Wind turbine energy efficiency state abnormality detection method and system

Abstract

The present invention relates to a method and system for detecting abnormal energy efficiency status of a wind turbine. The method comprises: obtaining pre-processed energy efficiency status operation data of the wind turbine at the current moment and energy efficiency status operation data of the wind turbine in a historical period, and determining the deviation matrix corresponding to the energy efficiency status operation data at the current moment and the energy efficiency status operation data in a historical period, inputting the deviation matrix corresponding to the current moment and the historical period into a pre-trained abnormal energy efficiency status detection model of the wind turbine, obtaining the abnormal energy efficiency status score of the wind turbine at the current moment, detecting whether the energy efficiency status of the wind turbine is abnormal based on the abnormal energy efficiency status score of the wind turbine at the current moment, and communicating the detection result. The technical solution provided by the present invention improves the accuracy of abnormal energy efficiency status detection of the wind turbine, while making abnormality detection more convenient and reducing the cost of abnormality detection.

Description

Technical Field

[0001] The present invention relates to the fields of artificial intelligence, deep learning, neural networks, new energy and abnormal energy efficiency status detection of wind turbines, and in particular to a method and system for abnormal energy efficiency status detection of wind turbines. Background Art

[0002] With the large-scale development of wind power, the daily maintenance of wind power equipment, including but not limited to the energy efficiency status of the unit, fault warning and other operation and maintenance challenges. Therefore, how to improve the daily monitoring efficiency of the efficiency status of wind turbines is a solution. With the widespread application of natural language processing (NLP) in wind power generation, it has attracted the attention of industry and academia. Traditional methods for abnormal detection of wind turbine energy efficiency status, such as machine learning methods, statistical probability, and single neural network models, have limitations with the development of artificial intelligence technology. First, the operation of wind turbines has dynamic characteristics; second, the data anomalies of the SCADA system of wind farms are more easily covered by their own fluctuations, resulting in low accuracy when using sensors, i.e. SCADA system data, to detect abnormalities in the energy efficiency status of wind turbines. Therefore, it is urgent to propose a method and system for accurate abnormal detection of wind turbine energy efficiency status using SCADA system data. Summary of the invention

[0003] The present application provides a method and system for detecting abnormal energy efficiency status of a wind turbine generator set, so as to at least solve the technical problem in the related art that abnormal energy efficiency status of a wind turbine generator set cannot be accurately detected.

[0004] The first embodiment of the present application provides a method for detecting abnormal energy efficiency status of a wind turbine generator set, the method comprising:

[0005] Acquire the energy efficiency status operation data of the wind turbine set at the current moment and the energy efficiency status operation data of the wind turbine set in the historical period, and pre-process the data;

[0006] Based on the preprocessed data, the reference value and deviation corresponding to the energy efficiency status operation data at the current moment and the reference value and deviation corresponding to the energy efficiency status operation data in the historical period are calculated respectively;

[0007] Based on the reference value and deviation corresponding to the energy efficiency state operation data at the current moment, a deviation matrix corresponding to the energy efficiency state operation data at the current moment is formed; based on the reference value and deviation corresponding to the energy efficiency state operation data in the historical period, a deviation matrix corresponding to the energy efficiency state operation data in the historical period is formed;

[0008] Input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into a pre-trained wind turbine energy efficiency state anomaly detection model to obtain an anomaly detection score for the energy efficiency state of the wind turbine at the current moment;

[0009] Detecting whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormality detection score of the energy efficiency state of the wind turbine generator set at the current moment, and communicating the detection result;

[0010] The pre-trained wind turbine energy efficiency status anomaly detection model is obtained by training based on the energy efficiency status operation data of the wind turbine in the historical period and the active power quality labels of the wind turbine in the historical period;

[0011] The pre-trained wind turbine energy efficiency status anomaly detection model includes: a teacher module, a student module and a fully connected module.

[0012] The second embodiment of the present application provides a wind turbine energy efficiency status abnormality detection system, the system comprising:

[0013] An acquisition module, used to acquire the energy efficiency status operation data of the wind turbine set at the current moment and the energy efficiency status operation data of the wind turbine set in the historical period, and pre-process the data;

[0014] A calculation module, used to calculate the reference value and deviation corresponding to the energy efficiency state operation data at the current moment and the reference value and deviation corresponding to the energy efficiency state operation data in the historical period based on the preprocessed data;

[0015] A matrix module, for forming a deviation matrix corresponding to the energy efficiency state operation data at the current moment based on the reference value and deviation corresponding to the energy efficiency state operation data at the current moment, and forming a deviation matrix corresponding to the energy efficiency state operation data in the historical period based on the reference value and deviation corresponding to the energy efficiency state operation data in the historical period;

[0016] A scoring module is used to input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into a pre-trained wind turbine energy efficiency state anomaly detection model to obtain an anomaly detection score for the energy efficiency state of the wind turbine at the current moment;

[0017] A detection module, used to detect whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormal detection score of the energy efficiency state of the wind turbine generator set at the current moment, and to communicate the detection result;

[0018] The pre-trained wind turbine energy efficiency status anomaly detection model is obtained by training based on the energy efficiency status operation data of the wind turbine in the historical period and the active power quality labels of the wind turbine in the historical period;

[0019] The pre-trained wind turbine energy efficiency status anomaly detection model includes: a teacher module, a student module and a fully connected module.

[0020] The third aspect of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the prediction method of the first aspect of the present application is implemented.

[0021] The fourth aspect of the present application provides a non-temporary computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the prediction method of the first aspect of the present application is implemented.

[0022] The technical solution provided by the embodiments of the present application brings at least the following beneficial effects:

[0023] The present invention provides a method and system for detecting abnormal energy efficiency status of a wind turbine set, wherein the method comprises: obtaining energy efficiency status operation data of the wind turbine set at the current moment and energy efficiency status operation data of the wind turbine set in a historical period, and preprocessing the data; calculating the reference value and deviation corresponding to the energy efficiency status operation data at the current moment and the reference value and deviation corresponding to the energy efficiency status operation data in a historical period based on the preprocessed data; forming a deviation matrix corresponding to the energy efficiency status operation data at the current moment based on the reference value and deviation corresponding to the energy efficiency status operation data at the current moment, and forming a deviation matrix corresponding to the energy efficiency status operation data in a historical period based on the reference value and deviation corresponding to the energy efficiency status operation data in a historical period; inputting the deviation matrix corresponding to the energy efficiency status operation data at the current moment and the deviation matrix corresponding to the energy efficiency status operation data in a historical period into a pre-trained abnormal energy efficiency status detection model for the wind turbine set, and obtaining an abnormal energy efficiency status detection score of the wind turbine set at the current moment; detecting whether the energy efficiency status of the wind turbine set is abnormal based on the abnormal energy efficiency status detection score of the wind turbine set at the current moment, and communicating the detection result. The technical solution provided by the present invention improves the accuracy of abnormal detection of energy efficiency status of wind turbines, while making abnormal detection more convenient and reducing the cost of abnormal detection.

[0024] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0026] Figure 1 It is a flow chart of a method for detecting abnormal energy efficiency status of a wind turbine generator set according to an embodiment of the present application;

[0027] Figure 2 It is a specific flow chart of a method for detecting abnormal energy efficiency status of a wind turbine generator set according to an embodiment of the present application;

[0028] Figure 3 It is a structural diagram of a wind turbine energy efficiency status abnormality detection system provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0030] The present application proposes a method and system for detecting abnormal energy efficiency status of a wind turbine set, wherein the method comprises: obtaining energy efficiency status operation data of the wind turbine set at the current moment and energy efficiency status operation data of the wind turbine set in a historical period, and preprocessing the data; calculating the reference value and deviation corresponding to the energy efficiency status operation data at the current moment and the reference value and deviation corresponding to the energy efficiency status operation data in a historical period based on the preprocessed data; forming a deviation matrix corresponding to the energy efficiency status operation data at the current moment based on the reference value and deviation corresponding to the energy efficiency status operation data at the current moment, and forming a deviation matrix corresponding to the energy efficiency status operation data in a historical period based on the reference value and deviation corresponding to the energy efficiency status operation data in a historical period; inputting the deviation matrix corresponding to the energy efficiency status operation data at the current moment and the deviation matrix corresponding to the energy efficiency status operation data in a historical period into a pre-trained abnormal energy efficiency status detection model of the wind turbine set, and obtaining an abnormal energy efficiency status detection score of the wind turbine set at the current moment; detecting whether the energy efficiency status of the wind turbine set is abnormal based on the abnormal energy efficiency status detection score of the wind turbine set at the current moment, and communicating the detection result. The technical solution provided by the present invention improves the accuracy of abnormal detection of energy efficiency status of wind turbines, while making abnormal detection more convenient and reducing the cost of abnormal detection.

[0031] Example 1

[0032] Figure 1A flow chart of a method for detecting abnormal energy efficiency status of a wind turbine set provided in an embodiment of the present disclosure, such as Figure 1 As shown, the method includes:

[0033] Step 1: Obtaining the energy efficiency status operation data of the wind turbine at the current moment and the energy efficiency status operation data of the wind turbine in the historical period, and preprocessing the data;

[0034] In the embodiment of the present disclosure, the step of acquiring the energy efficiency status operation data of the wind turbine set at the current moment and the energy efficiency status operation data of the wind turbine set in a historical period, and preprocessing the data, includes:

[0035] Abnormal data in the acquired energy efficiency status operation data of the wind turbine generator set at the current moment and the energy efficiency status operation data of the wind turbine generator set in a historical period are eliminated.

[0036] Step 2: Based on the preprocessed data, the reference value and deviation corresponding to the energy efficiency state operation data at the current moment and the reference value and deviation corresponding to the energy efficiency state operation data in the historical period are calculated respectively;

[0037] Step 3: constructing a deviation matrix corresponding to the energy efficiency state operation data at the current moment based on the reference value and deviation corresponding to the energy efficiency state operation data at the current moment, and constructing a deviation matrix corresponding to the energy efficiency state operation data in the historical period based on the reference value and deviation corresponding to the energy efficiency state operation data in the historical period;

[0038] Step 4: Input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into the pre-trained wind turbine energy efficiency state anomaly detection model to obtain the anomaly detection score of the wind turbine energy efficiency state at the current moment;

[0039] It should be noted that the pre-trained wind turbine energy efficiency status anomaly detection model is trained based on the energy efficiency status operation data of the wind turbine in the historical period and the active power quality labels of the wind turbine in the historical period;

[0040] It should be noted that the pre-trained wind turbine energy efficiency status anomaly detection model includes: a teacher module, a student module and a fully connected module.

[0041] In the embodiment of the present disclosure, the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period are input into a pre-trained wind turbine energy efficiency state anomaly detection model to obtain the anomaly detection score of the wind turbine energy efficiency state at the current moment, including:

[0042] Step F1: input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into the teacher module of the pre-trained wind turbine energy efficiency state anomaly detection model to obtain the attention map corresponding to the energy efficiency state operation data at the current moment;

[0043] Step F2: Migrating the attention map to the student module of the pre-trained wind turbine energy efficiency status anomaly detection model;

[0044] Step F3: inputting the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into the student module of the pre-trained wind turbine energy efficiency state anomaly detection model that has completed the migration, and obtaining the text features corresponding to the data of the wind turbine at the current moment;

[0045] Step F4: Input the text features corresponding to the data of the wind turbine set at the current moment into the fully connected module of the pre-trained wind turbine set energy efficiency state anomaly detection model and calculate using the Softmax function to obtain the anomaly detection score of the wind turbine set energy efficiency state at the current moment.

[0046] It should be noted that the student module of the wind turbine energy efficiency status anomaly detection model includes: a multi-channel convolutional neural network CNN layer and a BiLSTM+dropout layer.

[0047] In the disclosed embodiment, the training process of the teacher module of the wind turbine energy efficiency status abnormality detection model includes:

[0048] Obtain the deviation matrix corresponding to the energy efficiency status operation data of the wind turbine generator set in the historical period and the active power quality labels of the wind turbine generator set in the historical period to form training set sample data;

[0049] The training set sample data is input into the teacher module of the initial wind turbine energy efficiency status anomaly detection model, the mean square error loss MSE is used as the loss function of the model, and the model is trained using the adaptive matrix estimation Adam optimization algorithm to obtain a trained wind turbine energy efficiency status anomaly detection model teacher module, wherein the teacher module adopts the Bert model.

[0050] In the embodiment of the present disclosure, the training process of the student module of the wind turbine energy efficiency status abnormality detection model includes:

[0051] Obtain the deviation matrix corresponding to the energy efficiency status operation data of the wind turbine in the historical period and the active power quality label of the wind turbine in the historical period to form the training set sample data, and migrate the corresponding attention map when the teacher module training is completed to the student module of the initial wind turbine energy efficiency status anomaly detection model;

[0052] The training set sample data is input into the multi-channel convolutional neural network CNN layer and BiLSTM+dropout layer of the teacher-student module of the initial wind turbine energy efficiency status anomaly detection model after the transfer attention map, the mean square error loss MSE is used as the loss function of the model, and the model is trained using the adaptive matrix estimation Adam optimization algorithm to obtain the trained student module of the wind turbine energy efficiency status anomaly detection model.

[0053] It should be noted that the activation function ReLU function is used as the activation function of the teacher model and the student model.

[0054] Step 5: Detect whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormality detection score of the energy efficiency state of the wind turbine generator set at the current moment, and communicate the detection result;

[0055] In the embodiment of the present disclosure, the detecting whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormality detection score of the energy efficiency state of the wind turbine generator set at the current moment, and delivering the detection result as information includes:

[0056] Detecting whether the abnormality detection score is within a preset abnormality score range, if so, the energy efficiency state of the wind turbine generator set is abnormal, otherwise, the energy efficiency state of the wind turbine generator set is normal;

[0057] Communicate the inspection results;

[0058] Among them, the ways of information delivery include: text display, voice broadcast, outbound call terminal, SMS delivery, email delivery, smart speakers, voice wake-up and large-screen display.

[0059] The specific method of this application is illustrated by combining the above-mentioned abnormal detection method of energy efficiency status of wind turbines:

[0060] like Figure 2 The specific flow chart of the method for detecting abnormal energy efficiency status of a wind turbine generator set is shown, and the method comprises:

[0061] Step F1: data collection 101 and data preprocessing 102, respectively collect and preprocess the real-time monitoring data and historical data of the SCADA system. Considering that abnormal data will be stored in the SCADA system, the abnormal data processed due to sensor errors, null values, missing values, etc. are screened for data preprocessing, and combined with the wind turbine operation dependence on wind speed, the non- / driving direction generator bearing temperature, generator winding temperature, gearbox bearing temperature, gearbox oil temperature, wind turbine variable pitch motor 1 temperature, 2 temperature, 3 temperature are selected to form a deviation matrix as model input, and the baseline value calculation and deviation calculation of the input data are completed in advance to form a deviation matrix, and the active power quality is selected as the label, and then the trained student model is used to classify the test set to obtain the abnormal detection of wind turbine energy efficiency status.

[0062] Step F2: Complete the pre-training of the teacher module 103 to obtain the optimal module. The deviation matrix formed by the real data is input into the teacher module 103, and the teacher module is pre-trained to obtain the optimal attention map, including network model adjustment and optimization loss function. Here, the real data of the historical data in the past year is selected to form a large model training.

[0063] Step F3: Complete the pre-training of the student module 105 to obtain the optimal module, use the same real data hard label and the teacher module output attention map 104 to migrate to the student module 105 for pre-training, and use multi-channel convolution CNN+BiLSTM+dropout to further learn the spatiotemporal and optimal text features, including optimizing the loss function, and BiLSTM is good at capturing long-distance dependent text sequence information, so the optimal model can obtain high-level text features with optimal context information.

[0064] Step F4: Use the fully connected layer Softmax106 to calculate the anomaly detection results, input the real data into the student model to predict the anomaly detection results, and use the activation function ReLU function as the activation function of the teacher model and the student model.

[0065] Step F5: The abnormal energy efficiency status detection of the wind turbine is contacted and the result is displayed 107, wherein the contact method specifically includes: text display, voice broadcast, outbound terminal, SMS delivery, email contact, smart speaker, voice wake-up and large-screen display.

[0066] In summary, the wind turbine energy efficiency status abnormality detection method provided in this embodiment improves the accuracy of wind turbine energy efficiency status abnormality detection, while making abnormality detection more convenient and reducing abnormality detection costs.

[0067] Example 2

[0068] Figure 3The structure diagram of the wind turbine energy efficiency status abnormality detection system provided by the embodiment of the present disclosure is as follows: Figure 3 As shown, the system comprises:

[0069] An acquisition module, used to acquire the energy efficiency status operation data of the wind turbine set at the current moment and the energy efficiency status operation data of the wind turbine set in the historical period, and pre-process the data;

[0070] A calculation module, used to calculate the reference value and deviation corresponding to the energy efficiency state operation data at the current moment and the reference value and deviation corresponding to the energy efficiency state operation data in the historical period based on the preprocessed data;

[0071] A matrix module, for forming a deviation matrix corresponding to the energy efficiency state operation data at the current moment based on the reference value and deviation corresponding to the energy efficiency state operation data at the current moment, and forming a deviation matrix corresponding to the energy efficiency state operation data in the historical period based on the reference value and deviation corresponding to the energy efficiency state operation data in the historical period;

[0072] A scoring module is used to input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into a pre-trained wind turbine energy efficiency state anomaly detection model to obtain an anomaly detection score for the energy efficiency state of the wind turbine at the current moment;

[0073] A detection module, used to detect whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormal detection score of the energy efficiency state of the wind turbine generator set at the current moment, and to communicate the detection result;

[0074] The pre-trained wind turbine energy efficiency status anomaly detection model is obtained by training based on the energy efficiency status operation data of the wind turbine in the historical period and the active power quality labels of the wind turbine in the historical period;

[0075] The pre-trained wind turbine energy efficiency status anomaly detection model includes: a teacher module, a student module and a fully connected module.

[0076] In the embodiment of the present disclosure, the step of acquiring the energy efficiency status operation data of the wind turbine set at the current moment and the energy efficiency status operation data of the wind turbine set in a historical period, and preprocessing the data, includes:

[0077] Abnormal data in the acquired energy efficiency status operation data of the wind turbine generator set at the current moment and the energy efficiency status operation data of the wind turbine generator set in a historical period are eliminated.

[0078] In the embodiment of the present disclosure, the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period are input into a pre-trained wind turbine energy efficiency state anomaly detection model to obtain the anomaly detection score of the wind turbine energy efficiency state at the current moment, including:

[0079] Step T1: input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into the teacher module of the pre-trained wind turbine energy efficiency state anomaly detection model to obtain the attention map corresponding to the energy efficiency state operation data at the current moment;

[0080] Step T2: Migrating the attention map to the student module of the pre-trained wind turbine energy efficiency state anomaly detection model;

[0081] Step T3: inputting the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into the student module of the pre-trained wind turbine energy efficiency state anomaly detection model that has completed the migration, and obtaining the text features corresponding to the data of the wind turbine at the current moment;

[0082] Step T4: input the text features corresponding to the data of the wind turbine set at the current moment into the fully connected module of the pre-trained wind turbine set energy efficiency status anomaly detection model and calculate using the Softmax function to obtain the anomaly detection score of the wind turbine set energy efficiency status at the current moment.

[0083] In the disclosed embodiment, the student module of the wind turbine energy efficiency status abnormality detection model includes: a multi-channel convolutional neural network CNN layer and a BiLSTM+dropout layer.

[0084] It should be noted that the training process of the teacher module of the wind turbine energy efficiency status anomaly detection model includes:

[0085] Obtaining the deviation matrix corresponding to the energy efficiency status operation data of the wind turbine generator set in the historical period and the active power quality labels of the wind turbine generator set in the historical period to form training set sample data;

[0086] Input the training set sample data into the teacher module of the initial wind turbine energy efficiency state anomaly detection model, use the mean square error loss MSE as the loss function of the model, and use the adaptive matrix estimation Adam optimization algorithm to train the model to obtain a trained wind turbine energy efficiency state anomaly detection model teacher module;

[0087] Among them, the teacher module adopts the Bert model.

[0088] It should be noted that the training process of the student module of the wind turbine energy efficiency status abnormality detection model includes:

[0089] Obtain the deviation matrix corresponding to the energy efficiency status operation data of the wind turbine in the historical period and the active power quality label of the wind turbine in the historical period to form the training set sample data, and migrate the corresponding attention map when the teacher module training is completed to the student module of the initial wind turbine energy efficiency status anomaly detection model;

[0090] The training set sample data is input into the multi-channel convolutional neural network CNN layer and BiLSTM+dropout layer of the teacher-student module of the initial wind turbine energy efficiency status anomaly detection model after the transfer attention map, the mean square error loss MSE is used as the loss function of the model, and the model is trained using the adaptive matrix estimation Adam optimization algorithm to obtain the trained student module of the wind turbine energy efficiency status anomaly detection model.

[0091] In the embodiment of the present disclosure, the detecting whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormality detection score of the energy efficiency state of the wind turbine generator set at the current moment, and delivering the detection result as information includes:

[0092] Detecting whether the abnormality detection score is within a preset abnormality score range, if so, the energy efficiency state of the wind turbine generator set is abnormal, otherwise, the energy efficiency state of the wind turbine generator set is normal;

[0093] Communicate the inspection results;

[0094] Among them, the ways of information delivery include: text display, voice broadcast, outbound call terminal, SMS delivery, email delivery, smart speakers, voice wake-up and large-screen display.

[0095] In summary, the wind turbine energy efficiency status abnormality detection system provided in this embodiment improves the accuracy of wind turbine energy efficiency status abnormality detection, while making abnormality detection more convenient and reducing abnormality detection costs.

[0096] Example 3

[0097] In order to implement the above embodiments, the present disclosure also provides a computer device.

[0098] The computer device provided in this embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method in Embodiment 1 is implemented.

[0099] Example 4

[0100] In order to implement the above embodiments, the present disclosure also proposes a non-transitory computer-readable storage medium.

[0101] The computer device provided in this embodiment stores a computer program, and when the computer program is executed by a processor, the method in Embodiment 1 is implemented.

[0102] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0103] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0104] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for detecting abnormal energy efficiency status of a wind turbine generator set, characterized in that: The method comprises: Acquire the energy efficiency status operation data of the wind turbine set at the current moment and the energy efficiency status operation data of the wind turbine set in the historical period, and pre-process the data; Based on the preprocessed data, the reference value and deviation corresponding to the energy efficiency status operation data at the current moment and the reference value and deviation corresponding to the energy efficiency status operation data in the historical period are calculated respectively; Based on the reference value and deviation corresponding to the energy efficiency state operation data at the current moment, a deviation matrix corresponding to the energy efficiency state operation data at the current moment is formed; based on the reference value and deviation corresponding to the energy efficiency state operation data in the historical period, a deviation matrix corresponding to the energy efficiency state operation data in the historical period is formed; Input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into a pre-trained wind turbine energy efficiency state anomaly detection model to obtain an anomaly detection score for the energy efficiency state of the wind turbine at the current moment; Detecting whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormality detection score of the energy efficiency state of the wind turbine generator set at the current moment, and communicating the detection result; The pre-trained wind turbine energy efficiency status anomaly detection model is obtained by training based on the energy efficiency status operation data of the wind turbine in the historical period and the active power quality labels of the wind turbine in the historical period; The pre-trained wind turbine energy efficiency status anomaly detection model includes: a teacher module, a student module and a fully connected module.

2. The method according to claim 1, characterized in that The obtaining of the energy efficiency status operation data of the wind turbine set at the current moment and the energy efficiency status operation data of the wind turbine set in the historical period, and preprocessing the data, includes: Abnormal data in the acquired energy efficiency status operation data of the wind turbine generator set at the current moment and the energy efficiency status operation data of the wind turbine generator set in a historical period are eliminated.

3. The method according to claim 1, characterized in that The step of inputting the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into a pre-trained wind turbine energy efficiency state anomaly detection model to obtain an anomaly detection score for the energy efficiency state of the wind turbine at the current moment includes: Step F1: input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into the teacher module of the pre-trained wind turbine energy efficiency state anomaly detection model to obtain the attention map corresponding to the energy efficiency state operation data at the current moment; Step F2: Migrating the attention map to the student module of the pre-trained wind turbine energy efficiency status anomaly detection model; Step F3: inputting the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into the student module of the pre-trained wind turbine energy efficiency state anomaly detection model that has completed the migration, and obtaining the text features corresponding to the data of the wind turbine at the current moment; Step F4: Input the text features corresponding to the data of the wind turbine set at the current moment into the fully connected module of the pre-trained wind turbine set energy efficiency state anomaly detection model and calculate using the Softmax function to obtain the anomaly detection score of the wind turbine set energy efficiency state at the current moment.

4. The method according to claim 1, characterized in that The student module of the wind turbine energy efficiency status anomaly detection model includes: a multi-channel convolutional neural network CNN layer and a BiLSTM+dropout layer.

5. The method according to claim 4, characterized in that The training process of the teacher module of the wind turbine energy efficiency status anomaly detection model includes: Obtain the deviation matrix corresponding to the energy efficiency status operation data of the wind turbine generator set in the historical period and the active power quality labels of the wind turbine generator set in the historical period to form training set sample data; Input the training set sample data into the teacher module of the initial wind turbine energy efficiency state anomaly detection model, use the mean square error loss MSE as the loss function of the model, and use the adaptive matrix estimation Adam optimization algorithm to train the model to obtain a trained wind turbine energy efficiency state anomaly detection model teacher module; Among them, the teacher module adopts the Bert model.

6. The method according to claim 5, characterized in that The training process of the student module of the wind turbine energy efficiency status abnormality detection model includes: Obtain the deviation matrix corresponding to the energy efficiency status operation data of the wind turbine in the historical period and the active power quality label of the wind turbine in the historical period to form the training set sample data, and migrate the corresponding attention map when the teacher module training is completed to the student module of the initial wind turbine energy efficiency status anomaly detection model; The training set sample data is input into the multi-channel convolutional neural network CNN layer and BiLSTM+dropout layer of the student module of the initial wind turbine energy efficiency status anomaly detection model after the transfer attention map, the mean square error loss MSE is used as the loss function of the model, and the model is trained using the adaptive matrix estimation Adam optimization algorithm to obtain the trained student module of the wind turbine energy efficiency status anomaly detection model.

7. The method according to claim 1, characterized in that The detecting whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormality detection score of the energy efficiency state of the wind turbine generator set at the current moment, and delivering the detection result to information includes: Detecting whether the abnormality detection score is within a preset abnormality score range, if so, the energy efficiency state of the wind turbine generator set is abnormal, otherwise, the energy efficiency state of the wind turbine generator set is normal; Communicate the test results; Among them, the ways of information delivery include: text display, voice broadcast, outbound call terminal, SMS delivery, email delivery, smart speakers, voice wake-up and large-screen display.

8. A wind turbine energy efficiency status abnormality detection system, characterized in that: The system comprises: An acquisition module, used to acquire the energy efficiency status operation data of the wind turbine set at the current moment and the energy efficiency status operation data of the wind turbine set in the historical period, and pre-process the data; A calculation module, used to calculate the reference value and deviation corresponding to the energy efficiency state operation data at the current moment and the reference value and deviation corresponding to the energy efficiency state operation data in the historical period based on the preprocessed data; A matrix module, for forming a deviation matrix corresponding to the energy efficiency state operation data at the current moment based on the reference value and deviation corresponding to the energy efficiency state operation data at the current moment, and forming a deviation matrix corresponding to the energy efficiency state operation data in the historical period based on the reference value and deviation corresponding to the energy efficiency state operation data in the historical period; A scoring module is used to input the deviation matrix corresponding to the energy efficiency state operation data at the current moment and the deviation matrix corresponding to the energy efficiency state operation data in the historical period into a pre-trained wind turbine energy efficiency state anomaly detection model to obtain an anomaly detection score for the energy efficiency state of the wind turbine at the current moment; A detection module, used to detect whether the energy efficiency state of the wind turbine generator set is abnormal based on the abnormal detection score of the energy efficiency state of the wind turbine generator set at the current moment, and to communicate the detection result; The pre-trained wind turbine energy efficiency status anomaly detection model is obtained by training based on the energy efficiency status operation data of the wind turbine in the historical period and the active power quality labels of the wind turbine in the historical period; The pre-trained wind turbine energy efficiency status anomaly detection model includes: a teacher module, a student module and a fully connected module.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images