Equipment state prediction method and system based on virtual scene

Abstract

The invention discloses an equipment state prediction method and system based on a virtual scene, and the method and system can achieve the detailed mapping of all sub-bodies of equipment and the mutual relation of the sub-bodies through the construction and utilization of a digital twinborn model of the equipment, and provide a solid foundation for the subsequent prediction and analysis. Equipment operation data are collected and processed in real time, modal function prediction is carried out in combination with a long-short-term memory model, and the prediction accuracy and timeliness are effectively improved. Furthermore, by presetting the alarm grade table, the alarm result can be automatically generated, the alarm sub-body can be quickly positioned and highlighted in the virtual scene, and the response speed and efficiency of equipment maintenance are greatly improved. Meanwhile, the real-time performance and the accuracy of the digital twinborn model are ensured through the sub-body recovery function after the alarm is eliminated, and a powerful guarantee is provided for continuous and stable operation of equipment. According to the invention, efficient and accurate prediction and monitoring of the state of the target equipment are realized.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method and system for predicting device status based on a virtual scene. Background Art

[0002] With the development of the power system, the digital transformation of the power grid has become an important development direction of the power industry. Digital twin technology and virtual reality technology are the two core technologies in the metaverse technology, and are also key technologies for realizing the digital transformation of the power grid. Digital twin technology can build a digital model of the equipment, simulate the operating status of the equipment, and provide data support for the maintenance and management of the equipment. Virtual reality technology can create an immersive virtual scene, allowing users to intuitively feel the status and changes of the equipment.

[0003] Existing smart substations have used digital twin technology to build equipment models and predict equipment status using predictive analysis technology. However, these technologies still have some shortcomings. First, the existing digital twin technology cannot intuitively update the display effect in terms of equipment model status prediction, making it impossible for users to understand the status changes of the equipment in a timely manner. Secondly, the existing virtual reality technology is not closely integrated with smart substations, and there is a lack of deeper exploration of human-computer interaction, which makes it impossible for users to have richer interactions in virtual scenes.

[0004] To sum up, the device model constructed by digital twin technology is still difficult to interact efficiently with users in the virtual scenes constructed by virtual reality technology. The underlying problem is essentially the low efficiency of data processing. Summary of the invention

[0005] In view of the problem that the device models in the prior art are difficult to interact efficiently with users in virtual scenes, resulting in difficulty in accurately monitoring the alarm locations, the present invention provides a device status prediction method and system based on virtual scenes. By splitting the digital twin model, each sub-body is independent in the virtual scene and can be dynamically adjusted according to the alarm results, thereby improving the efficiency of information interaction in the virtual scene and facilitating the efficient discovery, monitoring and handling of problems.

[0006] The following is the technical solution of the present invention.

[0007] A device state prediction method based on a virtual scene comprises the following steps: S1: Obtain the digital twin model of the target device and split it into parts, record the positional relationship and geometric constraints between different parts, and use the virtual scene platform to map the digital twin model parts into the virtual scene; S2: collect the operating data of the target device in real time, perform preprocessing and data decomposition, and obtain several modal functions; S3: Use the long short-term memory model to predict the modal function, merge the prediction results corresponding to each modal function, and generate a prediction signal; S4: Generate an alarm result based on the prediction signal and the preset alarm level table; S5: Determine the alarm sub-body based on the alarm result, highlight the alarm sub-body from the digital twin model, and set the alarm sub-body to an active state. After the alarm is eliminated, restore the sub-body based on the position relationship and geometric constraints.

[0008] In the present invention, by constructing and utilizing the digital twin model of the equipment, the various sub-bodies of the equipment and their interrelationships can be mapped in detail, providing a solid foundation for subsequent predictive analysis. The equipment operation data is collected and processed in real time, and the modal function prediction is performed in combination with the long short-term memory model, which effectively improves the accuracy and timeliness of the prediction. Furthermore, by presetting the alarm level table, the alarm results can be automatically generated, the alarm sub-body can be quickly located, and it can be highlighted in the virtual scene, which greatly improves the response speed and efficiency of equipment maintenance. At the same time, the sub-body recovery function after the alarm is eliminated ensures the real-time and accuracy of the digital twin model, and provides a strong guarantee for the continuous and stable operation of the equipment. The present invention realizes efficient and accurate prediction and monitoring of the state of the target equipment.

[0009] Preferably, in S1, obtaining a digital twin model of the target device and splitting it includes: Obtain a digital twin model generated from the physical structure of the target device; According to the physical structure and function of different components in the target device, the digital twin model is decomposed into several sub-bodies, and the positional relationship and geometric constraints between different sub-bodies are recorded.

[0010] Preferably, in S1, mapping the digital twin model sub-body into the virtual scene using the virtual scene platform includes: In the virtual scene platform, each sub-body of the digital twin model is mapped to the corresponding virtual position according to the positional relationship and geometric constraints between different sub-bodies, ensuring that the layout of each sub-body in the virtual scene is consistent with the actual equipment; Setting display attributes for different sub-bodies, wherein the display attributes include alarm display and regular display; Add interactive features to the virtual scene, allowing users to view detailed information or perform troubleshooting by clicking or dragging sub-objects.

[0011] Preferably, the S2: real-time collection of operating data of the target device, preprocessing and data decomposition, to obtain a number of modal functions, including: Determine the data collection frequency according to the characteristics and predicted needs of the target device, and collect the operating data of the target device based on the data collection frequency; Clean the collected operation data, remove noise and outliers, and perform standardization to obtain pre-processed operation data; The preprocessed operating data is decomposed into multiple mode functions with specific frequencies and bandwidths using a variational mode decomposition algorithm.

[0012] Preferably, the method of decomposing the preprocessed operating data into a plurality of modal functions having specific frequencies and bandwidths by using a variational modal decomposition algorithm comprises: Set the key parameters of the variational mode decomposition algorithm, including bandwidth constraint parameters, noise tolerance parameters, mode number parameters, DC component parameters, initial frequency estimation method parameters, and convergence tolerance parameters; The decomposition process is transformed into a variational problem solving process. By introducing a quadratic penalty factor and a Lagrange multiplication operator, the constrained variational problem is transformed into an unconstrained variational problem. The goal is to minimize the sum of the estimated bandwidths of all modal functions while ensuring that the sum of the modal functions is similar to the original signal. The alternating direction multiplier method is used to solve the above variational problem. By alternately updating each mode function and its central frequency, as well as the Lagrange multiplication operator, the updating process is repeated to gradually approach the optimal solution until the convergence condition is met. After iterative optimization, a series of modal functions are output as decomposition results, each of which represents the vibration mode of the original signal within a specific frequency range.

[0013] Preferably, the S3: predicting the modal function using the long short-term memory model, merging the prediction results corresponding to each modal function, and generating a prediction signal, includes: Divide different modal functions into training sets and test sets; The LSTM model is trained using the training set data. During the training process, the LSTM model learns the mapping relationship between the modal function and the future state, adjusts the model parameters through the back propagation algorithm and the gradient descent method to minimize the prediction error, and monitors the change of the loss function during the training process to ensure that the model does not have overfitting or underfitting. Use the trained long short-term memory model to predict the modal functions in the test set and obtain the future state prediction results corresponding to each modal function; The prediction results of different modal functions are superimposed or combined according to preset weights to generate the final prediction signal.

[0014] Preferably, the step S4: generating an alarm result based on the prediction signal and the preset alarm level table includes: Establish a preset alarm level table based on the historical operation data of the target device, and divide the alarm levels into at least normal, caution, and warning, and determine the characteristic parameter range and / or threshold value corresponding to each alarm level; The prediction signal is compared with the characteristic parameter range and / or threshold in the preset alarm level table, and according to the comparison result, the sub-body associated with the prediction signal is classified into the corresponding alarm level to generate an alarm result.

[0015] Preferably, the S5: determining the alarm sub-body based on the alarm result, highlighting the alarm sub-body in the digital twin model, setting the alarm sub-body to an active state, and restoring the sub-body based on the position relationship and geometric constraints after the alarm is eliminated, includes: Determine different visual alarm display styles according to the alarm level. After determining the alarm sub-body, set the alarm sub-body display attributes according to the visual alarm display style corresponding to the alarm level to highlight the alarm sub-body. In the system interface, a special display area is allocated for the alarm result to display the detailed information of the corresponding alarm sub-body; Set the alarm sub-body to active state, add drag and zoom functions to the alarm sub-body, and allow users to drag or zoom the alarm sub-body with the mouse; After the alarm is eliminated, the alarm sub-body is deleted and regenerated in the digital twin model based on the position relationship and geometric constraints.

[0016] The present invention also provides a device state prediction system based on a virtual scene. When the system is running, the above-mentioned device state prediction method based on a virtual scene is executed.

[0017] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, the steps of the above-mentioned device state prediction method based on a virtual scene are implemented.

[0018] The substantial effects of the present invention include: By building and utilizing the digital twin model of the device, detailed mapping and efficient prediction of the device status are achieved. Specifically, the present invention first splits the digital twin model of the target device and maps it to the virtual scene, ensuring that the positional relationship and geometric constraints between the model sub-bodies are consistent with the actual device. This step provides a solid foundation for subsequent predictive analysis, making the prediction results more accurate and credible.

[0019] Based on the real-time collection and processing of equipment operation data, the present invention uses the variational modal decomposition algorithm to decompose the preprocessed data into multiple modal functions with specific frequencies and bandwidths, and further predicts these modal functions through the long-term and short-term memory model. This combination effectively improves the accuracy and timeliness of the prediction, making the prediction of the equipment status more consistent with the actual operation situation.

[0020] The present invention also realizes the function of automatically generating alarm results by presetting the alarm level table, and can quickly locate the alarm sub-body and highlight it in the virtual scene. This feature greatly improves the response speed and efficiency of equipment maintenance, allowing maintenance personnel to quickly and accurately locate and handle equipment failures.

[0021] Finally, the present invention also has the function of restoring the sub-body after the alarm is eliminated, ensuring the real-time and accuracy of the digital twin model. This function provides a strong guarantee for the continuous and stable operation of the equipment, making equipment maintenance and management more convenient and efficient.

[0022] In summary, the present invention achieves efficient and accurate prediction and monitoring of the state of target equipment, and provides a new solution for intelligent maintenance and management of equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution will be clearly and completely described below in combination with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] It should be understood that in various embodiments of the present invention, the size of the sequence number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0026] It should be understood that in the present invention, "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0027] It should be understood that in the present invention, "plurality" refers to two or more than two. "And / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "Contains A, B and C", "Contains A, B, C" means that A, B, and C are all included, "Contains A, B or C" means that one of A, B, and C is included, and "Contains A, B and / or C" means that any one, any two, or any three of A, B, and C are included.

[0028] The technical solution of the present invention is described in detail with specific embodiments below. The embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0029] Embodiment 1: A device state prediction method based on a virtual scene, comprising: Figure 1 The following steps are shown: S1: Obtain the digital twin model of the target device and split it into parts, record the positional relationship and geometric constraints between different parts, and use the virtual scene platform to map the digital twin model parts into the virtual scene, including: Obtain a digital twin model generated from the physical structure of the target device; According to the physical structure and function of different components in the target device, the digital twin model is decomposed into several sub-bodies, and the positional relationship and geometric constraints between different sub-bodies are recorded; In the virtual scene platform, each sub-body of the digital twin model is mapped to the corresponding virtual position according to the positional relationship and geometric constraints between different sub-bodies, ensuring that the layout of each sub-body in the virtual scene is consistent with the actual equipment; Setting display attributes for different sub-bodies, wherein the display attributes include alarm display and regular display; Add interactive features to the virtual scene, allowing users to view detailed information or perform troubleshooting by clicking or dragging sub-objects.

[0030] For example, taking a transformer as the target device, first obtain its digital twin model accurately generated from the physical structure. Then, according to the physical structure and functional characteristics of each component in the transformer, such as the core, winding, insulation, etc., the digital twin model is decomposed into several sub-bodies, and the positional relationship and geometric constraints between them are recorded in detail. In the virtual scene platform, according to these relationships and constraints, each sub-body is accurately mapped to the corresponding virtual position to ensure that the layout in the virtual scene is exactly the same as the actual transformer. At the same time, two attributes, alarm display and regular display, are set for different sub-bodies, and interactive functions are added to the virtual scene, so that users can easily view their detailed information or troubleshoot by clicking or dragging the sub-body.

[0031] S2: Collect the operating data of the target device in real time, perform preprocessing and data decomposition, and obtain several modal functions, including: Determine the data collection frequency according to the characteristics and predicted needs of the target device, and collect the operating data of the target device based on the data collection frequency; Clean the collected operation data, remove noise and outliers, and perform standardization to obtain pre-processed operation data; Set the key parameters of the variational mode decomposition algorithm, including bandwidth constraint parameters, noise tolerance parameters, mode number parameters, DC component parameters, initial frequency estimation method parameters, and convergence tolerance parameters; The decomposition process is transformed into a variational problem solving process. By introducing a quadratic penalty factor and a Lagrange multiplication operator, the constrained variational problem is transformed into an unconstrained variational problem. The goal is to minimize the sum of the estimated bandwidths of all modal functions while ensuring that the sum of the modal functions is similar to the original signal. The alternating direction multiplier method is used to solve the above variational problem. By alternately updating each mode function and its central frequency, as well as the Lagrange multiplication operator, the updating process is repeated to gradually approach the optimal solution until the convergence condition is met. After iterative optimization, a series of modal functions are output as decomposition results, each of which represents the vibration mode of the original signal within a specific frequency range.

[0032] For example, for transformer equipment, according to its working characteristics and forecasting requirements, a suitable data acquisition frequency is set, and the operating data of the transformer is collected in real time. In order to improve the data quality, the collected data is cleaned, noise and outliers are removed, and standardized. Then the key parameters of the variational modal decomposition algorithm are set, including bandwidth constraints, noise tolerance, number of modes, etc. By taking these parameters into consideration, the decomposition process is converted into a variational problem solving process. In order to solve this problem, this embodiment adopts the alternating direction multiplier method, which gradually approaches the optimal solution by alternatingly updating each modal function and its center frequency, as well as the Lagrange multiplication operator. Finally, a series of modal functions are obtained, each of which represents the vibration mode of the transformer within a specific frequency range, providing strong data support for subsequent analysis and prediction.

[0033] S3: Use the long short-term memory model to predict the modal function, merge the prediction results corresponding to each modal function, and generate a prediction signal, including: Divide different modal functions into training sets and test sets; The LSTM model is trained using the training set data. During the training process, the LSTM model learns the mapping relationship between the modal function and the future state, adjusts the model parameters through the back propagation algorithm and the gradient descent method to minimize the prediction error, and monitors the change of the loss function during the training process to ensure that the model does not have overfitting or underfitting. Use the trained long short-term memory model to predict the modal functions in the test set and obtain the future state prediction results corresponding to each modal function; The prediction results of different modal functions are superimposed or combined according to preset weights to generate the final prediction signal.

[0034] For example, in transformer state prediction, this embodiment divides the modal function data into a training set and a test set. The long short-term memory model is trained using the training set, and the model parameters are adjusted by back propagation and gradient descent to minimize the prediction error, and the loss function is monitored to avoid overfitting or underfitting. The trained model is then used to predict the test set to obtain the future state prediction results of each modal function. Finally, these results are superimposed according to the preset weights to generate the final transformer prediction signal.

[0035] S4: Based on the prediction signal and the preset alarm level table, an alarm result is generated, including: Establish a preset alarm level table based on the historical operation data of the target device, and divide the alarm levels into at least normal, caution, and warning, and determine the characteristic parameter range and / or threshold value corresponding to each alarm level; The prediction signal is compared with the characteristic parameter range and / or threshold in the preset alarm level table, and according to the comparison result, the sub-body associated with the prediction signal is classified into the corresponding alarm level to generate an alarm result.

[0036] For example, in this embodiment, based on the historical operation data of the transformer, a preset alarm level table is established for each component of the transformer that can collect status parameters, and the alarm level is divided into three levels: normal, caution, and alarm. The characteristic parameter range corresponding to each level is determined, such as a temperature range of 40-60°C for normal, 60-70°C for caution, and over 70°C for alarm. The predicted signal is compared with the preset range, and the relevant sub-body is classified into the corresponding alarm level according to the result, and an alarm result is generated.

[0037] S5: Determine the alarm sub-body based on the alarm result, highlight the alarm sub-body from the digital twin model, set the alarm sub-body to an active state, and restore the sub-body based on the position relationship and geometric constraints after the alarm is eliminated, including: Determine different visual alarm display styles according to the alarm level. After determining the alarm sub-body, set the alarm sub-body display attributes according to the visual alarm display style corresponding to the alarm level to highlight the alarm sub-body. In the system interface, a special display area is allocated for the alarm result to display the detailed information of the corresponding alarm sub-body; Set the alarm sub-body to active state, add drag and zoom functions to the alarm sub-body, and allow users to drag or zoom the alarm sub-body with the mouse; After the alarm is eliminated, the alarm sub-body is deleted and regenerated in the digital twin model based on the position relationship and geometric constraints.

[0038] For example, in this embodiment, for the alarm processing of transformer equipment, different colors and highlight display styles are first set according to the alarm level as a visual alarm display. When an alarm sub-body is detected, such as the temperature of the "high-voltage winding" exceeds 70°C, it will be highlighted in red, and detailed information of the sub-body, such as the current temperature, alarm level, etc., will be displayed in the alarm area of ​​the system interface. At the same time, the "high-voltage winding" sub-body is set to active, allowing the user to check its status in more detail by dragging or zooming with the mouse. Once the alarm is eliminated, for example, the temperature drops to a safe range, this embodiment will delete the alarm display, and regenerate and position the sub-body in the digital twin model based on the positional relationship and geometric constraints of the "high-voltage winding" and other components.

[0039] In this embodiment, by constructing and utilizing the digital twin model of the device, the various sub-bodies of the device and their interrelationships can be mapped in detail, providing a solid foundation for subsequent predictive analysis. The real-time acquisition and processing of equipment operation data, combined with the long-term and short-term memory model for modal function prediction, effectively improves the accuracy and timeliness of the prediction. Furthermore, by presetting the alarm level table, the alarm results can be automatically generated, the alarm sub-body can be quickly located, and highlighted in the virtual scene, greatly improving the response speed and efficiency of equipment maintenance. At the same time, the sub-body recovery function after the alarm is eliminated ensures the real-time and accuracy of the digital twin model, and provides a strong guarantee for the continuous and stable operation of the equipment. The present invention realizes efficient and accurate prediction and monitoring of the state of the target device.

[0040] Embodiment 2: This embodiment also provides a device state prediction system based on a virtual scene. When the system is running, the above-mentioned device state prediction method based on a virtual scene is executed.

[0041] The virtual scene-based equipment status prediction system of this embodiment mainly includes the following modules: a data acquisition module, which collects transformer operation data in real time; a data preprocessing module, which is responsible for data cleaning, denoising and standardization; a variational mode decomposition module, which decomposes the preprocessed data into several mode functions; a long short-term memory model prediction module, which uses the LSTM model to predict the mode function and generate a prediction signal; an alarm generation module, which generates an alarm result according to a preset alarm level table and highlights the alarm sub-body in the virtual scene; a virtual scene mapping and interaction module, which maps the transformer digital twin model to the virtual scene and adds interactive functions; and a system management and configuration module, which is responsible for the overall configuration and monitoring of the system.

[0042] Embodiment three: This embodiment also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, the steps of the above-mentioned device state prediction method based on a virtual scene are implemented.

[0043] The substantial effects of the above embodiments include: By building and utilizing the digital twin model of the device, detailed mapping and efficient prediction of the device status are achieved. Specifically, this embodiment first splits the digital twin model of the target device and maps it to the virtual scene, ensuring that the positional relationship and geometric constraints between the model sub-bodies are consistent with the actual device. This step provides a solid foundation for subsequent predictive analysis, making the prediction results more accurate and credible.

[0044] Based on the real-time collection and processing of equipment operation data, this embodiment uses the variational modal decomposition algorithm to decompose the preprocessed data into multiple modal functions with specific frequencies and bandwidths, and further predicts these modal functions through the long short-term memory model. This combination effectively improves the accuracy and timeliness of the prediction, making the prediction of the equipment status more consistent with the actual operation situation.

[0045] This embodiment also realizes the function of automatically generating alarm results by presetting the alarm level table, which can quickly locate the alarm sub-body and highlight it in the virtual scene. This feature greatly improves the response speed and efficiency of equipment maintenance, allowing maintenance personnel to quickly and accurately locate and handle equipment failures.

[0046] Finally, this embodiment also has the function of restoring the sub-body after the alarm is eliminated, ensuring the real-time and accuracy of the digital twin model. This function provides a strong guarantee for the continuous and stable operation of the equipment, making equipment maintenance and management more convenient and efficient.

[0047] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the specific device can be divided into different functional modules to complete all or part of the functions described above.

[0048] In the embodiments provided in the present application, it should be understood that the disclosed structures and methods can be implemented in other ways. For example, the embodiments of the structure described above are only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another structure, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, structures or units, which can be electrical, mechanical or other forms.

[0049] The units described as separate components may or may not be physically separated, and the components shown as units may be one physical unit or multiple physical units, that is, they may be located in one place or distributed in multiple different places. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0050] In addition, each functional unit in the embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0051] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0052] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A device state prediction method based on a virtual scene, characterized in that: The following steps are involved: S1: Obtain the digital twin model of the target device and split it into parts, record the positional relationship and geometric constraints between different parts, and use the virtual scene platform to map the digital twin model parts into the virtual scene; S2: collect the operating data of the target device in real time, perform preprocessing and data decomposition, and obtain several modal functions; S3: Use the long short-term memory model to predict the modal function, merge the prediction results corresponding to each modal function, and generate a prediction signal; S4: Generate an alarm result based on the prediction signal and the preset alarm level table; S5: Determine the alarm sub-body based on the alarm result, highlight the alarm sub-body from the digital twin model, and set the alarm sub-body to an active state. After the alarm is eliminated, restore the sub-body based on the position relationship and geometric constraints.

2. The device state prediction method based on virtual scene according to claim 1 is characterized in that: In S1, the digital twin model of the target device is obtained and split, including: Obtain a digital twin model generated from the physical structure of the target device; According to the physical structure and function of different components in the target device, the digital twin model is decomposed into several sub-bodies, and the positional relationship and geometric constraints between different sub-bodies are recorded.

3. The device state prediction method based on virtual scene according to claim 1 is characterized in that: In S1, the digital twin model sub-body is mapped to the virtual scene using the virtual scene platform, including: In the virtual scene platform, each sub-body of the digital twin model is mapped to the corresponding virtual position according to the positional relationship and geometric constraints between different sub-bodies, ensuring that the layout of each sub-body in the virtual scene is consistent with the actual equipment; Setting display attributes for different sub-bodies, wherein the display attributes include alarm display and regular display; Add interactive features to the virtual scene, allowing users to view detailed information or perform troubleshooting by clicking or dragging sub-objects.

4. The device state prediction method based on virtual scene according to claim 1 is characterized in that: S2: real-time collection of operating data of the target device, preprocessing and data decomposition, and obtaining several modal functions, including: Determine the data collection frequency according to the characteristics and predicted needs of the target device, and collect the operating data of the target device based on the data collection frequency; Clean the collected operation data, remove noise and outliers, and perform standardization to obtain pre-processed operation data; The preprocessed operating data is decomposed into multiple mode functions with specific frequencies and bandwidths using a variational mode decomposition algorithm.

5. The device state prediction method based on virtual scene according to claim 4 is characterized in that: The variational modal decomposition algorithm is used to decompose the preprocessed operating data into multiple modal functions with specific frequencies and bandwidths, including: Set the key parameters of the variational mode decomposition algorithm, including bandwidth constraint parameters, noise tolerance parameters, mode number parameters, DC component parameters, initial frequency estimation method parameters, and convergence tolerance parameters; The decomposition process is transformed into a variational problem solving process. By introducing a quadratic penalty factor and a Lagrange multiplication operator, the constrained variational problem is transformed into an unconstrained variational problem. The goal is to minimize the sum of the estimated bandwidths of all modal functions while ensuring that the sum of the modal functions is similar to the original signal. The alternating direction multiplier method is used to solve the above variational problem. By alternately updating each mode function and its central frequency, as well as the Lagrange multiplication operator, the updating process is repeated to gradually approach the optimal solution until the convergence condition is met. After iterative optimization, a series of modal functions are output as decomposition results, each of which represents the vibration mode of the original signal within a specific frequency range.

6. The device state prediction method based on virtual scene according to claim 1 is characterized in that: S3: predicting the modal function using the long short-term memory model, merging the prediction results corresponding to each modal function, and generating a prediction signal, including: Divide different modal functions into training sets and test sets; The LSTM model is trained using the training set data. During the training process, the LSTM model learns the mapping relationship between the modal function and the future state, adjusts the model parameters through the back propagation algorithm and the gradient descent method to minimize the prediction error, and monitors the change of the loss function during the training process to ensure that the model does not have overfitting or underfitting. Use the trained long short-term memory model to predict the modal functions in the test set and obtain the future state prediction results corresponding to each modal function; The prediction results of different modal functions are superimposed or combined according to preset weights to generate the final prediction signal.

7. The device state prediction method based on virtual scene according to claim 1 is characterized in that: S4: generating an alarm result based on the prediction signal and the preset alarm level table, including: Establish a preset alarm level table based on the historical operation data of the target device, and divide the alarm levels into at least normal, caution, and warning, and determine the characteristic parameter range and / or threshold value corresponding to each alarm level; The prediction signal is compared with the characteristic parameter range and / or threshold in the preset alarm level table, and according to the comparison result, the sub-body associated with the prediction signal is classified into the corresponding alarm level to generate an alarm result.

8. The device state prediction method based on virtual scene according to claim 1 is characterized in that: S5: determining the alarm sub-body based on the alarm result, highlighting the alarm sub-body in the digital twin model, setting the alarm sub-body to an active state, and restoring the sub-body based on the position relationship and geometric constraints after the alarm is eliminated, including: Determine different visual alarm display styles according to the alarm level. After determining the alarm sub-body, set the alarm sub-body display attributes according to the visual alarm display style corresponding to the alarm level to highlight the alarm sub-body. In the system interface, a special display area is allocated for the alarm result to display the detailed information of the corresponding alarm sub-body; the alarm sub-body is set to active state, and the drag and zoom functions are added to the alarm sub-body, allowing the user to drag the alarm sub-body or zoom the alarm sub-body with the mouse; After the alarm is eliminated, the alarm sub-body is deleted and regenerated in the digital twin model based on the position relationship and geometric constraints.

9. A device state prediction system based on virtual scenes, characterized in that: When the system is running, a device state prediction method based on a virtual scene as described in any one of claims 1 to 8 is executed.

10. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, it implements the steps of a device state prediction method based on a virtual scene as described in any one of claims 1 to 8.

Images