Offshore wind power multi-dimensional data comprehensive monitoring method and system

Abstract

The invention discloses an offshore wind power multi-dimensional data comprehensive monitoring method and system, and belongs to the field of data monitoring. The method comprises the steps that offshore wind power plant multi-dimensional data are collected and processed, and data features are extracted; constructing a wind power plant multi-dimensional data association network diagram and optimizing the association network diagram; predicting the operation state of the wind power plant and dynamically optimizing state prediction model parameters; constructing a digital twinborn model of the offshore wind plant for analogue simulation; and feeding back operation information of the wind power plant in real time and optimizing an operation strategy. According to the method, the burden of data transmission and processing can be reduced, the efficiency of a data monitoring system is improved, the calculation cost can be saved, the real-time data processing performance is improved, integral monitoring and maintenance are facilitated, comprehensive evaluation of the health state of the wind power equipment can be ensured, and the change of the state of the wind power equipment can be flexibly coped according to the real-time monitored data; the accuracy and the real-time performance of health monitoring are ensured, and the calculation time and the resource consumption are reduced.

Description

Technical Field

[0001] The present invention relates to the field of data monitoring, specifically to a method and system for comprehensive monitoring of multi-dimensional data of offshore wind power. Background Art

[0002] With the continuous growth of global energy demand and the increasing emphasis on environmental protection, offshore wind power, as a clean, renewable energy source, is rapidly developing worldwide. Offshore wind farms, due to their efficient utilization of offshore wind resources, have become an important solution to energy problems. However, due to the complex marine environment and the harsh operating conditions of wind turbines, the operation and maintenance of offshore wind farm equipment face numerous challenges, including equipment aging, wind speed fluctuations, and unstable power output. Therefore, it is of great significance to propose an efficient and intelligent multi-dimensional data integrated monitoring method for the real-time monitoring and data analysis of offshore wind turbines.

[0003] Existing offshore wind power multidimensional data comprehensive monitoring methods and systems have a high burden of data transmission and processing, which reduces the efficiency of the data monitoring system, increases computing costs, and is not conducive to overall monitoring and maintenance. In addition, existing offshore wind power multidimensional data comprehensive monitoring methods and systems cannot comprehensively assess the health status of wind power equipment, cannot flexibly respond to changes in the status of wind power equipment, and have low accuracy and real-time performance in health monitoring. To this end, we propose a method and system for offshore wind power multidimensional data comprehensive monitoring. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for comprehensive monitoring of multi-dimensional data of offshore wind power, which comprises the following steps:

[0005] Collect and process multi-dimensional data of offshore wind farms and extract data features;

[0006] Construct a wind farm multi-dimensional data association network diagram and optimize the association network diagram;

[0007] Predict the operating status of wind farms and dynamically optimize the parameters of the status prediction model;

[0008] Build a digital twin model of an offshore wind farm for simulation;

[0009] Provide real-time feedback of wind farm operation information and optimize operation strategies.

[0010] As a preferred solution of the comprehensive monitoring method of offshore wind power multidimensional data described in the present invention, the method of collecting and processing offshore wind farm multidimensional data and extracting data features is to pre-process the multidimensional data collected by various sensors and transmit it as input data to a pre-trained CNN network, the CNN network performs a convolution operation on each group of input data to obtain a feature map corresponding to each data, performs nonlinear processing on each group of feature maps through a ReLU activation function, repeatedly performs convolution and activation operations on the processed data until a preset number of iterations is reached, the CNN network output layer outputs the final extracted feature data, performs standardization processing on each group of extracted feature data, calculates the covariance matrix between each feature data, and performs eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors, sorts the eigenvectors from large to small according to the size of the eigenvalues, and selects the eigenvectors with eigenvalues ​​greater than a preset threshold as the principal component direction, projects the original feature data onto the selected principal component direction, and obtains the feature data after dimensionality reduction.

[0011] As a preferred embodiment of the method for comprehensive monitoring of offshore wind power multi-dimensional data described in the present invention, the method comprises: constructing a wind farm multi-dimensional data association network diagram and optimizing the association network diagram by collecting multi-dimensional data of the operation of each wind turbine in the offshore wind farm, including data on wind speed, temperature, humidity, and wind farm power output, and calculating the Euclidean distance between each group of wind turbines. Based on the collected wind turbine data, each wind turbine in the wind farm is used as a node, and the connection relationship between each wind turbine is used as an edge to construct a fully weighted graph, and each difference is used as the weight of the edge connecting the corresponding two groups of wind turbines;

[0012] By sorting the weights of all edges in the fully weighted graph in ascending order, a corresponding edge list is constructed. From the sorted edge list, the edges with the smallest weights are selected in sequence, and the nodes connected by the selected edges are regarded as a group of connected components. If the two wind turbines connected by the current edge are not in the same connected component, the current edge is added to the minimum spanning tree;

[0013] Repeatedly search and add the minimum edge until the minimum spanning tree connects all wind turbine nodes, and then output the final minimum spanning tree to obtain the multidimensional data association network diagram corresponding to the offshore wind farm.

[0014] As a preferred embodiment of the method for comprehensive monitoring of offshore wind power multi-dimensional data according to the present invention, predicting the operating status of a wind farm and dynamically optimizing prediction parameters includes extracting characteristic data from historical wind farm data from a generated multi-dimensional data association network diagram, including wind speed, wind direction, power generation power, blade angle, ambient temperature and humidity, and turbine speed, normalizing each set of characteristic data, dividing each set of processed characteristic data into a training set and a validation set, and constructing a state prediction model based on an LSTM model structure.

[0015] The training set is input into the state prediction model. Through forward propagation, the model input layer receives the training data. Multiple LSTM units in the LSTM layer extract the temporal dependencies in each set of training data. The LSTM extraction results are then input into the fully connected layer to generate the final prediction value. The output layer accepts the final prediction value, post-processes the prediction value, and inputs the prediction information of future wind speed or power generation.

[0016] As a preferred solution of the offshore wind power multi-dimensional data comprehensive monitoring method described in the present invention, wherein: the predicting of the wind farm operating status and the dynamic optimization of the prediction parameters further include measuring the difference between the model prediction value and the true value through the mean square error loss function;

[0017] Starting from the output layer of the state prediction model, the difference value is back-propagated layer by layer based on the chain rule until it is transmitted to the input layer. At the same time, the gradient descent algorithm is used to calculate the gradient value of the difference value for each network layer of the state prediction model. Based on the calculated gradient value, the parameters of each network layer of the state prediction model are updated through the SGD optimization algorithm. After each round of training, the performance indicators of the state prediction model are evaluated through the validation set, and the model hyperparameters are adjusted according to the performance indicators.

[0018] As a preferred solution of the method for comprehensive monitoring of offshore wind power multi-dimensional data according to the present invention, the method of predicting the operating status of a wind farm and dynamically optimizing the prediction parameters further includes:

[0019] The current feature data sequence after dimensionality reduction is input into the state prediction model. The state prediction model outputs future prediction values ​​through temporal dependency extraction of multi-layer LSTM and processing of the fully connected layer. The prediction values ​​are normalized through the Softmax function in the output layer to generate and output the actual wind speed or power generation.

[0020] The present invention provides the following technical solution: an offshore wind power multi-dimensional data integrated monitoring system, comprising: a data acquisition module, a data preprocessing module, an extraction and dimensionality reduction module, a construction optimization module, a prediction and analysis module, a status monitoring module, a virtual modeling module, a visual alarm module, a storage management module, and a decision support module;

[0021] The data acquisition module is used to obtain multi-dimensional data of offshore wind farms in real time;

[0022] The data preprocessing module is used to perform preliminary cleaning and processing on each group of acquired offshore wind farm multi-dimensional data;

[0023] The extraction and dimensionality reduction module is used to extract key features from the pre-processed multi-dimensional data of the offshore wind farm and reduce the dimensionality of the multi-dimensional feature data;

[0024] The construction optimization module constructs a wind farm association network diagram based on the processed multi-dimensional data of the offshore wind farm, and optimizes the wind farm association network diagram;

[0025] The prediction and analysis module analyzes the feature data after dimensionality reduction based on the constructed association network diagram to predict the future operating status of the offshore wind farm;

[0026] The status monitoring module monitors and evaluates the health status of wind power equipment based on real-time offshore wind farm multi-dimensional data and offshore wind farm multi-dimensional data prediction values;

[0027] The virtual modeling module simulates the operation process of the offshore wind farm based on real-time offshore wind farm multi-dimensional data and associated network diagrams, and simulates the power generation efficiency and equipment performance of the wind farm under different conditions;

[0028] The visual alarm module is used to visualize the real-time offshore wind farm multi-dimensional data, the predicted value of offshore wind farm multi-dimensional data and the health status of wind power equipment, automatically issue an alarm for abnormal data, and send a notification to the operation and maintenance personnel;

[0029] The storage management module is used to store and manage real-time offshore wind farm multi-dimensional data, associated network diagrams, offshore wind farm multi-dimensional data prediction values, and health status of wind power equipment;

[0030] The decision support module is used to provide operation and maintenance personnel with intelligent and optimal wind turbine scheduling solutions, equipment maintenance plans and operation optimization strategies based on the multi-dimensional data prediction values ​​of the offshore wind farm and the health status of the wind power equipment.

[0031] As a preferred embodiment of the offshore wind power multi-dimensional data integrated monitoring system of the present invention, the monitoring and evaluation of the state monitoring module includes taking the state characteristics or health indicators of each group of wind power equipment in the offshore wind farm as nodes in the state space, initializing the number of exploration bodies, the pheromone intensity of each node transfer path, the heuristic information, the pheromone volatility rate, and the weight coefficient;

[0032] Randomly initialize the position of each explorer, that is, the position of any node in the state space. Each explorer calculates the selection probability of each candidate node based on the pheromone strength and heuristic information between the current node and the candidate node. Each explorer randomly selects the next node based on the calculated selection probability distribution;

[0033] When the explorer moves to a new node, it records the current node in the taboo table and removes the corresponding node from the candidate nodes based on the nodes recorded in the taboo table. Each explorer moves between nodes through multiple path selections until it reaches the set end node and stops selecting. After completing a path search, the explorer updates the pheromone concentration on the path based on the effect of the corresponding search path.

[0034] The path is iterated through multiple exploration body path searches and pheromone updates until the pheromone change reaches the preset threshold, and the optimal evaluation path for the device health status is formed. The device health status prediction is updated based on the evaluation results of the optimal path, and a comprehensive evaluation is performed on the status nodes on the path to generate a health score for the device.

[0035] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned method for comprehensive monitoring of multi-dimensional data of offshore wind power are implemented.

[0036] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for comprehensive monitoring of multi-dimensional data of offshore wind power.

[0037] Beneficial effects of the present invention: The offshore wind power multi-dimensional data comprehensive monitoring method collects multi-dimensional data of the operation of each wind turbine in the offshore wind farm, including wind speed, temperature, humidity and wind farm power output data, and calculates the Euclidean distance between each group of wind turbines to determine the difference between the wind turbines. According to the collected wind turbine data, each wind turbine in the wind farm is used as a node, and the connection relationship between each wind turbine is used as an edge to construct a fully weighted graph, and each difference is used as the weight of the edge connecting the corresponding two groups of wind turbines. By sorting the weights of all edges in the fully weighted graph in ascending order, a corresponding edge list is constructed, and then the edges after sorting are sorted. In the edge list, the edges with the smallest weight are selected in turn, and the nodes connected by the selected edges are regarded as a group of connected components. If the two wind turbines connected by the edge are not in the same connected component, the edge is added to the minimum spanning tree, and the minimum edge is repeatedly searched and added until the minimum spanning tree connects all wind turbine nodes, and the final minimum spanning tree is output to obtain the multidimensional data association network diagram corresponding to the offshore wind farm. This can reduce the burden of data transmission and processing, improve the efficiency of the data monitoring system, save computing costs, and improve the performance of real-time data processing, which is convenient for overall monitoring and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 This is an overall flow chart of a method for comprehensive monitoring of multi-dimensional data of offshore wind power provided by the first embodiment of the present invention;

[0040] Figure 2 A schematic diagram of a multi-dimensional data association network diagram of a method for comprehensive monitoring of multi-dimensional data of offshore wind power provided by the first embodiment of the present invention;

[0041] Figure 3 This is a structural diagram of an offshore wind power multi-dimensional data integrated monitoring system provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0042] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0043] Example 1, with reference to Figures 1 and 2 According to one embodiment of the present invention, a method for comprehensive monitoring of multi-dimensional data of offshore wind power is provided, comprising:

[0044] Collect and process multi-dimensional data of offshore wind farms and extract data features;

[0045] Constructing a wind farm multi-dimensional data association network diagram and optimizing the association network diagram;

[0046] Predict wind farm operating status and dynamically optimize prediction parameters;

[0047] Build a digital twin model of an offshore wind farm for simulation;

[0048] Provide real-time feedback of wind farm operation information and optimize operation strategies.

[0049] Collect and process multi-dimensional data of offshore wind farms and extract data features.

[0050] Specifically, multi-dimensional data of offshore wind farms are collected and processed and data features are extracted.

[0051] Specifically, the multidimensional data collected by various sensors are preprocessed and transmitted as input data to the pre-trained CNN network. The CNN network performs a convolution operation on each set of input data to obtain the feature map corresponding to each data, and performs nonlinear processing on each set of feature maps through the ReLU activation function. Then, the processed data is repeatedly convolved and activated until the preset number of iterations is reached. The output layer of the CNN network outputs the final extracted feature data, and the extracted feature data are standardized. Then, the covariance matrix between the feature data is calculated, and the eigenvalue decomposition of the covariance matrix is ​​performed to obtain the eigenvalue and its corresponding eigenvector. According to the size of the eigenvalue, the eigenvectors are sorted from large to small, and the eigenvectors with eigenvalues ​​greater than the preset threshold are selected as the principal component directions. The original feature data are projected onto the selected principal component directions to obtain the feature data after dimensionality reduction.

[0052] In addition, in this embodiment, it should be noted that the historical multidimensional data of the offshore wind farm is extracted from the storage management module, and each group of historical multidimensional data is denoised, outliers are deleted, and missing values ​​are filled through the data preprocessing module, and each group of processed historical multidimensional data is normalized, and then each group of processed historical multidimensional data is divided into a training set and a test set, and the training set is input into the CNN network, and then the convolution layer in the CNN network performs a convolution operation on each group of input data to obtain the feature map corresponding to each data, and then the feature map of each group is pooled by the pooling layer, and the feature map of each group is nonlinearly transformed using the ReLU activation function. After processing, the processed data is repeatedly convolved, pooled, and activated until the preset number of iterations is reached. The extracted feature data is output through the CNN network output layer, and the loss value between each feature data and the real feature data is calculated through the cross entropy loss function. The calculated loss value is then propagated upward layer by layer starting from the CNN network output layer, and the gradient value of the loss value for each parameter of the CNN network is calculated. The CNN network parameters are then adjusted using the Adam optimizer. After each round of training, the performance of the CNN network is verified through the test set. If the loss value of the CNN network on the test set is higher than the preset threshold, the training is repeated, otherwise, the training is stopped.

[0053] Construct a wind farm multidimensional data association network diagram and optimize the association network diagram.

[0054] Specifically, refer to Figure 2 It can be seen that the multidimensional data of the operation of each wind turbine in the offshore wind farm, including wind speed, temperature, humidity and wind farm power output data, is collected, and the Euclidean distance between each group of wind turbines is calculated to determine the difference between the wind turbines. According to the collected wind turbine data, each wind turbine in the wind farm is used as a node, and the connection relationship between each wind turbine is used as an edge to construct a fully weighted graph, and each difference is used as the weight of the connecting edge of the corresponding two groups of wind turbines. By sorting the weights of all edges in the fully weighted graph in ascending order, a corresponding edge list is constructed, and then the edge with the smallest weight is selected from the sorted edge list in turn, and the nodes connected by the selected edge are regarded as a group of connected components. If the two wind turbines connected by the edge are not in the same connected component, the edge is added to the minimum spanning tree, and the minimum edge is repeatedly searched and added until the minimum spanning tree connects all wind turbine nodes and stops, and the final minimum spanning tree is output to obtain the multidimensional data association network diagram corresponding to the offshore wind farm.

[0055] Predict wind farm operating status and dynamically optimize prediction parameters.

[0056] Specifically, characteristic data in the historical data of the wind farm, including wind speed, wind direction, power generation, blade angle, ambient temperature and humidity, and turbine speed, are extracted from the generated multidimensional data association network diagram, and each group of characteristic data is normalized. The processed groups of characteristic data are divided into training sets and validation sets. Then, based on the LSTM model structure, a state prediction model is constructed, and the training set is input into the state prediction model. Through forward propagation, the model input layer receives the training data, and then multiple LSTM units in the LSTM layer extract the temporal dependencies in each group of training data. The LSTM extraction results are then input into the fully connected layer to generate the final prediction value. The output layer accepts the final prediction value and post-processes the prediction value, inputs the prediction information of future wind speed or power generation, and measures the difference between the model prediction value and the true value through the mean square error loss function. The difference value starts from the output layer of the state prediction model based on the chain rule. Back propagation is performed layer by layer until it is transmitted to the input layer. At the same time, the gradient descent algorithm is used to calculate the gradient value of the difference value for each network layer of the state prediction model, and based on the calculated gradient value, the parameters of each network layer of the state prediction model are updated through the SGD optimization algorithm. After each round of training, the performance indicators of the state prediction model are evaluated through the validation set, and the hyperparameters of the model are adjusted according to the performance indicators. The current feature data sequence after dimensionality reduction is input into the state prediction model. The state prediction model outputs the future prediction value through the temporal dependency extraction of multi-layer LSTM and the fully connected layer processing, and the prediction value is normalized by the Softmax function in the output layer to generate the actual wind speed or power generation and output it.

[0057] Build a digital twin model of the offshore wind farm for simulation.

[0058] Provide real-time feedback of wind farm operation information and optimize operation strategies.

[0059] Example 2, reference Figure 3 In one embodiment of the present invention, a system for a method for comprehensive monitoring of offshore wind power multi-dimensional data is provided, comprising: a data acquisition module, a data preprocessing module, an extraction and dimensionality reduction module, a construction and optimization module, a prediction and analysis module, a condition monitoring module, a virtual modeling module, a visualization and alarm module, a storage and management module, and a decision support module;

[0060] The data acquisition module is used to acquire multidimensional data of offshore wind farms in real time; the data preprocessing module is used to perform preliminary cleaning and processing on each group of acquired multidimensional data of offshore wind farms; the extraction and dimensionality reduction module is used to extract key features from the preprocessed multidimensional data of offshore wind farms and reduce the dimensionality of the multidimensional feature data.

[0061] The construction optimization module constructs the wind farm's associated network diagram based on the processed multi-dimensional data of the offshore wind farm and optimizes the wind farm's associated network diagram; the prediction analysis module analyzes the characteristic data after dimensionality reduction based on the constructed associated network diagram and predicts the future operating status of the offshore wind farm; the status monitoring module monitors and evaluates the health status of wind power equipment based on real-time offshore wind farm multi-dimensional data and the predicted values ​​of offshore wind farm multi-dimensional data.

[0062] Specifically, the state characteristics or health indicators of each group of wind power equipment in the offshore wind farm are respectively used as nodes in the state space. Then, the number of exploration bodies, the pheromone strength, heuristic information, pheromone volatility and weight coefficient of each node transfer path are initialized, and the position of each exploration body is randomly initialized, that is, the position of any node in the state space. Each exploration body calculates the selection probability of each candidate node based on the pheromone strength and heuristic information between the current node and the candidate node. Each exploration body randomly selects the next node according to the calculated selection probability distribution. When the exploration body moves to the new node, the node is recorded in the taboo table and the selection probability is calculated based on the selection probability distribution. The nodes recorded in the taboo table are removed from the corresponding candidate nodes. Each explorer moves between nodes through multiple path selections until it reaches the set end node and stops selecting. After completing a path search, the explorer updates the pheromone concentration on the path according to the effect of the corresponding search path. The path is iterated through multiple explorer path searches and pheromone updates until the pheromone change reaches the preset threshold and stops. The optimal evaluation path for the health status of the equipment is formed. The health status prediction of the equipment is updated according to the evaluation results of the optimal path, and the status nodes on the path are comprehensively evaluated to generate a health score for the equipment.

[0063] The virtual modeling module simulates the operation process of the offshore wind farm based on the real-time offshore wind farm multidimensional data and associated network diagrams, and simulates the power generation efficiency and equipment performance of the wind farm under different conditions; the visual alarm module is used to visualize the real-time offshore wind farm multidimensional data, the offshore wind farm multidimensional data predicted values ​​and the health status of the wind power equipment, automatically issues alarms for abnormal data, and sends notifications to the operation and maintenance personnel; the storage management module is used to store and manage the real-time offshore wind farm multidimensional data, associated network diagrams, the offshore wind farm multidimensional data predicted values ​​and the health status of the wind power equipment; the decision support module is used to provide the operation and maintenance personnel with intelligent and optimal wind turbine scheduling solutions, equipment maintenance plans and operation optimization strategies based on the offshore wind farm multidimensional data predicted values ​​and the health status of the wind power equipment.

[0064] It should be further explained that:

[0065] The state characteristics or health indicators of each group of wind power equipment in the offshore wind farm are respectively used as nodes in the state space, and then the positions of each exploration body are randomly initialized, that is, the position of any node in the state space. Each exploration body calculates the selection probability of each candidate node based on the pheromone strength and heuristic information between the current node and the candidate node. Each exploration body randomly selects the next node according to the calculated selection probability distribution. When the exploration body moves to the new node, the node is recorded in the taboo table, and the corresponding node in the candidate node is removed according to the node recorded in the taboo table. Each exploration body moves between nodes through multiple path selections until it reaches the set end node. Stop selection. After completing a path search, the explorer updates the pheromone concentration on the path according to the effect of the corresponding search path. The path is iterated through multiple explorer path searches and pheromone updates until the pheromone change reaches the preset threshold and stops. The optimal evaluation path of the equipment health status is formed. The equipment health status prediction is updated according to the evaluation results of the optimal path, and the status nodes on the path are comprehensively evaluated to generate the health score of the equipment. This can ensure a comprehensive evaluation of the health status of the wind power equipment. It can flexibly respond to changes in the status of the wind power equipment based on real-time monitoring data, ensure the accuracy and real-time performance of health monitoring, and reduce computing time and resource consumption.

[0066] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0067] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0068] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0069] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0070] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for comprehensive monitoring of multi-dimensional data of offshore wind power, characterized in that: include: Collect and process multi-dimensional data of offshore wind farms and extract data features; Construct a wind farm multi-dimensional data association network diagram and optimize the association network diagram; Predict the operating status of wind farms and dynamically optimize the parameters of the status prediction model; Build a digital twin model of an offshore wind farm for simulation; Provide real-time feedback of wind farm operation information and optimize operation strategies.

2. The method for comprehensive monitoring of multi-dimensional data of offshore wind power according to claim 1, characterized in that: The method for collecting and processing multidimensional data of offshore wind farms and extracting data features is to pre-process the multidimensional data collected by various sensors and transmit the pre-trained CNN network as input data, the CNN network performs a convolution operation on each group of input data to obtain a feature map corresponding to each data, performs nonlinear processing on each group of feature maps through a ReLU activation function, repeatedly performs convolution and activation operations on the processed data until a preset number of iterations is reached, the CNN network output layer outputs the final extracted feature data, performs standardization processing on each group of extracted feature data, calculates the covariance matrix between each feature data, and performs eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors, sorts the eigenvectors from large to small according to the size of the eigenvalues, selects the eigenvectors with eigenvalues ​​greater than a preset threshold as the principal component direction, projects the original feature data onto the selected principal component direction, and obtains the feature data after dimensionality reduction.

3. The method for comprehensive monitoring of offshore wind power multi-dimensional data according to claim 2, characterized in that: The method of constructing a wind farm multi-dimensional data association network diagram and optimizing the association network diagram is to collect multi-dimensional data of the operation of each wind turbine in the offshore wind farm, including wind speed, temperature, humidity and wind farm power output data, and calculate the Euclidean distance between each group of wind turbines. Based on the collected wind turbine data, each wind turbine in the wind farm is used as a node, and the connection relationship between each wind turbine is used as an edge to construct a fully weighted graph, and each difference is used as the weight of the edge connecting the corresponding two groups of wind turbines; By sorting the weights of all edges in the fully weighted graph in ascending order, a corresponding edge list is constructed. From the sorted edge list, the edges with the smallest weights are selected in sequence, and the nodes connected by the selected edges are regarded as a group of connected components. If the two wind turbines connected by the current edge are not in the same connected component, the current edge is added to the minimum spanning tree; Repeatedly search and add the minimum edge until the minimum spanning tree connects all wind turbine nodes, and then output the final minimum spanning tree to obtain the multidimensional data association network diagram corresponding to the offshore wind farm.

4. The method for comprehensive monitoring of multi-dimensional data of offshore wind power according to claim 3, characterized in that: The method of predicting the operating status of a wind farm and dynamically optimizing prediction parameters includes extracting characteristic data from historical wind farm data from a generated multidimensional data association network diagram, including wind speed, wind direction, generated power, blade angle, ambient temperature and humidity, and turbine speed, normalizing each set of characteristic data, dividing each set of processed characteristic data into a training set and a validation set, and constructing a state prediction model based on an LSTM model structure. The training set is input into the state prediction model. Through forward propagation, the model input layer receives the training data. Multiple LSTM units in the LSTM layer extract the temporal dependencies in each set of training data. The LSTM extraction results are then input into the fully connected layer to generate the final prediction value. The output layer accepts the final prediction value, post-processes the prediction value, and inputs the prediction information of future wind speed or power generation.

5. The method for comprehensive monitoring of multi-dimensional data of offshore wind power according to claim 4, characterized in that: The predicting of the wind farm operating status and dynamically optimizing the prediction parameters further includes measuring the difference between the model prediction value and the true value through a mean square error loss function; Starting from the output layer of the state prediction model, the difference value is back-propagated layer by layer based on the chain rule until it is transmitted to the input layer. At the same time, the gradient descent algorithm is used to calculate the gradient value of the difference value for each network layer of the state prediction model. Based on the calculated gradient value, the parameters of each network layer of the state prediction model are updated through the SGD optimization algorithm. After each round of training, the performance indicators of the state prediction model are evaluated through the validation set, and the model hyperparameters are adjusted according to the performance indicators.

6. The method for comprehensive monitoring of multi-dimensional data of offshore wind power according to claim 5, characterized in that: The predicting of wind farm operating status and dynamically optimizing prediction parameters also includes: The current feature data sequence after dimensionality reduction is input into the state prediction model. The state prediction model outputs future prediction values ​​through temporal dependency extraction of multi-layer LSTM and processing of the fully connected layer. The prediction values ​​are normalized through the Softmax function in the output layer to generate and output the actual wind speed or power generation.

7. A system using the offshore wind power multi-dimensional data integrated monitoring method according to any one of claims 1 to 6, characterized in that: It includes data acquisition module, data preprocessing module, extraction and dimensionality reduction module, construction optimization module, prediction analysis module, condition monitoring module, virtual modeling module, visual alarm module, storage management module and decision support module; The data acquisition module is used to obtain multi-dimensional data of offshore wind farms in real time; The data preprocessing module is used to perform preliminary cleaning and processing on each group of acquired offshore wind farm multi-dimensional data; The extraction and dimensionality reduction module is used to extract key features from the pre-processed multi-dimensional data of the offshore wind farm and reduce the dimensionality of the multi-dimensional feature data; The construction optimization module constructs a wind farm association network diagram based on the processed multi-dimensional data of the offshore wind farm, and optimizes the wind farm association network diagram; The prediction and analysis module analyzes the feature data after dimensionality reduction based on the constructed association network diagram to predict the future operating status of the offshore wind farm; The status monitoring module monitors and evaluates the health status of wind power equipment based on real-time offshore wind farm multi-dimensional data and offshore wind farm multi-dimensional data prediction values; The virtual modeling module simulates the operation process of the offshore wind farm based on real-time offshore wind farm multi-dimensional data and associated network diagrams, and simulates the power generation efficiency and equipment performance of the wind farm under different conditions; The visual alarm module is used to visualize the real-time offshore wind farm multi-dimensional data, the predicted value of offshore wind farm multi-dimensional data and the health status of wind power equipment, automatically issue an alarm for abnormal data, and send a notification to the operation and maintenance personnel; The storage management module is used to store and manage real-time offshore wind farm multi-dimensional data, associated network diagrams, offshore wind farm multi-dimensional data prediction values, and health status of wind power equipment; The decision support module is used to provide operation and maintenance personnel with intelligent and optimal wind turbine scheduling solutions, equipment maintenance plans and operation optimization strategies based on the multi-dimensional data prediction values ​​of the offshore wind farm and the health status of the wind power equipment.

8. The offshore wind power multi-dimensional data integrated monitoring system according to claim 7, characterized in that: The monitoring and evaluation of the state monitoring module includes taking the state characteristics or health indicators of each group of wind turbines in the offshore wind farm as nodes in the state space, initializing the number of exploration bodies, the pheromone intensity of each node transfer path, the heuristic information, the pheromone volatility rate and the weight coefficient; Randomly initialize the position of each explorer, that is, the position of any node in the state space. Each explorer calculates the selection probability of each candidate node based on the pheromone strength and heuristic information between the current node and the candidate node. Each explorer randomly selects the next node based on the calculated selection probability distribution; When the explorer moves to a new node, it records the current node in the taboo table and removes the corresponding node from the candidate nodes based on the nodes recorded in the taboo table. Each explorer moves between nodes through multiple path selections until it reaches the set end node and stops selecting. After completing a path search, the explorer updates the pheromone concentration on the path based on the effect of the corresponding search path. The path is iterated through multiple exploration body path searches and pheromone updates until the pheromone change reaches the preset threshold, and the optimal evaluation path for the device health status is formed. The device health status prediction is updated based on the evaluation results of the optimal path, and a comprehensive evaluation is performed on the status nodes on the path to generate a health score for the device.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the offshore wind power multi-dimensional data comprehensive monitoring method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for comprehensive monitoring of multi-dimensional data of offshore wind power are implemented as described in any one of claims 1 to 7.

Images