Digital twinning-based electromechanical system heterogeneous domain generalization fault diagnosis method and device

Abstract

The invention discloses an electromechanical system heterogeneous domain generalization fault diagnosis method and device based on digital twinning, and belongs to the technical field of electric data processing. The method comprises the following steps: constructing a digital twin model to simulate entity characteristics of an electromechanical system, generating virtual fault data and structured interconnection knowledge distributed across working conditions through the digital twin model, and reconstructing a K-class information domain according to the virtual fault data and entity data of the electromechanical system; the information source is divided into a source domain and a target domain, the source domain represents visible operation conditions, and the target domain represents invisible operation conditions; performing structured interconnection knowledge embedding on the source domains to obtain a knowledge embedding data set of the K-class source domains; and constructing a heterogeneous domain generalization network, and training the heterogeneous domain generalization network by utilizing the knowledge embedding data set of the K-type source domain and the K-type target domain data set to obtain a fault diagnosis model so as to perform fault diagnosis on invisible operation condition information to be diagnosed. According to the invention, fault diagnosis can be carried out on the electromechanical system under the limitation of invisible operation conditions.

Description

Technical Field

[0001] This invention relates to a method and apparatus for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins, belonging to the field of electrical data processing technology. Background Technology

[0002] Two common problems in fault diagnosis are the data imbalance between normal and fault states, and the varying data distribution across operating conditions. To address these issues, a number of variations on data generation and transfer learning have emerged, most of which follow a single data-driven paradigm. While these methods may be effective for entities with relatively simple fault modes and operating conditions (such as bearings and gearboxes), they are quite challenging for electromechanical systems (EMS). The extended fault modes and diverse operating conditions in EMS result in severe scarcity and significant differences in fault data across operating conditions, even leading to unseen operating conditions with no fault data.

[0003] There are no publicly available methods, systems, or devices for diagnosing faults in invisible operating conditions. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, the purpose of this invention is to provide a method and apparatus for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins, which can perform fault diagnosis on electromechanical systems under invisible operating conditions.

[0005] To achieve the stated objectives, this invention provides a method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins, comprising: S01: Construct a full-system digital twin model to simulate the physical characteristics of the electromechanical system; S02: Virtual fault data and structured interconnected knowledge distributed across operating conditions are generated through a full-system digital twin model. K types of information domains are reconstructed based on the virtual fault data and electromechanical system entity data. Each information domain represents an operating condition of an electromechanical system entity. The information source is divided into source domain and target domain. The source domain represents visible operating conditions, and the target domain represents invisible operating conditions. Structured interconnected knowledge is embedded into the source domain to obtain a knowledge embedding dataset of K types of source domains. S03: Construct a heterogeneous domain generalization network, and train the heterogeneous domain generalization network using the knowledge embedding dataset of K types of source domains and the dataset of K types of target domains to obtain a fault diagnosis model; S04: Perform fault diagnosis on the invisible operating condition information to be diagnosed using the fault diagnosis model.

[0006] To achieve the stated objectives, this invention provides a heterogeneous domain generalized fault diagnosis device for electromechanical systems based on digital twins. The device includes: a full-system digital twin model for simulating the physical characteristics of the electromechanical system; a training dataset acquisition unit that generates virtual fault data and structured interconnected knowledge distributed across operating conditions using the full-system digital twin model, and reconstructs K types of information domains based on the virtual fault data and electromechanical system entity data. Each information domain represents an operating condition of the electromechanical system entity. The information sources are divided into source domains and target domains, where the source domains represent visible operating conditions and the target domains represent invisible operating conditions. Structured interconnected knowledge is embedded into the source domains to obtain a knowledge embedding dataset of K types of source domains; and a fault diagnosis model that uses the knowledge embedding dataset of K types of source domains and the dataset of K types of target domains to train a heterogeneous domain generalization network for fault diagnosis of invisible operating condition information.

[0007] Compared with existing technologies, this invention provides a heterogeneous domain generalized fault diagnosis method and apparatus for electromechanical systems based on digital twins. Driven by digital twins, it generates two modalities of information: virtual fault data and structured interconnection knowledge distributed across operating conditions. This not only enriches the scale of the available dataset but also the types of available datasets. Based on this, a heterogeneous domain generalization network (HDGN) is constructed to achieve generalized fault diagnosis from both virtual fault data and structured interconnection knowledge distributed across operating conditions. By embedding prior structured interconnection knowledge distributed across operating conditions, the domain-invariant consistency of the virtual fault data is stably maintained. Driven by specificity and latent feature similarity loss, multi-channel domain-specific discriminative representations are adaptively learned from the virtual fault data embedded with domain-invariant knowledge. With the help of digital twin-integrated HDGN, the generalization ability for invisible operating conditions is gradually improved, exhibiting good stability and adaptability. This invention can effectively improve the accuracy of imbalance fault diagnosis under invisible operating conditions. Attached Figure Description

[0008] Figure 1 This is a flowchart of the heterogeneous domain generalization fault diagnosis method for electromechanical systems based on digital twins provided by the present invention. Detailed Implementation

[0009] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0010] First Embodiment

[0011] Figure 1 This is a flowchart of the heterogeneous domain generalization fault diagnosis method for electromechanical systems based on digital twins provided by the present invention, such as... Figure 1 As shown, the present invention provides a method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins, comprising: S01: Construct a full-system digital twin model to simulate the physical characteristics of the electromechanical system; S02: Virtual fault data and structured interconnected knowledge distributed across operating conditions are generated through a full-system digital twin model. K types of information domains are reconstructed based on the virtual fault data and electromechanical system entity data. Each information domain represents an operating condition of an electromechanical system entity. The information source is divided into source domain and target domain. The source domain represents visible operating conditions, and the target domain represents invisible operating conditions. Structured interconnected knowledge is embedded into the source domain to obtain a knowledge embedding dataset of K types of source domains. S03: Construct a heterogeneous domain generalization network, and train the heterogeneous domain generalization network using the knowledge embedding dataset of K types of source domains and the dataset of K types of target domains to obtain a fault diagnosis model; S04: Perform fault diagnosis on the invisible operating condition information to be diagnosed using the fault diagnosis model.

[0012] In the first embodiment, the entity description of the electromechanical system (EMS) is divided into three types: element description (ED), data description (DD), and relation description (RD).

[0013] The element descriptions (EDs) of an EMS entity include: structural dimension elements. Physical dimension elements Operating condition dimension elements and fault dimension elements The intrinsic mechanism elements describing the characteristics of EMS entities, including structural dimension elements. Includes: structural components and structural details Physical dimension elements Including control principles Electromagnetic principles Mechanical principles and fluid principles Working condition dimension elements Including the power generation process Power transmission process and the process of running under load Fault dimension elements Including fault location Cause of the malfunction Impact of the fault and symptoms of failure .

[0014] The data description (DD) of an EMS entity includes: dynamic data description. and static data description This describes sensor measurement-related information for measuring EMS entities, including dynamic data description. Includes: rapidly changing data and slowly changing data Static data description Includes: sensor name Sensor position Sensor sampling frequency Sensor base and signal preprocessing methods .

[0015] EMS entity data relationship description (RD) includes: inline relationships and interconnection Among them, internal relationships Includes: structural dimension elements and physical dimension elements Coupling between Describes the intrinsic relationship between spatial structure and physics; structural dimension elements Physical dimension elements and working condition dimension elements Coupling between This describes how to integrate the physical behaviors of some unit-level entities into functional domain-level operating conditions; structural dimension elements. Physical dimension elements Operating condition dimension elements and fault dimension elements Coupling between This describes how a normally functioning entity evolves into a faulty entity through coupling with faulty elements. Interconnection relationships. This includes the mutual coupling relationship between the power functional domain and the transmission functional domain. The mutual coupling relationship between the transmission functional domain and the load functional domain The mutual coupling relationship between the load functional domain and the power functional domain .

[0016] In the first embodiment, a full-system digital twin model of the electromechanical system is constructed using a hierarchical collaborative modeling method, including: S01-1: Construct digital twin models sequentially for the functional domains of the electromechanical system. , respectively representing the spatial level sub-model, behavioral level sub-model, process level sub-model, and state level sub-model; for the i-th level sub-model First, inherit from the previous level. As a baseline model.

[0017] In the first embodiment, the spatial sub-model ,in, Representation of spatial level submodel The main body, for a given structural dimension element Using spatial geometric modeling algorithms Mechanistic models for constructing spatial-level sub-models to statically reflect the structural characteristics of entities, such as Boolean modeling and spline curve modeling; data-driven reverse engineering algorithms. It involves quantitatively constructing numerical space models, such as point cloud scanning and raster projection; and utilizing fusion algorithms. , Will be The geometric dimensions are updated and optimized to achieve a more objective digital representation of the physical spatial structure.

[0018] In the first embodiment, the behavior-level sub-model Inherited spatial submodel : , Represents behavioral sub-models The main body, for physical elements Employing multidisciplinary modeling algorithms Achieving physical virtualization, such as finite element analysis and differential equations; modifying data-driven algorithms using parameter optimization techniques. Parameters, such as least squares method; using a fusion algorithm based on steady-state analysis. right and Integrate.

[0019] In the first embodiment, the process-level sub-model is the third level of the hierarchical sub-model. It enables the interaction between local behavioral-level sub-models to virtualize the operational conditions of entities. Process-level sub-model for: It describes multiple local behavior-level sub-models. The logical interaction between them. Given logical operating condition elements, the mechanism-based element-driven process algorithm... Different disciplines These are connected to form operating conditions, such as bond graphs and Petri network models. Given transient data elements across different operating conditions, Used to update different The transfer parameters are used to achieve more accurate process transfer. Unlike steady-state analysis, The key focus is and The transient fusion between them focuses on solving and optimizing across working conditions. Its time-varying characteristics.

[0020] In the first embodiment, the state-level sub-model M4 is the molecular model at the last level, which virtualizes functional domain entities under different fault modes through fault physics and deep generation models: ,in, Represents the main body of the state-level sub-model; fault physics-driven algorithm According to the fault element Fault injection can be performed, such as by using slotting deformation on the process-level sub-model driven by the finite element method. Fault injection is performed. (Deep generation algorithm) Intrinsic fault characteristics are learned directly from the entity's measurement data, such as through generative adversarial networks and variational encoders. Fusion algorithms... Utilizing various pre-training fine-tuning and transfer learning techniques to the above and Optimize.

[0021] In the first embodiment, constructing a full-system digital twin model of the electromechanical system using a hierarchical collaborative modeling method further includes: S01-2: Using the functional modeling interface to integrate digital twin models Each is encapsulated into a compatible model functional unit. In the middle; and the model functional units Integrating into the digital twin model. In the first embodiment, guided by the description of the interconnection relationships between the three core functional domains—power functional domain, transmission functional domain, and load functional domain—[the following is done]: Integrated into the digital twin model, where, The mathematical description is as follows: , in, , These are sub-models representing the electromechanical system entities in the power functional domain, transmission functional domain, and load functional domain, respectively. express The set of states of the running state; They represent The set of input variables and the set of output variables; Represent the initial state; using extrapolation equations , For driving from arrive The equation is performed. For time intervals; Let be the state variable at time t. Let be the input variable at time t. This is the equation for generating virtual data output at time t; It is the solver that drives the equations.

[0022] In the first embodiment, each sub-model is encapsulated as a model functional unit (FMU), which includes a sub-digital twin model, a sub-environment, and a solver. Guided by interconnection, the whole system digital twin model is integrated from the FMUs of three core functional domain entities, which include interactive variables and collaborative environments. With the help of timing orchestration algorithms (such as Gauss-Seidel), the entire system can be iteratively driven to generate virtual data in timing fault states and operating states.

[0023] The heterogeneous domain generalization fault diagnosis method for electromechanical systems based on digital twins provided by this invention also includes: S05: Consistency between the electromechanical system entity and the digital twin model is ensured through a consistency metric. This consistency metric is achieved by comparing the entity data of the electromechanical system measured by sensors with the virtual data generated by the digital twin model. It can be implemented through distance measurement, similarity measurement, etc. If the consistency metric result exceeds a predefined threshold, a consistency update of the digital twin model will be initiated.

[0024] The heterogeneous domain generalization fault diagnosis method for electromechanical systems based on digital twins provided by this invention also includes: S06: Updating the digital twin model through consistent updates: for spatial sub-models Reverse engineering methods are used to reconstruct the spatial geometric parameters of the computer electrical system entity, generating a new spatial-level sub-model. : , In the formula, and These are the parameters before and after the spatial-level sub-model update, respectively.

[0025] This invention updates the behavioral-level sub-model using a constrained parameter estimation algorithm based on steady-state data. Generate new behavioral sub-models : , In the formula, , and yes The upper and lower limits; and These are the steady-state data and behavioral sub-models of the electromechanical system entity. The generated data.

[0026] This invention updates the process-level sub-model using a constrained parameter estimation algorithm based on transient data. Generate new process-level sub-models : , In the formula, , and yes The upper and lower limits; and These are entity data and process-level sub-models of electromechanical systems under fault conditions. The generated data.

[0027] This invention employs an implicit deep generative model and uses the gradient descent algorithm to update the state-level sub-models. Generate new state-level sub-models .

[0028] Furthermore, the system-wide digital twin model itself is a highly structured knowledge container from which knowledge related to fault propagation can be extracted, such as spatial structure knowledge, physical mechanism knowledge, state process knowledge, and mutual coupling knowledge. Digital twin model knowledge describes the potential impact of fault propagation within the electromechanical system from multiple dimensions.

[0029] In the first embodiment, given a full-system digital twin model, multiple information domains are reconstructed using two modalities: structured interconnected knowledge distributed across operating conditions and virtual fault data. This supports HDGN in generalizing from multi-source domains to invisible target domains. The reconstructed source domains are represented as follows: , In the formula, and These represent the source domain based on electromechanical system entities and the source domain based on virtual fault data, respectively. The structured interconnection knowledge, extracted from the power sub-digital twin model, the transmission sub-digital twin model, and the load sub-digital twin model, describes the distribution pattern of sensor data and the operating conditions of each entity's functional area. ,in Based on the operating conditions of the electromechanical system entity, ,in Operating conditions based on DT; ,in, for One of the samples; ,in, for One of the samples; Greater than or equal to .

[0030] In the first embodiment, by embedding structured interconnected knowledge distributed across operating conditions into the source domain, the domain difference components in the numerical data of different domains are stably eliminated, and the domain invariant consistency is maintained, thereby stimulating the ability of the fault diagnosis model to perform subsequent data-driven generalization learning.

[0031] This invention conducts fault diagnosis based on HDGN under invisible operating conditions, and maintains domain-invariant consistency in the embedding of structured interconnection knowledge distributed across operating conditions. Based on the power sub-digital twin model, the transmission sub-digital twin model, and the load sub-digital twin model, it extracts structured interconnection knowledge distributed across operating conditions. .

[0032] Based on the extracted structured interconnect knowledge distributed across operating conditions This transforms the data into embeddable, cross-condition distributed structured interconnected knowledge to eliminate condition-sensitive components and preserve domain-invariant consistency. By embedding cross-condition distributed structured interconnected knowledge, entity source domain data is preserved. and digital twin source domain data Domain-invariant consistency in the training process stimulates specific discriminative learning.

[0033] In the first embodiment, the fault diagnosis model is obtained by training a heterogeneous domain generalization network using a K-class source domain knowledge embedding dataset and a K-class target domain dataset, including: S3-1-01: Obtain the domain-specific training set of the k-th type of source domain knowledge embedding data , , The k-th source domain is respectively the first Each source domain knowledge is embedded into the training data and its real-world domain specificity. ; S3-1-02: From Domain-Specific Training Sets Obtain a training data set ; S3-1-03: Obtain from the k-th class of heterogeneous source domain learners Estimated specific features , These are the parameters to be optimized for the k-th domain-specific learner. S3-1-04: According to and Calculate the loss function Determine the loss function If it is the minimum, output the optimal parameter. Otherwise, according to the loss function Update parameters Return to step S3-1-02 and start from the training set. Extract a training data set and continue training the domain-specific learner for the k-th class. S3-1-05: Repeat steps S3-1-01 to S3-1-04 to obtain a K-class domain-specific learner with better performance. , , These are the embedded data of the k-th source domain knowledge and the estimated specific features extracted by the k-th domain-specific learner, respectively.

[0034] In the first embodiment, the method of training a heterogeneous domain generalization network using knowledge embedding data from K types of source domains and K types of target domains to obtain a fault diagnosis model further includes: S3-1-06: Constructing Domain-Specific Penalty Loss : , In the formula, Let be the covariance matrix of the discriminant features of the k-th source domain; It is the covariance matrix of the discriminant features of the j-th source domain. For norm dimension; It is the Frobenius norm; and Domain-specific training sets domain-specific training set The number of samples; It is a diagonal identity matrix; , ; S3-1-07: Determine the loss If the minimum is found, output the optimal parameters. Then proceed to step S3-1-08; otherwise, based on the loss... Update parameters Return to step S3-1-02, and then from the training set Take another set of training data; S3-1-08: Repeat steps S3-1-01 to S3-1-07 to obtain the K-class domain-specific optimization learner. , , These represent the embedded data of the k-th source domain knowledge and the estimated specific features extracted by the k-th domain-specific optimization learner, respectively.

[0035] In the first embodiment, the method of training a heterogeneous domain generalization network using a K-class source domain knowledge embedding dataset and a K-class target domain dataset to obtain a fault diagnosis model further includes: S3-2-01: Obtain the implicit feature training set of the k-th target domain , , The k-th target domain is the first One training data and the true implicit features; S3-2-02: From the implicit feature training set Obtain a training data set ; S3-2-03: Obtain from the k-th class implicit feature learner Estimation of implicit features , For the parameters to be optimized in the k-th class of implicit feature learners; S3-2-04: According to and Calculate the loss function Determine the loss function If it is the minimum, output the optimal parameter. Otherwise, according to the loss function Update parameters Then return to step S3-2-02, from the training set Take another set of training data and continue training the implicit feature learner. The normalized activation function; S3-2-05: Repeat steps S3-2-01 to S3-2-04 to obtain the K-class target domain implicit feature optimization learner. , , These represent the target domain data of class k and the estimated implicit features extracted by the learner based on the implicit features of class k, respectively.

[0036] In the first embodiment, the method of training a heterogeneous domain generalization network using a K-class source domain knowledge embedding dataset and a K-class target domain dataset to obtain a fault diagnosis model further includes: S3-2-06: Based on domain learner Obtain each and The domain is then normalized to obtain the normalized domain: and , Normalized activation function; These are the input features and parameters to be optimized for the domain learner, respectively. S3-2-07 Constructing a Domain Similarity Penalty Loss Function : , In the formula, The cross-entropy loss function; S3-2-08: Determine the loss function If the minimum value is found, output the optimized parameters of the domain learner. Otherwise, based on the loss function Update the parameters of the domain learner Then, the updated parameters are used to replace the original parameters, and the process returns to step S3-2-06 to continue training the domain learner.

[0037] In the first embodiment, the method of training a heterogeneous domain generalization network using K types of source domain knowledge embedding data and K types of target domain data to obtain a fault diagnosis model further includes: S3-3-01 is based on the following formula: and Weighting: , In the formula, and As weight.

[0038] In the first embodiment, the method of training a heterogeneous domain generalization network using K types of source domain knowledge embedding data and K types of target domain data to obtain a fault diagnosis model further includes: S3-3-02: Obtain the training set for source domain fault identification of the k-th class. , The true state label of the k-th source domain; S3-3-03: From the source domain fault identification training set Obtain a training data set ; S3-3-04: Obtain from the k-th type of source domain fault learner Estimated state label , The parameters to be optimized are those of the fault learner for the k-th source domain. S3-3-05: Calculate the weighted aggregation state label according to the following formula: , In the formula, , S3-3-06: According to and Calculate the loss function Determine the loss function If it is the minimum, output the optimal parameter. Output the fault learner for the k-th target domain. Otherwise, according to the loss function renew Then return to step S3-3-02, from the training set Take another piece of data and continue training the fault learner for the k-th target domain. S3-3-07: Repeat steps S3-3-02 to S3-3-06 to obtain the K-class target domain fault identification optimization learner.

[0039] In the first embodiment, fault diagnosis of the target domain using a fault diagnosis model includes: S4-1: Obtain invisible operating condition information according to the following formula Estimated implicit features: , S4-2: Obtain the implicit features according to the following formula Normalization domain: ; S4-3: Estimate the state label according to the following formula: ; S4-4: Obtain the weighted aggregation state label according to the following formula: ; S4-5: Obtain the estimated fault state label for invisible operating condition information according to the following formula: .

[0040] The first embodiment of the present invention also provides a system including a storage medium and one or more processors, the storage medium being used to store a computer program that can be invoked by one or more processors to implement the above-described method.

[0041] The first embodiment of the present invention also provides a computer program product, which is program code written in a computer language according to the above method, and the program code can be called and executed by one or more processors.

[0042] The first embodiment of this invention provides a heterogeneous domain generalization fault diagnosis method for electromechanical systems based on digital twins. Driven by digital twins, it generates two modalities of information: virtual fault data and structured interconnection knowledge distributed across operating conditions. This not only enriches the scale but also the types of available datasets. Based on this, a High-Quality Domain Generalization Network (HDGN) is constructed to achieve generalized fault diagnosis from both virtual fault data and structured interconnection knowledge distributed across operating conditions. By embedding prior structured interconnection knowledge distributed across operating conditions, the domain-invariant consistency of the virtual fault data is stably maintained. Driven by specificity and latent feature similarity loss, multi-channel domain-specific discriminative representations are adaptively learned from the virtual fault data embedded with domain-invariant knowledge. With the help of digital twin-integrated HDGN, the generalization ability for invisible operating conditions is gradually improved, exhibiting good stability and adaptability. This invention can effectively improve the accuracy of imbalance fault diagnosis under invisible operating conditions.

[0043] Second Embodiment

[0044] The second embodiment of the present invention only describes the content that differs from the first embodiment; the same content will not be repeated.

[0045] The second embodiment of the present invention provides a heterogeneous domain generalized fault diagnosis device for electromechanical systems based on digital twins, comprising: a full-system digital twin model for simulating the entity characteristics of the electromechanical system; a training dataset acquisition unit that generates virtual fault data and structured interconnected knowledge distributed across operating conditions through the full-system digital twin model, and reconstructs K types of information domains based on the virtual fault data and electromechanical system entity data, each information domain representing an operating condition of the electromechanical system entity, dividing the information source into source domains and target domains, wherein the source domains represent visible operating conditions and the target domains represent invisible operating conditions; embedding structured interconnected knowledge into the source domains to obtain a knowledge embedding dataset of K types of source domains; and a fault diagnosis model that uses the knowledge embedding dataset of K types of source domains and the dataset of K types of target domains to train a heterogeneous domain generalization network for fault diagnosis of invisible operating condition information.

[0046] The beneficial effects of the second embodiment of the present invention are the same as those of the first embodiment, and will not be described again here.

[0047] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins, characterized in that, include: S01: Construct a full-system digital twin model to simulate the physical characteristics of the electromechanical system; S02: Virtual fault data and structured interconnected knowledge distributed across operating conditions are generated through a full-system digital twin model. K types of information domains are reconstructed based on the virtual fault data and electromechanical system entity data. Each information domain represents an operating condition of an electromechanical system entity. The information source is divided into a source domain and a target domain. The source domain represents visible operating conditions, and the target domain represents invisible operating conditions. By performing structured interconnection knowledge embedding on the source domains, a knowledge embedding dataset of K types of source domains is obtained; S03: Construct a heterogeneous domain generalization network, and train the heterogeneous domain generalization network using the knowledge embedding dataset of K types of source domains and the dataset of K types of target domains to obtain a fault diagnosis model; S04: Perform fault diagnosis on the invisible operating condition information to be diagnosed using the fault diagnosis model.

2. The method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins according to claim 1, characterized in that, The whole system digital twin model includes a power digital twin sub-model, a transmission digital twin sub-model, and a load digital twin sub-model.

3. The method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins according to claim 1, characterized in that, The fault diagnosis model obtained by training a heterogeneous domain generalization network using a K-class source domain knowledge embedding dataset and a K-class target domain dataset includes: S3-1-01: Obtain the domain-specific training set of the k-th type of source domain knowledge embedding data , , The k-th source domain is respectively the first Each source domain knowledge is embedded into the training data and its real-world domain specificity. ; S3-1-02: From Domain-Specific Training Sets Obtain a training data set ; S3-1-03: Obtain from the k-th class of heterogeneous source domain learners Estimated specific features , These are the parameters to be optimized for the k-th domain-specific learner; S3-1-04: According to and Calculate the loss function Determine the loss function If it is the minimum, output the optimal parameter. Otherwise, according to the loss function Update parameters Return to step S3-1-02 and start from the training set. Extract a training data set and continue training the domain-specific learner for the k-th class. S3-1-05: Repeat steps S3-1-01 to S3-1-04 to obtain a K-class domain-specific learner with better performance. , , These are the embedded data of the k-th source domain knowledge and the estimated specific features extracted by the k-th domain-specific learner, respectively.

4. The method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins according to claim 3, characterized in that, The fault diagnosis model obtained by training a heterogeneous domain generalization network using knowledge embedding data from K types of source domains and K types of target domains also includes: S3-1-06: Constructing a Domain-Specific Penalty Loss Function : ， In the formula, Let be the covariance matrix of the discriminant features of the k-th source domain; It is the covariance matrix of the discriminant features of the j-th source domain. For norm dimension; It is the Frobenius norm; and Domain-specific training sets domain-specific training set The number of samples; It is a diagonal identity matrix; , ; S3-1-07: Determine the loss function If the minimum is found, output the optimal parameters. Then proceed to step S3-1-08; otherwise, based on the loss function... Update parameters Return to step S3-1-02, and then from the training set Take another set of training data; S3-1-08: Repeat steps S3-1-01 to S3-1-07 to obtain the K-class domain-specific optimization learner. , , These represent the embedded data of the k-th source domain knowledge and the estimated specific features extracted by the k-th domain-specific optimization learner, respectively.

5. The method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins according to claim 4, characterized in that, The fault diagnosis model obtained by training a heterogeneous domain generalization network using a K-class source domain knowledge embedding dataset and a K-class target domain dataset also includes: S3-2-01: Obtain the implicit feature training set of the k-th target domain , , The k-th target domain is the first One training data and the true implicit features; S3-2-02: From the implicit feature training set Obtain a training data set ; S3-2-03: Obtain from the k-th class implicit feature learner Estimation of implicit features , For the parameters to be optimized in the k-th class of implicit feature learners; S3-2-04: According to and Calculate the loss function Determine the loss function If it is the minimum, output the optimal parameter. Otherwise, according to the loss function Update parameters Then return to step S3-2-02, from the training set Take another set of training data and continue training the implicit feature learner. The normalized activation function; S3-2-05: Repeat steps S3-2-01 to S3-2-04 to obtain the K-class target domain implicit feature optimization learner. , , These represent the target domain data of class k and the estimated implicit features extracted by the learner based on the implicit features of class k, respectively.

6. The method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins according to claim 5, characterized in that, The fault diagnosis model obtained by training a heterogeneous domain generalization network using a K-class source domain knowledge embedding dataset and a K-class target domain dataset also includes: S3-2-06: Based on domain learner Obtain each and The domain is then normalized to obtain the normalized domain: and , Normalized activation function; These are the input features and parameters to be optimized for the domain learner, respectively. S3-2-07 Constructing a Domain Similarity Penalty Loss Function : ， In the formula, The cross-entropy loss function; S3-2-08: Determine the loss function If the minimum value is found, output the optimized parameters of the domain learner. Otherwise, based on the loss function Update the parameters of the domain learner Then, the updated parameters are used to replace the original parameters, and the process returns to step S3-2-06 to continue training the domain learner.

7. The method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins according to claim 6, characterized in that, The fault diagnosis model obtained by training a heterogeneous domain generalization network using K types of source domain knowledge embedding data and K types of target domain data also includes: S3-3-01 is based on the following formula: and Weighting: , In the formula, and As weight.

8. The method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins according to claim 7, characterized in that, Also includes: S3-3-02: Obtain the training set for source domain fault identification of the k-th class. , The true state label of the k-th source domain; S3-3-03: From the source domain fault identification training set Obtain a training data set ; S3-3-04: Obtain from the k-th type of source domain fault learner Estimated state label , The parameters to be optimized are those of the fault learner for the k-th source domain. S3-3-05: Calculate the weighted aggregation state label according to the following formula: ， In the formula, , S3-3-06: According to and Calculate the loss function Determine the loss function If it is the minimum, output the optimal parameter. Output the fault learner for the k-th target domain. Otherwise, according to the loss function renew Then return to step S3-3-02, from the training set Take another piece of data and continue training the fault learner for the k-th target domain. S3-3-07: Repeat steps S3-3-02 to S3-3-06 to obtain the K-class target domain fault identification optimization learner.

9. The method for heterogeneous domain generalization fault diagnosis of electromechanical systems based on digital twins according to claim 8, characterized in that, Fault diagnosis of the target domain using a fault diagnosis model includes: S4-1: Obtain invisible operating condition information according to the following formula Estimated implicit features: ； S4-2: Obtain the implicit features according to the following formula Normalization domain: ; S4-3: Estimate the state label according to the following formula: ； S4-4: Obtain the weighted aggregation state label according to the following formula: ； S4-5: Obtain the estimated fault state label for invisible operating condition information according to the following formula: 。 10. A heterogeneous domain generalized fault diagnosis device for electromechanical systems based on digital twins, characterized in that, include: A full-system digital twin model, used to simulate the physical characteristics of electromechanical systems; The training dataset acquisition unit generates virtual fault data and structured interconnected knowledge distributed across operating conditions through a full-system digital twin model. It then reconstructs K types of information domains based on the virtual fault data and electromechanical system entity data. Each information domain represents an operating condition of an electromechanical system entity. The information source is divided into a source domain and a target domain. The source domain represents visible operating conditions, and the target domain represents invisible operating conditions. By performing structured interconnection knowledge embedding on the source domains, a knowledge embedding dataset of K types of source domains is obtained; The fault diagnosis model is trained on a heterogeneous domain generalization network using a knowledge embedding dataset of K types of source domains and a dataset of K types of target domains, in order to diagnose faults based on invisible operating condition information.

Images