Operation maintenance management system and method for wind power generation equipment

Abstract

The invention relates to the technical field of intelligent management, and particularly discloses an operation maintenance management system and method for wind power generation equipment, and the system employs an artificial intelligence technology based on deep learning to monitor and analyze the operation data and environment data of the wind power generation equipment, and captures the operation state time sequence change characteristics of the wind power generation equipment. And whether the performance of the equipment is abnormal or not is intelligently judged based on the time sequence response characteristics of the operation state of the wind power generation equipment relative to the environment data. Therefore, real-time monitoring, performance evaluation and intelligent maintenance of the wind power generation equipment can be realized, so that the operation efficiency and stability of the wind power generation equipment are improved.

Description

Technical Field

[0001] This application relates to the field of intelligent management technology, and more specifically, to an operation and maintenance management system and method for a wind power generation device. Background Art

[0002] Wind power generation refers to the process of converting the kinetic energy of wind into electrical energy. A wind power generation device can utilize the power of wind and drive a generator to generate electrical energy through the rotation of the wind turbine blades of a wind power generation unit. Specifically, when the wind blows towards the wind turbine blades of a wind power generation unit, due to the special shape of the blades, a part of the kinetic energy of the wind is converted into the rotational kinetic energy of the blades, and the rotational kinetic energy further drives the generator connected to the wind turbine to rotate, thereby converting the kinetic energy into electrical energy. With the increasing demand for renewable energy in society, wind power generation, as a green and environmentally friendly energy supply method, has been widely applied globally.

[0003] Since wind power generation devices are usually built in open outdoor environments, they are inevitably affected by natural factors such as wind, temperature, and humidity. To ensure the stability of the devices, operation and maintenance management are required. However, the traditional maintenance management methods for wind power devices usually rely on manual inspections and regular maintenance. This method is not only inefficient but also has problems such as untimely monitoring and low accuracy, making it difficult to achieve real-time monitoring and early warning. In addition, since wind power generation devices are usually installed in remote areas, it is particularly difficult to perform real-time monitoring and maintenance of the devices.

[0004] Therefore, an optimized operation and maintenance management system and method for wind power generation devices are expected. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. Embodiments of this application provide an operation and maintenance management system and method for a wind power generation device, which utilize artificial intelligence technology based on deep learning to monitor and analyze the operation data and environmental data of the wind power generation device, capture the time-series change characteristics of the operation state of the wind power generation device and the time-series change characteristics of the environmental data, and intelligently judge whether there is an abnormality in the performance of the device based on the time-series response characteristics of the operation state of the wind power generation device relative to the environmental data. In this way, real-time monitoring, performance evaluation, and intelligent maintenance of the wind power generation device can be achieved to improve the operation efficiency and stability of the wind power generation device.

[0006] Correspondingly, according to one aspect of this application, an operation and maintenance management system for a wind power generation device is provided, which includes:

[0007] A wind power generation information acquisition module, which is used to acquire the operation data of a wind power generation device at multiple predetermined time points within a predetermined time period, as well as the environmental data at multiple predetermined time points within the predetermined time period. Among them, the operation data includes the power, wind turbine speed, current and temperature of the device, and the environmental data includes the wind speed, wind direction and air pressure;

[0008] A device operation status time series feature extraction module, which is used to perform time series correlation coding on the operation data at the multiple predetermined time points to obtain a device operation status time series feature vector;

[0009] An environmental time series feature extraction module, which is used to extract time series features from the environmental data at the multiple predetermined time points to obtain an environmental data time series feature vector;

[0010] A time series correlation response module, which is used to perform correlation coding and correlation feature enhancement on the device operation status time series feature vector and the environmental data time series feature vector to obtain a significant device operation status - environmental time series response feature matrix;

[0011] A device performance analysis module, which is used to determine whether there is an abnormality in the performance of the wind power generation device based on the significant device operation status - environmental time series response feature matrix.

[0012] In the above operation and maintenance management system of the wind power generation device, the device operation status time series feature extraction module includes: an operation data regularization unit, which is used to regularize the operation data at the multiple predetermined time points to obtain an operation data input matrix; an operation status time series feature extraction unit, which is used to pass the operation data input matrix through a device operation status time series correlation encoder based on a convolutional neural network to obtain the device operation status time series feature vector.

[0013] In the above operation and maintenance management system of the wind power generation device, the data regularization unit is used to: arrange the operation data at the multiple predetermined time points in a time dimension and a sample dimension to form the operation data input matrix.

[0014] In the above operation and maintenance management system of the wind power generation device, the environmental time series feature extraction module includes: an environmental data regularization unit, which is used to arrange the environmental data at the multiple predetermined time points in a time dimension and a sample dimension to form an environmental data input matrix; an environmental time series feature extraction unit, which is used to pass the environmental data input matrix through an environmental time series feature extractor based on a convolutional neural network to obtain the environmental data time series feature vector.

[0015] In the above operation and maintenance management system of the wind power generation equipment, the time series correlation response module includes: an association unit, configured to perform association encoding on the time series feature vector of the equipment operation state and the time series feature vector of the environmental data to obtain an equipment operation state-environment time series response feature matrix; a feature saliency unit, configured to pass the equipment operation state-environment time series response feature matrix through a feature saliency detector based on a spatial attention layer to obtain the saliency equipment operation state-environment time series response feature matrix.

[0016] In the above operation and maintenance management system of the wind power generation equipment, the association unit includes: a dynamic optimization subunit, configured to perform differential entropy quantization dynamic optimization on the time series feature vector of the equipment operation state and the time series feature vector of the environmental data respectively to obtain an optimized time series feature vector of the equipment operation state and an optimized time series feature vector of the environmental data; a feature fusion subunit, configured to perform weighted fusion on the optimized time series feature vector of the equipment operation state and the optimized time series feature vector of the environmental data to obtain an equipment operation state-environment time series response feature vector; an autocorrelation subunit, configured to multiply the equipment operation state-environment time series response feature vector by its own transpose to obtain the equipment operation state-environment time series response feature matrix.

[0017] In the above operation and maintenance management system of the wind power generation equipment, the dynamic optimization subunit is configured to: calculate a self-similarity matrix of the time series feature vector of the equipment operation state, and perform key dimension reduction on the self-similarity matrix of the time series feature vector of the equipment operation state to obtain a set of intrinsic component encoding vectors of the time series feature vector of the equipment operation state; input the set of intrinsic component encoding vectors of the time series feature vector of the equipment operation state into a sequence encoder based on a forward LSTM model to obtain a set of intrinsic component context association encoding vectors of the time series feature vector of the equipment operation state; calculate the bit-by-bit fluctuation entropy between each pair of corresponding intrinsic component context association encoding vectors and intrinsic component encoding vectors of the time series feature vector of the equipment operation state in the set of intrinsic component context association encoding vectors of the time series feature vector of the equipment operation state and the set of intrinsic component encoding vectors of the time series feature vector of the equipment operation state to obtain a set of bit-by-bit fluctuation entropy; perform weight processing based on the Softmax function on the set of bit-by-bit fluctuation entropy to obtain a set of bit-by-bit transformation entropy adjustment parameters; based on the set of bit-by-bit transformation entropy adjustment parameters, fuse the set of time series feature vectors of the equipment operation state to obtain an optimized time series feature vector of the equipment operation state.

[0018] In the above-mentioned operation and maintenance management system of a wind power generation device, the feature saliency unit includes: a deep convolutional encoding subunit, configured to perform deep convolutional encoding on the device operating state-environment time series response feature matrix using the convolutional layer of the feature saliency device to obtain an initial convolutional feature map; a spatial attention generation subunit, configured to input the initial convolutional feature map into the spatial attention branch of the spatial attention mechanism model to obtain a spatial attention map; an activation subunit, configured to pass the spatial attention map through a Softmax activation function to obtain a spatial attention feature map; a spatial attention application subunit, configured to calculate the element-wise multiplication of the spatial attention feature map and the initial convolutional feature map to obtain a spatially enhanced device operating state-environment time series response feature map; a pooling subunit, configured to perform average pooling processing on the spatially enhanced device operating state-environment time series response feature map along the channel dimension to obtain the saliency device operating state-environment time series response feature matrix.

[0019] In the above-mentioned operation and maintenance management system of a wind power generation device, the device performance analysis module is configured to: pass the saliency device operating state-environment time series response feature matrix through a classifier to obtain a classification result, and the classification result is used to determine whether there is an abnormality in the performance of the wind power generation device.

[0020] According to another aspect of the present application, there is provided an operation and maintenance management method for a wind power generation device, which includes:

[0021] Obtain the operation data of the wind power generation device at multiple predetermined time points within a predetermined time period, and the environmental data at multiple predetermined time points within the predetermined time period, wherein the operation data includes the power, wind turbine speed, current, and temperature of the device, and the environmental data includes the wind speed, wind direction, and air pressure;

[0022] Perform time series correlation encoding on the operation data at the multiple predetermined time points to obtain a device operating state time series feature vector;

[0023] Perform time series feature extraction on the environmental data at the multiple predetermined time points to obtain an environmental data time series feature vector;

[0024] Perform correlation encoding and correlation feature enhancement on the device operating state time series feature vector and the environmental data time series feature vector to obtain a saliency device operating state-environment time series response feature matrix;

[0025] Based on the saliency device operating state-environment time series response feature matrix, determine whether there is an abnormality in the performance of the wind power generation device.

[0026] Compared with the prior art, the operation and maintenance management system and method of the wind power generation equipment provided by the present application utilize artificial intelligence technology based on deep learning to monitor and analyze the operation data and environmental data of the wind power generation equipment, capture the time-series change characteristics of the operation state of the wind power generation equipment and the time-series change characteristics of the environmental data, and intelligently judge whether there is an abnormality in the performance of the equipment based on the time-series response characteristics of the operation state of the wind power generation equipment relative to the environmental data. In this way, real-time monitoring, performance evaluation and intelligent maintenance of the wind power generation equipment can be realized to improve the operation efficiency and stability of the wind power generation equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By describing the embodiments of the present application in more detail with reference to the accompanying drawings, the above and other objects, features and advantages of the present application will become more obvious. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0028] Figure 1 It is a block diagram of an operation and maintenance management system of a wind power generation equipment according to an embodiment of the present application.

[0029] Figure 2 It is a schematic diagram of the architecture of an operation and maintenance management system of a wind power generation equipment according to an embodiment of the present application.

[0030] Figure 3 It is a block diagram of a device operation state time-series feature extraction module in an operation and maintenance management system of a wind power generation equipment according to an embodiment of the present application.

[0031] Figure 4 It is a block diagram of an environmental time-series feature extraction module in an operation and maintenance management system of a wind power generation equipment according to an embodiment of the present application.

[0032] Figure 5 It is a block diagram of a time-series correlation response module in an operation and maintenance management system of a wind power generation equipment according to an embodiment of the present application.

[0033] Figure 6 It is a flowchart of an operation and maintenance management method of a wind power generation equipment according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0035] Figure 1It is a block diagram of an operation and maintenance management system for a wind power generation device according to an embodiment of the present application.

[0036] Figure 2 It is a schematic diagram of the architecture of an operation and maintenance management system for a wind power generation device according to an embodiment of the present application. As Figure 1 and Figure 2 shown, the operation and maintenance management system 100 for a wind power generation device according to an embodiment of the present application includes: a wind power generation information acquisition module 110, configured to acquire operation data of the wind power generation device at multiple predetermined time points within a predetermined time period, and environmental data at multiple predetermined time points within the predetermined time period, wherein the operation data includes the power, wind turbine speed, current, and temperature of the device, and the environmental data includes wind speed, wind direction, and air pressure; a device operation state time series feature extraction module 120, configured to perform time series correlation encoding on the operation data at the multiple predetermined time points to obtain a device operation state time series feature vector; an environmental time series feature extraction module 130, configured to perform time series feature extraction on the environmental data at the multiple predetermined time points to obtain an environmental data time series feature vector; a time series correlation response module 140, configured to perform correlation encoding and correlation feature enhancement on the device operation state time series feature vector and the environmental data time series feature vector to obtain a significant device operation state - environmental time series response feature matrix; and a device performance analysis module 150, configured to determine whether there is an abnormality in the performance of the wind power generation device based on the significant device operation state - environmental time series response feature matrix.

[0037] As mentioned in the above background art, wind power generation is a green and renewable energy conversion method. In recent years, with the intensification of global climate change and energy crisis, wind power generation has received extensive attention and application globally. Since the power generation efficiency and stability of wind power generation devices are of great significance to people's quality of life and social and economic development, higher requirements have been put forward for the operation and maintenance management of wind power generation devices. However, traditional operation and maintenance management methods often rely on manual inspections and regular maintenance, suffering from problems such as low efficiency, difficulty in achieving real-time monitoring and early warning, etc. In addition, wind power generation devices are mostly installed in remote areas, such as mountains, coastal areas, etc., where the climate conditions are harsh, bringing great difficulties to the real-time monitoring and maintenance of the devices.

[0038] In view of the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to monitor and analyze the operation data and environmental data of wind power generation equipment, capture the time-series change characteristics of the operation state of wind power generation equipment and the time-series change characteristics of environmental data, and intelligently judge whether there is an abnormality in the performance of the equipment based on the time-series response characteristics of the operation state of the wind power generation equipment relative to environmental data. In this way, real-time monitoring, performance evaluation, and intelligent maintenance of wind power generation equipment can be achieved to improve the operation efficiency and stability of wind power generation equipment.

[0039] In the operation and maintenance management system 100 of the above wind power generation equipment, the wind power generation information acquisition module 110 is used to acquire the operation data of the wind power generation equipment at multiple predetermined time points within a predetermined time period, and the environmental data at multiple predetermined time points within the predetermined time period. Among them, the operation data includes the power, wind turbine speed, current, and temperature of the equipment, and the environmental data includes wind speed, wind direction, and air pressure. It should be understood that wind power generation equipment drives the rotation of the wind turbine blades by using the power of the wind, thereby driving the generator to generate electric energy. That is to say, the power generation efficiency of wind power generation equipment is closely related to meteorological factors. Therefore, in the technical solution of this application, the working state and power generation efficiency of the wind power generation equipment are characterized by acquiring the power, wind turbine speed, current, and temperature data during the operation of the wind power generation equipment. At the same time, the corresponding wind speed, wind direction, and air pressure data are acquired, and the correlation between the operation state of the equipment and environmental factors is established by comprehensively analyzing the operation data and environmental data of the equipment, so as to identify abnormal performance of the equipment.

[0040] In the operation and maintenance management system 100 of the above wind power generation equipment, the equipment operation state time-series feature extraction module 120 is used to perform time-series correlation encoding on the operation data at the multiple predetermined time points to obtain an equipment operation state time-series feature vector. Specifically, Figure 3 It is a block diagram of the equipment operation state time-series feature extraction module in the operation and maintenance management system of the wind power generation equipment according to an embodiment of this application. As Figure 3 shown, the equipment operation state time-series feature extraction module 120 includes: an operation data regularization unit 121, which is used to regularize the operation data at the multiple predetermined time points to obtain an operation data input matrix; an operation state time-series feature extraction unit 122, which is used to pass the operation data input matrix through an equipment operation state time-series correlation encoder based on a convolutional neural network to obtain the equipment operation state time-series feature vector.

[0041] Specifically, the operation data regularization unit 121 is configured to regularize the operation data at the multiple predetermined time points to obtain an operation data input matrix. In a specific example of the present application, the processing method of regularizing the operation data at the multiple predetermined time points to obtain an operation data input matrix is to arrange the operation data at the multiple predetermined time points into the operation data input matrix according to the time dimension and the sample dimension. It should be understood that the operation data of wind power generation equipment often has time dependence, that is, the data at the previous moment may affect the state at subsequent moments. Therefore, in order to retain the time sequence information of the operation data of wind power generation equipment and consider the time dependence between the operation data at different time points, the operation data at the multiple predetermined time points is further arranged according to the time dimension and the sample dimension to integrate the time sequence correlation information between each operation parameter, so as to facilitate capturing the dynamic change characteristics of the operation state of wind power generation equipment over time and improving the model's understanding ability of the operation state of wind power generation equipment.

[0042] Specifically, the operation state time sequence feature extraction unit 122 is configured to obtain the device operation state time sequence feature vector by passing the operation data input matrix through a device operation state time sequence correlation encoder based on a convolutional neural network. It should be understood that a convolutional neural network (CNN) is a feedforward neural network that can automatically learn the feature information in the input data through the backpropagation algorithm. In the technical solution of the present application, a device operation state time sequence correlation encoder based on a convolutional neural network is used to process the operation data input matrix. The device operation state time sequence correlation encoder can use the convolutional layer of the convolutional neural network model to extract the local structure pattern of the operation data input matrix through a sliding convolution operation, mine the local correlation information of each operation parameter in the time dimension, better capture the time sequence correlation between the device operation data, and help understand the evolution process of the device operation state. At the same time, the device operation state time sequence correlation encoder can achieve feature dimensionality reduction and abstraction by using the pooling layer and the fully connected layer of the convolutional neural network model, which helps to abstract more discriminative feature representations in the operation data input matrix, reduce the influence of data dimension and noise, reduce the risk of overfitting, and improve the generalization ability of the model.

[0043] In the above operation and maintenance management system 100 of wind power generation equipment, the environmental time sequence feature extraction module 130 is configured to extract time sequence features from the environmental data at the multiple predetermined time points to obtain an environmental data time sequence feature vector. Specifically, Figure 4 The block diagram of the environmental time sequence feature extraction module in the operation and maintenance management system of wind power generation equipment according to an embodiment of the present application. As Figure 4As shown, the environmental time series feature extraction module 130 includes: an environmental data regularization unit 131, which is used to arrange the environmental data of the multiple predetermined time points into an environmental data input matrix according to the time dimension and the sample dimension; an environmental time series feature extraction unit 132, which is used to pass the environmental data input matrix through an environmental time series feature extractor based on a convolutional neural network to obtain the environmental data time series feature vector.

[0044] Specifically, the environmental data regularization unit 131 is used to arrange the environmental data at the multiple predetermined time points according to the time dimension and the sample dimension into an environmental data input matrix. It should be understood that the environmental data at the multiple predetermined time points also have time dependence. Therefore, similarly, the environmental data at the multiple predetermined time points are arranged according to the time dimension and the sample dimension to retain the time sequence of the environmental data and integrate the time sequence correlation information between various environmental parameters, so as to better represent the dynamic changes of the operating environment of the wind power generation equipment and improve the model's ability to understand environmental factors.

[0045] Specifically, the environmental time series feature extraction unit 132 is used to pass the environmental data input matrix through an environmental time series feature extractor based on a convolutional neural network to obtain the environmental data time series feature vector. It should be understood that a convolutional neural network (CNN) can automatically learn feature information in input data. Similarly, in the technical solution of the present application, an environmental time series feature extractor based on a convolutional neural network is used to process the environmental data input matrix. The environmental time series feature extractor can extract the local feature representation of the environmental data input matrix by performing convolution and pooling operations on the environmental data input matrix, and mine the local correlation information of each environmental parameter in the time dimension, so as to better understand the changing process of the operating environment of the wind power generation equipment.

[0046] In the above-mentioned wind power generation equipment operation and maintenance management system 100, the time series association response module 140 is used to perform association coding and association feature enhancement on the equipment operation state time series feature vector and the environment data time series feature vector to obtain a significant equipment operation state-environment time series response feature matrix. Specifically, Figure 5 FIG. 1 is a block diagram of a timing-related response module in an operation and maintenance management system of a wind power generation device according to an embodiment of the present application. Figure 5 As shown, the timing association response module 140 includes: an association unit 141, which is used to associate and encode the device operating state timing feature vector and the environment data timing feature vector to obtain a device operating state-environment timing response feature matrix; a feature saliency unit 142, which is used to pass the device operating state-environment timing response feature matrix through a feature saliency device based on a spatial attention layer to obtain the salient device operating state-environment timing response feature matrix.

[0047] Specifically, the association unit 141 is configured to perform association encoding on the time-series feature vector of the device operation state and the time-series feature vector of the environmental data to obtain a device operation state-environment time-series response feature matrix. It should be understood that the operation state of a wind power generation device is often affected by environmental conditions. That is to say, there is a certain association response relationship between the operation state of a wind power generation device and environmental data. Therefore, in order to extract the time-series response relationship between the device operation state and environmental factors for evaluating the performance of a wind power generation device, the time-series feature vector of the device operation state and the time-series feature vector of the environmental data are further subjected to association encoding.

[0048] Specifically, the association unit 141 includes: a dynamic optimization subunit, configured to perform differential entropy quantization dynamic optimization on the time-series feature vector of the device operation state and the time-series feature vector of the environmental data respectively to obtain an optimized time-series feature vector of the device operation state and an optimized time-series feature vector of the environmental data; a feature fusion subunit, configured to perform weighted fusion on the optimized time-series feature vector of the device operation state and the optimized time-series feature vector of the environmental data to obtain a device operation state-environment time-series response feature vector; an autocorrelation subunit, configured to multiply the device operation state-environment time-series response feature vector by its own transpose to obtain the device operation state-environment time-series response feature matrix.

[0049] More specifically, the dynamic optimization subunit is configured to: calculate the self-similarity matrix of the time-series feature vector of the device operation state, and perform key dimension reduction on the self-similarity matrix of the time-series feature vector of the device operation state to obtain a set of inherent component encoding vectors of the time-series feature vector of the device operation state, which is expressed by the formula:

[0050]

[0051] where V represents the time-series feature vector of the device operation state, T represents the transpose of the vector, M z represents the self-similarity matrix, U represents the set of inherent component encoding vectors of the time-series feature of the device operation state, v1, v2, v m respectively represent the first, second, and mth inherent component encoding vectors of the time-series feature of the device operation state, Λ represents the diagonal matrix of the time-series feature of the device operation state after key dimension reduction, and λ1, λ m respectively represent the first and mth eigenvalues on the diagonal of the diagonal matrix of the time-series feature of the device operation state.

[0052] That is, by calculating the self-similarity matrix, the mutual correlation strength of the time-series feature vectors of the device operating state can be quantified at different time lags, revealing its dynamic coupling law, such as the time-series synchronization between power fluctuations and wind speed changes. Through this analysis, a basis for subsequent feature dimensionality reduction is provided, and the introduction of key dimensionality reduction (PCA) solves the problems of low computational efficiency and noise interference caused by excessive redundant information in high-dimensional data. Specifically, PCA maps the time-series feature vectors of the device operating state to a low-dimensional space by extracting the principal components in the self-similarity matrix, that is, the orthogonal directions with the largest variance, while retaining the most core time-series change patterns in the data, strengthening the significant correlation features between the device operating state and the environmental response, thereby improving the sensitivity and accuracy of anomaly detection.

[0053] More specifically, the dynamic optimization subunit is further configured to: input the set of the inherent component encoding vectors of the time-series feature vectors of the device operating state into a sequence encoder based on a forward LSTM model to obtain a set of context-correlated encoding vectors of the inherent components of the time-series feature vectors of the device operating state, which is expressed by the formula:

[0054] F = LSTM([v1, v2, …, v m ) = [s1, s2, …, s m

[0055] where LSTM represents the forward LSTM model, F represents the set of context-correlated encoding vectors of the inherent components of the time-series feature vectors of the device operating state, and s1, s2, s m represent the first, second, and m-th context-correlated encoding vectors of the inherent components of the time-series feature vectors of the device operating state.

[0056] That is, by introducing the forward LSTM model, the time-series dependence relationships between the inherent component encoding vectors of the time-series feature vectors of each device operating state can be captured layer by layer in a sequence modeling manner, especially the long-range context correlation. Specifically, through the gating mechanism of LSTM, the historical context information is dynamically integrated, and the set of discrete inherent component encoding vectors of the time-series feature vectors of the device operating state is transformed into a set of context-correlated encoding vectors of the inherent components of the time-series feature vectors of the device operating state with strong time-series semantics, which not only retains the core feature expression ability after PCA processing, but also strengthens the representation accuracy of nonlinear dynamic interactions, significantly improving the comprehensiveness and accuracy of device performance anomaly determination.

[0057] ​More specifically, the dynamic optimization subunit is further configured to: calculate the bit-by-bit fluctuation entropy between each corresponding device operation state time-series feature vector inherent component context correlation coding vector and device operation state time-series feature vector inherent component coding vector in the set of device operation state time-series feature vector inherent component context correlation coding vectors and the set of device operation state time-series feature vector inherent component coding vectors, which is expressed by the formula:

[0058]

[0059] where ∧ represents a logical operator, w represents a bit-by-bit comparison function, and v i represents the i-th device operation state time-series feature inherent component coding vector, represents the feature value at the j-th position of the i-th device operation state time-series feature inherent component coding vector, s i represents the i-th device operation state time-series feature inherent component context correlation coding vector, represents the feature value at the j-th position of the i-th device operation state time-series feature inherent component context correlation coding vector, ε represents a predetermined threshold, and ε can be set to 1.2. Of course, there is no specific limitation in this example and it can be adjusted according to the actual situation. r i represents the i-th bit-by-bit fluctuation matching feature vector, represents the feature value at the j-th position of the i-th bit-by-bit fluctuation matching feature vector, L represents the length of the i-th bit-by-bit fluctuation matching feature vector, and e i represents the i-th bit-by-bit fluctuation entropy.

[0060] That is, through the calculation of the bit-by-bit fluctuation entropy, the bit-level difference between the set of device operation state time-series feature vector inherent component context correlation coding vectors and the device operation state time-series feature vector inherent component coding vectors in the binary coding can be quantified, and the specific pattern of information gain or perturbation can be captured. A high entropy value may indicate a drastic change in the key information bits, while a low entropy value may reflect the stability of redundant information. In this way, the feature difference is converted into a quantifiable entropy value signal, providing a basis for differential weight allocation for subsequent weight dynamic adjustment and feature fusion, significantly improving the characterization granularity of feature differences, and enhancing the sensitivity of anomaly detection and the ability to suppress false alarms.

[0061] More specifically, the dynamic optimization subunit is further configured to: perform weight processing on the set of bit-by-bit fluctuation entropy based on the Softmax function to obtain a set of bit-by-bit transformation entropy adjustment parameters, which is expressed by the formula:

[0062] a i = softmax(e i )

[0063] Among them, softmax represents the normalized exponential function, and a i represents the i-th bitwise transformation entropy adjustment parameter.

[0064] That is, by introducing the Softmax function, the discrete entropy value is transformed into the bitwise transformation entropy adjustment parameter in the form of a probability distribution. Its exponential mechanism amplifies the weight gap between the high-entropy value component and the low-entropy value component, so that the inherent components that have a significant impact on the device performance obtain higher optimization weights, and thus are strongly focused on in subsequent feature fusion.

[0065] More specifically, the dynamic optimization subunit is further configured to: based on the set of the bitwise transformation entropy adjustment parameters, fuse the set of the device operation state time series feature vectors to obtain an optimized device operation state time series feature vector, which is expressed by the formula:

[0066]

[0067] where v f represents the optimized device operation state time series feature vector.

[0068] That is, through the way of weight processing, feature fusion no longer depends on fixed rules, but dynamically adjusts the contribution degree of each dimension according to the degree of information perturbation, so as to construct a more discriminative and robust optimized device operation state time series feature vector, significantly improving the information density and interpretability of the optimized device operation state time series feature vector, and reducing the interference of noise features on model decision-making. Here, the processing process of performing differential entropy quantization dynamic optimization on the environmental data time series feature vector to obtain the optimized environmental data time series feature vector can also refer to the above processing process for the device operation state time series feature vector.

[0069] Specifically, the feature saliency unit 142 is configured to obtain a saliency device operating state - environmental time - series response feature matrix by using a feature saliency detector based on a spatial attention layer for the device operating state - environmental time - series response feature matrix. It should be understood that each local spatial position in the device operating state - environmental time - series response feature matrix represents different features. However, some of these features are important and some are irrelevant. Therefore, in order to utilize the important feature information in the spatial domain of the device operating state - environmental time - series response feature matrix to enhance its feature expression ability, a feature saliency detector based on a spatial attention layer is further used to perform feature enhancement processing on the device operating state - environmental time - series response feature matrix in the spatial dimension. Among them, the spatial attention layer can measure the importance of features at different local spatial positions in the device operating state - environmental time - series response feature matrix, thereby realizing the saliency of features. Specifically, the spatial attention layer calculates the attention weight of each local spatial position of the device operating state - environmental time - series response feature matrix to represent the importance of the feature at that position, thereby realizing the highlighting of important features and the suppression of irrelevant features. In this way, the saliency device operating state - environmental time - series response feature matrix obtained through the processing of the spatial attention layer has higher feature expression ability, which helps to improve the accuracy of subsequent performance evaluation of wind power generation equipment.

[0070] In a specific example of the present application, the feature saliency unit 142 includes: a depth - convolution encoding subunit, configured to perform depth - convolution encoding on the device operating state - environmental time - series response feature matrix by using the convolution layer of the feature saliency detector to obtain an initial convolution feature map; a spatial attention generation subunit, configured to input the initial convolution feature map into the spatial attention branch of the spatial attention mechanism model to obtain a spatial attention map; an activation subunit, configured to obtain a spatial attention feature map by passing the spatial attention map through the Softmax activation function; a spatial attention application subunit, configured to calculate the element - wise multiplication of the spatial attention feature map and the initial convolution feature map to obtain a spatially enhanced device operating state - environmental time - series response feature map; a pooling subunit, configured to perform average pooling processing on the spatially enhanced device operating state - environmental time - series response feature map along the channel dimension to obtain the saliency device operating state - environmental time - series response feature matrix.

[0071] In the above-mentioned operation and maintenance management system 100 of wind power generation equipment, the equipment performance analysis module 150 is used to determine whether the performance of the wind power generation equipment is abnormal based on the significant equipment operation state-environment time series response feature matrix. In a specific example of the present application, the implementation method of determining whether the performance of the wind power generation equipment is abnormal based on the significant equipment operation state-environment time series response feature matrix is ​​to pass the significant equipment operation state-environment time series response feature matrix through a classifier to obtain a classification result, and the classification result is used to determine whether the performance of the wind power generation equipment is abnormal. It should be understood that the classifier is a machine learning model, and its working principle is to map the input features to the corresponding category label. In the technical solution of the present application, the significant equipment operation state-environment time series response feature matrix is ​​input into the classifier for classification operation. The classifier can learn the feature pattern in the significant equipment operation state-environment time series response feature matrix, dig out the performance law of the wind power generation equipment, and associate the learned feature pattern with the performance standard of the wind power generation equipment, and then judge whether the performance of the wind power generation equipment is abnormal, which provides an important basis for the operation and maintenance of the wind power generation equipment. For example, when the classification results indicate that there is an abnormality in equipment performance, maintenance personnel can take appropriate maintenance measures in a timely manner to prevent production losses caused by equipment failure and provide support for the optimized operation and management of the equipment.

[0072] In summary, the operation and maintenance management system of the wind power generation equipment according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to monitor and analyze the operation data and environmental data of the wind power generation equipment, captures the time series change characteristics of the operation state of the wind power generation equipment, and the time series change characteristics of the environmental data, and intelligently judges whether there is an abnormality in the performance of the equipment based on the time series response characteristics of the operation state of the wind power generation equipment relative to the environmental data. In this way, real-time monitoring, performance evaluation and intelligent maintenance of wind power generation equipment can be achieved to improve the operation efficiency and stability of wind power generation equipment.

[0073] Figure 6 FIG. 1 is a flow chart of a method for operating, maintaining and managing a wind power generation device according to an embodiment of the present application. Figure 6As shown, the operation and maintenance management method of a wind power generation device according to an embodiment of the present application includes the steps of: S110, obtaining the operation data of the wind power generation device at multiple predetermined time points within a predetermined time period, and the environmental data at multiple predetermined time points within the predetermined time period, where the operation data includes the power, wind turbine speed, current, and temperature of the device, and the environmental data includes the wind speed, wind direction, and air pressure; S120, performing time series correlation coding on the operation data at the multiple predetermined time points to obtain a time series feature vector of the device operation state; S130, performing time series feature extraction on the environmental data at the multiple predetermined time points to obtain a time series feature vector of the environmental data; S140, performing correlation coding and correlation feature enhancement on the time series feature vector of the device operation state and the time series feature vector of the environmental data to obtain a significant device operation state - environmental time series response feature matrix; S150, based on the significant device operation state - environmental time series response feature matrix, determining whether there is an abnormality in the performance of the wind power generation device.

[0074] Here, those skilled in the art can understand that the specific operations of each step in the above operation and maintenance management method of the wind power generation device have been described in detail above with reference to Figures 1 to 5 the description of the operation and maintenance management system of the wind power generation device, and therefore, the repeated description thereof will be omitted.

[0075] In the above embodiments, the descriptions of the various embodiments have their own focuses. For parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems, and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0076] In addition, in each embodiment of the present invention, the functional modules can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a hardware plus software functional module.

[0077] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be encompassed within the present invention. Any reference signs in the claims should not be construed as limiting the claims concerned.

[0078] In addition, it is obvious that the term "comprising" does not exclude other units or steps, and the singular does not exclude the plural. A plurality of units recited in the system claims can also be implemented by one unit through software or hardware.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An operation and maintenance management system for wind power generation equipment, characterized in that: include: A wind power generation information acquisition module, used to acquire the operation data of the wind power generation equipment at multiple predetermined time points within a predetermined time period, and the environmental data at multiple predetermined time points within the predetermined time period, wherein the operation data includes the power, wind rotor speed, current and temperature of the equipment, and the environmental data includes wind speed, wind direction and air pressure; A device operation state time series feature extraction module, used for performing time series correlation coding on the operation data of the plurality of predetermined time points to obtain a device operation state time series feature vector; An environmental time series feature extraction module, used for extracting time series features from the environmental data at the plurality of predetermined time points to obtain an environmental data time series feature vector; A time series association response module, used for performing association coding and association feature enhancement on the time series feature vector of the device operation state and the time series feature vector of the environment data to obtain a significant device operation state-environment time series response feature matrix; The equipment performance analysis module is used to determine whether there is any abnormality in the performance of the wind power generation equipment based on the significant equipment operating state-environment time series response characteristic matrix.

2. The operation and maintenance management system of wind power generation equipment according to claim 1, characterized in that: The device operation status time series feature extraction module includes: An operation data regularization unit, used for performing data regularization on the operation data at the plurality of predetermined time points to obtain an operation data input matrix; The operating state timing feature extraction unit is used to input the operating data into a matrix and pass it through a device operating state timing association encoder based on a convolutional neural network to obtain the device operating state timing feature vector.

3. The operation and maintenance management system of wind power generation equipment according to claim 2, characterized in that: The data regularization unit is used to: The operation data of the plurality of predetermined time points are arranged into the operation data input matrix according to the time dimension and the sample dimension.

4. The operation and maintenance management system of wind power generation equipment according to claim 3, characterized in that: The environmental temporal feature extraction module comprises: An environmental data regularization unit, used to arrange the environmental data of the plurality of predetermined time points into an environmental data input matrix according to a time dimension and a sample dimension; The environmental time series feature extraction unit is used to input the environmental data into a matrix and pass it through an environmental time series feature extractor based on a convolutional neural network to obtain the environmental data time series feature vector.

5. The operation and maintenance management system of wind power generation equipment according to claim 4, characterized in that: The timing association response module includes: an associating unit, configured to perform associative coding on the device operation state time series feature vector and the environment data time series feature vector to obtain a device operation state-environment time series response feature matrix; The feature saliency unit is used to obtain the salient device operating state-environmental temporal response feature matrix by passing the device operating state-environmental temporal response feature matrix through a feature saliency device based on a spatial attention layer.

6. The operation and maintenance management system of wind power generation equipment according to claim 5, characterized in that: The association unit comprises: A dynamic optimization subunit, used to perform differential entropy quantization dynamic optimization on the device operation state timing feature vector and the environment data timing feature vector respectively to obtain an optimized device operation state timing feature vector and an optimized environment data timing feature vector; A feature fusion subunit performs weighted fusion on the optimized device operation state time series feature vector and the optimized environment data time series feature vector to obtain a device operation state-environment time series response feature vector; The autocorrelation subunit is used to multiply the device operation state-environment timing response characteristic vector by its own transposition to obtain the device operation state-environment timing response characteristic matrix.

7. The operation and maintenance management system of wind power generation equipment according to claim 6, characterized in that: The dynamic optimization subunit is used for: Calculating the self-similarity matrix of the device operation state time series feature vector, and performing key dimension reduction on the self-similarity matrix of the device operation state time series feature vector to obtain a set of intrinsic component encoding vectors of the device operation state time series feature vector; Inputting the set of intrinsic component encoding vectors of the device operation status time series feature vector into a sequence encoder based on a forward LSTM model to obtain a set of intrinsic component context-related encoding vectors of the device operation status time series feature vector; Calculate the bit-by-bit fluctuation entropy between the set of intrinsic component context-associated coding vectors of the device operation state time series feature vector and each corresponding set of intrinsic component context-associated coding vectors of the device operation state time series feature vector and the intrinsic component coding vector of the device operation state time series feature vector in the set to obtain a set of bit-by-bit fluctuation entropy; Performing a weighting process based on a Softmax function on the set of bit-by-bit fluctuation entropies to obtain a set of bit-by-bit transformation entropy adjustment parameters; Based on the set of bit-by-bit transformation entropy adjustment parameters, the set of device operating state time series feature vectors is fused to obtain an optimized device operating state time series feature vector.

8. The operation and maintenance management system of wind power generation equipment according to claim 7, characterized in that: The feature salient unit comprises: A deep convolutional coding subunit, used for performing deep convolutional coding on the device operating state-environmental timing response feature matrix using the convolutional layer of the feature salient device to obtain an initial convolutional feature map; A spatial attention generating subunit, configured to input the initial convolutional feature map into the spatial attention branch of the spatial attention mechanism model to obtain a spatial attention map; An activation subunit, configured to pass the spatial attention map through a Softmax activation function to obtain a spatial attention feature map; A spatial attention applying subunit, used for calculating the point-by-point multiplication of the spatial attention feature map and the initial convolution feature map to obtain a spatial enhancement device operation state-environment time series response feature map; The pooling subunit is used to perform mean pooling processing on the spatial enhancement device operating state-environment time series response feature map along the channel dimension to obtain the significant device operating state-environment time series response feature matrix.

9. The operation and maintenance management system of wind power generation equipment according to claim 8, characterized in that: The equipment performance analysis module is used to: The significant equipment operating state-environment time series response feature matrix is ​​passed through a classifier to obtain a classification result, and the classification result is used to determine whether the performance of the wind power generation equipment is abnormal.

10. A method for operation, maintenance and management of wind power generation equipment, characterized in that: include: Acquire operation data of the wind power generation equipment at multiple predetermined time points within a predetermined time period, and environmental data at multiple predetermined time points within the predetermined time period, wherein the operation data includes power, wind rotor speed, current and temperature of the equipment, and the environmental data includes wind speed, wind direction and air pressure; Performing time series correlation coding on the operation data of the plurality of predetermined time points to obtain a time series feature vector of the equipment operation state; Extracting time series features from the environmental data at the plurality of predetermined time points to obtain a time series feature vector of the environmental data; Performing association coding and association feature enhancement on the device operation state time series feature vector and the environment data time series feature vector to obtain a significant device operation state-environment time series response feature matrix; Based on the significant equipment operating state-environment time series response characteristic matrix, it is determined whether the performance of the wind power generation equipment is abnormal.

Images