimize and regulate the correction process and reduce the generation of erroneous information.

[0061] S2: Use a multi-layer perceptron to perform spatial fusion processing on the feature data to obtain a feature data set, specifically:

[0062] S201: setting category weights for feature data of the excavator based on preset feature categories. In this embodiment, the preset feature categories include working environment, power performance and working performance of the excavator, extracting Pearson correlation coefficients between feature data, marking feature data whose Pearson correlation coefficients are greater than a preset coefficient threshold, obtaining multiple marked feature data, and setting association weights for the marked feature data respectively according to the respective Pearson correlation coefficients of the multiple marked feature data;

[0063] S202: Based on the category weights and the association weights, a multi-layer perceptron is used to perform spatial fusion processing on the feature data to obtain a feature data set.

[0064] It should be noted that the Pearson correlation coefficient is used to measure the linear correlation between two feature data, and its value is between -1 and 1. Usually, the Pearson correlation coefficient between two variables is defined as the quotient of the product of the covariance and the standard deviation between the two variables (or the normalized covariance).

[0065] Specifically, in this embodiment, based on the working environment, power performance and working performance of the excavator, category weights are set for the feature data. For example, the category weight of the feature data belonging to the excavator working environment category is 20%, the category weight of the feature data belonging to the power performance category is 40%, and the category weight of the feature data belonging to the working performance category is 40%. The actual category weight depends on the actual situation. Since the Pearson correlation coefficient of some feature data is relatively high, it means that its correlation is relatively high. Therefore, feature data whose Pearson correlation coefficient exceeds the preset coefficient threshold are extracted and marked as marked feature data, and association weights are set for the marked feature data according to the Pearson correlation coefficients of each of the multiple marked feature data. The higher the Pearson correlation coefficient, the higher the association weight. No association weight is set for unmarked feature data.

[0066] S3: Use the feature data in the feature data set to construct a feature undirected graph, input the feature undirected graph into the graph neural network and process it in combination with the multi-head attention mechanism to obtain the importance score of the feature undirected graph node, filter a preset number of feature data according to the importance score, and build an excavator relationship model based on the parameter indicators corresponding to the feature data. Among them, the feature data is used as the node of the undirected graph, and the association relationship between the feature data generates the edge of the undirected graph, specifically:

[0067] S301: construct an undirected graph according to the feature data set, use each feature data in the feature data set as a node of the undirected graph, obtain the association relationship between each node according to the Pearson correlation coefficient, generate the edge of the undirected graph based on the association relationship between each node, and obtain a feature undirected graph associated with each feature data;

[0068] S302: Obtain the maximum mutual information coefficient between different nodes of the feature undirected graph, and construct an adjacency matrix of the feature undirected graph according to the maximum mutual information coefficient;

[0069] S303: inputting the adjacency matrix of the feature undirected graph into the graph neural network, the graph neural network outputs neighbor node vectors of multiple neighbor nodes, and obtains multi-head attention coefficients of multiple neighbor nodes in the adjacency matrix based on the multi-head attention mechanism;

[0070] S304: Multiply the neighbor node vectors of multiple neighbor nodes by the corresponding multi-head attention coefficients and average them to obtain the importance score of the feature undirected graph node.

[0071] It should be noted that in this embodiment, the multi-head attention mechanism is used to calculate the importance score of the undirected graph node, which is a conventional technical means. The multi-head attention mechanism is a mechanism widely used in the Transformer model, which captures different feature representations through multiple independent attention heads. In graph neural networks, the multi-head attention mechanism is used to process the relationship between nodes and information transfer.

[0072] Specifically, in this embodiment, the maximum mutual information coefficient between different nodes of the undirected graph is used to measure the correlation between two nodes, and its range is between 0 and 1. When it is 0, it means that the two nodes are completely independent, and when it is 1, it means that there is a complete dependency relationship between the two nodes. The adjacency matrix of the characteristic undirected graph is constructed according to the maximum mutual information coefficient by creating an N×N matrix (N is the number of nodes). This matrix will be used as the adjacency matrix of the characteristic undirected graph. For each pair of nodes (X, Y), if their maximum mutual information coefficient exceeds a certain threshold, the corresponding positions (i, j) and (j, i) in the adjacency matrix are set to 1, indicating that there is an edge between the two nodes;

[0073] At the same time, the adjacency matrix of the feature undirected graph and the feature matrix of the node (the feature matrix of the node is an N×F matrix, F is the number of features of each node) are input into the graph neural network, and the graph neural network outputs neighbor node vectors of multiple neighbor nodes. At the same time, the multi-head attention mechanism is set in the graph neural network, and multi-head attention operations are performed on each node for the neighbor nodes to obtain the multi-head attention coefficients of multiple neighbor nodes in the adjacency matrix.

[0074] S4: Build an initial excavator relationship model based on a deep neural network, input feature data and importance scores into the initial excavator relationship model for training, and obtain the final excavator relationship model after training, specifically:

[0075] S401: Obtain an autoencoder network, use the autoencoder network as an input layer of a deep neural network model, and obtain an initial excavator relationship model;

[0076] S402: Optimizing the network structure of the initial excavator relationship model using a particle swarm algorithm to obtain an initial excavator relationship model with an optimal structure, specifically:

[0077] S403: Initialize the particle swarm algorithm parameters, particle position and velocity information, use the output accuracy of the initial excavator relationship model as fitness, and obtain the individual extreme value and global extreme value of the particle;

[0078] S404: performing iterative optimization based on a simulated annealing algorithm, and updating particle positions and velocities using simulated annealing neighborhood search;

[0079] S405: When the preset termination condition is met, the optimization is stopped, the optimal structural parameters of the autoencoder network are obtained according to the optimal particle position, and an initial excavator relationship model of the optimal structure is constructed based on the optimal structural parameters. In this embodiment, the structural parameters of the autoencoder network include the number of network layers, neural network weights and thresholds.

[0080] S406: inputting the feature data and the importance score into the initial excavator relationship model for training, and obtaining the excavator relationship model that is ultimately used to generate the excavator structure correlation coefficient after training.

[0081] Specifically, in this embodiment, the excavator structure correlation coefficient is used to characterize the corresponding correlation relationship between the excavator operating conditions and structural components.

[0082] S5: inputting the feature data to be processed into the excavator relationship model to output the excavator structure association coefficient of the association relationship between the feature data and the excavator structural components, and modifying the excavator three-dimensional model using the excavator structure association coefficient;

[0083] S6: running the excavator three-dimensional model in the excavator simulation working environment, and returning various performance indicators and reward values ​​of the excavator according to the running status of the excavator three-dimensional model;

[0084] Specifically, in this embodiment, by running the constructed visual three-dimensional model of the excavator in the excavator simulation working environment, it is convenient to obtain whether the various performance indicators of the excavator after preliminary correction after operation reach the preset goals and meet the actual situation, and output the reward value according to each performance indicator to correct the excavator three-dimensional model.

[0085] S7: The excavator three-dimensional model is modified using the excavator's various performance indicators and reward values, and the modified excavator three-dimensional model is then input into the excavator relationship model to output the excavator structure correlation coefficient. The optimal excavator three-dimensional model is obtained through repeated iterative modification until the reward value is stable.

[0086] It should be noted that, in this embodiment, the reward value of the deep neural network comes from the judgment of various performance indicators generated by the three-dimensional model of the excavator running in the simulated working environment of the excavator. For example, if the moving speed is too slow in the working environment, the corresponding production reward value is used to optimize the characteristic data and component structure parameters that affect the driving speed in the excavator.

[0087] Specifically, in this embodiment, the feature data of the excavator is obtained, the feature data is spatially fused using a multilayer perceptron, and a feature undirected graph is constructed based on the feature data. The feature undirected graph is input into a graph neural network and processed in combination with a multi-head attention mechanism to obtain the importance scores of the nodes of the feature undirected graph. Finally, a neural network model is used to combine the feature data and the importance scores to construct an excavator relationship model for generating an excavator structure correlation coefficient, thereby solving the related problem that the relationship between the optimal dynamic performance and the design parameters has not been established in the prior art, thereby affecting the design of the excavator structure. At the same time, the corrected excavator twin structure is placed in the constructed excavator simulation working environment for operation, so that various performance indicators of the excavator can be obtained. Based on the deviation of each performance indicator, the optimization reward value used to correct the structural parameters and feature data during operation is obtained.

[0088] Example 2

[0089] The present invention also provides an optimization system for digital design of an excavator, the system comprising:

[0090] The feature data acquisition module 100 is used to acquire the feature data of the excavator, build a three-dimensional model of the excavator based on the digital twin, and create a simulated working environment for the excavator;

[0091] The feature data set construction module 200 is used to perform spatial fusion processing on the feature data using a multi-layer perceptron to obtain a feature data set;

[0092] An undirected graph construction module 300 is used to construct a feature undirected graph using feature data in a feature data set, input the feature undirected graph into a graph neural network for processing in combination with a multi-head attention mechanism, and obtain importance scores of nodes in the feature undirected graph, wherein the feature data are used as nodes of the undirected graph, and the associations between the feature data generate edges of the undirected graph;

[0093] A model training module 400 is used to construct an initial excavator relationship model based on a deep neural network, input feature data and importance scores into the initial excavator relationship model for training, and obtain the excavator relationship model that is ultimately used to generate an excavator structure correlation coefficient after training;

[0094] The structure correction module 500 is used to input the feature data to be processed into the excavator relationship model to output the excavator structure correlation coefficient of the correlation relationship between the feature data and the excavator structural components, and to correct the excavator three-dimensional model using the excavator structure correlation coefficient;

[0095] The simulation operation module 600 is used to operate the excavator three-dimensional model in the excavator simulation working environment, and return various performance indicators and reward values ​​of the excavator according to the operation status of the excavator three-dimensional model;

[0096] The structure correction module 500 is also used to correct the excavator three-dimensional model using the excavator's various performance indicators and reward values, and then input the corrected excavator three-dimensional model into the excavator relationship model to output the excavator structure correlation coefficient. After repeated iterative corrections until the reward value is stable, the optimal excavator three-dimensional model is obtained.

[0097] It should be noted that the modules in the system of Example 2 correspond to the steps in the method of Example 1. The steps in the method of Example 1 have been described in detail in Example 1. In this Example 2, the contents of the modules in the system will not be described in detail.

[0098] Example 3

[0099] This embodiment further provides a computer device, including a memory 1005 and a processor 1001, wherein the memory 1005 stores a computer program, and the processor 1001 implements the steps of any one of the above methods when executing the computer program.

[0100] It should be noted that the processor 1001 is used to execute the steps in the above method embodiments according to the instructions in the program code. Alternatively, the processor 1001 implements the functions of the above systems / modules / units in the system embodiments when executing the computer program.

[0101] Specifically, in this embodiment, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the memory 1005, and are executed by the processor 1001 to complete the present application. One or more modules / units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0102] The terminal device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor 1001 and a memory 1005. Those skilled in the art will appreciate that this does not constitute a limitation on the terminal device, and may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device may also include an input / output device 1003, a network access device 1002, a bus 1006, etc.

[0103] The processor 1001 may be a central processing unit (CPU), or other general-purpose processors 1001, digital signal processors 1001 (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor 1001 may be a microprocessor or any conventional processor, etc.

[0104] The memory 1005 may be an internal storage unit of the terminal device, such as a hard disk or memory of the terminal device. The memory 1005 may also be an external storage device 1004 of the terminal device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal device. Further, the memory 1005 may also include both the internal storage unit of the terminal device and the external storage device 1004. The memory 1005 is used to store computer programs and other programs and data required by the terminal device. The memory 1005 may also be used to temporarily store data that has been output or is to be output.

[0105] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0106] Example 4

[0107] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the above methods are implemented.

[0108] Among them, the computer readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, system or device, or any combination of the above. More specific examples of computer readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk. Random Access Memory (RAM), Read-Only Memory (ROM), Erasable Programmable Read Only Memory (EPROM), registers, hard disks, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or the above in a suitable combination, or any other form of computer readable storage medium known in the art.

[0109] An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application-specific integrated circuit (ASIC). In an embodiment of the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system, or device.

[0110] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An optimization method for digital design of an excavator, characterized in that the method include: Obtain the characteristic data of the excavator, build a 3D model of the excavator based on digital twins, and create a simulated working environment for the excavator; Use a multi-layer perceptron to perform spatial fusion processing on the feature data to obtain a feature data set; The feature data in the feature data set is used to construct a feature undirected graph, and the maximum mutual information coefficient between different nodes of the feature undirected graph is obtained to construct the adjacency matrix of the feature undirected graph. The adjacency matrix of the feature undirected graph is input into the graph neural network and processed in combination with the multi-head attention mechanism to obtain the importance score of the undirected graph node, where the feature data is used as the node of the undirected graph, and the relationship between the nodes generates the edge of the undirected graph; An initial excavator relationship model is constructed based on the autoencoder network as the input layer of the deep neural network model, and the network structure of the initial excavator relationship model is optimized by using the particle swarm algorithm to obtain the initial excavator relationship model with the optimal structure. The feature data and importance score are input into the initial excavator relationship model for training, and after training, the excavator relationship model finally used to generate the excavator structure correlation coefficient is obtained; Inputting the feature data to be processed into the excavator relationship model to output the excavator structure correlation coefficient of the correlation relationship between the feature data and the excavator structure components, and using the excavator structure correlation coefficient to correct the excavator three-dimensional model; Run the excavator 3D model in the excavator simulation working environment, and return various performance indicators and reward values ​​of the excavator according to the operation status of the excavator 3D model; The excavator's three-dimensional model is corrected using the excavator's various performance indicators and reward values, and then the corrected excavator's three-dimensional model is input into the excavator's relationship model to output the excavator's structural correlation coefficient. The optimal excavator three-dimensional model is obtained through repeated iterative corrections until the reward value is stable.

2. The optimization method for digital design of an excavator according to claim 1, characterized in that: The feature data is spatially fused using a multi-layer perceptron to obtain a feature data set, specifically: Setting category weights for feature data of the excavator based on preset feature categories, extracting Pearson correlation coefficients between feature data, marking feature data whose Pearson correlation coefficients are greater than a preset coefficient threshold, obtaining a plurality of marked feature data, and setting association weights for the marked feature data respectively according to the respective Pearson correlation coefficients of the plurality of marked feature data; Based on the category weight and association weight, the feature data is spatially fused using a multi-layer perceptron to obtain a feature data set.

3. The optimization method for digital design of an excavator according to claim 2, characterized in that: The feature data in the feature data set is used to construct a feature undirected graph, and the maximum mutual information coefficient between different nodes of the feature undirected graph is obtained to construct the adjacency matrix of the feature undirected graph. The adjacency matrix of the feature undirected graph is input into the graph neural network and processed in combination with the multi-head attention mechanism to obtain the importance score of the undirected graph node. Among them, the feature data is used as the node of the undirected graph, and the association relationship between the nodes generates the edge of the undirected graph, which is specifically: An undirected graph is constructed according to the feature data set, each feature data in the feature data set is used as a node of the undirected graph, the correlation relationship between each node is obtained according to the Pearson correlation coefficient, and the edge of the undirected graph is generated based on the correlation relationship between each node, so as to obtain a feature undirected graph associated with each feature data; Obtain the maximum mutual information coefficient between different nodes of the feature undirected graph, and construct the adjacency matrix of the feature undirected graph according to the maximum mutual information coefficient; The adjacency matrix of the feature undirected graph is input into the graph neural network, which outputs neighbor node vectors of multiple neighbor nodes and obtains the multi-head attention coefficients of multiple neighbor nodes in the adjacency matrix based on the multi-head attention mechanism; The neighbor node vectors of multiple neighbor nodes are multiplied by the corresponding multi-head attention coefficients and then averaged to obtain the importance score of the feature undirected graph node.

4. The optimization method for digital design of an excavator according to claim 1, characterized in that: Based on the autoencoder network as the input layer of the deep neural network model, the initial excavator relationship model is constructed, and the network structure of the initial excavator relationship model is optimized using the particle swarm algorithm to obtain the initial excavator relationship model with the optimal structure. The feature data and importance score are input into the initial excavator relationship model for training. After training, the excavator relationship model used to generate the excavator structure correlation coefficient is obtained, which is specifically: Obtain an autoencoder network, use the autoencoder network as the input layer of a deep neural network model, and obtain an initial excavator relationship model; The particle swarm algorithm is used to optimize the network structure of the initial excavator relationship model and obtain the initial excavator relationship model with the optimal structure. The feature data and importance scores are input into the initial excavator relationship model for training, and after training, the excavator relationship model finally used to generate the excavator structure association coefficient is obtained.

5. The optimization method for digital design of an excavator according to claim 4, characterized in that: The particle swarm algorithm is used to optimize the network structure of the initial excavator relationship model to obtain the initial excavator relationship model with the optimal structure, which is as follows: Initialize the particle swarm algorithm parameters, particle position and velocity information, use the output accuracy of the initial excavator relationship model as the fitness, and obtain the individual extreme value and global extreme value of the particle; Based on the simulated annealing algorithm, it is repeated to optimize, and the particle position and velocity are updated by using simulated annealing neighborhood search; When the preset termination condition is met, the optimization is stopped, the optimal structural parameters of the autoencoder network are obtained according to the optimal particle position, and the initial excavator relationship model of the optimal structure is constructed based on the optimal structural parameters.

6. The optimization method for digital design of an excavator according to claim 2, characterized in that: The preset feature categories include excavator working environment, power performance and working performance.

7. The optimization method for digital design of an excavator according to claim 1, characterized in that: The simulated working environment of the excavator includes farmland, mountains, cities and rivers.

8. An optimization system for digital design of excavators, characterized in that: The system comprises: The feature data acquisition module is used to acquire the feature data of the excavator, build a three-dimensional model of the excavator based on the digital twin, and create a simulated working environment for the excavator; A feature data set construction module is used to perform spatial fusion processing on feature data using a multi-layer perceptron to obtain a feature data set; An undirected graph construction module is used to construct a feature undirected graph using the feature data in the feature data set, obtain the maximum mutual information coefficient between different nodes of the feature undirected graph to construct the adjacency matrix of the feature undirected graph, and input the adjacency matrix of the feature undirected graph into the graph neural network and process it in combination with the multi-head attention mechanism to obtain the importance scores of the nodes of the undirected graph, wherein the feature data is used as the nodes of the undirected graph, and the association relationship between the nodes generates the edges of the undirected graph; A model training module is used to construct an initial excavator relationship model based on the autoencoder network as the input layer of the deep neural network model, optimize the network structure of the initial excavator relationship model using a particle swarm algorithm to obtain an initial excavator relationship model with an optimal structure, input feature data and importance scores into the initial excavator relationship model for training, and obtain an excavator relationship model that is ultimately used to generate an excavator structure correlation coefficient after training; A structure correction module is used to input the feature data to be processed into the excavator relationship model to output the excavator structure correlation coefficient of the correlation relationship between the feature data and the excavator structural components, and to correct the excavator three-dimensional model using the excavator structure correlation coefficient; The simulation operation module is used to run the excavator 3D model in the excavator simulation working environment and return various performance indicators and reward values ​​of the excavator according to the operation status of the excavator 3D model; The structure correction module is also used to correct the excavator three-dimensional model using the excavator's various performance indicators and reward values, and then input the corrected excavator three-dimensional model into the excavator relationship model to output the excavator structure correlation coefficient. After repeated iterative corrections until the reward value is stable, the optimal excavator three-dimensional model is obtained.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images