Digital twin data processing method and system based on artificial intelligence

Abstract

The present invention discloses an artificial intelligence-based digital twin data processing method and system, which belongs to the field of data processing technology. An artificial intelligence algorithm is used to construct a digital twin data processing model, and an intelligent optimization algorithm is used to train the digital twin data processing model to obtain a trained digital twin data processing model. Then, during the operation of the digital twin model, the trained digital twin data processing model is used to process the real-time operation parameters of the digital twin model, thereby realizing the automatic processing of digital twin data and effectively improving the processing efficiency and accuracy of digital twin data.

Description

Technical Field

[0001] The present invention belongs to the field of data processing technology, and specifically relates to a digital twin data processing method and system based on artificial intelligence. Background Art

[0002] A digital twin model is a virtual model that accurately digitally represents a physical entity, system, process, or service in the real world. This model aims to simulate the behavior, performance, and structure of the actual object, enabling analysis, prediction, and optimization without a physical prototype. Generally, digital twin models serve as a data synchronization and visualization tool to facilitate operational monitoring and debugging of equipment. However, manual operation monitoring, while eliminating the need for on-site equipment inspections, can be time-consuming and labor-intensive, leading to inadequate monitoring. Summary of the Invention

[0003] The present invention provides an artificial intelligence-based digital twin data processing method and system to solve the problems that existing digital twin models only provide data monitoring but do not automatically process information, resulting in time-consuming and labor-intensive operations and inadequate monitoring.

[0004] In one aspect, the present invention provides a digital twin data processing method based on artificial intelligence, comprising:

[0005] Build a digital twin model corresponding to the target machine device and synchronize the operating parameters of the target machine device to the digital twin model to run the digital twin model;

[0006] Use artificial intelligence algorithms to build a digital twin data processing model, and use intelligent optimization algorithms to train the digital twin data processing model to obtain the trained digital twin data processing model;

[0007] Collect the real-time operating parameters corresponding to the digital twin model, and use the trained digital twin data processing model to process the real-time operating parameters to obtain parameter processing results;

[0008] When the parameter processing results meet the preset requirements, corresponding labels and label-associated real-time operating parameters are generated on the digital twin model to complete the processing of the digital twin data.

[0009] Furthermore, an artificial intelligence algorithm is used to construct a digital twin data processing model, including: using a CNN algorithm, an R-CNN algorithm, a Fast R-CNN algorithm, a YOLO algorithm and / or a BP Net algorithm to construct a digital twin data processing model.

[0010] Furthermore, an intelligent optimization algorithm is used to train the digital twin data processing model to obtain the trained digital twin data processing model, including:

[0011] Initialize the hyperparameters of the digital twin data processing model to generate a population containing multiple parameter individuals;

[0012] For each parameter individual in the population, apply the parameter individual to the digital twin data processing model, and use the cross entropy loss function to obtain the loss function value corresponding to the parameter individual;

[0013] According to the loss function value corresponding to the parameter individual, the optimal parameter individual is obtained, and the optimal parameter individual is updated using the normal distribution jump strategy to obtain the updated optimal parameter individual;

[0014] For any parameter individual, perform a local fast guided search on the parameter individual according to the optimal parameter individual to obtain the parameter individual after the local fast guided search;

[0015] For any parameter individual after local rapid guided search, multi-information learning search is performed on the parameter individual to obtain the parameter individual after multi-information learning search;

[0016] For any parameter individual after multi-information learning search, perform adaptive chain search on the parameter individual to obtain the parameter individual after adaptive chain search;

[0017] Determine whether the training end conditions are met. If so, obtain the trained digital twin data processing model based on the parameter individuals after the adaptive chain search. Otherwise, return to the step of obtaining the loss function value.

[0018] Furthermore, for each parameter individual in the population, the parameter individual is applied to the digital twin data processing model, and the cross entropy loss function is used to obtain the loss function value corresponding to the parameter individual, including:

[0019] For each parameter individual in the population, apply the parameter individual to the digital twin data processing model to obtain the digital twin data processing model after the parameters are applied;

[0020] Use historical operating parameters as input to the digital twin data processing model to obtain the actual output of the digital twin data processing model;

[0021] The data labels corresponding to the historical operating parameters are used as the expected output of the digital twin data processing model. Based on the expected output and actual output of the digital twin data processing model, the cross entropy loss function is used to obtain the loss function value corresponding to the parameter individual.

[0022] Furthermore, according to the loss function value corresponding to the parameter individual, the optimal parameter individual is obtained, and the optimal parameter individual is updated using the normal distribution jump strategy to obtain the updated optimal parameter individual, including:

[0023] According to the loss function value corresponding to the parameter individual, the parameter individual with the smallest loss function value is determined as the optimal parameter individual;

[0024] The random jump factor generated by normal distribution is:

[0025]

[0026]

[0027]

[0028] According to the random jump factor, the optimal parameter individual is updated, and the optimal parameter individual after the update is obtained: ;in, Indicates the k The optimal parameter individual in the training process d dimensional parameters, represents the optimal parameter individual after update d dimensional parameters, represents the random jump factor, R represents the adaptive update coefficient, and , represents the first constant term, represents the first random parameter individual except the optimal parameter individual d dimensional parameters, represents the first random jump control factor, v represents the second random jump control factor, Represents a random number between (0,1), Satisfies normal distribution , Satisfies normal distribution , represents the first intermediate parameter, represents the second constant term and is set to 1; represents the gamma function.

[0029] Furthermore, for any parameter individual, a local rapid guided search is performed on the parameter individual according to the optimal parameter individual to obtain the parameter individual after the local rapid guided search, including:

[0030] Generate random bootstrap control factors between (0,1) based on the standard normal distribution;

[0031] According to the random bootstrap control factor and the optimal parameter individual, a local fast bootstrap search is performed on the parameter individual, and the parameter individual after the local fast bootstrap search is obtained as follows: ;in, Indicates the k The mth parameter individual in the training process, represents the optimal parameter individual, represents the random bootstrap control factor, and ||*|| represents obtaining the Euclidean distance.

[0032] Furthermore, for any parameter individual after the local rapid guided search, a multi-information learning search is performed on the parameter individual to obtain a parameter individual after the multi-information learning search, including:

[0033] For any parameter individual after local rapid guided search, randomly match the parameter individual with another parameter individual to obtain the paired parameter individual corresponding to each parameter individual;

[0034] According to the paired parameter individuals corresponding to the parameter individuals, the parameter individuals are studied and searched, and the first learning information is obtained as follows: ;in, Represents the first learning information, Indicates the k The nth parameter individual in the training process, Represents parameter individuals The corresponding pairing parameter individuals, represents the first learning coefficient, Represents a random number between (0,1);

[0035] According to the historical optimal value corresponding to the parameter individual, the parameter individual is studied and searched, and the second learning information is obtained as follows: ;in, represents the second learning information, Represents parameter individuals The corresponding historical optimal value is represents the second learning coefficient, Represents a random number between (0,1);

[0036] According to the optimal parameter individual, the parameter individual is studied and searched, and the third learning information is obtained as follows: ;in, represents the third learning information, represents the optimal parameter individual, represents the third learning coefficient, Represents a random number between (0,1);

[0037] According to the first learning information, the second learning information and the third learning information, the parameter individuals after the multi-information learning search are obtained as follows: ;in, Represents the parameter individual after multi-information learning search .

[0038] Furthermore, for any parameter individual after the multi-information learning search, an adaptive chain search is performed on the parameter individual to obtain the parameter individual after the adaptive chain search, including:

[0039] For any parameter individual after multi-information learning search, obtain the loss function value corresponding to the parameter individual, and arrange the parameter individuals in order from small to large according to the loss function value;

[0040] Based on the parameter individuals after arrangement, an adaptive chain search is performed on the parameter individuals, and the parameter individuals after the adaptive chain search are obtained as follows: ;in, Indicates the first i parameter individuals, Represents the parameter individual after adaptive chain search , Indicates the first i -1 parameter individual, and when i When 1, Set to The historical optimal value of represents the first chain search factor, represents the second chain search factor, represents the fourth random number between (0,1), represents the fifth random number between (0,1), represents the adaptive inertia weight, and ; represents the maximum value of the adaptive inertia weight, Represents the minimum value of the adaptive inertia weight, and K represents the preset maximum number of training times.

[0041] Furthermore, when the parameter processing results meet the preset requirements, corresponding tags and tag-associated real-time operating parameters are generated on the digital twin model, including:

[0042] When the parameter processing result shows that an anomaly exists, a label is generated on the digital twin model, and the real-time operating parameters are associated on the secondary interface of the label.

[0043] On the other hand, the present invention provides an artificial intelligence-based digital twin data processing system, comprising: a digital twin module, a model building module, a parameter processing module, and a result processing module;

[0044] The digital twin module is used to build a digital twin model corresponding to the target machine device and synchronize the operating parameters of the target machine device to the digital twin model to run the digital twin model;

[0045] The model construction module is used to construct a digital twin data processing model using an artificial intelligence algorithm, and to train the digital twin data processing model using an intelligent optimization algorithm to obtain a trained digital twin data processing model;

[0046] The parameter processing module is used to collect real-time operating parameters corresponding to the digital twin model and process the real-time operating parameters using the trained digital twin data processing model to obtain parameter processing results;

[0047] The result processing module is used to generate corresponding tags and tag-associated real-time operating parameters on the digital twin model when the parameter processing results meet the preset requirements, thereby completing the processing of the digital twin data.

[0048] The present invention provides an artificial intelligence-based digital twin data processing method and system, which constructs a digital twin data processing model by adopting an artificial intelligence algorithm, and trains the digital twin data processing model by adopting an intelligent optimization algorithm to obtain a trained digital twin data processing model. Then, during the operation of the digital twin model, the trained digital twin data processing model is used to process the real-time operating parameters of the digital twin model, thereby realizing the automatic processing of digital twin data and effectively improving the processing efficiency and accuracy of digital twin data. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0050] Figure 1 A flowchart of a digital twin data processing method based on artificial intelligence provided by an embodiment of the present invention.

[0051] Figure 2 A schematic structural diagram of an artificial intelligence-based digital twin data processing system provided in an embodiment of the present invention.

[0052] The above drawings have shown specific embodiments of the present invention, which will be described in more detail below. These drawings and the text description are not intended to limit the scope of the present invention in any way, but rather to illustrate the concept of the present invention to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0053] Exemplary embodiments will be described in detail herein, examples of which are illustrated in the accompanying drawings. In the following description, when referring to the drawings, like numbers in different figures represent like or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present invention, as detailed in the appended claims.

[0054] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0055] like Figure 1 As shown, an embodiment of the present invention provides a digital twin data processing method based on artificial intelligence, including:

[0056] S1. Build a digital twin model corresponding to the target machine device and synchronize the operating parameters of the target machine device to the digital twin model to run the digital twin model;

[0057] The target machine equipment can be various devices, such as motors, robotic arms, etc. Digital twin technology can be used to build a digital twin model corresponding to the target machine equipment, so as to realize online monitoring or simulated operation of the target machine equipment.

[0058] S2. Use artificial intelligence algorithms to build a digital twin data processing model, and use intelligent optimization algorithms to train the digital twin data processing model to obtain a trained digital twin data processing model;

[0059] In an embodiment of the present invention, an artificial intelligence algorithm is used to construct a digital twin data processing model, including: using a CNN (convolutional neural network) algorithm, an R-CNN algorithm, a Fast R-CNN algorithm, a YOLO algorithm and / or a BP Net (back propagation neural network) algorithm to construct a digital twin data processing model.

[0060] However, it is worth noting that the above-mentioned artificial intelligence algorithm is only used as an example of an embodiment of the present invention. Other artificial intelligence algorithms can also be used to construct a digital twin data processing model. Similarly, one artificial intelligence algorithm can be used for data processing, or multiple artificial intelligence algorithms can be used for data processing. For example, when it is necessary to process the digital twin data of a motor and monitor motor failures, the CNN algorithm can be used to identify the motor data twin data, so that it can quickly detect whether the motor is faulty, and it can also be used for motor debugging. When it is necessary to process the digital twin data of a robotic arm, it is possible to process operating parameters and image parameters, so they need to be processed separately. The YOLO algorithm can be used to process image parameters, and the CNN algorithm can be used to process operating parameters.

[0061] S3. Collect the real-time operating parameters corresponding to the digital twin model, and use the trained digital twin data processing model to process the real-time operating parameters to obtain parameter processing results;

[0062] After training, the digital twin data processing model has the ability to identify digital twin data. Therefore, it can collect real-time operating parameters for identification, thereby determining the operating status, thus obtaining the parameter processing results. For example, when processing motor digital twin data and needing to identify motor faults, the parameter processing results can be the normal motor category or the specific motor fault type, while the real-time operating parameters can be operating current, operating voltage, operating speed, operating vibration frequency, and so on.

[0063] S4. When the parameter processing results meet the preset requirements, corresponding labels and label-associated real-time operating parameters are generated on the digital twin model to complete the processing of the digital twin data.

[0064] For example, the preset requirements can set a specific fault type. When the parameter processing result is the same as the specific fault type, it proves that the motor has failed. Therefore, a label can be generated at a specific location on the motor digital twin model. Assuming that there is a problem with the rotor, a label can be generated on the rotor in the motor digital twin model. After clicking the label, a secondary menu can pop up. This secondary menu records the real-time operating parameters at the time of the fault, thereby improving the processing efficiency of the digital twin model data.

[0065] In an embodiment of the present invention, an intelligent optimization algorithm is used to train a digital twin data processing model to obtain a trained digital twin data processing model, including:

[0066] Initialize the hyperparameters of the digital twin data processing model to generate a population containing multiple parameter individuals;

[0067] Each hyperparameter has its corresponding upper and lower bounds, and can be randomly initialized between the upper and lower bounds to generate parameter individuals. After repeated initialization, a population can be generated. For example, when optimizing the hyperparameters of a CNN algorithm, its weight parameters can be initialized.

[0068] For each parameter individual in the population, apply the parameter individual to the digital twin data processing model, and use the cross entropy loss function to obtain the loss function value corresponding to the parameter individual;

[0069] According to the loss function value corresponding to the parameter individual, the optimal parameter individual is obtained, and the optimal parameter individual is updated using the normal distribution jump strategy to obtain the updated optimal parameter individual;

[0070] For any parameter individual, perform a local fast guided search on the parameter individual according to the optimal parameter individual to obtain the parameter individual after the local fast guided search;

[0071] For any parameter individual after local rapid guided search, multi-information learning search is performed on the parameter individual to obtain the parameter individual after multi-information learning search;

[0072] For any parameter individual after multi-information learning search, perform adaptive chain search on the parameter individual to obtain the parameter individual after adaptive chain search;

[0073] Determine whether the training end conditions are met. If so, obtain the trained digital twin data processing model based on the parameter individuals after the adaptive chain search. Otherwise, return to the step of obtaining the loss function value.

[0074] Optionally, after each search, out-of-bounds processing can be performed on individual parameters to ensure parameter validity.

[0075] In the prior art, there is a problem of easily falling into local optimality during the process of optimizing hyperparameters. Therefore, an embodiment of the present invention provides a new intelligent optimization algorithm to solve the problems existing in the prior art and improve the processing accuracy of digital twin data.

[0076] In an embodiment of the present invention, for each parameter individual in the population, the parameter individual is applied to the digital twin data processing model, and the cross entropy loss function is used to obtain the loss function value corresponding to the parameter individual, including:

[0077] For each parameter individual in the population, apply the parameter individual to the digital twin data processing model to obtain the digital twin data processing model after the parameters are applied;

[0078] Use historical operating parameters as input to the digital twin data processing model to obtain the actual output of the digital twin data processing model;

[0079] The data labels corresponding to the historical operating parameters are used as the expected output of the digital twin data processing model. Based on the expected output and actual output of the digital twin data processing model, the cross entropy loss function is used to obtain the loss function value corresponding to the parameter individual.

[0080] It is worth noting that in addition to using the cross entropy loss function to obtain the loss function value, you can also use functions such as root mean square loss and mean square loss to obtain the loss function value.

[0081] In an embodiment of the present invention, the optimal parameter individual is obtained according to the loss function value corresponding to the parameter individual, and the optimal parameter individual is updated using a normal distribution jump strategy to obtain the updated optimal parameter individual, including:

[0082] According to the loss function value corresponding to the parameter individual, the parameter individual with the smallest loss function value is determined as the optimal parameter individual;

[0083] The random jump factor generated by normal distribution is:

[0084]

[0085]

[0086]

[0087] According to the random jump factor, the optimal parameter individual is updated, and the optimal parameter individual after the update is obtained: ;in, Indicates the k The optimal parameter individual in the training process d dimensional parameters, represents the optimal parameter individual after update d dimensional parameters, represents the random jump factor, R represents the adaptive update coefficient, and , represents the first constant term, represents the first random parameter individual except the optimal parameter individual d dimensional parameters, represents the first random jump control factor, v represents the second random jump control factor, Represents a random number between (0,1), Satisfies normal distribution , Satisfies normal distribution , represents the first intermediate parameter, represents the second constant term and is set to 1; represents the gamma function.

[0088] The normal distribution jump strategy provided by the embodiment of the present invention can effectively perturb the optimal parameter individuals, thereby improving the global optimization ability of the algorithm and reducing the complexity of the algorithm.

[0089] Optionally, in the process of updating the optimal parameter individuals, a greedy algorithm or an annealing simulation algorithm may be used for control, thereby ensuring the convergence speed of the algorithm.

[0090] In the embodiment of the present invention, for any parameter individual, a local rapid guided search is performed on the parameter individual according to the optimal parameter individual to obtain the parameter individual after the local rapid guided search, including:

[0091] Generate random bootstrap control factors between (0,1) based on the standard normal distribution;

[0092] According to the random bootstrap control factor and the optimal parameter individual, a local fast bootstrap search is performed on the parameter individual, and the parameter individual after the local fast bootstrap search is obtained as follows: ;in, Indicates the k The mth parameter individual in the training process, represents the optimal parameter individual, represents the random bootstrap control factor, and ||*|| represents obtaining the Euclidean distance.

[0093] The local fast guided search provided by the embodiment of the present invention can enable all parameter individuals to quickly search around the optimal position, which can effectively improve the search speed of the algorithm.

[0094] In the embodiment of the present invention, for any parameter individual after the local rapid guided search, a multi-information learning search is performed on the parameter individual to obtain the parameter individual after the multi-information learning search, including:

[0095] For any parameter individual after local rapid guided search, randomly match the parameter individual with another parameter individual to obtain the paired parameter individual corresponding to each parameter individual;

[0096] According to the paired parameter individuals corresponding to the parameter individuals, the parameter individuals are studied and searched, and the first learning information is obtained as follows: ;in, Represents the first learning information, Indicates the k The nth parameter individual in the training process, Represents parameter individuals The corresponding pairing parameter individuals, represents the first learning coefficient, Represents a random number between (0,1);

[0097] According to the historical optimal value corresponding to the parameter individual, the parameter individual is studied and searched, and the second learning information is obtained as follows: ;in, represents the second learning information, Represents parameter individuals The corresponding historical optimal value is represents the second learning coefficient, Represents a random number between (0,1);

[0098] According to the optimal parameter individual, the parameter individual is studied and searched, and the third learning information is obtained as follows: ;in, represents the third learning information, represents the optimal parameter individual, represents the third learning coefficient, Represents a random number between (0,1);

[0099] According to the first learning information, the second learning information and the third learning information, the parameter individuals after the multi-information learning search are obtained as follows: ;in, Represents the parameter individual after multi-information learning search .

[0100] The multi-information learning search provided by the embodiment of the present invention can continuously move toward a better position that has been explored during the information interaction process, thereby improving the algorithm's search capability and local search capability.

[0101] In an embodiment of the present invention, for any parameter individual after multi-information learning search, an adaptive chain search is performed on the parameter individual to obtain the parameter individual after the adaptive chain search, including:

[0102] For any parameter individual after multi-information learning search, obtain the loss function value corresponding to the parameter individual, and arrange the parameter individuals in order from small to large according to the loss function value;

[0103] Based on the parameter individuals after arrangement, an adaptive chain search is performed on the parameter individuals, and the parameter individuals after the adaptive chain search are obtained as follows: ;in, Indicates the first i parameter individuals, Represents the parameter individual after adaptive chain search , Indicates the first i -1 parameter individual, and when i When 1, Set to The historical optimal value of represents the first chain search factor, represents the second chain search factor, represents the fourth random number between (0,1), represents the fifth random number between (0,1), represents the adaptive inertia weight, and ; represents the maximum value of the adaptive inertia weight, Represents the minimum value of the adaptive inertia weight, and K represents the preset maximum number of training times.

[0104] The adaptive chain search provided by the embodiment of the present invention can enable all parameter individuals to perform collaborative search and perform chain vortex search in the solution space, thereby improving the solution space search capability and thus enhancing the global search capability.

[0105] In an embodiment of the present invention, when the parameter processing result meets the preset requirements, a corresponding label and label-associated real-time operating parameters are generated on the digital twin model, including:

[0106] When the parameter processing result shows that an anomaly exists, a label is generated on the digital twin model, and the real-time operating parameters are associated on the secondary interface of the label.

[0107] like Figure 2 As shown, an embodiment of the present invention provides a digital twin data processing system based on artificial intelligence, comprising: a digital twin module 1, a model building module 2, a parameter processing module 3 and a result processing module 4;

[0108] The digital twin module 1 is used to build a digital twin model corresponding to the target machine device and synchronize the operating parameters of the target machine device to the digital twin model to run the digital twin model;

[0109] The model construction module 2 is used to construct a digital twin data processing model using an artificial intelligence algorithm, and train the digital twin data processing model using an intelligent optimization algorithm to obtain a trained digital twin data processing model;

[0110] The parameter processing module 3 is used to collect the real-time operating parameters corresponding to the digital twin model, and use the trained digital twin data processing model to process the real-time operating parameters to obtain parameter processing results;

[0111] The result processing module 4 is used to generate corresponding tags and tag-associated real-time operating parameters on the digital twin model when the parameter processing results meet the preset requirements, thereby completing the processing of the digital twin data.

[0112] An embodiment of the present invention provides an artificial intelligence-based digital twin data processing system that can execute the above-mentioned method and technical solution. Its principles and beneficial effects are similar and will not be repeated here.

[0113] The present invention provides an artificial intelligence-based digital twin data processing method and system, which constructs a digital twin data processing model by adopting an artificial intelligence algorithm, and trains the digital twin data processing model by adopting an intelligent optimization algorithm to obtain a trained digital twin data processing model. Then, during the operation of the digital twin model, the trained digital twin data processing model is used to process the real-time operating parameters of the digital twin model, thereby realizing the automatic processing of digital twin data and effectively improving the processing efficiency and accuracy of digital twin data.

[0114] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0115] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0116] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0117] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1The steps for the function specified in one or more boxes.

[0118] Those skilled in the art will understand that all or part of the steps in implementing the above facts and methods can be completed by instructing relevant hardware through a program, and the program involved or the program can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: the corresponding method steps are then brought out, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0119] 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. A digital twin data processing method based on artificial intelligence, characterized in that: include: Build a digital twin model corresponding to the target machine device and synchronize the operating parameters of the target machine device to the digital twin model to run the digital twin model; Use artificial intelligence algorithms to build a digital twin data processing model, and use intelligent optimization algorithms to train the digital twin data processing model to obtain the trained digital twin data processing model; Collect the real-time operating parameters corresponding to the digital twin model, and use the trained digital twin data processing model to process the real-time operating parameters to obtain parameter processing results; When the parameter processing results meet the preset requirements, corresponding tags and tag-associated real-time operating parameters are generated on the digital twin model to complete the processing of the digital twin data; The digital twin data processing model is trained using an intelligent optimization algorithm to obtain the trained digital twin data processing model, including: Initialize the hyperparameters of the digital twin data processing model to generate a population containing multiple parameter individuals; For each parameter individual in the population, apply the parameter individual to the digital twin data processing model, and use the cross entropy loss function to obtain the loss function value corresponding to the parameter individual; According to the loss function value corresponding to the parameter individual, the optimal parameter individual is obtained, and the optimal parameter individual is updated using the normal distribution jump strategy to obtain the updated optimal parameter individual; For any parameter individual, perform a local fast guided search on the parameter individual according to the optimal parameter individual to obtain the parameter individual after the local fast guided search; For any parameter individual after local rapid guided search, multi-information learning search is performed on the parameter individual to obtain the parameter individual after multi-information learning search; For any parameter individual after multi-information learning search, perform adaptive chain search on the parameter individual to obtain the parameter individual after adaptive chain search; Determine whether the training end conditions are met. If so, obtain the trained digital twin data processing model based on the parameter individuals after the adaptive chain search. Otherwise, return to the step of obtaining the loss function value.

2. The digital twin data processing method based on artificial intelligence according to claim 1, characterized in that: Use artificial intelligence algorithms to build digital twin data processing models, including: using CNN algorithm, R-CNN algorithm, Fast R-CNN algorithm, YOLO algorithm and / or BP Net algorithm to build digital twin data processing models.

3. The digital twin data processing method based on artificial intelligence according to claim 1 is characterized in that: For each parameter individual in the population, the parameter individual is applied to the digital twin data processing model, and the cross entropy loss function is used to obtain the loss function value corresponding to the parameter individual, including: For each parameter individual in the population, apply the parameter individual to the digital twin data processing model to obtain the digital twin data processing model after the parameters are applied; Use historical operating parameters as input to the digital twin data processing model to obtain the actual output of the digital twin data processing model; The data labels corresponding to the historical operating parameters are used as the expected output of the digital twin data processing model. Based on the expected output and actual output of the digital twin data processing model, the cross entropy loss function is used to obtain the loss function value corresponding to the parameter individual.

4. The digital twin data processing method based on artificial intelligence according to claim 3 is characterized in that: According to the loss function value corresponding to the parameter individual, the optimal parameter individual is obtained, and the normal distribution jump strategy is used to update the optimal parameter individual to obtain the updated optimal parameter individual, including: According to the loss function value corresponding to the parameter individual, the parameter individual with the smallest loss function value is determined as the optimal parameter individual; The random jump factor generated by normal distribution is: According to the random jump factor, the optimal parameter individual is updated, and the optimal parameter individual after the update is obtained: ;in, Indicates the k The optimal parameter individual in the training process d dimensional parameters, represents the optimal parameter individual after update d dimensional parameters, represents the random jump factor, R represents the adaptive update coefficient, and , represents the first constant term, represents the first random parameter individual except the optimal parameter individual d dimensional parameters, represents the first random jump control factor, v represents the second random jump control factor, Represents a random number between (0,1), Satisfies normal distribution , Satisfies normal distribution , represents the first intermediate parameter, represents the second constant term and is set to 1; represents the gamma function.

5. The artificial intelligence-based digital twin data processing method according to claim 4, characterized in that: For any parameter individual, a local fast guided search is performed on the parameter individual according to the optimal parameter individual to obtain the parameter individual after the local fast guided search, including: Generate random bootstrap control factors between (0,1) based on the standard normal distribution; According to the random bootstrap control factor and the optimal parameter individual, a local fast bootstrap search is performed on the parameter individual, and the parameter individual after the local fast bootstrap search is obtained as follows: ;in, Indicates the k The mth parameter individual in the training process, represents the optimal parameter individual, represents the random bootstrap control factor, and ||*|| represents obtaining the Euclidean distance.

6. The artificial intelligence-based digital twin data processing method according to claim 5, characterized in that: For any parameter individual after local rapid guided search, multi-information learning search is performed on the parameter individual to obtain the parameter individual after multi-information learning search, including: For any parameter individual after local rapid guided search, randomly match the parameter individual with another parameter individual to obtain the paired parameter individual corresponding to each parameter individual; According to the paired parameter individuals corresponding to the parameter individuals, the parameter individuals are studied and searched, and the first learning information is obtained as follows: ;in, Represents the first learning information, Indicates the k The nth parameter individual in the training process, Represents parameter individuals The corresponding pairing parameter individuals, represents the first learning coefficient, Represents a random number between (0,1); According to the historical optimal value corresponding to the parameter individual, the parameter individual is studied and searched, and the second learning information is obtained as follows: ;in, represents the second learning information, Represents parameter individuals The corresponding historical optimal value is represents the second learning coefficient, Represents a random number between (0,1); According to the optimal parameter individual, the parameter individual is studied and searched, and the third learning information is obtained as follows: ;in, represents the third learning information, represents the optimal parameter individual, represents the third learning coefficient, Represents a random number between (0,1); According to the first learning information, the second learning information and the third learning information, the parameter individuals after the multi-information learning search are obtained as follows: ;in, Represents the parameter individual after multi-information learning search .

7. The artificial intelligence-based digital twin data processing method according to claim 6, characterized in that: For any parameter individual after multi-information learning search, perform adaptive chain search on the parameter individual to obtain the parameter individual after adaptive chain search, including: For any parameter individual after multi-information learning search, obtain the loss function value corresponding to the parameter individual, and arrange the parameter individuals in order from small to large according to the loss function value; Based on the parameter individuals after arrangement, an adaptive chain search is performed on the parameter individuals, and the parameter individuals after the adaptive chain search are obtained as follows: ;in, Indicates the first i parameter individuals, Represents the parameter individual after adaptive chain search , Indicates the first i -1 parameter individual, and when i When 1, Set to The historical optimal value of represents the first chain search factor, represents the second chain search factor, represents the fourth random number between (0,1), represents the fifth random number between (0,1), represents the adaptive inertia weight, and ; represents the maximum value of the adaptive inertia weight, Represents the minimum value of the adaptive inertia weight, and K represents the preset maximum number of training times.

8. The artificial intelligence-based digital twin data processing method according to claim 1, characterized in that: When the parameter processing results meet the preset requirements, corresponding tags and tag-associated real-time operating parameters are generated on the digital twin model, including: When the parameter processing result shows that an anomaly exists, a label is generated on the digital twin model, and the real-time operating parameters are associated on the secondary interface of the label.

9. An artificial intelligence-based digital twin data processing system, which is capable of executing the artificial intelligence-based digital twin data processing method according to any one of claims 1 to 8, characterized in that: include: Digital twin module, model building module, parameter processing module and result processing module; The digital twin module is used to build a digital twin model corresponding to the target machine device and synchronize the operating parameters of the target machine device to the digital twin model to run the digital twin model; The model construction module is used to construct a digital twin data processing model using an artificial intelligence algorithm, and to train the digital twin data processing model using an intelligent optimization algorithm to obtain a trained digital twin data processing model; The parameter processing module is used to collect real-time operating parameters corresponding to the digital twin model and process the real-time operating parameters using the trained digital twin data processing model to obtain parameter processing results; The result processing module is used to generate corresponding tags and tag-associated real-time operating parameters on the digital twin model when the parameter processing results meet the preset requirements, thereby completing the processing of the digital twin data.

Images