Digital Twin-Based Switching Operation Management System and Method for Offshore Wind Farms

Abstract

Disclosed is a switching operation management system and method for an offshore wind farm based on digital twin. First, the time series of the operation state data of the offshore wind farm is arranged into an operation state multi-parameter time series collaborative input matrix. Then, the operation state multi-parameter time series collaborative input matrix is subjected to feature extraction by an operation state time series feature extractor to obtain an operation state multi-parameter time series correlation feature vector. Next, the obtained time series of load demands is arranged and time series feature analysis is performed to obtain a power grid load time series correlation feature vector. Then, the operation state multi-parameter time series correlation feature vector and the power grid load time series correlation feature vector are fused. Finally, the recommended switching order label is determined based on the obtained operation state-load demand collaborative semantic correlation feature and is displayed on a display screen. In this way, real-time monitoring, prediction, and optimization of the physical system can be achieved.

Description

Technical Field

[0001] The present application relates to the field of offshore wind farms, and more specifically, to a switching operation management system and method for offshore wind farms based on digital twin. Background Art

[0002] An offshore wind farm refers to a wind power farm built on the ocean, which can utilize offshore wind energy for power generation and has the advantages of high efficiency, energy conservation, environmental protection, etc. With the development of renewable energy and the increasing demand for clean energy, the construction and operation of offshore wind farms have gradually become an important energy supply method. However, the operation of offshore wind farms also faces some challenges, such as changes in wind speed and direction, fluctuations in grid load, and fan failures. These factors will affect the stability and reliability of offshore wind farms. Therefore, in order to ensure the safe operation of offshore wind farms, reasonable switching operation management is required.

[0003] However, traditional switching operation management systems for offshore wind farms usually rely on empirical rules and the empirical judgment of operation and maintenance personnel to determine the switching operation sequence. This method may be limited by personal experience and subjective judgment, and is prone to misjudgment or unable to fully consider the complex operating conditions of offshore wind turbines and fluctuations in grid load.

[0004] Therefore, an optimized switching operation management system for offshore wind farms based on digital twin is desired. Summary of the Invention

[0005] In view of this, the present application proposes a switching operation management system and method for offshore wind farms based on digital twin, which can utilize digital twin technology to collect the operation status data and load demand of offshore wind farms in real time through monitoring, and introduce data processing and analysis algorithms at the backend to perform collaborative analysis of these data, so as to judge the switching sequence and formulate a plan.

[0006] According to an aspect of the present application, there is provided a switching operation management system for an offshore wind farm based on digital twin, which includes:

[0007] A data acquisition module for obtaining the time series of the operation status data and the time series of the load demand of the offshore wind farm, wherein the operation status data includes the fan output, wind direction, and wind speed, and the load demand is the grid load;

[0008] A multi-parameter time series data integration module for arranging the time series of the operation status data into a multi-parameter time series collaborative input matrix according to the time dimension and the parameter sample dimension;

[0009] The operating state multi-parameter time-series collaborative feature extraction module is used to extract features from the operating state multi-parameter time-series collaborative input matrix through an operating state time-series feature extractor based on a deep neural network model to obtain an operating state multi-parameter time-series correlation feature vector;

[0010] The power grid load time-series feature extraction module is used to arrange the load demand time series into a power grid load time-series input vector according to the time dimension, and then perform time-series feature analysis on the power grid load time-series input vector to obtain a power grid load time-series correlation feature vector;

[0011] The operating state - load demand collaborative semantic association coding module is used to fuse the operating state multi-parameter time-series correlation feature vector and the power grid load time-series correlation feature vector to obtain an operating state - load demand collaborative semantic association feature; and

[0012] The switching sequence determination and display module is used to determine a recommended switching sequence label based on the operating state - load demand collaborative semantic association feature, and display the recommended switching sequence label on a display screen.

[0013] According to another aspect of the present application, a switching operation management method for an offshore wind farm based on digital twin is provided, which includes:

[0014] Obtain the operating state data time series and load demand time series of the offshore wind farm, where the operating state data includes wind turbine output, wind direction, and wind speed, and the load demand is the power grid load;

[0015] Arrange the operating state data time series into an operating state multi-parameter time-series collaborative input matrix according to the time dimension and parameter sample dimension;

[0016] Extract features from the operating state multi-parameter time-series collaborative input matrix through an operating state time-series feature extractor based on a deep neural network model to obtain an operating state multi-parameter time-series correlation feature vector;

[0017] Arrange the load demand time series into a power grid load time-series input vector according to the time dimension, and then perform time-series feature analysis on the power grid load time-series input vector to obtain a power grid load time-series correlation feature vector;

[0018] Fuse the operating state multi-parameter time-series correlation feature vector and the power grid load time-series correlation feature vector to obtain an operating state - load demand collaborative semantic association feature; and

[0019] Based on the operating state - load demand collaborative semantic association feature, determine a recommended switching sequence label, and display the recommended switching sequence label on a display screen.

[0020] In this application, first, the time series of the operation status data of the offshore wind farm is arranged into an operation status multi-parameter time series collaborative input matrix. Then, the operation status time series feature extractor extracts features from the operation status multi-parameter time series collaborative input matrix to obtain an operation status multi-parameter time series correlation feature vector. Next, the obtained load demand time series is arranged and time series feature analysis is performed to obtain a power grid load time series correlation feature vector. Then, the operation status multi-parameter time series correlation feature vector and the power grid load time series correlation feature vector are fused. Finally, based on the obtained operation status-load demand collaborative semantic association feature, the recommended switching sequence label is determined and displayed on the display screen. In this way, real-time monitoring, prediction, and optimization of the physical system can be achieved.

[0021] Other features and aspects of the present application will become clear from the following detailed description of the present application with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The drawings included in and constituting a part of the specification illustrate exemplary embodiments, features, and aspects of the present application together with the specification and are used to explain the principles of the present application.

[0023] Figure 1 A block diagram showing a digital twin-based switching operation management system for an offshore wind farm according to an embodiment of the present application.

[0024] Figure 2 A flowchart showing a digital twin-based switching operation management method for an offshore wind farm according to an embodiment of the present application.

[0025] Figure 3 A schematic diagram showing the architecture of a digital twin-based switching operation management method for an offshore wind farm according to an embodiment of the present application.

[0026] Figure 4 An application scenario diagram showing a digital twin-based switching operation management system for an offshore wind farm according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall also fall within the protection scope of the present application.

[0028] As shown in this application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0029] Various exemplary embodiments, features, and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the drawings denote elements having the same or similar functions. Although various aspects of the embodiments are shown in the drawings, the drawings do not have to be drawn to scale unless otherwise specified.

[0030] In addition, for a better description of the present application, numerous specific details are given in the following detailed implementation. Those skilled in the art should understand that the present application can also be implemented without certain specific details. In some instances, methods, means, elements, and circuits well-known to those skilled in the art are not described in detail to highlight the gist of the present application.

[0031] It should be understood that the operating state and load demand of an offshore wind farm are key factors affecting the operation and power generation efficiency of the wind farm. To ensure the safe operation of the wind farm and optimize the power generation benefit, it is necessary to monitor and analyze the operating state data such as the output of the wind turbines, wind direction, and wind speed, and formulate a reasonable switching scheme according to the grid load demand. Moreover, since digital twin is a technology that uses data-driven models to simulate the behavior and performance of real physical systems, it can achieve real-time monitoring, prediction, and optimization of physical systems.

[0032] Based on this, Figure 1 A block diagram of a digital-twin-based switching operation management system for an offshore wind farm according to an embodiment of the present application is shown. As Figure 1As shown in the figure, the switching operation management system 100 for an offshore wind farm based on digital twin according to an embodiment of the present application includes: a data acquisition module 110, configured to obtain the time series of the operation state data and the time series of the load demand of the offshore wind farm, where the operation state data includes the fan output, the wind direction and the wind speed, and the load demand is the grid load; an operation state multi-parameter time series data integration module 120, configured to arrange the time series of the operation state data into an operation state multi-parameter time series collaborative input matrix according to the time dimension and the parameter sample dimension; an operation state multi-parameter time series collaborative feature extraction module 130, configured to perform feature extraction on the operation state multi-parameter time series collaborative input matrix through an operation state time series feature extractor based on a deep neural network model to obtain an operation state multi-parameter time series correlation feature vector; a grid load time series feature extraction module 140, configured to arrange the time series of the load demand into a grid load time series input vector according to the time dimension, and then perform time series feature analysis on the grid load time series input vector to obtain a grid load time series correlation feature vector; an operation state-load demand collaborative semantic association encoding module 150, configured to fuse the operation state multi-parameter time series correlation feature vector and the grid load time series correlation feature vector to obtain an operation state-load demand collaborative semantic association feature; and a switching sequence determination and display module 160, configured to determine a recommended switching sequence label based on the operation state-load demand collaborative semantic association feature, and display the recommended switching sequence label on a display screen.

[0033] It should be understood that the data acquisition module 110 is responsible for collecting the operation state data of the wind farm, such as the fan output, the wind direction and the wind speed, as well as the grid load demand. The function of the operation state multi-parameter time series data integration module 120 is to integrate the collected operation state data into a form suitable for subsequent processing and analysis. The goal of the operation state multi-parameter time series collaborative feature extraction module 130 is to extract key features from the operation state data for subsequent analysis and decision-making. The function of the grid load time series feature extraction module 140 is to extract relevant features from the grid load data for subsequent analysis and decision-making. The purpose of the operation state-load demand collaborative semantic association encoding module 150 is to provide a basis for subsequent switching sequence determination by encoding and fusing the association features of the operation state and the load demand. The function of the switching sequence determination and display module 160 is to determine the optimal switching sequence according to the analyzed feature information and display the result to the operator so that they can perform corresponding operations. These modules together constitute the switching operation management system for an offshore wind farm based on digital twin. By collecting, integrating, extracting and analyzing the operation state data and load demand data of the offshore wind farm, and fusing relevant features, a recommended sequence label for the switching operation is finally provided to improve the operation efficiency and safety of the wind farm.

[0034] Specifically, the technical concept of this application is to utilize digital twin technology to collect the operation status data (such as fan output, wind direction, and wind speed) and load demand (grid load) of an offshore wind farm through real-time monitoring, and introduce data processing and analysis algorithms at the backend to conduct collaborative analysis of these data, so as to judge the switching sequence and formulate a plan. In this way, the switching plan can be optimized according to the operation status and load demand of the offshore wind farm, thereby improving the operation efficiency and power generation capacity of the offshore wind farm, and ensuring the safe operation of the offshore wind farm at the same time.

[0035] Specifically, in the technical solution of this application, first, a time series of operation status data and a time series of load demand of the offshore wind farm are obtained, where the operation status data includes fan output, wind direction, and wind speed, and the load demand is the grid load. Then, considering that the operation status data of the offshore wind farm includes fan output, wind direction, and wind speed data, these operation status data items not only have dynamic change rules in the time dimension, but also there is a time-series collaborative correlation relationship between these operation status data items. Therefore, in order to capture the trend and pattern of the operation status of the offshore wind farm changing with time, in the technical solution of this application, the time series of operation status data needs to be arranged into an operation status multi-parameter time-series collaborative input matrix according to the time dimension and parameter sample dimension, so as to integrate the distribution information of each operation status data item in the time dimension and sample dimension. In this way, the correlation between different operation status items of the offshore wind farm can be taken into account to better analyze the interaction and influence between them.

[0036] Then, the operation status multi-parameter time-series collaborative input matrix is subjected to feature mining through an operation status time-series feature extractor based on a convolutional neural network model to extract the multi-parameter time-series collaborative correlation feature information between the operation status data items of the offshore wind farm, thereby obtaining an operation status multi-parameter time-series correlation feature vector.

[0037] Correspondingly, in the operation status multi-parameter time-series collaborative feature extraction module 130, the deep neural network model is a convolutional neural network model, that is, the operation status time-series feature extractor based on the deep neural network model is an operation status time-series feature extractor based on a convolutional neural network model.

[0038] It is worth mentioning that the Convolutional Neural Network (CNN) is a deep learning model. The core idea of the convolutional neural network is a network structure composed of convolutional layers, pooling layers, and fully connected layers, which extracts features and classifies the input data. The following are several key components of the convolutional neural network: 1. Convolutional Layer: The convolutional layer is the core component of the convolutional neural network, which extracts features from the input data through a series of convolutional operations. The convolutional operation uses a set of learnable filters (also called convolutional kernels) to perform a sliding window-based convolutional operation on the input data, thereby extracting local features. The convolutional layer can extract different features through multiple filters. 2. Pooling Layer: The pooling layer is used to downsample the feature map output by the convolutional layer, reducing the size and number of parameters of the feature map. Common pooling operations include Max Pooling and Average Pooling, which respectively select the maximum value or average value in the local area as the value after pooling. 3. Fully Connected Layer: The fully connected layer flattens the feature map output by the pooling layer into a one-dimensional vector and performs classification or regression through a series of fully connected operations. The role of the fully connected layer is to map the extracted features to the probability distribution of the target category or numerical prediction. The convolutional neural network constructs a deep network by stacking multiple convolutional layers, pooling layers, and fully connected layers, and using non-linear activation functions (such as ReLU), thereby realizing the extraction of high-level features and complex pattern recognition of the input data. In the running state time series feature extractor based on the convolutional neural network, it can extract key time series features from the multi-parameter time series collaborative input matrix of the running state for subsequent analysis and decision-making. Through convolutional operations, the network can automatically learn features at different time scales, thereby better capturing the time series correlation of the running state data. Such feature extraction ability helps to improve the system's understanding and representation of the running state data, providing more accurate and reliable recommendations for subsequent switching sequence determination.

[0039] Furthermore, considering the load demand, for example, the grid load also has dynamic change characteristics in the time dimension, that is, each load demand in the load demand time series has a time series correlation relationship in the time dimension. Therefore, in the technical solution of this application, the load demand time series is arranged in the time dimension as a grid load time series input vector and then undergoes feature mining in the grid load time series feature extractor based on a one-dimensional convolutional layer to extract the time series feature information of the grid load in the time dimension, thereby obtaining a grid load time series correlation feature vector.

[0040] Accordingly, the grid load time series feature extraction module 140 is configured to: arrange the load demand time series in the time dimension as the grid load time series input vector, and then obtain the grid load time series correlation feature vector through the grid load time series feature extractor based on the one-dimensional convolutional layer. It is worth mentioning that the one-dimensional convolutional layer (1D Convolutional Layer) is a type of layer in the convolutional neural network, which is used to process data with sequence structure, such as text, time series, etc. Compared with the traditional two-dimensional convolutional layer (used to process images), the one-dimensional convolutional layer only considers the local relationship in one dimension when processing data. The operation of the one-dimensional convolutional layer is similar to that of the two-dimensional convolutional layer, but only performs the sliding window-style convolutional operation in one dimension. Specifically, the one-dimensional convolutional layer uses a set of learnable filters (also called convolutional kernels) to perform convolutional operations on the input data, so as to extract local features. The size of the filter is usually specified by the user, which can control the width of the convolutional window. The convolutional operation of the one-dimensional convolutional layer can capture the local patterns and features in the input data, so it has certain advantages when processing sequence data. For example, in the grid load time series feature extraction module, the one-dimensional convolutional layer can extract the time series correlation features of the grid load from the load demand time series. Through the sliding window-style convolutional operation, the one-dimensional convolutional layer can automatically learn and capture the features at different time scales, so as to improve the understanding and representation ability of the grid load time series data. The grid load time series feature extractor based on the one-dimensional convolutional layer can be constructed by stacking multiple one-dimensional convolutional layers, pooling layers and fully connected layers to achieve the feature extraction and classification of the grid load time series data. The use of the one-dimensional convolutional layer can help the system extract the key time series correlation features from the load demand time series, and provide more accurate and useful information for subsequent analysis and decision-making.

[0041] It should be understood that in an offshore wind farm, the operating state and load demand of the offshore wind farm are interrelated. For example, changes in the operating state parameters will affect the load demand, and changes in the load demand will also feedback on the operating state of the offshore wind farm. Therefore, in order to better understand and analyze the operation of the wind farm, it is necessary to fuse the temporal characteristics of the operating state and load demand of the offshore wind farm. It should be understood that the dimensionality reduction sparse fusion module can reduce the high-dimensional feature vector to a lower dimension and retain important information. Therefore, in the technical solution of this application, the dimensionality reduction sparse fusion module is further used to fuse the multi-parameter temporal correlation feature vector of the operating state and the temporal correlation feature vector of the grid load to obtain the operating state-load demand collaborative semantic correlation feature vector. In particular, the dimensionality reduction sparse fusion module generally adopts sparse coding and dimensionality reduction techniques, and through learning the sparse representation and dimensionality reduction projection of the data, transforms the original feature vector into a more compact and meaningful feature representation. This can reduce the dimensionality of the features, remove redundant information, and retain important features. At the same time, the dimensionality reduction sparse fusion module can also consider the correlation relationship between the multi-parameter temporal correlation features of the operating state and the temporal features of the grid load, so as to better capture the collaborative semantic correlation between the operating state and load demand of the offshore wind farm. That is to say, by using the dimensionality reduction sparse fusion module, the features of the operating state and load demand of the offshore wind farm can be effectively fused, so as to more comprehensively reflect and utilize the relationship between the operating state and load demand of the wind farm, and provide more accurate and comprehensive information for subsequent classification and decision-making.

[0042] Correspondingly, the operating state-load demand collaborative semantic correlation coding module 150 includes: a collaborative semantic correlation analysis unit, configured to use the dimensionality reduction sparse fusion module to fuse the multi-parameter temporal correlation feature vector of the operating state and the temporal correlation feature vector of the grid load to obtain the operating state-load demand collaborative semantic correlation feature vector; and a collaborative semantic feature optimization unit, configured to optimize the operating state-load demand collaborative semantic correlation feature vector to obtain the optimized operating state-load demand collaborative semantic correlation feature vector as the operating state-load demand collaborative semantic correlation feature.

[0043] It should be understood that in the operating state - load demand collaborative semantic association encoding module 150, there are two key units, namely the collaborative semantic association analysis unit and the collaborative semantic feature optimization unit. The collaborative semantic association analysis unit uses a dimensionality reduction sparse fusion module to fuse two feature vectors. This fusion process can model and capture the correlation between the operating state and the load demand, thereby obtaining a more comprehensive and integrated feature representation. The collaborative semantic feature optimization unit can further extract and strengthen the useful information in the features through feature optimization, remove redundancy or noise, and make the features more accurate and discriminative. The optimized feature vector can better reflect the collaborative relationship between the operating state and the load demand, providing a more reliable feature input for the subsequent determination of the switching sequence. These two units work together to fuse and optimize the correlation features of the operating state and the load demand, obtaining the operating state - load demand collaborative semantic association features. This feature can comprehensively consider the relationship between the operating state and the load demand, providing a more comprehensive and accurate feature representation. Such a feature representation helps the system better understand and analyze the correlation between the operating state and the load demand, thereby providing a more accurate and reliable recommendation for the determination of the switching sequence.

[0044] Among them, in one example, the collaborative semantic association analysis unit is used to: use a dimensionality reduction sparse fusion module to fuse the operating state multi - parameter time - series association feature vector and the power grid load time - series association feature vector with the following fusion formula to obtain the operating state - load demand collaborative semantic association feature vector; where the fusion formula is:

[0045] Z = X * + Y *

[0046] X * = W x X

[0047] Y * = W y Y

[0048] Among them, X is the operating state multi - parameter time - series association feature vector, Y is the power grid load time - series association feature vector, W x and W y are respectively the transformation matrix of the operating state multi - parameter time - series association feature vector and the transformation matrix of the power grid load time - series association feature vector, X * and Y * are respectively the between - class scatter vectors of the operating state multi - parameter time - series association feature vector and the power grid load time - series association feature vector, and Z is the operating state - load demand collaborative semantic association feature vector.

[0049] Subsequently, the optimized operation status - load demand collaborative semantic association feature vector is passed through a classifier to obtain a classification result, and the classification result is used to represent the recommended switching sequence label. That is to say, the label of the classifier is the recommended switching sequence label. Therefore, after obtaining the classification result, the switching sequence can be judged and a scheme can be formulated based on the classification result, and the recommended switching sequence label is displayed on the display screen. In this way, the switching scheme can be optimized according to the operation status and load demand of the offshore wind farm, thereby improving the operation efficiency and power generation capacity of the offshore wind farm, and ensuring the safe operation of the offshore wind farm at the same time.

[0050] Correspondingly, the switching sequence determination and display module 160 includes: a switching sequence label recommendation unit, configured to pass the optimized operation status - load demand collaborative semantic association feature vector through a classifier to obtain a classification result, and the classification result is used to represent the recommended switching sequence label; and a switching sequence display unit, configured to display the recommended switching sequence label on the display screen.

[0051] Specifically, in one example, the switching sequence label recommendation unit includes: a fully connected encoding sub - unit, configured to perform fully connected encoding on the optimized operation status - load demand collaborative semantic association feature vector using the fully connected layer of the classifier to obtain an encoded classification feature vector; and a classification sub - unit, configured to input the encoded classification feature vector into the Softmax classification function of the classifier to obtain the classification result.

[0052] It should be understood that the role of the classifier is to use the given categories and known training data to learn classification rules and the classifier, and then classify (or predict) unknown data. Logistics regression, SVM, etc. are often used to solve binary classification problems. For multi - class classification problems, logistics regression or SVM can also be used, but it requires multiple binary classifications to form a multi - class classification, which is prone to errors and has low efficiency. The commonly used multi - class classification method is the Softmax classification function.

[0053] In the above technical solution, the multi-parameter time-series correlation feature vector of the operating state represents the sample-time series cross-dimensional local correlation feature of the operating state data, and the time-series correlation feature vector of the power grid load represents the local time-series correlation feature of the load demand. Therefore, the multi-parameter time-series correlation feature vector of the operating state and the time-series correlation feature vector of the power grid load have variability under the correlation calculation dimension based on different feature expressions of heterogeneous data. Therefore, in order to improve the fusion effect of the multi-parameter time-series correlation feature vector of the operating state and the time-series correlation feature vector of the power grid load when using the dimensionality reduction sparse fusion module, and under the category determination of the classifier, the present application preferably optimizes the fused operating state-load demand collaborative semantic correlation feature vector, for example, denoted as V.

[0054] Specifically, in an example, the collaborative semantic feature optimization unit includes: adding the eigenvalue of each position of the operating state-load demand collaborative semantic correlation feature vector V to the square root of the length of the operating state-load demand collaborative semantic correlation feature vector V and the reciprocal of the square root of the two-norm of the operating state-load demand collaborative semantic correlation feature vector V, and then taking the sum value as the exponent of the exponential function with the natural constant as the base. Then, multiplying the eigenvalue of each position of the operating state-load demand collaborative semantic correlation feature vector V by the one-norm of the operating state-load demand collaborative semantic correlation feature vector V and the weighted hyperparameter, and adding the product value to the above exponential value to obtain the optimized operating state-load demand collaborative semantic correlation feature vector.

[0055] In this way, by using the structured norm representation of the operation state - load demand collaborative semantic association feature vector V as the local canonical coordinates for each eigenvalue of the operation state - load demand collaborative semantic association feature vector V, the rotation offset of the overall vector distribution of the operation state - load demand collaborative semantic association feature vector V relative to the eigenvalues is determined for the offset prediction direction of each eigenvalue centered thereon, and eigenvalue constraints are imposed using the bounding box of the vector eigenvalue distribution of the operation state - load demand collaborative semantic association feature vector V, so as to achieve the stacked fusion of the eigenvalues of the operation state - load demand collaborative semantic association feature vector V according to the variability of the semantic content between the operation state multi - parameter time - series association feature vector V1 and the power grid load time - series association feature vector V2 in different calculation dimensions, thereby enhancing the fusion effect of the operation state multi - parameter time - series association feature vector V1 and the power grid load time - series association feature vector V2, and thus improving the accuracy of the classification result obtained by the classifier for the operation state - load demand collaborative semantic association feature vector. In this way, the switching operation plan can be optimized according to the operation state and load demand of the offshore wind farm, thereby improving the operation efficiency and power generation capacity of the offshore wind farm while ensuring the safe operation of the offshore wind farm.

[0056] In summary, the digital - twin - based switching operation management system 100 of the embodiment of the present application is elucidated, which can achieve real - time monitoring, prediction, and optimization of the physical system.

[0057] As described above, the digital - twin - based switching operation management system 100 of the embodiment of the present application can be implemented in various terminal devices, such as a server with a digital - twin - based switching operation management algorithm for an offshore wind farm. In one example, the digital - twin - based switching operation management system 100 of the offshore wind farm can be integrated into the terminal device as a software module and / or a hardware module. For example, the digital - twin - based switching operation management system 100 of the offshore wind farm can be a software module in the operating system of the terminal device, or can be an application program developed for the terminal device; of course, the digital - twin - based switching operation management system 100 of the offshore wind farm can also be one of the many hardware modules of the terminal device.

[0058] Alternatively, in another example, the digital - twin - based switching operation management system 100 of the offshore wind farm and the terminal device can also be separate devices, and the digital - twin - based switching operation management system 100 of the offshore wind farm can be connected to the terminal device through a wired and / or wireless network and transmit interaction information in accordance with a predefined data format.

[0059] Figure 2A flowchart showing a method for managing switching operations of an offshore wind farm based on digital twins according to an embodiment of the present application. Figure 3 A schematic diagram showing the system architecture of a method for managing switching operations of an offshore wind farm based on digital twins according to an embodiment of the present application. As Figure 2 and Figure 3 shown, the method for managing switching operations of an offshore wind farm based on digital twins according to an embodiment of the present application includes: S110, obtaining a time series of operation state data and a time series of load demand of the offshore wind farm, where the operation state data includes wind turbine output, wind direction, and wind speed, and the load demand is the grid load; S120, arranging the time series of operation state data in a multi-parameter time series collaborative input matrix according to the time dimension and the parameter sample dimension; S130, extracting features from the multi-parameter time series collaborative input matrix of the operation state through an operation state time series feature extractor based on a deep neural network model to obtain a multi-parameter time series correlation feature vector of the operation state; S140, arranging the time series of load demand in a grid load time series input vector according to the time dimension, and then performing time series feature analysis on the grid load time series input vector to obtain a grid load time series correlation feature vector; S150, fusing the multi-parameter time series correlation feature vector of the operation state and the grid load time series correlation feature vector to obtain an operation state-load demand collaborative semantic correlation feature; and S160, based on the operation state-load demand collaborative semantic correlation feature, determining a recommended switching order label and displaying the recommended switching order label on a display screen.

[0060] In a possible implementation, the deep neural network model is a convolutional neural network model.

[0061] Here, those skilled in the art can understand that the specific operations of each step in the above method for managing switching operations of an offshore wind farm based on digital twins have been described in detail above with reference to Figure 1 the description of the system for managing switching operations of an offshore wind farm based on digital twins, and therefore, the repeated description thereof will be omitted.

[0062] Figure 4 A diagram showing an application scenario of a system for managing switching operations of an offshore wind farm based on digital twins according to an embodiment of the present application. As Figure 4 shown, in this application scenario, first, a time series of operation state data of the offshore wind farm (for example, Figure 4 D1 shown in Figure 4as shown in D2), where the operating state data includes the fan output, wind direction, and wind speed, and the load demand is the grid load. Then, the time series of the operating state data and the time series of the load demand are input into a server (e.g., Figure 4 as shown in S), where the server can use the digital twin-based switching operation management algorithm for the offshore wind farm to process the time series of the operating state data and the time series of the load demand to obtain a classification result representing the recommended switching order label.

[0063] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory including computer program instructions, and the above computer program instructions can be executed by a processing component of the device to complete the above method.

[0064] This application can be a system, a method, and / or a computer program product. The computer program product can include a computer-readable storage medium having thereon computer-readable program instructions for causing a processor to implement various aspects of this application.

[0065] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of an instruction, and the module, program segment, or part of an instruction includes one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions marked in the block may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0066] The above has described the embodiments of this application. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, the practical application, or the improvement of the technology in the market, or to enable other ordinary skill in the art in the technical field to understand the disclosed embodiments.

Claims

1. A switching operation management system for an offshore wind farm based on digital twin, characterized in that, Including: A data acquisition module, configured to obtain the time series of the operation status data and the time series of the load demand of an offshore wind farm, wherein the operation status data includes the output of the wind turbines, the wind direction, and the wind speed, and the load demand is the grid load; An operation status multi-parameter time series data integration module, configured to arrange the time series of the operation status data into an operation status multi-parameter time series collaborative input matrix according to the time dimension and the parameter sample dimension; An operation status multi-parameter time series collaborative feature extraction module, configured to extract features from the operation status multi-parameter time series collaborative input matrix through an operation status time series feature extractor based on a deep neural network model to obtain an operation status multi-parameter time series correlation feature vector; A grid load time series feature extraction module, configured to arrange the time series of the load demand into a grid load time series input vector according to the time dimension, and then perform time series feature analysis on the grid load time series input vector to obtain a grid load time series correlation feature vector; An operation status - load demand collaborative semantic association encoding module, configured to fuse the operation status multi-parameter time series correlation feature vector and the grid load time series correlation feature vector to obtain an operation status - load demand collaborative semantic association feature; and A switching sequence determination and display module, configured to determine a recommended switching sequence label based on the operation status - load demand collaborative semantic association feature, and display the recommended switching sequence label on a display screen.

2. The digital twin-based switching operation management system for an offshore wind farm according to claim 1, wherein, The deep neural network model is a convolutional neural network model.

3. The digital-twin-based switching operation management system for an offshore wind farm according to claim 2, wherein, The grid load time series feature extraction module is configured to: Arrange the time series of the load demand into the grid load time series input vector according to the time dimension, and then obtain the grid load time series correlation feature vector through a grid load time series feature extractor based on a one-dimensional convolutional layer.

4. The digital twin-based switching operation management system for an offshore wind farm according to claim 3, wherein, The operation status - load demand collaborative semantic association encoding module includes: A collaborative semantic association analysis unit, configured to use a dimensionality reduction sparse fusion module to fuse the operation status multi-parameter time series correlation feature vector and the grid load time series correlation feature vector to obtain an operation status - load demand collaborative semantic association feature vector; and A collaborative semantic feature optimization unit, configured to optimize the features of the operation status - load demand collaborative semantic association feature vector to obtain an optimized operation status - load demand collaborative semantic association feature vector as the operation status - load demand collaborative semantic association feature.

5. The digital twin-based switching operation management system for an offshore wind farm according to claim 4, wherein The collaborative semantic association analysis unit is configured to: Use a dimensionality reduction sparse fusion module to fuse the operation status multi-parameter time series correlation feature vector and the grid load time series correlation feature vector according to the following fusion formula to obtain the operation status - load demand collaborative semantic association feature vector; wherein, the fusion formula is: Z = X * + Y * X * = W x X Y * = W y Y Among them, X is the multi-parameter time-series correlation feature vector of the operating state, Y is the time-series correlation feature vector of the grid load, W x and W y are respectively the transformation matrix of the multi-parameter time-series correlation feature vector of the operating state and the transformation matrix of the time-series correlation feature vector of the grid load, X * and Y * are respectively the between-class scatter vectors of the multi-parameter time-series correlation feature vector of the operating state and the between-class scatter vector of the time-series correlation feature vector of the grid load, and Z is the operating state-load demand collaborative semantic correlation feature vector.

6. The digital twin-based switching operation management system for an offshore wind farm according to claim 5, wherein The switching sequence determination and display module includes: A switching sequence label recommendation unit, configured to obtain a classification result by passing the optimized operation status - load demand collaborative semantic association feature vector through a classifier, and the classification result is used to represent the recommended switching sequence label; and A switching sequence display unit, configured to display the recommended switching sequence label on a display screen.

7. The digital twin-based switching operation management system for an offshore wind farm according to claim 6, wherein, The switching sequence label recommendation unit includes: A fully connected encoding sub-unit, configured to perform fully connected encoding on the optimized operation state-load demand collaborative semantic association feature vector by using the fully connected layer of the classifier to obtain an encoded classification feature vector; and A classification sub-unit, configured to input the encoded classification feature vector into the Softmax classification function of the classifier to obtain the classification result.

8. A switching operation management method for an offshore wind farm based on digital twin, characterized in that, It includes: Obtain the operation state data time series and the load demand time series of the offshore wind farm, wherein the operation state data includes the fan output, the wind direction and the wind speed, and the load demand is the grid load; Arrange the operation state data time series into an operation state multi-parameter time series collaborative input matrix according to the time dimension and the parameter sample dimension; Extract features from the operation state multi-parameter time series collaborative input matrix through an operation state time series feature extractor based on a deep neural network model to obtain an operation state multi-parameter time series association feature vector; After arranging the load demand time series into a grid load time series input vector according to the time dimension, perform time series feature analysis on the grid load time series input vector to obtain a grid load time series association feature vector; Fuse the operation state multi-parameter time series association feature vector and the grid load time series association feature vector to obtain an operation state-load demand collaborative semantic association feature; and Based on the operation state-load demand collaborative semantic association feature, determine the recommended switching sequence label, and display the recommended switching sequence label on the display screen.

9. The method for managing switching operations of an offshore wind farm based on digital twin according to claim 8, wherein The deep neural network model is a convolutional neural network model.

Images