Abnormality determination method and device for wind turbine generator, medium, equipment and program product

Abstract

The invention relates to a wind turbine generator abnormity determination method and device, a medium, equipment and a program product, and relates to the technical field of wind power. The method comprises the following steps: acquiring operation data of a plurality of first wind turbine generators from a sensor; calculating the deviation between the operation data of the first wind turbine generator and the average value of the operation data of the first wind turbine generator, and determining the first wind turbine generator with the deviation exceeding a preset deviation range as a second wind turbine generator; and determining an abnormal wind turbine generator in the second wind turbine generator based on an isolated forest algorithm. According to the method, the accuracy of determining the abnormal wind turbine generator from the plurality of wind turbine generators can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of wind power technology, and in particular to a method, device, medium, equipment and program product for determining abnormality of a wind turbine. Background Art

[0002] With the rapid development of the new energy industry, the operation and maintenance management of new energy generators faces huge challenges. The current operation and maintenance technology mainly relies on the analysis of historical data of wind turbines, which has two main problems: first, the lack of effective use of real-time operation data, making it difficult to capture the dynamic changes of equipment status in a timely manner; second, the accuracy of the early warning system is not high, and false alarms and missed alarms are prone to occur. These problems seriously affect the safe and stable operation and economic benefits of the new energy power generation system. Therefore, it is of great practical significance to develop an intelligent fault diagnosis and early warning system that can analyze data in real time and improve the accuracy of early warning. Summary of the invention

[0003] In order to overcome the problems existing in the related art, the present application provides a method, device, medium, equipment and program product for determining abnormality of a wind turbine.

[0004] According to a first aspect of the present application, a method for determining an abnormality of a wind turbine is provided, comprising: Acquire operation data of a first wind turbine generator set from a sensor, wherein the first wind turbine generator set has a plurality of wind turbine generator sets; Calculating the deviation between the operating data of the first wind turbine group and the average value of the operating data thereof, and determining the first wind turbine group whose deviation exceeds a preset deviation range as the second wind turbine group; Based on the isolation forest algorithm, abnormal wind turbines in the second wind turbines are determined.

[0005] Optionally, the determining the abnormal wind turbines in the second wind turbines based on the isolation forest algorithm includes: constructing an isolated forest according to the operating data of the second wind turbine generator set; Based on the isolation forest, determining an anomaly score for each second wind turbine group; The second wind turbine generator set having an abnormality score greater than a preset abnormality threshold is determined as the abnormal wind turbine generator set.

[0006] Optionally, the operation data is multi-dimensional data, each dimension corresponds to a different feature of the second wind turbine generator set, and the step of constructing an isolation forest according to the operation data of the second wind turbine generator set includes: Determine a plurality of different feature sequences, wherein the feature sequences are permutations and combinations of features of each dimension of the operating data, and the feature permutations and combinations between two feature sequences are different; For each characteristic sequence, a subsample is randomly selected from all second wind turbines; Based on each feature sequence and its corresponding sub-sample, an isolated tree is constructed to obtain a plurality of isolated trees, and the plurality of isolated trees constitute the isolation forest.

[0007] Optionally, constructing an isolation tree based on each feature sequence and its corresponding subsample includes: Selecting target features of the nodes to be divided in order from the feature sequence, and randomly selecting a split point within the value range of the target features, each of the nodes to be divided corresponds to a different second wind turbine generator set; According to the target feature and the segmentation point, the node to be divided is divided into a left child node and a right child node, the left child node includes a node corresponding to the second wind turbine group whose target feature value is less than the segmentation point, and the right child node includes a node corresponding to the second wind turbine group whose target feature value is greater than or equal to the segmentation point; Determine the left child node as a new node to be divided, return to execute the target feature of randomly selecting the node to be divided corresponding to the second wind turbine group, and randomly select a split point within the value range of the target feature until the depth of the isolated tree reaches a preset maximum depth or the left child node contains only one node corresponding to the second wind turbine group; The right child node is determined as a new node to be divided, and the target feature of randomly selecting the node to be divided corresponding to the second wind turbine group is returned to execute, and a split point is randomly selected within the value range of the target feature until the depth of the isolated tree reaches the preset maximum depth or the right child node contains only one node corresponding to the second wind turbine group.

[0008] Optionally, determining the abnormal score of each second wind turbine generator set based on the isolation forest includes: For each second wind turbine group, determining a path length of the target second wind turbine group in each isolated tree; An abnormal score corresponding to the target second wind turbine group is determined according to an average value of all path lengths of the target second wind turbine group.

[0009] Optionally, the method further includes determining abnormal characteristics of each abnormal wind turbine generator set based on a Gaussian algorithm, specifically including: For each characteristic of the target abnormal wind turbine, perform the following steps: According to the historical normal operation data of the target characteristic of the target abnormal wind turbine group, a Gaussian distribution model corresponding to the target characteristic is established; Determining the probability density of the target feature according to the Gaussian distribution model; When the probability density is less than a preset probability density, the target feature is determined to be the abnormal feature.

[0010] According to a second aspect of the present application, a device for determining abnormality of a wind turbine is provided, comprising: A data acquisition module, used for acquiring operation data of a first wind turbine generator set from a sensor, wherein the first wind turbine generator set has a plurality of wind turbine generator sets; a mean deviation module, used to calculate the deviation between the operating data of the first wind turbine group and the average value of the operating data, and determine the first wind turbine group whose deviation exceeds a preset deviation range as the second wind turbine group; The isolation forest module is used to determine the abnormal wind turbines in the second wind turbines based on the isolation forest algorithm.

[0011] According to the third aspect of the present application, a non-temporary computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods provided in the first aspect of the present application are implemented.

[0012] According to a fourth aspect of the present application, there is provided an electronic device, including: a memory having a computer program stored thereon; A processor is used to execute the computer program in the memory to implement the steps of any one of the methods provided in the first aspect of the present application.

[0013] According to a fifth aspect of the present application, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of any one of the methods provided in the first aspect of the present application.

[0014] The above technical solution collects operating data through sensors, calculates the deviation of data from the average value, and screens out abnormal candidate units through preset thresholds. Then, the isolation forest algorithm is used for further analysis to accurately identify the real abnormal units. Through multiple screening, abnormal units can be effectively and quickly identified from a large number of wind turbines, improving the efficiency and accuracy of abnormality detection.

[0015] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings: Figure 1 is a flow chart of a method for determining abnormality of a wind turbine generator set according to an exemplary embodiment; Figure 2 is a block diagram of a device for determining abnormality of a wind turbine generator set according to an exemplary embodiment; Figure 3It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0017] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0018] It should be noted that all actions of acquiring signals, information or data in the present disclosure are carried out in compliance with the relevant data protection laws and policies of the country where the device is located and with the authorization given by the owner of the corresponding device.

[0019] Herein, the terms "first", "second", etc. are only used to distinguish one element from another element, and do not require or imply any actual relationship or order between these elements. In fact, the first element can also be called the second element, and vice versa.

[0020] Figure 1 The present invention is a flow chart of a method for determining abnormality of a wind turbine according to an exemplary embodiment. The method can be applied to determining abnormality of a wind turbine, and specifically can determine an abnormal wind turbine from multiple wind turbines. The method comprises the following steps: In step 110, operating data of a first wind turbine generator set is acquired from a sensor, and there are a plurality of the first wind turbine generator sets; In step 120, the deviation between the operating data of the first wind turbine group and the average value of the operating data is calculated, and the first wind turbine group whose deviation exceeds a preset deviation range is determined as the second wind turbine group; In step 130, abnormal wind turbines in the second wind turbines are determined based on the isolation forest algorithm.

[0021] Here, the first wind turbine refers to a plurality of selected wind turbines to be detected, which is the initial group of detection objects. The operating data may include data such as wind speed, temperature, rotation speed, output power, voltage, current, blade angle, etc. These data are collected by sensors and obtained after preprocessing (such as data cleaning and normalization). The average value of the operating data is the average value calculated for a certain operating data (such as temperature) of the selected plurality of wind turbines to be detected. The deviation is the difference between the operating data of a certain wind turbine and the average value. The preset deviation range is a threshold range set to determine whether the deviation is abnormal. The second wind turbine is a wind turbine whose deviation exceeds the preset range and is marked as an object requiring further detection. The isolation forest algorithm is a machine learning algorithm for anomaly detection, which is used in this method to further identify abnormal units from the second wind turbine. The abnormal wind turbine is a wind turbine that is finally determined to be abnormal.

[0022] The above method collects the operating data of multiple wind turbines, calculates the average value and the deviation of each unit, selects the unit with abnormal deviation as the second wind turbine, uses the isolation forest algorithm to further analyze the second wind turbine, and finally determines the abnormal wind turbine. Through multiple screening, the abnormal unit can be effectively and quickly identified from a large number of wind turbines, improving the efficiency and accuracy of anomaly detection.

[0023] In one embodiment, determining the abnormal wind turbines among the second wind turbines based on the isolation forest algorithm may include: constructing an isolation forest based on the operating data of the second wind turbines; determining the abnormality score of each second wind turbine based on the isolation forest; and determining the second wind turbine whose abnormality score is greater than a preset abnormality threshold as the abnormal wind turbine.

[0024] Here, Isolation Forest is an unsupervised learning algorithm for anomaly detection. Its basic principle is that abnormal data points (abnormal wind turbines) are more likely to be "isolated" than normal data points. Anomalies are detected by building multiple decision trees (i.e. "forests") and calculating the anomaly score of each data point (the second wind turbine). The anomaly score is a measure of the degree of abnormality of a data point. The higher the score, the more likely the data point is an anomaly. The preset anomaly threshold is a set threshold used to determine whether a data point is an anomaly. If the anomaly score of a data point exceeds this threshold, it is considered an anomaly.

[0025] The above technical solution uses the operating data of the second wind turbine to construct an isolation forest model. Using the constructed isolation forest model, an anomaly score is calculated for each second wind turbine. The anomaly score is compared with a preset anomaly threshold. If the anomaly score of a wind turbine is greater than the threshold, it is determined to be an abnormal wind turbine. By combining the statistical analysis of mean deviation and isolation forest, more accurate identification of wind turbine anomalies can be achieved.

[0026] In one embodiment, the operating data is multidimensional data, each dimension corresponds to a different feature of the second wind turbine set, and constructing an isolation forest based on the operating data of the second wind turbine set may include: determining a plurality of different feature sequences, the feature sequences being permutations and combinations of features of each dimension of the operating data, and the feature permutations and combinations between each pair of feature sequences are different; for each feature sequence, randomly selecting a subsample from all second wind turbine sets; and constructing an isolation tree based on each feature sequence and its corresponding subsample to obtain a plurality of isolated trees, wherein the plurality of isolated trees constitute the isolation forest.

[0027] For example, assume that the operating data of a wind turbine includes the following dimensions: wind speed, power generation, rotation speed, and temperature. Features are the dimensions of the operating data of the wind turbine, such as wind speed, power generation, rotation speed, and temperature in the above example. Assuming that the operating data includes the above four features, possible feature sequences include: [wind speed, power generation, rotation speed, temperature], [power generation, wind speed, temperature, rotation speed], [temperature, rotation speed, wind speed, power generation].

[0028] The isolation tree is the basic building block in the isolation forest algorithm. It is a special decision tree that splits data by randomly selecting features and split points until the data points are isolated or the specified tree height is reached.

[0029] For example, assuming there are 10 second wind turbines, each with data of the above 4 features, 20 different feature sequences may be generated. For each sequence, 5 wind turbines are randomly selected as subsamples. Then, an isolation tree is constructed for each feature sequence and the corresponding subsample. Finally, 20 isolated trees are obtained, which together form an isolation forest.

[0030] The above technical solution first generates multiple different feature sequences, then randomly selects some wind turbines as subsamples for each feature sequence, then constructs an isolation tree for each feature sequence and subsample, and finally combines all constructed isolation trees into an isolation forest. By constructing an isolation forest through multi-angle and multi-sample feature combinations, the accuracy and robustness of anomaly detection can be improved.

[0031] In one embodiment, the construction of an isolation tree based on each feature sequence and its corresponding subsample may include: selecting target features of the nodes to be divided in order from the feature sequence, and randomly selecting a split point within the value range of the target feature, wherein each node in the nodes to be divided corresponds to a different second wind turbine generator set; dividing the nodes to be divided into a left child node and a right child node according to the target feature and the split point, wherein the left child node includes a node corresponding to the second wind turbine generator set whose target feature value is less than the split point, and the right child node includes a node corresponding to the second wind turbine generator set whose target feature value is greater than or equal to the split point; dividing the left child node into a right child node; The node is determined as a new node to be divided, and the target feature of the node to be divided corresponding to the second wind turbine group is returned to be executed, and a split point is randomly selected within the value range of the target feature until the depth of the isolated tree reaches the preset maximum depth or the left child node contains only one node corresponding to the second wind turbine group; the right child node is determined as a new node to be divided, and the target feature of the node to be divided corresponding to the second wind turbine group is returned to be executed, and a split point is randomly selected within the value range of the target feature until the depth of the isolated tree reaches the preset maximum depth or the right child node contains only one node corresponding to the second wind turbine group.

[0032] Here, the node to be divided is the node that needs to be divided in the current partition in the isolated tree. Assuming that there are 5 second wind turbines (numbered 1-5), initially, the root node is a node to be divided that contains all 5 wind turbines. The target feature is the feature used for segmentation at the current node. Assuming the feature sequence is [wind speed, power generation, speed], the target feature selected for the first time may be wind speed, the second time may be power generation, and so on. The value range of the target feature is the minimum to maximum value of all samples of the feature in the current node. The split point is a value randomly selected within this range. For example, the wind speed range is 5-15m / s, and 10m / s may be randomly selected as the split point.

[0033] For example, assuming the split point is 10m / s, the wind speeds of wind turbines 1 and 2 are less than 10m / s, and those of 3, 4, and 5 are greater than or equal to 10m / s, then the left child node contains wind turbines 1 and 2, and the right child node contains wind turbines 3, 4, and 5. The preset maximum depth is the maximum number of layers allowed in the isolated tree, which is used to limit the growth of the tree and can be the number of the second wind turbine minus one. If the preset maximum depth is 4, then the isolated tree can only have 4 layers at most.

[0034] The recursive construction process is performed on the left and right child nodes until the depth of the isolated tree reaches the preset maximum depth or the node contains only one wind turbine. For example, the first split can split [1, 2, 3, 4, 5] into [1, 2] and [3, 4, 5]. The second split of the left node can split [1, 2] into [1] and [2]. The second split of the right node can split [3, 4, 5] into [3, 4] and [5].

[0035] The above technical solution first selects the target feature and the split point, then divides the current node into left and right child nodes according to the split point, and finally repeats the first two steps for the left and right child nodes until the termination condition is reached. By randomly selecting features and split points, a diversified decision tree can be constructed to improve the accuracy and robustness of anomaly detection, thereby better identifying anomalies in complex data.

[0036] In one embodiment, determining the anomaly score of each second wind turbine group based on the isolation forest may include: for each second wind turbine group, determining the path length of the target second wind turbine group in each isolated tree; and determining the anomaly score corresponding to the target second wind turbine group based on the average value of all path lengths of the target second wind turbine group.

[0037] Here, the path length refers to the number of edges from the root node of the isolated tree to a specific data point (the second target wind turbine). For example, assuming that the target wind turbine is in an isolated tree, and it has gone through 4 splits from the root node to the leaf node where the wind turbine is located, then its path length in this tree is 4. The average path length refers to the arithmetic mean of the path lengths of the target wind turbine in all isolated trees. Assuming there are 3 isolated trees, and the path lengths of the target wind turbine in these trees are 4, 5, and 6 respectively, then its average path length is (4+5+6) / 3=5.

[0038] The anomaly score can be calculated based on the average path length. For example, the anomaly score can be 2^(-E(h(x)) / c(n)), where E(h(x)) is the average path length and c(n) is a normalization factor for the data size. For example, if the average path length is 5 and c(n)=10, then the anomaly score may be 2^(-5 / 10)≈0.758.

[0039] The above technical solution first traverses all isolated trees for each second wind turbine, then calculates the path length of the target wind turbine in each isolated tree, then calculates the average of all path lengths, and finally calculates the anomaly score based on the average path length. The above solution can more accurately and robustly quantify the degree of anomaly of each second wind turbine, thereby improving the reliability of anomaly detection.

[0040] In one embodiment, the method may further include determining the abnormal characteristics of each abnormal wind turbine based on a Gaussian algorithm, which may specifically include: performing the following steps for each characteristic of the target abnormal wind turbine: establishing a Gaussian distribution model corresponding to the target characteristic based on the historical normal operation data of the target characteristic of the target abnormal wind turbine; determining the probability density of the target characteristic based on the Gaussian distribution model; and determining that the target characteristic is the abnormal characteristic when the probability density is less than a preset probability density.

[0041] Here, the Gaussian algorithm is a statistical method based on normal distribution (Gaussian distribution) for describing and analyzing data. The abnormal feature refers to the abnormal feature in the operation data of the abnormal wind turbine. The target abnormal wind turbine is the specific abnormal wind turbine being analyzed. The target feature refers to the specific feature being analyzed. The historical normal operation data is the historical operation data collected by the target abnormal wind turbine under normal operating conditions.

[0042] The Gaussian distribution model is a statistical model built using historical normal data to describe the normal distribution of features. Assuming that the target feature of the target abnormal wind turbine is wind speed, the average value of the historical normal data of wind speed is 10m / s and the standard deviation is 2m / s, a Gaussian distribution model with a mean of 10 and a standard deviation of 2 can be constructed. The probability density is the probability of a specific value appearing in a Gaussian distribution. For example, using the above Gaussian distribution model, the probability density of a wind speed of 8m / s can be calculated. If the probability density of the target feature is lower than the preset probability density, the target feature can be considered to be an abnormal feature. For example, if the probability density of the target feature wind speed of 15m / s is the preset probability density, the wind speed can be judged to be an abnormal feature.

[0043] The above technical solution first collects the historical normal data of the target abnormal wind turbine, then establishes a Gaussian model for each feature, then calculates the probability density of the corresponding feature value, then compares the probability density with the preset probability density, and finally identifies the abnormal feature. By establishing a statistical model of Gaussian distribution, the abnormal features of abnormal wind turbines can be effectively identified, thereby improving the accuracy and reliability of anomaly detection.

[0044] Figure 2 1 is a block diagram of a wind turbine abnormality determination device 200 according to an exemplary embodiment. The wind turbine abnormality determination device 200 includes: A data acquisition module 210, configured to acquire operating data of a first wind turbine generator set from a sensor, wherein the first wind turbine generator set has a plurality of wind turbine generator sets; A mean deviation module 220, configured to calculate the deviation between the operating data of the first wind turbine generator set and the average value of the operating data thereof, and determine the first wind turbine generator set whose deviation exceeds a preset deviation range as the second wind turbine generator set; The isolation forest module 230 is used to determine abnormal wind turbines in the second wind turbines based on the isolation forest algorithm.

[0045] Optionally, the isolation forest module 230 includes: An isolated forest building module, used for building an isolated forest according to the operation data of the second wind turbine generator set; an abnormal score determination module, used for determining an abnormal score of each second wind turbine generator set based on the isolation forest; The abnormal unit determination module is used to determine the second wind turbine unit whose abnormality score is greater than a preset abnormality threshold as the abnormal wind turbine unit.

[0046] Optionally, the operation data is multi-dimensional data, each dimension corresponds to a different feature of the second wind turbine generator set, and the isolated forest establishment module includes: A feature sequence determination module, used to determine a plurality of different feature sequences, wherein the feature sequences are permutations and combinations of features of each dimension of the operation data, and the feature permutations and combinations between each two feature sequences are different; A sub-sample determination module is used to randomly select a sub-sample from all second wind turbines for each characteristic sequence; The isolated tree determination module is used to construct an isolated tree based on each feature sequence and its corresponding sub-sample to obtain multiple isolated trees, and the multiple isolated trees constitute the isolation forest.

[0047] Optionally, the isolated tree determination module is specifically configured to: Selecting target features of the nodes to be divided in order from the feature sequence, and randomly selecting a split point within the value range of the target features, each of the nodes to be divided corresponds to a different second wind turbine generator set; According to the target feature and the segmentation point, the node to be divided is divided into a left child node and a right child node, the left child node includes a node corresponding to the second wind turbine group whose target feature value is less than the segmentation point, and the right child node includes a node corresponding to the second wind turbine group whose target feature value is greater than or equal to the segmentation point; Determine the left child node as a new node to be divided, return to execute the target feature of randomly selecting the node to be divided corresponding to the second wind turbine group, and randomly select a split point within the value range of the target feature until the depth of the isolated tree reaches a preset maximum depth or the left child node contains only one node corresponding to the second wind turbine group; The right child node is determined as a new node to be divided, and the target feature of randomly selecting the node to be divided corresponding to the second wind turbine group is returned to execute, and a split point is randomly selected within the value range of the target feature until the depth of the isolated tree reaches the preset maximum depth or the right child node contains only one node corresponding to the second wind turbine group.

[0048] Optionally, the anomaly score determination module is specifically configured to: For each second wind turbine group, determining a path length of the target second wind turbine group in each isolated tree; An abnormal score corresponding to the target second wind turbine group is determined according to an average value of all path lengths of the target second wind turbine group.

[0049] Optionally, the device 200 includes a Gaussian distribution analysis module, specifically used for: For each characteristic of the target abnormal wind turbine, perform the following steps: According to the historical normal operation data of the target characteristic of the target abnormal wind turbine group, a Gaussian distribution model corresponding to the target characteristic is established; Determining the probability density of the target feature according to the Gaussian distribution model; When the probability density is less than a preset probability density, the target feature is determined to be the abnormal feature.

[0050] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0051] Figure 3 FIG. 1 is a block diagram of an electronic device 300 according to an exemplary embodiment. Figure 3 As shown, the electronic device 300 may include: a processor 301 , a memory 302 . The electronic device 300 may also include one or more of a multimedia component 303 , an input / output (I / O) interface 304 , and a communication component 305 .

[0052] The processor 301 is used to control the overall operation of the electronic device 300 to complete all or part of the steps in the above-mentioned pressure relief effect evaluation method. The memory 302 is used to store various types of data to support the operation of the electronic device 300. For example, these data may include instructions for any application or method used to operate on the electronic device 300, and application-related data, such as contact data, messages sent and received, pictures, audio, video, etc. The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 303 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 302 or sent through the communication component 305. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 304 provides an interface between the processor 301 and other interface modules, and the other interface modules may be keyboards, mice, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 305 is used for wired or wireless communication between the electronic device 300 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC or other 5G, etc., or a combination of one or more of them, is not limited here. Therefore, the corresponding communication component 305 may include: Wi-Fi module, Bluetooth module, NFC module, etc.

[0053] In an exemplary embodiment, the electronic device 300 can be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to perform the above-mentioned pressure relief effect evaluation method.

[0054] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned pressure relief effect evaluation method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 302 including program instructions, and the above-mentioned program instructions can be executed by the processor 301 of the electronic device 300 to complete the above-mentioned pressure relief effect evaluation method.

[0055] In another exemplary embodiment, a computer program is further provided. The computer program product is a computer program that can be executed by a programmable device. The computer program has a code portion for executing the above-mentioned pressure relief effect evaluation method when executed by the programmable device.

[0056] The preferred embodiments of the present application are described in detail above in conjunction with the accompanying drawings; however, the present application is not limited to the specific details in the above embodiments. Within the technical concept of the present application, a variety of simple modifications can be made to the technical solution of the present application, and these simple modifications all fall within the protection scope of the present application.

[0057] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, this application will not further describe various possible combinations.

[0058] In addition, the various implementation modes of the present application may be arbitrarily combined, and as long as they do not violate the concept of the present application, they should also be regarded as the contents disclosed in the present application.

Claims

1. A method for determining abnormality of a wind turbine generator set, characterized in that: include: Acquire operation data of a first wind turbine generator set from a sensor, wherein the first wind turbine generator set has a plurality of wind turbine generator sets; Calculating the deviation between the operating data of the first wind turbine group and the average value of the operating data thereof, and determining the first wind turbine group whose deviation exceeds a preset deviation range as the second wind turbine group; Based on the isolation forest algorithm, abnormal wind turbines in the second wind turbines are determined.

2. The method according to claim 1, characterized in that The determining of abnormal wind turbines in the second wind turbines based on the isolation forest algorithm includes: constructing an isolated forest according to the operating data of the second wind turbine generator set; Based on the isolation forest, determining an anomaly score for each second wind turbine group; The second wind turbine generator set having an abnormality score greater than a preset abnormality threshold is determined as the abnormal wind turbine generator set.

3. The method according to claim 2, characterized in that The operation data is multi-dimensional data, each dimension of which corresponds to a different feature of the second wind turbine generator set. The isolation forest is constructed according to the operation data of the second wind turbine generator set, including: Determine a plurality of different feature sequences, wherein the feature sequences are permutations and combinations of features of each dimension of the operating data, and the feature permutations and combinations between two feature sequences are different; For each characteristic sequence, a subsample is randomly selected from all second wind turbines; Based on each feature sequence and its corresponding sub-sample, an isolated tree is constructed to obtain a plurality of isolated trees, and the plurality of isolated trees constitute the isolation forest.

4. The method according to claim 3, characterized in that The step of constructing an isolation tree based on each feature sequence and its corresponding subsample includes: Selecting target features of the nodes to be divided in order from the feature sequence, and randomly selecting a split point within the value range of the target features, each of the nodes to be divided corresponds to a different second wind turbine generator set; According to the target feature and the segmentation point, the node to be divided is divided into a left child node and a right child node, the left child node includes a node corresponding to the second wind turbine group whose target feature value is less than the segmentation point, and the right child node includes a node corresponding to the second wind turbine group whose target feature value is greater than or equal to the segmentation point; Determine the left child node as a new node to be divided, return to execute the target feature of randomly selecting the node to be divided corresponding to the second wind turbine group, and randomly select a split point within the value range of the target feature until the depth of the isolated tree reaches a preset maximum depth or the left child node contains only one node corresponding to the second wind turbine group; The right child node is determined as a new node to be divided, and the target feature of randomly selecting the node to be divided corresponding to the second wind turbine group is returned to execute, and a split point is randomly selected within the value range of the target feature until the depth of the isolated tree reaches the preset maximum depth or the right child node contains only one node corresponding to the second wind turbine group.

5. The method according to claim 3, characterized in that: The determining, based on the isolation forest, an abnormal score of each second wind turbine generator set comprises: For each second wind turbine group, determining a path length of the target second wind turbine group in each isolated tree; An abnormal score corresponding to the target second wind turbine group is determined according to an average value of all path lengths of the target second wind turbine group.

6. The method according to any one of claims 2 to 5, characterized in that: The method further includes determining the abnormal characteristics of each abnormal wind turbine generator set based on a Gaussian algorithm, specifically including: For each characteristic of the target abnormal wind turbine, perform the following steps: According to the historical normal operation data of the target characteristic of the target abnormal wind turbine group, a Gaussian distribution model corresponding to the target characteristic is established; Determining the probability density of the target feature according to the Gaussian distribution model; When the probability density is less than a preset probability density, the target feature is determined to be the abnormal feature.

7. A device for determining abnormality of a wind turbine generator set, characterized in that: include: A data acquisition module, used for acquiring operation data of a first wind turbine generator set from a sensor, wherein the first wind turbine generator set has a plurality of wind turbine generator sets; a mean deviation module, used to calculate the deviation between the operating data of the first wind turbine group and the average value of the operating data, and determine the first wind turbine group whose deviation exceeds a preset deviation range as the second wind turbine group; The isolation forest module is used to determine the abnormal wind turbines in the second wind turbines based on the isolation forest algorithm.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

9. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images