A Multidimensional Data Fusion Analysis Method and Device for Digital Twins

Abstract

The present application provides a multi-dimensional data fusion analysis method and device for digital twins, which relates to the technical field of data analysis and includes: constructing M fault prediction models in a target power plant; performing multi-level data fusion modeling on multiple power plant systems; loading the M fault prediction models into multiple system-level digital twins; performing multi-level fault transfer analysis on the target power plant; rendering fault conduction on the initial twin of the power plant according to the power plant fault state transition diagram to obtain the digital twin of the power plant; loading M pieces of real-time device operation data into the M fault prediction models, and after obtaining the operation states of M devices, visualizing the operation states of M devices in the digital twin of the power plant. Through the present application, the technical problem in the prior art that the operation and maintenance efficiency of the power plant is low due to the lack of effective fusion of multi-dimensional data can be solved. By constructing the digital twin of the power plant, the visualization of the device operation state is realized, and the operation and maintenance efficiency of the power plant is improved.

Description

Technical Field

[0001] This application relates to the technical field of data analysis, and in particular, to a multi-dimensional data fusion analysis method and device for digital twins. Background Art

[0002] With the continuous expansion of the scale of modern power plants and other complex industrial systems, the number of devices is increasing day by day, and the operating conditions are becoming more and more complex. The traditional operation and maintenance management method is difficult to meet the requirements of high efficiency and real-time. As an innovative device management method, digital twin technology can map the operating state of actual devices to a digital platform through a virtual model, providing real-time visualization. However, most of the existing digital twins focus on the virtual mapping at the device level, lacking the integrated processing of cross-device and multi-dimensional data, and are prone to problems such as exceeding the safety distance limit of devices and inaccurate positioning of potential faults, which in turn affects the operation and maintenance management efficiency of power plants.

[0003] In summary, there is a technical problem in the prior art that due to the complex multi-source heterogeneous data involved in the power plant system and the lack of deep integration across devices and systems, the operation and maintenance efficiency of the power plant is relatively low. Summary of the Invention

[0004] The purpose of this application is to provide a multi-dimensional data fusion analysis method and device for digital twins, so as to solve the technical problem in the prior art that due to the complex multi-source heterogeneous data involved in the power plant system and the lack of deep integration across devices and systems, the operation and maintenance efficiency of the power plant is relatively low.

[0005] In view of the above problems, this application provides a multi-dimensional data fusion analysis method and device for digital twins.

[0006] First aspect, the present application provides a multi-dimensional data fusion analysis method for digital twins. The multi-dimensional data fusion analysis method for digital twins is implemented through a multi-dimensional data fusion analysis device for digital twins. Among them, the multi-dimensional data fusion analysis method for digital twins includes: constructing M fault prediction models for M power plant equipment in the target power plant by using an LSTM network, where the M power plant equipment has M equipment source identifiers; after collecting multi-dimensional data from multiple power plant systems to obtain multiple system-level multi-dimensional data, using the multiple system-level multi-dimensional data for multi-level data fusion modeling to generate multiple system-level digital twins; according to the M equipment source identifiers, loading the M fault prediction models into the multiple system-level digital twins to complete the construction of the initial power plant twin; performing multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram; performing fault conduction rendering on the initial power plant twin according to the power plant fault state transition diagram to obtain a power plant digital twin; after loading the M real-time device operation data of the M power plant equipment into the M fault prediction models of the power plant digital twin to obtain M device operation states, visualizing the M device operation states in the power plant digital twin.

[0007] Second aspect, the present application also provides a multi-dimensional data fusion analysis device for digital twins, which is used to execute a multi-dimensional data fusion analysis method for digital twins as described in the first aspect. Among them, the multi-dimensional data fusion analysis device for digital twins includes: a fault prediction model construction module, which is used to construct M fault prediction models for M power plant devices in the target power plant by using an LSTM network, where the M power plant devices have M device source identifiers; a digital twin modeling module, which is used to perform multi-level data fusion modeling by using the multiple system-level multi-dimensional data after collecting multi-dimensional data of multiple power plant systems to generate multiple system-level digital twins; a power plant twin construction module, which is used to load the M fault prediction models into the multiple system-level digital twins according to the M device source identifiers to complete the construction of the initial power plant twin; a fault transfer analysis module, which is used to perform multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram; a fault conduction rendering module, which is used to perform fault conduction rendering on the initial power plant twin according to the power plant fault state transition diagram to obtain a power plant digital twin; an operation state visualization module, which is used to visualize the operation states of the M devices in the power plant digital twin after loading the real-time device operation data of the M power plant devices into the M fault prediction models of the power plant digital twin to obtain the operation states of the M devices.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0009] By adopting an LSTM network to construct M fault prediction models for M power plant equipment in a target power plant, where the M power plant equipment has M equipment source identifiers; after collecting multi-dimensional data from multiple power plant systems to obtain multiple system-level multi-dimensional data, using the multiple system-level multi-dimensional data for multi-level data fusion modeling to generate multiple system-level digital twins; according to the M equipment source identifiers, loading the M fault prediction models into the multiple system-level digital twins to complete the construction of the initial power plant twin; conducting multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram; performing fault conduction rendering on the initial power plant twin according to the power plant fault state transition diagram to obtain a power plant digital twin; after loading the M real-time equipment operation data of the M power plant equipment into the M fault prediction models of the power plant digital twin to obtain M equipment operation states, visualizing the M equipment operation states in the power plant digital twin. That is to say, by separately constructing LSTM fault prediction models for multiple devices, using multi-level data fusion to model and fuse multi-dimensional data at multiple system levels to generate multiple system-level digital twins, integrating the established fault prediction models into the digital twins, realizing real-time monitoring and optimization of the entire power plant system, conducting fault conduction rendering through multi-level fault transfer analysis, loading the real-time operation data of the equipment into the fault prediction models, and visualizing the states, showing the latest state of each device in the digital twin, realizing the visualization of the equipment operation state, and improving the operation and maintenance efficiency of the power plant.

[0010] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the following specifically gives the specific implementation manners of this application. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of this application, nor is it used to limit the scope of this application. Other features of this application will become easily understood through the following description. Brief Description of the Drawings

[0011] In order to more clearly illustrate the technical solutions in this application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0012] Figure 1 It is a schematic flow chart of a multi-dimensional data fusion analysis method for a digital twin-oriented power plant in this application;

[0013] Figure 2This is a schematic structural diagram of a multi-dimensional data fusion analysis device for a digital twin body in this application.

[0014] Explanation of reference numerals: Fault prediction model construction module 11, digital twin body modeling module 12, power plant twin body construction module 13, fault transfer analysis module 14, fault conduction rendering module 15, operation status visualization module 16. Specific implementation mode

[0015] This application provides a multi-dimensional data fusion analysis method and device for a digital twin body, which solves the technical problem in the prior art that due to the complex multi-source heterogeneous data involved in the power plant system and the lack of deep-level fusion across devices and systems, the operation and maintenance efficiency of the power plant is relatively low. By respectively constructing LSTM fault prediction models for multiple devices, using multi-level data fusion to model and fuse multi-dimensional data at multiple system levels, generating multiple system-level digital twin bodies, integrating the established fault prediction models into the digital twin bodies, realizing real-time monitoring and optimization of the entire power plant system, conducting fault conduction rendering through multi-level fault transfer analysis, loading the real-time operation data of the device into the fault prediction model, performing status visualization, and displaying the latest status of each device in the digital twin body, the visualization of the device operation status is realized, and the operation and maintenance efficiency of the power plant is improved.

[0016] Next, the technical solutions in this application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the example embodiments described here. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this application. In addition, it should be noted that for the sake of description, only the parts related to this application are shown in the accompanying drawings rather than all.

[0017] Example 1, please refer to the attached Figure 1 , this application provides a multi-dimensional data fusion analysis method for a digital twin body. Among them, the multi-dimensional data fusion analysis method for a digital twin body is applied to a multi-dimensional data fusion analysis device for a digital twin body. The multi-dimensional data fusion analysis method for a digital twin body specifically includes the following steps:

[0018] S100: Use the LSTM network to construct M fault prediction models for M power plant devices in the target power plant, where the M power plant devices have M device source identifiers.

[0019] Specifically, there are M power plant devices in the target power plant, and each device has different working environments, failure modes, and operating characteristics. To build personalized prediction models for different devices, an LSTM network model needs to be built separately for each device, so that the failure prediction ability of each device will be more accurate. Each device will have a unique device source identifier, which can be the device ID number, device type, device location, etc., used to distinguish the data of different devices and ensure that the data sources of each device are correctly identified and processed. During model training and prediction, these identifiers are used to index the data corresponding to each device to ensure that the model does not confuse the operating characteristics of different devices.

[0020] For each device, prepare the corresponding historical failure operation data. The historical failure data of each device contains multiple features (such as temperature, pressure, vibration, load, etc.), and these data can be used to predict the failure trend and failure type of the device. According to the nodes where failures occur, the historical failure data of each device is segmented, and the dynamic slicing window is used for normalizing the segmentation results to obtain M groups of sample failure operation data. For each power plant device, define an LSTM network structure, including the number of layers, the number of neurons, activation functions, etc. The size of the input layer should be consistent with the size of the sliding window. Set multiple LSTM layers to capture the long-term dependencies in the time series data. The size of the output layer depends on the prediction task, including the failure trend and failure type of the device.

[0021] Divide the dataset of each device into a training set, a validation set, and a test set. Use the training set data to train the LSTM model of each device, adjust the model parameters such as the learning rate and batch size through the validation set, and use the test set to evaluate the performance of the model, such as accuracy, recall, etc. The specific training process is explained in detail in the refinement of the subsequent corresponding steps. For the sake of simplicity of the specification, only a brief introduction is given here. According to the M groups of sample failure operation data, complete the construction of M failure evolution prediction units and M failure type identification units, and cascade them to obtain M failure prediction models. By building failure prediction models independently for each device, customized predictions can be made according to the working characteristics and failure modes of the devices. Maintenance or replacement of devices can be carried out in advance according to the failure prediction results, avoiding unnecessary downtime, improving the utilization rate of the devices, and reducing the maintenance cost.

[0022] S200: After collecting multi-dimensional data from multiple power plant systems and obtaining multiple system-level multi-dimensional data, use the multiple system-level multi-dimensional data for multi-level data fusion modeling to generate multiple system-level digital twins.

[0023] Specifically, the target power plant includes multiple power systems, that is, systems with different functions or regions, and each system is responsible for a part of the power plant's functions, such as the power generation system, the cooling system, the control system, etc. Each power plant system contains multiple devices, and the operating data of each device is closely related to the overall performance of the system. Data collection is carried out on multiple power plant systems through various methods such as sensors, device monitoring systems, and environmental monitoring systems, and multiple system-level multi-dimensional data is obtained, including point cloud data, real-scene data, power plant design information, etc. corresponding to each power system.

[0024] For multiple system-level multi-dimensional data, multi-level data fusion modeling is carried out. That is to say, on the basis of multiple system-level multi-dimensional data, through data processing and fusion at different levels, different data sources from each system and device are integrated. Through the comprehensive processing of these multi-level data, an efficient model reflecting the overall state of the power plant is constructed.

[0025] Due to the large scale and complexity of the power plant system, a unified large digital twin may be difficult to process the data of all devices and systems. To improve the computing efficiency and support real-time data update, the digital twin can be split into multiple sub-twins, and each sub-twin is responsible for the modeling and simulation of a specific subsystem or device. A distributed computing architecture is adopted, that is, a large digital twin is decomposed into multiple small and independent sub-twins, and each sub-twin is responsible for the modeling and simulation of a specific power system or subsystem. Each subsystem can operate independently on different servers or computing nodes, and these servers exchange data and cooperate through a high-speed network. In short, each power plant system corresponds to a digital twin, which is constructed through the multi-dimensional data corresponding to the power plant system. The dependent claim corresponding to this step takes the power generation system as an example to construct the digital twin. The construction of the digital twin for other power systems in the target power plant is similar to that of the power generation system and will not be repeated for explanation.

[0026] Through the above steps, multiple sub-twins are processed and fused with data, and finally multiple system-level digital twins are formed, providing functions such as real-time monitoring, performance analysis, and fault diagnosis for each subsystem or the entire power plant of the power plant. Through the distributed computing architecture, the huge data processing tasks of the power plant can be dispersed to multiple computing nodes for parallel processing, greatly improving the computing efficiency and response speed, enabling the power plant system to update its digital twin in real time and reflecting the latest operating state. The distributed computing architecture enables the digital twin of the power plant to have good scalability. As the equipment and systems of the power plant increase, new sub-twins can be easily added without large-scale modification or reconstruction of the existing system. Each sub-twin can display the device status, operating efficiency, and key performance indicators through a visual interface, helping the operation and maintenance personnel to timely understand the operating conditions of the power plant and make corresponding decisions, not only improving the operation and maintenance efficiency, but also effectively preventing faults and reducing the equipment downtime.

[0027] S300: According to the M device source identifiers, load the M fault prediction models into the multiple system-level digital twins to complete the construction of the initial twin of the power plant.

[0028] Further, S300 of the present application includes:

[0029] Conduct a complexity analysis of fault prediction based on the M historical fault operation data, and calculate and output multiple system-level prediction complexities according to the analysis results; refer to the multiple system-level prediction complexities, allocate computing nodes to the multiple power plant systems in the distributed computing architecture to obtain K groups of power systems of K computing nodes; after loading the M fault prediction models into the multiple system-level digital twins according to the M device source identifiers, use the K groups of power systems as replication guides to copy the multiple system-level digital twins to the K computing nodes to complete the construction of the initial twin of the power plant.

[0030] Specifically, conduct a complexity analysis of the M historical fault operation data to evaluate the difficulty and complexity of fault prediction. For example, by statistically analyzing the frequency, type, and distribution of fault occurrences, predicting the amount of data to be processed, and the complexity problems that the model may encounter during operation (such as too large data dimensions, too long calculation time, etc.), evaluate the difficulty of fault prediction. According to the analysis results, calculate and output multiple system-level prediction complexities. According to the analysis results of the multiple system-level prediction complexities, allocate computing nodes to each power plant system in the distributed computing architecture. Dynamically allocate computing resources according to the prediction complexity of each power plant system. For systems with higher complexity, allocate more computing nodes, and for systems with lower complexity, allocate fewer computing nodes, which helps to improve resource utilization and reduce calculation time. Accordingly, obtain K computing nodes of K groups of power systems.

[0031] According to the device source identifier, load M fault prediction models into multiple system-level digital twins. The fault prediction model of each device corresponds to its source identifier, ensuring that a personalized prediction model is provided for each power plant device. Subsequently, copy the multiple system-level digital twins to the corresponding K computing nodes according to K groups of power systems. Each node will run the corresponding fault prediction model. Distribute the constructed multiple system-level digital twins to K computing nodes according to the K groups of power systems of the power plant. Each computing node runs one or more groups of digital twins of the power plant system and performs collaborative computing through a distributed computing architecture to ensure efficient cooperation between computing nodes and complete the real-time modeling and simulation of the power plant system.

[0032] Through the above steps, the construction of the initial digital twin of the power plant is completed. Each computing node simulates the operating state, fault prediction, performance evaluation, etc. of the power plant in real time by running the digital twin model. The digital twins of the entire power plant work together through distributed computing nodes to form a comprehensive and accurate simulation model of the power plant system. The distributed computing architecture enables the system to work in parallel on multiple nodes, ensuring that when some nodes fail, other nodes can still operate normally, thereby enhancing the stability and reliability of the entire power plant digital twin.

[0033] S400: Conduct multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram.

[0034] Furthermore, S400 of this application includes:

[0035] Conduct fault cross-system conduction analysis on the target power plant to obtain a system-level state transition diagram; conduct fault cross-device conduction analysis on the target power plant according to the M historical fault operation data to obtain multiple device-level state transition diagrams of multiple power plant systems; obtain a power plant fault state transition diagram through multi-dimensional data fusion of the system-level state transition diagram and multiple device-level state transition diagrams.

[0036] Specifically, according to the equipment and system structure of the target power plant, a cross-system fault conduction analysis is carried out. By modeling the interactions between equipment and fault conduction paths in multiple power systems, the chains of possible fault conduction between equipment are identified. For example, a boiler fault may cause an interruption in steam supply, which in turn affects the normal operation of the steam turbine. The cross-system fault conduction analysis is to analyze the fault conduction mechanism between different power systems and study how the operation of other systems is affected when a fault occurs in a certain system. Since the target power plant contains multiple independent systems, the equipment within each system may also have a chain reaction due to faults, resulting in faults in other equipment or systems. The system-level state transition diagram represents the transition process between different states of each system in the power plant (such as the power generation system, the power transmission system, etc.). Each state represents a working state of the system (such as normal, faulty, repaired, etc.), and the transition represents the process of the system transitioning from one state to another.

[0037] The equipment-level state transition diagram is similar to the system-level state transition diagram. The difference is that the equipment-level state transition diagram is for the specific equipment within the power plant (such as boilers, steam turbines, transformers, etc.) for state transition analysis. Each power plant equipment will have different states under different working conditions, and the transitions between these states are triggered by the working conditions and fault states of the equipment. A fault conduction analysis is carried out for each power plant equipment (such as boilers, steam turbines, transformers, etc.) to establish an equipment-level state transition diagram. Each equipment may have different behavior patterns and fault transfer mechanisms in different states (such as normal, warning, faulty, etc.).

[0038] Through multi-dimensional data fusion, the data of the equipment-level state transition diagram and the system-level state transition diagram are integrated to obtain the power plant fault state transition diagram. During the fusion process, it is necessary to combine the fault data, operation status data, and historical fault data from different systems and equipment to obtain a comprehensive fault conduction path. After completing the data fusion, a power plant fault state transition diagram is generated, which integrates the fault state transition information of multiple equipment, shows the mutual influence between equipment, the fault conduction path between systems, and the possible fault transfer process during the operation of the entire power plant. By constructing the power plant fault state transition diagram, the mutual influence and fault conduction path between each system and equipment can be comprehensively understood, and when a fault occurs, it can quickly respond and make rapid adjustments according to the fault state transition diagram to minimize the downtime and economic losses.

[0039] S500: Render the fault conduction of the initial power plant twin according to the power plant fault state transition diagram to obtain the power plant digital twin.

[0040] Specifically, the constructed power plant fault state transition diagram is rendered onto the initial power plant twin to obtain the power plant digital twin. That is to say, each fault state, equipment state, and their transfer paths in the power plant fault state transition diagram are also integrated onto the initial power plant twin, and the conduction process of faults is presented in the digital twin. According to the data in the fault state transition diagram, specific equipment or systems are selected as the starting points of faults in the digital twin. According to the fault state transfer rules of the equipment, the conduction process of faults is simulated in the digital twin. Through graphical rendering technology, the fault conduction process will be presented visually in the digital twin. The rendering engine can display information such as the working state of the equipment, the time nodes of fault occurrence, and the fault conduction path. Conducting fault conduction simulation in the digital twin helps power plant operation and maintenance personnel identify key equipment and fault conduction paths, and timely discover potential fault chains. The power plant digital twin is a virtual copy of the power plant constructed through digital technology, simulating the structure, equipment, systems, and operating states of the power plant, capable of reflecting the operating state of the power plant, the health status of the equipment, and fault information in real time, and can be used for fault early warning, diagnosis, maintenance, and optimization. By performing fault conduction rendering on the initial power plant twin, the conduction process of fault states can be simulated and presented in the digital twin, and real-time visual display of each system and equipment in the power plant can be carried out, helping the power plant improve the efficiency of fault diagnosis and emergency response.

[0041] S600: After loading the M real-time device operation data of the M power plant devices into the M fault prediction models of the power plant digital twin to obtain the M device operation states, perform visualization of the M device operation states in the power plant digital twin.

[0042] Furthermore, S600 of the present application includes:

[0043] Collect pixel data for the M power plant devices to obtain the M real-time device operation data; in the power plant digital twin, perform fault feature analysis on the M real-time device operation data using the M fault prediction models and output the M device operation states; perform fault feature aggregation on the M historical fault operation data to obtain M sets of fault operation performances; traverse the M sets of fault operation performances using the M device operation states and schedule M real-time operation performances; use the real-time rendering engine to perform visualization of the M device operation states for the M real-time operation performances in the power plant digital twin; perform fault transfer analysis on the M real-time operation performances in the power plant fault state transition diagram and perform visualization of the fault state transfer trend in the power plant digital twin.

[0044] Specifically, pixel data collection is performed on M power plant devices, that is, various real-time data of the M power plant devices are collected through devices such as sensors to obtain M real-time device operation data, including temperature, pressure, load, vibration, etc. In the power plant digital twin, M fault prediction models corresponding to the M power plant devices are used to analyze the fault characteristics of the M real-time device operation data, judge the fault risk of the devices, and predict the possible fault occurrence time to obtain M device operation states, including fault risk (the possibility of a fault occurring in the current operation state) and predicted fault time (the predicted time of fault occurrence).

[0045] Fault feature aggregation is performed on the M historical fault operation data to analyze the performance of the devices in historical faults, including the working parameters of the devices and the time nodes of fault occurrence. The fault operation performance refers to the performance of the device when a fault occurs, including the change of the device's working data and the characteristics of the fault appearance, and presents the abnormal conditions of the device when the fault occurs in a graphical or image form. The M sets of fault operation performance refer to the external manifestations of the device in different fault stages obtained through the analysis of historical fault data, such as the different characteristics of the device in the initial, middle, and end stages of the fault. According to the M device operation states, traverse the M sets of fault operation performance and schedule the real-time operation performance to monitor the state change and potential fault risk of the device.

[0046] Using a real-time rendering engine, the M real-time operation performances are displayed in the power plant digital twin in a visual form to help operation and maintenance personnel intuitively understand the operation status of each device. Perform fault transfer analysis on the real-time operation performance of the device, model and analyze the fault propagation path of the device and the system through a fault state transition diagram, and identify how a fault may conduct from one device to another. Present the results of the fault state transfer analysis in the digital twin and display the evolution process of the fault state in a dynamic graph. For example, the transition process of the device from the normal state to the warning state and then to the fault state will be shown in the visual interface to help operation and maintenance personnel better perform fault diagnosis and prevention.

[0047] Through real-time device operation data collection, fault feature analysis and device operation state output, historical fault data analysis and real-time operation performance scheduling, device operation state visualization and fault state transfer trend analysis, the operation state of power plant devices can be comprehensively monitored and managed, potential fault risks can be discovered and responded to in a timely manner, and the operation and maintenance efficiency and reliability of the power plant can be improved.

[0048] Furthermore, the present application S100 includes:

[0049] Collect device data from the target power plant to obtain M historical fault operation data of M power plant devices; after splitting the M historical fault operation data based on the fault occurrence nodes, use a sliding slice window to normalize the splitting results to obtain M sets of sample fault operation data, where the sliding slice window has K sample splitting steps; use an LSTM network to construct M fault evolution prediction units for the M power plant devices; use a CNN network to construct M fault type identification units for the M power plant devices; according to the K sample splitting steps, decompose the M sets of sample fault operation data, and use the decomposition results to optimize the parameter tuning of the M fault evolution prediction units and M fault type identification units; map and cascade the optimized M fault evolution prediction units and M fault type identification units to complete the construction of the M fault prediction models.

[0050] Specifically, collect data from the devices of the target power plant, obtain the fault operation data of M power plant devices, and get M historical fault operation data, including information such as the operating status of the devices, environmental conditions, historical fault records, and sensor data. The target power plant is a complex facility composed of multiple interconnected power systems, and each system contains multiple power devices. For example, the power generation system includes boilers, steam turbines, generators, etc., the power transmission system includes transformers, cables, distribution equipment, etc., the cooling system includes cooling towers, water pumps, etc., and the auxiliary system includes pumping stations, fans, compressors, etc.

[0051] For the historical fault operation data of M devices, split the data according to the fault occurrence nodes. The fault occurrence nodes are the marked fault occurrence points in the historical fault data. Whenever a device fails, record the time node of the fault occurrence, the fault type, and the fault impact, etc. data. Use a sliding slice window to normalize the splitting results, convert each data slice to a unified scale, and facilitate subsequent analysis. The K sample splitting steps are the number of samples spanned each time the sliding window slides. K is the window size, that is, the length of the data considered each time slicing (usually the number of samples in the time series). The sliding slice window slides in the data segment according to the given step size and is split into multiple small time windows (samples). These small windows will be normalized to ensure that the data is compared and trained on the same scale. The sliding slice window is a time series data processing method that divides the time series data by setting the window size (for example, K samples) and continuously sliding this window in the data sequence. Each slide will generate a new window and process the data therein.

[0052] Normalize each sample data after cutting. The purpose of normalization is to eliminate the dimensional differences of different devices and different measurement data, map the data to the interval [0, 1], or make the mean of the data 0 and the variance 1 through standardization. After sliding slice window segmentation and normalization, finally M groups of sample fault operation data of M devices are obtained. Each group of samples represents the state data of a device within a specific time period, and the features of each sample have been standardized, which is suitable for training the prediction model.

[0053] For each device (a total of M devices), use the LSTM network to construct a fault evolution prediction unit. The LSTM network can learn the time evolution law of device faults based on historical data (including fault evolution data and steady-state data), and predict the fault evolution process of the device in the future period of time. Taking the first fault evolution prediction unit as an example, obtain the corresponding first group of sample fault operation data from the M groups of sample fault operation data. Among them, the first group of historical fault operation data includes the first group of sample fault evolution data and the first group of sample fault steady-state data, including various sensor data (such as temperature, pressure, vibration, etc.) when the device fails and the stable operation state data after the device fails.

[0054] Use the first group of sample fault evolution data and the first group of sample fault steady-state data as the training data of the LSTM network to train the LSTM model and help the LSTM learn the evolution process after the device fails. The training data will form inputs and outputs according to the time series, and the LSTM model will learn the time dependence of these data to predict the future development trend of faults. During the training process, the LSTM network will adjust the network weights according to the input data (fault evolution data and steady-state data) to better predict the future fault evolution path of the device. Use the test set to evaluate the generalization ability of the model, calculate the prediction error index, such as the mean square error, and accordingly optimize the model parameters until the model converges. Take the model at this time as the first fault evolution prediction unit. Use the M groups of sample fault operation data as the training data to complete the construction of M fault evolution prediction units for M power plant devices.

[0055] Extract the first set of sample fault types from the steady-state data of the first set of sample faults. The sample fault type refers to the characteristic pattern or type exhibited by the equipment after a fault occurs. Each fault type usually has unique operating characteristics, such as overheating, overload, abnormal vibration, etc., which helps to identify the nature and severity of the fault. CNN is a deep learning algorithm widely used in image recognition and classification tasks. In fault type recognition, CNN can automatically learn the important features in the equipment fault data and classify different fault types based on these features. Design the model structure of CNN according to the first set of sample fault steady-state data, including the input layer, convolutional layer, pooling layer, and fully connected layer. Design the input layer according to the dimension of the fault data, and set multi-channel input according to the multi-dimensional signals input. Adjust the size and number of convolutional kernels according to requirements, and use multiple convolutional layers to extract fault features. Add a pooling layer (such as max pooling) after the convolutional layer to reduce the feature dimension and extract key information. Finally, add a fully connected layer, and the number of output nodes corresponds to the number of categories of the fault type.

[0056] Initialize appropriate batch size (such as 32 or 64), number of training epochs (such as 100), learning rate, etc. Use the first set of sample fault steady-state data and the first set of sample fault types as training data and input them into the CNN network for training. During the training process, the CNN network will extract the key features in the input data through multiple convolutional and pooling layers and classify the fault types based on these features. Adjust and optimize the model structure and parameters according to the problems encountered in actual applications. Use the validation set to monitor the performance changes during the model training process to prevent overfitting. Deploy the trained CNN model as the first fault type recognition unit, receive new fault steady-state data, and output the type of the equipment fault based on the previously trained model. Use the M sets of sample fault operation data as training data to complete the construction of M fault type recognition units for M power plant equipment.

[0057] Decompose the M sets of sample fault operation data according to K sample segmentation steps. The fault operation data is usually time series data, including the historical operating states of the equipment, such as multiple parameters like temperature, pressure, vibration, etc. Through a sliding slice window, divide the original time series data into multiple samples according to the set step size K, and each sample contains K data points. According to the decomposition results, tune and optimize the parameters of the M fault evolution prediction units (LSTM networks) and the M fault type recognition units (CNN networks). The purpose of parameter tuning and optimization is to improve the prediction accuracy by adjusting the hyperparameters of the model. The process of parameter tuning includes selecting appropriate parameters such as learning rate, optimizer, number of network layers, number of neurons, etc. to ensure that the model can effectively learn the features in the data.

[0058] Hyperparameter tuning and optimization refer to adjusting the hyperparameters of a model (such as learning rate, number of network layers, number of neurons, etc.) during the model training process to improve the model performance. Hyperparameter tuning can be accomplished through methods such as grid search, random search, or Bayesian optimization, with the aim of finding a set of the most suitable parameter configurations to maximize the prediction accuracy and robustness of the model. The decomposition result refers to the sub-datasets generated by splitting multiple faulty operation data (for example, using the sliding window method) during the data processing, which contains information on different time periods of the original data. For example, in the hyperparameter tuning and optimization of an LSTM network, the learning rate is adjusted (such as from 0.001 to 0.01) to select the learning rate most suitable for the current dataset. Then, the number of layers and neurons of the LSTM network are adjusted, and different network configurations are tested to improve the accuracy and robustness of the model. Similarly, in a CNN network, hyperparameters such as the size of the convolutional kernel, the number of pooling layers, and the number of convolutional layers can be adjusted.

[0059] After completing the hyperparameter tuning and optimization of the fault evolution prediction unit and the fault type recognition unit, the next step is to cascade the two optimized models. Cascade optimization means jointly mapping the fault evolution prediction unit and the fault type recognition unit after hyperparameter tuning and optimization to form a complete fault prediction model. Cascade optimization refers to combining multiple optimized models and making them cooperate with each other through the optimization process to construct the final fault prediction model. In the fault prediction task, the LSTM model is responsible for predicting the evolution process of equipment faults (such as how temperature and pressure change over time), while the CNN model is responsible for identifying the specific types of equipment faults (such as overheating or mechanical faults). The combination of the two can achieve comprehensive fault prediction, that is, it can not only know when the equipment fault will occur but also determine the type of the fault. Through the above steps, the construction of M fault prediction models for M devices is finally completed. The fault prediction model for each device consists of a fault evolution prediction unit (LSTM network) and a fault type recognition unit (CNN network). These two modules work together to accurately predict the operating state of the device, helping the power plant to carry out fault early warning and operation and maintenance decision-making.

[0060] By jointly using the LSTM and CNN networks, it is possible to fully utilize the time dependence and feature information in the historical data of equipment faults, providing more accurate fault prediction and type recognition. The LSTM network can predict the evolution trend of faults, while the CNN can accurately classify the fault types. The hyperparameter tuning and optimization of the model improve the robustness and stability of the fault prediction model. Through accurate fault prediction and type recognition, potential faults can be discovered in a timely manner, reducing the sudden shutdown of equipment and improving the reliability and stability of power plant equipment, ensuring the safe and efficient operation of the power plant.

[0061] Furthermore, this application also includes the following steps:

[0062] Decompose the first set of sample fault operation data into the first set of sample fault evolution data and the first set of sample fault steady-state data, where the time span of the sample fault evolution data is K-1 sample segmentation steps; use the first set of sample fault evolution data and the first set of sample fault steady-state data as training data to construct the first fault evolution prediction unit of the first power plant equipment; after extracting the first set of sample fault types from the first set of sample fault steady-state data, use the first set of sample fault steady-state data and the first set of sample fault types as training data to construct the first fault type recognition unit of the first power plant equipment; cascade the first fault evolution prediction unit and the first fault type recognition unit to complete the construction of the first fault prediction model; and so on, use the M historical fault operation data as training data and use the LSTM network to construct the M fault prediction models.

[0063] Specifically, according to the first set of sample fault operation data, decompose it into the first set of sample fault evolution data and the first set of sample fault steady-state data. Among them, the sample fault evolution data refers to the data of the equipment's state change over time after the fault occurs, reflecting the fault degree, health status, etc. of the equipment at different time points; the sample fault steady-state data is the stable operation state data of the equipment after the fault occurs for a period of time. Although the equipment has failed, its state has been relatively stable and no longer changes violently. The time span of the sample fault evolution data is K-1 sample segmentation steps, that is, a longer period of data is determined through a sliding window to accurately capture the evolution and steady-state characteristics of the equipment fault. The time span is based on the setting of the sliding slice window, subtracting 1 from the K sample segmentation steps as the time range of each evolution data set. For example, if K is set to 100, the time span is 99 steps (that is, the process data from the initial stage of the fault to its stable stage).

[0064] Use the first set of sample fault evolution data and the first set of sample fault steady-state data as training data to train the LSTM network model to obtain the first fault evolution prediction unit. Through its time series modeling ability, the LSTM network can effectively learn the law of the equipment fault evolution and then predict the evolution process of the equipment fault. The specific training process is described in detail in the construction steps of the above-mentioned M fault evolution prediction units and is similar here.

[0065] Extract the first set of sample fault types from the first set of sample fault steady-state data, including overheating, wear, mechanical faults, etc. The extracted fault types are used as labels and, together with the fault steady-state data, are used as training data to train the CNN network, thereby constructing the fault type recognition unit. The CNN processes the data through the convolutional layer and the pooling layer to identify different fault patterns. The specific training process is described in detail in the construction steps of the above-mentioned M fault type recognition units and is similar here.

[0066] After the training of the LSTM network and the CNN network is completed, these two units are cascaded, and the comprehensive prediction of equipment failures is achieved through the combined action of the two models. The evolution process of equipment failures and the identification of failure types are combined to form a complete prediction system. The above steps are continuously repeated for M historical failure operation data. According to the failure occurrence nodes, the data is segmented, and the sliding slice window is used to normalize the segmentation results, obtaining M sets of sample failure operation data. The M sets of sample failure operation data are used as training data to construct M failure prediction models for M power plant equipment, thereby constructing a dedicated failure prediction model for each equipment. Since the failure modes of each equipment are different, by training a separate failure prediction model for each equipment, a personalized prediction model can be generated based on the historical failure data of the equipment, avoiding a one-size-fits-all approach and improving the pertinence and effectiveness of the prediction.

[0067] Further, S200 of this application includes:

[0068] By performing laser scanning on the power generation system, point cloud data of the power generation system is obtained; by performing photogrammetry on the power generation system, real-scene data of the power generation system is obtained; after obtaining the power plant design information of the target power plant through interaction, a three-dimensional design model of the power generation system is constructed according to the power plant design information, and then the model pose is calibrated using the point cloud data of the power generation system to obtain a spatial model of the power generation system. Among them, the point cloud data of the power generation system, the real-scene data of the power generation system, and the power plant design information constitute multi-dimensional data of the power generation system; the Scale-Invariant Feature Transform (SIFT) is used to perform feature matching on the spatial model of the power generation system and the real-scene data of the power generation system to obtain real-scene spatial alignment coordinates; according to the real-scene spatial alignment coordinates, the real-scene data of the power generation system and the spatial model of the power generation system are fused between virtual and real, and then a real-time rendering engine is introduced to complete the construction of the digital twin of the power generation system; and so on, multi-level data fusion modeling is performed using the multi-system multi-dimensional data to generate the multi-system digital twins.

[0069] Specifically, laser scanning is performed on multiple power generation systems of the target power plant to obtain point cloud data of the power generation system. Laser scanning is a method of obtaining high-precision three-dimensional data of an object's surface through laser technology. For example, lidar emits laser beams and receives reflected signals to generate high-density point cloud data. The point cloud data of the power generation system is a set of data points with three-dimensional coordinates obtained through laser scanning or other three-dimensional scanning methods. Each point represents a position on the surface of the scanned object, and point cloud data is often used for three-dimensional modeling and digital representation. At the same time, photogrammetry is performed on the power generation system through photography technology, usually using photos from multiple angles, and the three-dimensional information of the object is reconstructed through image processing technology. Photogrammetry is carried out to obtain the real-scene data of the power generation system, including the appearance of the equipment, the installation location, the surrounding environment, etc.

[0070] By interacting with power plant designers or operation and maintenance personnel, obtain the design information of the target power plant, including equipment specifications, installation locations, configurations, system drawings, etc., which can provide detailed data on the overall structure of the power plant. Construct a 3D design model of the power generation system based on the power plant design information to ensure that the digital model is consistent with the design documents. The 3D design model refers to a 3D digital model of equipment or systems constructed based on power plant design information through computer-aided design (CAD) or other modeling software, representing the geometric form, installation location of the equipment, and its relationship with other equipment.

[0071] Use the point cloud data of the power generation system for model pose calibration. By comparing the spatial positions of the point cloud data and the design model, adjust the position, rotation angle, etc. of the design model to make it exactly aligned with the actual physical position. After model pose calibration, an accurate spatial model, i.e., the power generation system spatial model, is obtained, which accurately reflects the actual state of the power generation system. Multidimensional data refers to data containing multiple types and sources, such as point cloud data, real scene data, and design information. The multidimensional data of the power generation system consists of power generation system point cloud data, power generation system real scene data, and power plant design information.

[0072] Use the SIFT algorithm to perform feature matching on the power generation system spatial model and real scene data, thereby obtaining real scene spatial alignment coordinates, ensuring the precise spatial alignment of the virtual model and the actually captured real scene data. SIFT (Scale-Invariant Feature Transform) is a computer vision algorithm that can extract key feature points from images. These feature points have scale invariance and rotation invariance and can be used to match images from different perspectives. Through the SIFT algorithm, key feature points (such as the four corner points of the boiler) can be extracted from the power generation system real scene data and these features can be matched with the corresponding features in the point cloud data to achieve the precise docking of the virtual model and the real data. The real scene spatial alignment coordinates refer to aligning the real scene data in the real world with the spatial data of the virtual 3D model through a feature matching algorithm, thereby ensuring the precise spatial matching of the virtual model and the real scene.

[0073] Fuse the real - world data of the power generation system with the spatial model of the power generation system, align the real - world data with the virtual model in space and time, so that the virtual model can reflect the exact state of the real world. Virtual - real fusion refers to combining a virtual 3D model with real - world data (such as point cloud data or real - scene images obtained through laser scanning and photogrammetry) to form a highly unified and coordinated digital twin. The goal of virtual - real fusion is to completely align the virtual model and real - world data in terms of space, scale, etc. for further analysis or display. Use a real - time rendering engine to render the fused data. A real - time rendering engine is a computer graphics engine that can generate and display 3D images in real time and is usually used in dynamic interactive applications. In the construction of a digital twin, a real - time rendering engine can fuse a virtual 3D model and real - time data (such as sensor data, monitoring data, etc.) and present it to the user, usually used for simulation, visualization, and real - time monitoring.

[0074] Through virtual - real fusion and a real - time rendering engine, a complete digital twin of the power generation system is obtained, which not only includes the 3D design model of the power generation system but also incorporates data from the real world (such as real - time sensor data, image data, etc.), can reflect the operating state of the power generation system in real time, and has high interactivity and visualization functions. Repeat this process to obtain multiple system - level multi - dimensional data of multiple power generation systems in the target power plant, perform multi - level data fusion modeling, and construct corresponding multiple system - level digital twins.

[0075] By combining laser scanning, photogrammetry, and design information, the 3D data of the power generation system can be accurately obtained, and through attitude calibration and feature matching, the spatial alignment between the virtual model and the actual equipment can be ensured, thus achieving high - precision virtual - real fusion. Use a real - time rendering engine to dynamically render the fused digital twin, enabling users to observe the operating state of the equipment, environmental changes, etc. in real time and improving the real - time response ability of operation and maintenance.

[0076] In summary, the multi - dimensional data fusion analysis method for a digital twin provided by this application has the following technical effects:

[0077] By adopting an LSTM network to construct M fault prediction models for M power plant equipment in a target power plant, where the M power plant equipment have M equipment source identifiers; after collecting multi-dimensional data from multiple power plant systems to obtain multiple system-level multi-dimensional data, using the multiple system-level multi-dimensional data for multi-level data fusion modeling to generate multiple system-level digital twins; according to the M equipment source identifiers, loading the M fault prediction models into the multiple system-level digital twins to complete the construction of the initial power plant twin; performing multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram; performing fault conduction rendering on the initial power plant twin according to the power plant fault state transition diagram to obtain a power plant digital twin; after loading the M real-time equipment operation data of the M power plant equipment into the M fault prediction models of the power plant digital twin to obtain M equipment operation states, performing visualization of the M equipment operation states in the power plant digital twin. That is to say, by separately constructing LSTM fault prediction models for multiple devices, using multi-level data fusion to model and fuse multi-dimensional data at multiple system levels to generate multiple system-level digital twins, integrating the established fault prediction models into the digital twins, realizing real-time monitoring and optimization of the entire power plant system, performing fault conduction rendering through multi-level fault transfer analysis, loading the real-time operation data of the equipment into the fault prediction models, and performing state visualization, and displaying the latest state of each device in the digital twin, realizing visualization of the equipment operation state and improving the operation and maintenance efficiency of the power plant.

[0078] Embodiment 2. Based on the same inventive concept as the multi-dimensional data fusion analysis method for a digital twin in the foregoing Embodiment 1, the present application also provides a multi-dimensional data fusion analysis device for a digital twin. Please refer to the attached Figure 2 , the multi-dimensional data fusion analysis device for a digital twin includes:

[0079] A fault prediction model construction module 11, where the fault prediction model construction module 11 is used to construct M fault prediction models for M power plant equipment in a target power plant by using an LSTM network. Among them, the M power plant equipment has M equipment source identifiers; a digital twin modeling module 12, where the digital twin modeling module 12 is used to perform multi-dimensional data collection on multiple power plant systems to obtain multiple system-level multi-dimensional data, and then use the multiple system-level multi-dimensional data for multi-level data fusion modeling to generate multiple system-level digital twins; a power plant twin construction module 13, where the power plant twin construction module 13 is used to load the M fault prediction models into the multiple system-level digital twins according to the M equipment source identifiers to complete the construction of the initial power plant twin; a fault transfer analysis module 14, where the fault transfer analysis module 14 is used to perform multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram; a fault conduction rendering module 15, where the fault conduction rendering module 15 is used to perform fault conduction rendering on the initial power plant twin according to the power plant fault state transition diagram to obtain a power plant digital twin; an operating state visualization module 16, where the operating state visualization module 16 is used to load the M real-time device operating data of the M power plant equipment into the M fault prediction models of the power plant digital twin to obtain M device operating states, and then perform visualization of the M device operating states in the power plant digital twin.

[0080] Further, the fault prediction model construction module 11 in the multi-dimensional data fusion analysis device for digital twins is further used for:

[0081] Collect device data of a target power plant to obtain M historical fault operation data of M power plant equipment; after segmenting the M historical fault operation data based on the fault occurrence nodes, use a sliding slice window to normalize the segmentation results to obtain M groups of sample fault operation data, where the sliding slice window has K sample segmentation steps; use an LSTM network to construct M fault evolution prediction units for the M power plant equipment; use a CNN network to construct M fault type recognition units for the M power plant equipment; decompose the M groups of sample fault operation data according to the K sample segmentation steps, and then use the decomposition results to optimize the parameter tuning of the M fault evolution prediction units and M fault type recognition units; map and cascade the optimized M fault evolution prediction units and M fault type recognition units to complete the construction of the M fault prediction models.

[0082] Further, the fault prediction model construction module 11 in the multi-dimensional data fusion analysis device for digital twins is further used for:

[0083] Decompose the first set of sample fault operation data into the first set of sample fault evolution data and the first set of sample fault steady-state data, where the time span of the sample fault evolution data is K - 1 sample segmentation steps; use the first set of sample fault evolution data and the first set of sample fault steady-state data as training data to construct the first fault evolution prediction unit of the first power plant equipment; after extracting the first set of sample fault types from the first set of sample fault steady-state data, use the first set of sample fault steady-state data and the first set of sample fault types as training data to construct the first fault type recognition unit of the first power plant equipment; cascade the first fault evolution prediction unit and the first fault type recognition unit to complete the construction of the first fault prediction model; and so on, use the M historical fault operation data as training data and use the LSTM network to construct the M fault prediction models.

[0084] Further, the digital twin modeling module 12 in the multidimensional data fusion analysis device for a digital twin also is further configured to:

[0085] Obtain the point cloud data of the power generation system by laser scanning the power generation system; obtain the real scene data of the power generation system by photogrammetry of the power generation system; after interactively obtaining the power plant design information of the target power plant, construct a three-dimensional design model of the power generation system according to the power plant design information, and then use the point cloud data of the power generation system to perform model attitude calibration to obtain the spatial model of the power generation system, where the point cloud data of the power generation system, the real scene data of the power generation system, and the power plant design information constitute the multi-dimensional data of the power generation system; use SIFT to perform feature matching on the spatial model of the power generation system and the real scene data of the power generation system to obtain the real-scene space alignment coordinates; according to the real-scene space alignment coordinates, perform virtual-real fusion on the real scene data of the power generation system and the spatial model of the power generation system, and then introduce a real-time rendering engine to complete the construction of the digital twin of the power generation system; and so on, use the multi-system multi-dimensional data for multi-level data fusion modeling to generate the multi-system digital twins.

[0086] Further, the power plant twin construction module 13 in the multidimensional data fusion analysis device for a digital twin also is further configured to:

[0087] Perform fault prediction complexity analysis based on the M historical fault operation data, and calculate and output multiple system-level prediction complexities according to the analysis results; with reference to the multiple system-level prediction complexities, perform computing node allocation for the multiple power plant systems in a distributed computing architecture to obtain K groups of power systems for the K computing nodes; according to the M device source identifiers, after loading the M fault prediction models into the multiple system-level digital twins, using the K groups of power systems as replication guides, copy the multiple system-level digital twins to the K computing nodes to complete the construction of the initial twin of the power plant.

[0088] Further, the fault transfer analysis module 14 in the multidimensional data fusion analysis device for digital twins is further configured to:

[0089] Perform fault cross-system conduction analysis on the target power plant to obtain a system-level state transition diagram; perform fault cross-device conduction analysis on the target power plant according to the M historical fault operation data to obtain multiple device-level state transition diagrams for the multiple power plant systems; through multi-dimensional data fusion of the system-level state transition diagram and the multiple device-level state transition diagrams, obtain a power plant fault state transition diagram.

[0090] Further, the operation state visualization module 16 in the multidimensional data fusion analysis device for digital twins is further configured to:

[0091] Collect pixel data for the M power plant devices to obtain the M real-time device operation data; in the power plant digital twin, use the M fault prediction models to perform fault feature analysis on the M real-time device operation data and output the M device operation states; perform fault feature aggregation on the M historical fault operation data to obtain M groups of fault operation performances; use the M device operation states to traverse the M groups of fault operation performances and schedule M real-time operation performances; use the real-time rendering engine to visualize the M real-time operation performances in the power plant digital twin for the M device operation states; perform fault transfer analysis on the M real-time operation performances in the power plant fault state transition diagram and visualize the analysis results in the power plant digital twin for the fault state transition trend.

[0092] The various embodiments in this specification are described in a progressive manner, and the key point of each embodiment is the difference from other embodiments. The foregoing Figure 1The multi-dimensional data fusion analysis method and specific example in the first embodiment are equally applicable to the multi-dimensional data fusion analysis device for digital twins in this embodiment. Through the foregoing detailed description of the multi-dimensional data fusion analysis method for digital twins, those skilled in the art can clearly understand the multi-dimensional data fusion analysis device for digital twins in this embodiment. Therefore, for the sake of simplicity of the specification, it will not be elaborated herein. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0093] The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0094] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is also intended to include these changes and modifications.

Claims

1. A multi-dimensional data fusion and analysis method for digital twins, characterized in that, Including: Using an LSTM network to construct M fault prediction models for M power plant devices in a target power plant, where the M power plant devices have M device source identifiers; After collecting multi-dimensional data from multiple power plant systems to obtain multiple system-level multi-dimensional data, using the multiple system-level multi-dimensional data for multi-level data fusion modeling to generate multiple system-level digital twins; According to the M device source identifiers, loading the M fault prediction models into the multiple system-level digital twins to complete the construction of the initial power plant twin; Performing multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram; Performing fault conduction rendering on the initial power plant twin according to the power plant fault state transition diagram to obtain a power plant digital twin; After loading the M real-time device operation data of the M power plant devices into the M fault prediction models of the power plant digital twin to obtain M device operation states, visualizing the M device operation states in the power plant digital twin; The using an LSTM network to construct M fault prediction models for M power plant devices in a target power plant includes: Collecting device data from the target power plant to obtain M historical fault operation data of the M power plant devices; After segmenting the M historical fault operation data based on the fault occurrence nodes, using a sliding slice window to normalize the segmentation results to obtain M sets of sample fault operation data, where the sliding slice window has K sample segmentation steps; Using an LSTM network to construct M fault evolution prediction units for the M power plant devices; Using a CNN network to construct M fault type identification units for the M power plant devices; According to the K sample segmentation steps, decomposing the M sets of sample fault operation data, and using the decomposition results to optimize the parameter tuning of the M fault evolution prediction units and M fault type identification units; Mapping and cascading the optimized M fault evolution prediction units and M fault type identification units to complete the construction of the M fault prediction models.

2. The multi-dimensional data fusion analysis method for digital twins according to claim 1, characterized in that After collecting multi-dimensional data from multiple power plant systems to obtain multiple system-level multi-dimensional data, using the multiple system-level multi-dimensional data for multi-level data fusion modeling to generate multiple system-level digital twins, including: Obtaining point cloud data of the power generation system by laser scanning the power generation system; Obtaining real scene data of the power generation system by photogrammetry of the power generation system; After interactively obtaining the power plant design information of the target power plant, constructing a three-dimensional design model of the power generation system according to the power plant design information, and then using the point cloud data of the power generation system for model attitude calibration to obtain a spatial model of the power generation system, where the point cloud data of the power generation system, the real scene data of the power generation system, and the power plant design information constitute the multi-dimensional data of the power generation system; Using SIFT to perform feature matching on the spatial model of the power generation system and the real scene data of the power generation system to obtain real scene space alignment coordinates; According to the real scene space alignment coordinates, performing virtual-real fusion on the real scene data of the power generation system and the spatial model of the power generation system, and then introducing a real-time rendering engine to complete the construction of the digital twin of the power generation system; And so on, using the multiple system-level multi-dimensional data for multi-level data fusion modeling to generate the multiple system-level digital twins.

3. The multi-dimensional data fusion analysis method for digital twins according to claim 1, wherein Perform multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram, including: Perform fault cross-system conduction analysis on the target power plant to obtain a system-level state transition diagram; Perform fault cross-device conduction analysis on the target power plant according to the M historical fault operation data to obtain multiple device-level state transition diagrams of multiple power plant systems; Obtain the power plant fault state transition diagram by performing multi-dimensional data fusion on the system-level state transition diagram and multiple device-level state transition diagrams.

4. The multi-dimensional data fusion analysis method for digital twins according to claim 3, characterized in that, After loading the M real-time device operation data of the M power plant devices into the M fault prediction models of the power plant digital twin to obtain the M device operation states, perform visualization of the M device operation states in the power plant digital twin, including: Collect pixel data for the M power plant devices to obtain the M real-time device operation data; In the power plant digital twin, use the M fault prediction models to perform fault feature analysis on the M real-time device operation data and output the M device operation states; Perform fault feature aggregation on the M historical fault operation data to obtain M sets of fault operation performances; Traverse the M sets of fault operation performances with the M device operation states to schedule M real-time operation performances; Use the real-time rendering engine to perform visualization of the M device operation states with the M real-time operation performances in the power plant digital twin; Perform fault transfer analysis on the M real-time operation performances in the power plant fault state transition diagram and perform visualization of the fault state transition trend in the power plant digital twin for the analysis results.

5. The multi-dimensional data fusion analysis method for digital twins according to claim 1, characterized in that Including: Decompose the first set of sample fault operation data into the first set of sample fault evolution data and the first set of sample fault steady-state data, where the time span of the sample fault evolution data is K-1 sample segmentation steps; Use the first set of sample fault evolution data and the first set of sample fault steady-state data as training data to construct the first fault evolution prediction unit of the first power plant device; After extracting the first set of sample fault types of the first set of sample fault steady-state data, use the first set of sample fault steady-state data and the first set of sample fault types as training data to construct the first fault type recognition unit of the first power plant device; Cascade the first fault evolution prediction unit and the first fault type recognition unit to complete the construction of the first fault prediction model; And so on, use the M historical fault operation data as training data and use the LSTM network to construct the M fault prediction models.

6. The multi-dimensional data fusion analysis method for digital twins according to claim 1, characterized in that According to the M device source identifiers, load the M fault prediction models into the multiple system-level digital twins to complete the construction of the initial power plant twin, including: Perform fault prediction complexity analysis according to the M historical fault operation data and calculate and output multiple system-level prediction complexities according to the analysis results; With reference to the multiple system-level prediction complexities, computing node allocation for the multiple power plant systems is performed in a distributed computing architecture to obtain K sets of power systems for K computing nodes; After loading the M fault prediction models into the multiple system-level digital twins according to the M device source identifiers, using the K sets of power systems as replication guidance, the multiple system-level digital twins are replicated to the K computing nodes to complete the construction of the initial power plant twin.

7. A multi-dimensional data fusion and analysis device for digital twins, characterized in that, For implementing the steps of the method for multi-dimensional data fusion analysis for a digital twin according to any one of claims 1 to 6, the apparatus for multi-dimensional data fusion analysis for a digital twin includes: A fault prediction model construction module, which is configured to construct M fault prediction models for M power plant devices in a target power plant using an LSTM network, where the M power plant devices have M device source identifiers; A digital twin modeling module, which is configured to perform multi-level data fusion modeling using the multiple system-level multi-dimensional data after collecting multi-dimensional data of multiple power plant systems to generate multiple system-level digital twins; A power plant twin construction module, which is configured to load the M fault prediction models into the multiple system-level digital twins according to the M device source identifiers to complete the construction of the initial power plant twin; A fault transfer analysis module, which is configured to perform multi-level fault transfer analysis on the target power plant to obtain a power plant fault state transition diagram; A fault conduction rendering module, which is configured to perform fault conduction rendering on the initial power plant twin according to the power plant fault state transition diagram to obtain a power plant digital twin; An operating state visualization module, which is configured to visualize the operating states of the M devices after loading the real-time device operating data of the M power plant devices into the M fault prediction models of the power plant digital twin.

Images