A power plant equipment feature operation state trend analysis method, system, device and storage medium

Abstract

The application provides a power station equipment characteristic operation state trend analysis method, system, device and storage medium, the method comprises the following steps: S1, obtaining the online data of the first characteristic in the power station equipment, obtaining the standard data through preprocessing; S2, disassembling and remodeling the power station equipment to obtain the analysis model of digital twinning; S3, the nonlinear relationship between different nodes in the analysis model is obtained by adversarial training on different nodes in the analysis model; S4, the nonlinear relationship is given to the analysis model to obtain the prediction model, and the standard data is input into the prediction model to obtain the online prediction result. The dynamic network model is constructed, and the nonlinear relationship function between nodes is generated, which significantly improves the interaction description accuracy of each component characteristic of the equipment; at the same time, through the real-time data updating and deviation correction mechanism, the robustness and real-time performance of the model are enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of trend analysis, and in particular relates to a method, system, equipment and storage medium for analyzing the trend of characteristic operating status of power station equipment. Background Art

[0002] With the rapid development of modern industry and energy systems, power plant equipment, as a core component of power generation and transmission, faces a significant challenge. The reliability and stability of its operating status are directly linked to the safety and efficiency of the entire power system. Recent advances in digital technology, sensor networks, and the Internet of Things (IoT) have enabled real-time monitoring of equipment operating status and big data analysis. Digital twin technology, in particular, offers a novel solution for predicting and analyzing equipment operating status by dynamically mapping virtual and physical devices. Furthermore, machine learning and deep learning technologies are increasingly being used in industrial fault detection and status prediction. These technologies can extract complex features from large amounts of historical and real-time data, enabling intelligent analysis of equipment operating status.

[0003] Currently, numerous technologies have been applied to monitoring and trend analysis of the operating status of power plant equipment. For example, trend prediction methods based on statistical models can provide simple predictions for a single variable, methods based on physical modeling can analyze the operating behavior of a device through its internal principles, and technologies based on neural networks can learn complex relationships between multiple variables. However, these methods still have limitations in terms of real-time equipment status, multi-node correlation, and prediction accuracy. In particular, traditional condition monitoring technologies mostly rely on static models or simple linear analysis, making it difficult to capture the nonlinear changes caused by the dynamic environment or complex interactions between components during power plant equipment operation. Summary of the Invention

[0004] The first object of the present invention is to provide a method for analyzing the trend of characteristic operating status of power plant equipment in response to the above-mentioned problems.

[0005] To this end, the above-mentioned purpose of the present invention is achieved through the following technical solutions:

[0006] A method for analyzing the trend of characteristic operating status of power station equipment includes the following steps:

[0007] S1. Obtain online data of a first feature of power station equipment and obtain standard data through preprocessing;

[0008] S2. Disassemble and remodel the power plant equipment to obtain a digital twin analysis model;

[0009] S3. By performing adversarial training on different nodes in the analysis model, a nonlinear relationship between different nodes in the analysis model is obtained;

[0010] S4. Assign the nonlinear relationship to the analysis model to obtain a prediction model, input the standard data into the prediction model, and obtain the online prediction result.

[0011] While adopting the above technical solutions, the present invention may also adopt or combine the following technical solutions:

[0012] As a preferred technical solution of the present invention: in step S1, the first feature includes temperature information and vibration information.

[0013] As a preferred technical solution of the present invention: Step S1 further includes the following sub-steps:

[0014] S11. Obtain the standard deviation, mean, and range from historical data and combine them with the online data collected by the sensor to obtain standardized high-dimensional data. The formula is as follows:

[0015]

[0016] Where, represents the standard deviation of the first feature, represents the mean of the first feature, represents the range of the first characteristic, Online data representing the first feature, Indicates the current moment;

[0017] S12. When the equipment starts running, continuously record online data and update the real-time standard deviation, mean, and range. Intercept the online data within a time window to obtain the standard data representation of the time series. The formula is as follows:

[0018]

[0019] Where, represents the sequence from time ta to time t, a represents the length of the time window, Represents the standard data at time t;

[0020] S13. In the time window, the real-time first characteristic data is corrected by a preset verification algorithm to obtain standard data in each sensor.

[0021] As a preferred technical solution of the present invention: step S2 further includes the following sub-steps:

[0022] S21. Disassemble the device into components, treat each component as a node, and connect the nodes to indicate that the first feature has a transfer relationship between the nodes, thereby obtaining a dynamic graph network;

[0023] S22. By continuously working on the status of each node, the first feature is transmitted in the dynamic graph network, thereby realizing trend analysis of the characteristic operation status of the device;

[0024] S23. During the analysis process, the real-time first features of some nodes are updated by acquiring online data. Meanwhile, the first features of all nodes are updated based on the updated first features. Some nodes include nodes equipped with sensors, which acquire online data of the first features through the sensors.

[0025] As a preferred technical solution of the present invention: in step S3, adversarial training includes the generator G and the discriminator D, including the following sub-steps:

[0026] S31, using the output of the generator as a nonlinear relationship function of the first feature between the two nodes;

[0027] The generator includes, input: Representation node The first characteristic, Representation node The first feature of ; output: Represents the generated nonlinear relationship function;

[0028] The discriminator includes: Input: the true relationship function obtained by fitting historical data , the relation function generated by the generator ; Output: the probability that the input relationship function is a true relationship;

[0029] S32, respectively, with the goal of minimizing the loss function, update the parameters of the generator G and the discriminator D;

[0030] Fix the generator G and update the parameters of the discriminator D to make the discrimination results more accurate;

[0031] Fix the discriminator D and update the parameters of the generator G to make the generated function closer to the true relationship;

[0032] S33. Repeat the training until the function generated by the generator cannot be distinguished by the discriminator, and the adversarial balance is achieved;

[0033] S34. Use the trained generator to generate a nonlinear relationship function to describe the interactive relationship of the first feature between real nodes.

[0034] As a preferred technical solution of the present invention: S41, assigning a nonlinear relationship function to the analysis model, where a node represents each component and a line represents the nonlinear relationship function of the first feature between the components;

[0035] S42: Using the standard data collected by the sensor as the first feature initial data, and using the prediction model to obtain the initial value of the first feature of each node;

[0036] S43. During the continuous operation phase, obtain the nodes H={h1, h2, ..., hn} that generate the first feature in the device, and obtain the first feature generated by each node in the set H at the next moment through a physical mathematical model;

[0037] Where H represents the set of nodes that generate the first feature, hn represents the nth node that generates the first feature, and n represents the number of nodes that generate the first feature;

[0038] S44. Using the prediction model, the first feature generated by each node in the set H is transferred between the nodes according to the nonlinear relationship function, thereby obtaining the first feature of each node at the next time t1;

[0039] S45, integrating the first features of each node to obtain a first feature set of each component of the device, performing a comprehensive analysis using a preset neural network algorithm to obtain a comprehensive assessment result of the state trend at the next moment t1;

[0040] S46. Based on the first feature set of each device component, the first feature of each device component at the next time t1 is updated; based on the first feature set of each device component at the next time t1, the first feature generated by each node in the set H at the next time t2 is obtained, and the first feature of each node at time t2 is predicted using the prediction model; a comprehensive analysis is performed using a preset neural network algorithm to obtain a comprehensive assessment result of the state trend at time t2;

[0041] S47, cyclically updating the first characteristics of each component of the equipment, predicting the first characteristics of each node, and comprehensively evaluating the status trend, thereby analyzing the characteristic operating status trend of the equipment within the set time interval;

[0042] S48. During the process of analyzing the trend of the equipment characteristic operating status, by acquiring online data in real time, updating the first characteristics of some nodes in the prediction model in real time, and updating the first characteristics of other nodes using the connections in the prediction model;

[0043] S49: Using the updated data as comparison data, while completing the update of the first feature of all nodes, analyze the prediction results: if the deviation between the current prediction result and the comparison data exceeds a preset threshold, re-analyze the device feature operating status trend within the time interval starting from the current time;

[0044] If the deviation between the current prediction result and the comparison data does not exceed the preset threshold, only the current comparison data will be updated, and the analysis results of the equipment characteristic operation status trend within the time interval will be retained.

[0045] The second object of the present invention is to provide a power plant equipment characteristic operating status trend analysis system, comprising the following modules:

[0046] The data acquisition module is used to obtain online data of the first characteristic of the power station equipment and obtain standard data after preprocessing;

[0047] The analytical model building module is used to disassemble and remodel power plant equipment to obtain a digital twin analytical model;

[0048] An analysis module, configured to obtain nonlinear relationships between different nodes in the analysis model by performing adversarial training on different nodes in the analysis model;

[0049] The prediction module is used to assign nonlinear relationships to the analysis model to obtain a prediction model, input standard data into the prediction model, and obtain online prediction results.

[0050] The third object of the present invention is to provide an electronic device, comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus.

[0051] a memory for storing a computer program;

[0052] A processor is used to execute the computer program stored in the memory to implement the steps of the power plant equipment characteristic operating status trend analysis method as described above.

[0053] Another object of the present invention is to provide a non-volatile storage medium, in which an executable program is stored. When the executable program is executed by a processor, the steps of the power plant equipment characteristic operating status trend analysis method as described above are implemented.

[0054] Compared with the existing technology, the present invention has the following beneficial effects: by combining digital twin technology with generative adversarial networks (GANs), high-precision analysis and real-time prediction of the operating status trends of power station equipment are achieved; a dynamic network model is constructed and nonlinear relationship functions are generated between nodes, which significantly improves the accuracy of the interactive description of the characteristics of various components of the equipment; at the same time, through real-time data updates and deviation correction mechanisms, the robustness and real-time performance of the model are enhanced; comprehensive assessment of the overall status trends of the equipment provides more reliable technical support for equipment operation and maintenance, effectively solving the problems of insufficient real-time performance and accuracy in the existing technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of the method for analyzing the characteristic operating status trend of power plant equipment provided by the present invention.

[0056] Figure 2 This is a framework diagram of the power plant equipment characteristic operation status trend analysis system provided by the present invention. DETAILED DESCRIPTION

[0057] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] like Figure 1 As shown, a method for analyzing the characteristic operating status trend of power station equipment includes the following steps:

[0059] S1. Obtain online data of a first characteristic of power station equipment and obtain standard data through preprocessing; the first characteristic includes but is not limited to temperature information and vibration information.

[0060] S11. Obtain the standard deviation, mean, and range from historical data and combine them with the online data collected by the sensor to obtain standardized high-dimensional data. The formula is as follows:

[0061]

[0062] Where, represents the standard deviation of the first feature, represents the mean of the first feature, represents the range of the first characteristic, Online data representing the first feature, Indicates the current moment;

[0063] S12. When the equipment starts running, continuously record online data and update the real-time standard deviation, mean, and range. Intercept the online data within a time window to obtain the standard data representation of the time series. The formula is as follows:

[0064]

[0065] Where, represents the sequence from time ta to time t, a represents the length of the time window, Represents the standard data at time t;

[0066] S13. In the time window, the real-time first characteristic data is corrected by a preset verification algorithm to obtain standard data in each sensor.

[0067] You can use time series models (such as the sliding average or ARIMA model) to model historical sensor data and correct the real-time first feature data. Alternatively, you can use pre-trained machine learning models (such as SVM or neural networks) to classify the data, determine whether it is an outlier, and then correct it.

[0068] By calculating statistics such as standard deviation, mean, and range, combined with online data collection, we remove noise, outliers, and data drift to ensure the stability and reliability of input data. High-quality standard data can significantly reduce the impact of data fluctuations on subsequent analysis results, improving model robustness and prediction accuracy.

[0069] Indicators such as standard deviation, mean, and range are recorded and updated through time windows to reflect the dynamic trends of the equipment's primary characteristics over different time periods. Dynamic feature capture helps describe real-time changes in the operating status of power plant equipment and supports the model's identification of abnormalities or changing trends in equipment status. Serialized standard data is captured within a time window to construct a time series representation, enhancing the model's ability to learn from temporal changes in features. Real-time data is corrected using a preset verification algorithm to reduce the impact of sensor measurement errors and environmental interference on the data, ensuring that each sensor provides accurate standard data. This data correction mechanism improves the credibility of input data and avoids model prediction bias caused by input errors.

[0070] S2. Disassemble and remodel the power plant equipment to obtain a digital twin analysis model;

[0071] S21. Disassemble the device into components, treat each component as a node, and connect the nodes to indicate that the first feature has a transfer relationship between the nodes, thereby obtaining a dynamic graph network;

[0072] S22. By continuously working on the status of each node, the first feature is transmitted in the dynamic graph network, thereby realizing trend analysis of the characteristic operation status of the device;

[0073] S23. During the analysis process, the real-time first features of some nodes are updated by acquiring online data. Meanwhile, the first features of all nodes are updated based on the updated first features. Some nodes include nodes equipped with sensors, which acquire online data of the first features through the sensors.

[0074] By breaking down the device into its components and constructing a dynamic graph network of nodes and connections, the goal is to capture the complex interactions between device components, particularly the dynamic transmission patterns of primary characteristics (such as temperature and vibration) between different components. This approach can accurately reflect the nonlinear changes and interactions of device characteristics during operation, providing a precise dynamic modeling foundation for trend analysis of device characteristic operating states. The approach emphasizes updating the characteristics of some nodes using online data, and using the updated characteristic values ​​to drive updates to other nodes, thereby adjusting the state of the global network. This dynamic update mechanism ensures that the model can promptly reflect the real-time status of the device and accurately predict future trends based on the current state, thereby improving the model's real-time performance and predictive capabilities.

[0075] In real-world applications, some equipment components may not be suitable for direct sensor installation (for example, due to space constraints or cost considerations). However, some key components can directly obtain real-time data on the first characteristic through sensors. By designing a mechanism for updating some nodes, the model can infer the status of the entire network using limited sensor data, significantly improving analysis efficiency and reducing reliance on comprehensive sensor deployment.

[0076] S3. By performing adversarial training on different nodes in the analysis model, a nonlinear relationship between different nodes in the analysis model is obtained; the adversarial training includes the generator G and the discriminator D, and includes the following sub-steps:

[0077] S31, using the output of the generator as a nonlinear relationship function of the first feature between the two nodes;

[0078] The generator includes, input: Representation node The first characteristic, Representation node The first feature of ; output: Represents the generated nonlinear relationship function;

[0079] The discriminator includes: Input: the true relationship function obtained by fitting historical data , the relation function generated by the generator ; Output: the probability that the input relationship function is a true relationship;

[0080] S32, respectively, with the goal of minimizing the loss function, update the parameters of the generator G and the discriminator D;

[0081] Fix the generator G and update the parameters of the discriminator D to make the discrimination results more accurate;

[0082] Fix the discriminator D and update the parameters of the generator G to make the generated function closer to the true relationship;

[0083] S33. Repeat the training until the function generated by the generator cannot be distinguished by the discriminator, and the adversarial balance is achieved;

[0084] S34. Use the trained generator to generate a nonlinear relationship function to describe the interactive relationship of the first feature between real nodes.

[0085] This relationship function effectively captures the complex dynamics of feature transfer between nodes, including nonlinearity, time-varying characteristics, and coupling effects between nodes. This modeling approach avoids the limitations of traditional linear relationship descriptions and provides a solid foundation for accurate analysis of device status trends.

[0086] The training process is based on the overall characteristics of the entire device, rather than simply treating two nodes in isolation. When generating the nonlinear relationship function between nodes, not only is the transmission pattern of the primary characteristic (such as temperature or vibration) analyzed between the two nodes, but also the internal characteristics of the nodes are comprehensively considered. These include energy changes indirectly caused by other factors: mechanical vibration within the node may generate heat, affecting node temperature; and frictional heat generation: friction between components has a direct impact on temperature characteristics. Adversarial training allows these objective influencing factors to be factored into the nonlinear relationship function.

[0087] This ensures that the nonlinear relationship function generated by the generator can simultaneously reflect the interaction relationship between nodes and the changes in the internal characteristics of the nodes themselves, thereby improving the physical rationality and prediction accuracy of the model.

[0088] Through adversarial training between the generator and the discriminator, the generator is continuously optimized, and the generated nonlinear relationship function gradually approaches the characteristic relationships between real nodes. At the same time, the discriminator improves the generator's ability by distinguishing between the real relationship function and the generated function. When the adversarial equilibrium is finally reached, the nonlinear relationship function output by the generator has strong expressive power and can accurately describe complex feature interactions between nodes.

[0089] S4. Assign the nonlinear relationship to the analysis model to obtain a prediction model, input the standard data into the prediction model, and obtain the online prediction result.

[0090] S41, assigning a nonlinear relationship function to the analysis model, where a node represents each component and a line represents the nonlinear relationship function of the first feature between the components;

[0091] S42, using the standard data collected by the sensor as the first feature initial data, and using the prediction model to obtain the initial value of the first feature of each node (the sensor only has a part of the nodes);

[0092] S43. During the continuous operation phase, obtain the nodes H={h1, h2, ..., hn} that generate the first feature in the device, and obtain the first feature generated by each node in the set H at the next moment through a physical mathematical model;

[0093] Where H represents the set of nodes that generate the first feature, hn represents the nth node that generates the first feature, and n represents the number of nodes that generate the first feature;

[0094] Nodes that generate primary features are those device components that actively contribute to or drive the primary feature (such as temperature or vibration). They serve as direct sources of the primary feature, rather than simply transmitting it. The characteristics of these nodes are often directly related to the device's operating principle. For the temperature feature, this node is the driver, not the gear friction. However, in this case, gear friction also generates heat, a problem overcome through adversarial training as described above.

[0095] S44. Using the prediction model, the first feature generated by each node in the set H is transferred between the nodes according to the nonlinear relationship function, thereby obtaining the first feature of each node at the next time t1;

[0096] S45, integrating the first features of each node to obtain a first feature set of each component of the device, performing a comprehensive analysis using a preset neural network algorithm to obtain a comprehensive assessment result of the state trend at the next moment t1;

[0097] S46. Based on the first feature set of each device component, the first feature of each device component at the next time t1 is updated; based on the first feature set of each device component at the next time t1, the first feature generated by each node in the set H at the next time t2 is obtained, and the first feature of each node at time t2 is predicted using the prediction model; a comprehensive analysis is performed using a preset neural network algorithm to obtain a comprehensive assessment result of the state trend at time t2;

[0098] S47, cyclically updating the first characteristics of each component of the equipment, predicting the first characteristics of each node, and comprehensively evaluating the status trend, thereby analyzing the characteristic operating status trend of the equipment within the set time interval;

[0099] S48. During the process of analyzing the trend of the equipment characteristic operating status, by acquiring online data in real time, updating the first characteristics of some nodes in the prediction model in real time, and updating the first characteristics of other nodes using the connections in the prediction model;

[0100] S49: Using the updated data as comparison data, while completing the update of the first feature of all nodes, analyze the prediction results: if the deviation between the current prediction result and the comparison data exceeds a preset threshold, re-analyze the device feature operating status trend within the time interval starting from the current time;

[0101] If the deviation between the current prediction result and the comparison data does not exceed the preset threshold, only the current comparison data will be updated, and the analysis results of the equipment characteristic operation status trend within the time interval will be retained.

[0102] The transfer of characteristics (such as temperature and vibration) between power plant equipment components is often nonlinear and dynamic. This transfer is influenced by multiple factors, including mechanical coupling, heat conduction paths, and varying operating conditions. By imbuing the analytical model with nonlinear relationship functions, the predictive model can accurately describe these complex interactions. This nonlinear relationship function empowers the predictive model with enhanced expressive power, capturing the nonlinear variations in characteristic transfer and avoiding the simplistic assumptions of characteristic transfer in traditional linear models, thereby improving the accuracy of model predictions.

[0103] Constructing a dynamic graph network through nodes and connections can dynamically update the operating status of equipment, predict the changing trends of each component's primary characteristic, and reflect the operating status of the entire equipment. Power plant equipment operates in a complex environment, and the status of each component may be affected by external disturbances or changes in operating conditions. This requires the predictive model to be able to update and respond in real time. Nonlinear relationship functions can dynamically adapt to changes in real-time data, providing flexibility to the model.

[0104] The prediction model leverages nonlinear relationships between nodes to dynamically transfer node features and derive the first eigenvalue of each node at the next moment. The first feature sets of each node are integrated to form the first feature set of the entire device. These features are then analyzed using a pre-defined neural network algorithm (such as a deep learning model) to generate a comprehensive assessment of the device's status. This analysis is used to predict the device's status trend at the next moment (e.g., t1). By continuously updating the feature data of device components, feature prediction and comprehensive assessment are performed for each node, enabling real-time prediction of device status. This process is cyclical; at each moment t1, t2, ..., the next state is continuously derived and compared with historical data to optimize the prediction.

[0105] The core goal of this solution is to accurately predict the operating status trend of the equipment by combining nonlinear relationship functions between nodes and real-time data updates, and to dynamically adjust according to the actual operating status of the equipment (such as temperature, vibration, etc.).

[0106] The Deep Gated Recurrent Unit (DGRU) is a time series data prediction model used in deep learning. It is a variant of the Recurrent Neural Network (RNN) and can effectively process sequential data and capture long-term dependencies. In this design, improvements to the channeled DGRU enable it to more effectively handle trend prediction. Due to the complex nonlinear interactions between device components, we introduced a Graph Neural Network (GNN) based on the DGRU to help handle nonlinear transfer relationships between nodes. Graph neural networks can transfer information between nodes and model nonlinear relationships between nodes based on a graph structure. This combination can effectively capture the mutual influence between nodes, especially the dynamic collaboration between device components (nodes).

[0107] The standard DGRU uses a fixed update and reset gating mechanism. The parameters of each gate (such as the update gate weight matrix) are static, obtained through a specific training process, and remain unchanged throughout the training process. However, such a fixed mechanism may not capture the dynamic changes of device components under different operating conditions. Therefore, in this design, the gating mechanism is replaced by nonlinear relationships, eliminating the static parameter limitations of traditional models. Through the generator (G) in a generative adversarial network (GAN), this design introduces nonlinear relationship functions for the interaction between each node (device component) in the device. These relationship functions can adaptively adjust the update process of each node state based on the different device operating states and the dynamic characteristics of the components.

[0108] The output of the nonlinear relationship function is not a simple weighted average or gating mechanism, but rather dynamically adjusts the state of each node through the complex relationships between nodes learned by the generator. In other words, the nonlinear relationship function replaces the role of the traditional update gate and reset gate and directly affects the state update of each node. Once the generator learns the appropriate nonlinear relationship, it will use these relationship functions to update the state of the device nodes at each time step. By applying the nonlinear relationship function to the state transfer between device components, the generator dynamically determines the state changes of each node at different time steps without the need for fixed gating control. This means that the state change of a node not only depends on the state of the previous moment, but also takes into account the interaction with other nodes.

[0109] Nonlinear relationship functions primarily address the complex interactions between device nodes. They no longer rely on fixed weighting parameters (such as the weight matrix in a gating mechanism). Specifically, each device component may respond differently under different operating conditions. Simple weighted averages may fail to capture these complex interactions. Nonlinear relationship functions, however, can adaptively adjust to changes in real-time data. During the status update process of a device component, the nonlinear relationship function can dynamically adjust its status update based on the actual operating status of the current node. This allows device state changes to no longer be static, but to respond to changes over time and operating conditions.

[0110] In the process of analyzing the trend of equipment characteristic operation status, by acquiring online data in real time, the real-time first characteristics of some nodes are updated in the prediction model, and the first characteristics of other nodes are updated using the connections in the prediction model.

[0111] Using the updated data as comparison data, while completing the first feature update for all nodes, analyze the prediction results. If the deviation between the current prediction result and the comparison data exceeds a preset threshold, re-analyze the device feature operating status trend within the time interval starting from the current moment. If the deviation between the current prediction result and the comparison data does not exceed the preset threshold, only the current comparison data is updated, and the analysis results of the device feature operating status trend within the time interval are retained.

[0112] During equipment operation, various nodes are subject to varying operating conditions, and their characteristics (such as temperature, vibration, and pressure) are constantly changing. By acquiring real-time online data and feeding it into the prediction model, the primary characteristics of some nodes are updated, reflecting the current state changes of various equipment components in real time. Furthermore, by leveraging the nonlinear relationships within the prediction model (i.e., the transfer relationships between nodes), the characteristics of updated nodes can be used to infer and update the state characteristics of other nodes. This real-time update mechanism ensures immediate feedback on equipment status, enabling the prediction model to more accurately capture various dynamic changes in equipment operation.

[0113] After updating the characteristics of each node, the updated node data is used as comparison data and compared with the prediction results to determine the deviation between the current prediction result and the historical comparison data. If the deviation between the prediction result and the comparison data exceeds the preset threshold, it indicates that the current device status is abnormal or the prediction result has a large error, and the status trend analysis needs to be re-performed to ensure the accuracy of the prediction result. If the deviation does not exceed the threshold, it means that the device status change meets the expectations of the prediction model. The current analysis result can be maintained and only the current comparison data can be updated. The purpose of this design is to avoid frequent re-analysis by setting a reasonable threshold, while being able to promptly identify anomalies in device operation and dynamically adjust the analysis strategy based on actual conditions.

[0114] Predicting the status of device characteristics is a continuous and dynamic process. Real-time performance is particularly important for devices that require rapid response, such as power generation equipment. This design utilizes a mechanism for controlling comparative data deviations and dynamically updating data to adjust device characteristic trend predictions in real time, ensuring both accurate and timely analysis results. Furthermore, threshold control avoids ineffective reanalysis, ensuring analysis efficiency and system stability.

[0115] like Figure 2 As shown, the present invention also provides a power plant equipment characteristic operation status trend analysis system, including the following modules:

[0116] The data acquisition module is used to obtain online data of the first characteristic of the power station equipment and obtain standard data after preprocessing;

[0117] The analytical model building module is used to disassemble and remodel power plant equipment to obtain a digital twin analytical model;

[0118] An analysis module, configured to obtain nonlinear relationships between different nodes in the analysis model by performing adversarial training on different nodes in the analysis model;

[0119] The prediction module is used to assign nonlinear relationships to the analysis model to obtain a prediction model, input standard data into the prediction model, and obtain online prediction results.

[0120] The present invention also provides an electronic device, which includes a processor, a communication interface, a memory and a communication bus. The processor, the communication interface and the memory communicate with each other via the communication bus.

[0121] Memory, memory is used to store computer programs;

[0122] The processor is used to execute the computer program stored in the memory to implement the steps of the power plant equipment characteristic operating status trend analysis method as described above.

[0123] The present invention also provides a non-volatile storage medium, in which an executable program is stored. When the executable program is executed by a processor, the steps of the method for analyzing the characteristic operating status trend of power station equipment as described above are implemented.

[0124] Thus far, the technical solutions of the present invention have been described in conjunction with the specific experimental processes shown in the accompanying drawings. However, the scope of protection of the present invention is not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A method for analyzing the trend of characteristic operating status of power station equipment, characterized in that: The steps include: S1. Obtain online data of a first feature of power station equipment and obtain standard data through preprocessing; S2. Disassemble and remodel the power plant equipment to obtain a digital twin analysis model; S3. By performing adversarial training on different nodes in the analysis model, a nonlinear relationship between different nodes in the analysis model is obtained; S4. Assign the nonlinear relationship to the analysis model to obtain a prediction model, input the standard data into the prediction model, and obtain the online prediction result; Step S4 specifically includes the following steps: S41, assigning a nonlinear relationship function to the analysis model, where a node represents each component and a line represents the nonlinear relationship function of the first feature between the components; S42: Using the standard data collected by the sensor as the first feature initial data, and using the prediction model to obtain the initial value of the first feature of each node; S43. During the continuous operation phase, obtain the nodes H={h1, h2, ..., hn} that generate the first feature in the device, and obtain the first feature generated by each node in the set H at the next moment through a physical mathematical model; Where H represents the set of nodes that generate the first feature, hn represents the nth node that generates the first feature, and n represents the number of nodes that generate the first feature; S44. Using the prediction model, the first feature generated by each node in the set H is transferred between the nodes according to the nonlinear relationship function, thereby obtaining the first feature of each node at the next time t1; S45, integrating the first features of each node to obtain a first feature set of each component of the device, performing a comprehensive analysis using a preset neural network algorithm to obtain a comprehensive assessment result of the state trend at the next moment t1; S46. Based on the first feature set of each device component, the first feature of each device component at the next time t1 is updated; based on the first feature set of each device component at the next time t1, the first feature generated by each node in the set H at the next time t2 is obtained, and the first feature of each node at time t2 is predicted using the prediction model; a comprehensive analysis is performed using a preset neural network algorithm to obtain a comprehensive assessment result of the state trend at time t2; S47, cyclically updating the first characteristics of each component of the equipment, predicting the first characteristics of each node, and comprehensively evaluating the status trend, thereby analyzing the characteristic operating status trend of the equipment within the set time interval; S48. During the process of analyzing the trend of the equipment characteristic operating status, by acquiring online data in real time, updating the first characteristics of some nodes in the prediction model in real time, and updating the first characteristics of other nodes using the connections in the prediction model; S49: Use the updated data as comparison data, and analyze the prediction results while completing the update of the first features of all nodes.

2. The method according to claim 1, wherein: In step S1, the first feature includes temperature information and vibration information.

3. The method according to claim 1, wherein: Step S1 also includes the following sub-steps: S11. Obtain the standard deviation, mean, and range from historical data and combine them with the online data collected by the sensor to obtain standardized high-dimensional data. The formula is as follows: ； Where, represents the standard deviation of the first feature, represents the mean of the first feature, represents the range of the first characteristic, Online data representing the first feature, Indicates the current moment; S12. When the equipment starts running, continuously record online data and update the real-time standard deviation, mean, and range. Intercept the online data within a time window to obtain the standard data representation of the time series. The formula is as follows: ； Where, represents the sequence from time ta to time t, a represents the length of the time window, Represents the standard data at time t; S13. In the time window, the real-time first characteristic data is corrected by a preset verification algorithm to obtain standard data in each sensor.

4. The method according to claim 1, wherein: Step S2 also includes the following sub-steps: S21. Disassemble the device into components, treat each component as a node, and connect the nodes to indicate that the first feature has a transfer relationship between the nodes, thereby obtaining a dynamic graph network; S22. By continuously working on the status of each node, the first feature is transmitted in the dynamic graph network, thereby realizing trend analysis of the characteristic operation status of the device; S23. During the analysis process, the real-time first features of some nodes are updated by acquiring online data. Meanwhile, the first features of all nodes are updated based on the updated first features. Some nodes include nodes equipped with sensors, which acquire online data of the first features through the sensors.

5. The method according to claim 1, wherein: In step S3, adversarial training includes the generator G and the discriminator D, and includes the following sub-steps: S31, using the output of the generator as a nonlinear relationship function of the first feature between the two nodes; The generator includes, input: Representation node The first characteristic, Representation node The first characteristic of Output: Represents the generated nonlinear relationship function; The discriminator includes: Input: the true relationship function obtained by fitting historical data , the relation function generated by the generator ; Output: the probability that the input relationship function is a true relationship; S32, respectively, with the goal of minimizing the loss function, update the parameters of the generator G and the discriminator D; Fix the generator G and update the parameters of the discriminator D to make the discrimination results more accurate; Fix the discriminator D and update the parameters of the generator G to make the generated function closer to the true relationship; S33. Repeat the training until the function generated by the generator cannot be distinguished by the discriminator, and the adversarial balance is achieved; S34. Use the trained generator to generate a nonlinear relationship function to describe the interactive relationship of the first feature between real nodes.

6. The method according to claim 1, wherein: Step S4 also includes the following sub-steps: If the deviation between the current prediction result and the comparison data exceeds the preset threshold, the analysis of the equipment characteristic operation status trend within the time interval will be repeated starting from the current moment; If the deviation between the current prediction result and the comparison data does not exceed the preset threshold, only the current comparison data will be updated, and the analysis results of the equipment characteristic operation status trend within the time interval will be retained.

7. A power plant equipment characteristic operation status trend analysis system, characterized in that: Includes the following modules: The data acquisition module is used to obtain online data of the first characteristic of the power station equipment and obtain standard data after preprocessing; The analytical model building module is used to disassemble and remodel power plant equipment to obtain a digital twin analytical model; An analysis module, configured to obtain nonlinear relationships between different nodes in the analysis model by performing adversarial training on different nodes in the analysis model; The prediction module is used to assign nonlinear relationships to the analysis model to obtain a prediction model. Standard data is input into the prediction model to obtain the online prediction results. The prediction module is executed through the following steps: S41, assigning a nonlinear relationship function to the analysis model, where a node represents each component and a line represents the nonlinear relationship function of the first feature between the components; S42: Using the standard data collected by the sensor as the first feature initial data, and using the prediction model to obtain the initial value of the first feature of each node; S43. During the continuous operation phase, obtain the nodes H={h1, h2, ..., hn} that generate the first feature in the device, and obtain the first feature generated by each node in the set H at the next moment through a physical mathematical model; Where H represents the set of nodes that generate the first feature, hn represents the nth node that generates the first feature, and n represents the number of nodes that generate the first feature; S44. Using the prediction model, the first feature generated by each node in the set H is transferred between the nodes according to the nonlinear relationship function, thereby obtaining the first feature of each node at the next time t1; S45, integrating the first features of each node to obtain a first feature set of each component of the device, performing a comprehensive analysis using a preset neural network algorithm to obtain a comprehensive assessment result of the state trend at the next moment t1; S46. Based on the first feature set of each device component, the first feature of each device component at the next time t1 is updated; based on the first feature set of each device component at the next time t1, the first feature generated by each node in the set H at the next time t2 is obtained, and the first feature of each node at time t2 is predicted using the prediction model; a comprehensive analysis is performed using a preset neural network algorithm to obtain a comprehensive assessment result of the state trend at time t2; S47, cyclically updating the first characteristics of each component of the equipment, predicting the first characteristics of each node, and comprehensively evaluating the status trend, thereby analyzing the characteristic operating status trend of the equipment within the set time interval; S48. During the process of analyzing the trend of the equipment characteristic operating status, by acquiring online data in real time, updating the first characteristics of some nodes in the prediction model in real time, and updating the first characteristics of other nodes using the connections in the prediction model; S49: Use the updated data as comparison data, and analyze the prediction results while completing the update of the first features of all nodes.

8. An electronic device comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus, wherein: a memory for storing a computer program; A processor, wherein the processor is configured to execute a computer program stored in a memory to implement the steps of the method for analyzing a trend of characteristic operating status of power plant equipment according to any one of claims 1 to 6.

9. A non-volatile storage medium, characterized in that: The non-volatile storage medium stores an executable program, and when the executable program is executed by the processor, the steps of the method for analyzing the characteristic operating status trend of power station equipment according to any one of claims 1 to 6 are implemented.

Images