Wind power blade fault detection method based on image data

Abstract

The invention discloses a wind power blade fault detection method based on image data, and relates to the field of blade faults, and the method comprises the following steps: obtaining blade motion data, obtaining blade rhythm data according to the blade motion data, and obtaining blade behavior degree according to the blade rhythm data; according to the blade behavior degree, a personalized wind turbine generator blade instruction is sent to the unmanned aerial vehicle detection group, and according to the personalized wind turbine generator blade instruction, the unmanned aerial vehicle detection group obtains unit blade image data of the wind turbine generator blades; according to the unit blade image data, obtaining micro-aggregation image data and a blade micro-aggregation degree, and according to the blade micro-aggregation degree, analyzing the micro-aggregation image data to obtain a blade covering factor; obtaining blade fingerprint state data according to the blade covering factor and the micro-gathering image data, obtaining a blade fault damage value according to the fingerprint state data, and carrying out wind turbine generator blade maintenance management through the blade fault damage value; the operation and maintenance efficiency of the wind power plant can be improved.

Description

Technical Field

[0001] The invention relates to the field of blade failures, and in particular to a method for detecting wind turbine blade failures based on image data. Background Art

[0002] With the increase in global demand for renewable energy, wind power generation has developed rapidly around the world. As the core equipment of wind farms, the efficient operation and long-term stability of wind turbines are crucial to the entire power production system. However, due to the harsh operating environment of wind turbines, damage and failure of wind turbine blades are inevitable. For example, the accuracy and stability of detection equipment are insufficient, and the fault diagnosis method is single. How to quickly and accurately detect wind turbine blade failures has become a key issue in improving wind power reliability and reducing maintenance costs. To this end, a wind turbine blade fault detection method based on image data is provided. Summary of the invention

[0003] In order to solve the above technical problems, the object of the present invention is to provide a wind turbine blade fault detection method based on image data, comprising the following steps: Step S1: setting up a drone detection group to obtain blade motion data of wind turbine blades, and obtaining blade rhythm data based on the blade motion data, and performing rhythm judgment on the blade motion data based on the blade rhythm data to obtain blade behavior degree; Step S2: sending a personalized wind turbine blade instruction to the drone detection group according to the blade behavior degree, and the drone detection group acquires the wind turbine blade image data according to the personalized wind turbine blade instruction; Step S3: obtaining micro-aggregation image data and blade micro-aggregation degree according to the blade image data of the unit, analyzing the micro-aggregation image data according to the blade micro-aggregation degree, and obtaining the blade coverage factor; Step S4: extract the micro-aggregate image data in a targeted manner through the blade texture factor to obtain the blade fingerprint status data, analyze the fingerprint status data to obtain the blade fault damage value, and perform wind turbine blade maintenance management through the blade fault damage value.

[0004] Furthermore, a process of setting up a drone detection group to obtain blade motion data of wind turbine blades and obtaining blade rhythm data based on the blade motion data includes: The drone detection group includes several drone detection groups, each of which is provided with n detection drones; the blade motion data includes blade motion image data and blade motion time; a detection cycle is set, and the drone detection collects blade motion data of all wind turbine blades according to the detection cycle; According to the blade motion data, the blade initial mark is obtained; and according to the blade initial mark, the blade motion time and the blade motion image data, the blade vibration amplitude and the blade vibration time are obtained; according to the detection period and the blade vibration amplitude, the blade vibration frequency is obtained, and according to the blade vibration amplitude, the blade vibration time, the blade initial mark and the blade vibration frequency, the blade rhythm data is obtained.

[0005] Furthermore, according to the blade rhythm data, the blade motion data is subjected to rhythm judgment, and the process of obtaining the blade behavior degree includes: According to the blade initial mark, detection period, blade vibration amplitude and blade vibration frequency of the blade rhythm data, the blade amplitude difference value and the blade frequency difference value are obtained; according to the blade amplitude difference value and the blade frequency difference value, the blade behavior degree is obtained.

[0006] Furthermore, according to the blade behavior degree, a personalized wind turbine blade instruction is sent to the UAV detection group. According to the personalized wind turbine blade instruction, the process in which the UAV detection group obtains the wind turbine blade image data includes: According to the blade motion data, the blade behavior parameter coefficient is set; according to the blade behavior degree and the blade behavior parameter coefficient, the wind turbine blade area block is obtained, and according to the wind turbine positioning information and the wind turbine blade area block, the personalized wind turbine blade instruction is sent to the UAV detection group; according to the personalized wind turbine blade instruction, the unit blade image data of the wind turbine blade is obtained, and according to the blade initial mark and the unit blade image, the blade initial association relationship is obtained, and the unit blade image data includes the image data of each wind turbine blade area and the blade initial association relationship.

[0007] Further, the process of obtaining micro-aggregation image data and blade micro-aggregation degree according to the blade image data of the unit includes: According to the initial association relationship of the blades of the unit blade image data, the unit blade image group data is obtained; according to the unit blade image group data, the unit blade lossless image group data is obtained; and according to the unit blade image group data and the unit blade lossless image group data, the micro-aggregate image data is obtained, and according to the initial association relationship of the blades, the micro-aggregate image group data is obtained; the extraction sample is set; according to the extraction sample, the unit blade image group data and the micro-aggregate image group data, the unit blade image sample and the micro-aggregate image sample are obtained; and the blade sample amount and the micro-aggregate sample amount are obtained; according to the micro-aggregate sample amount and the blade sample amount, the blade micro-aggregate degree is obtained.

[0008] Furthermore, the process of analyzing the micro-aggregation image data according to the leaf micro-aggregation degree and obtaining the leaf texture factor includes: If the blade micro-aggregation degree is not equal to the blade behavior degree, the grayscale processing is performed on the lossless image group data of the unit blades until the blade micro-aggregation degree is equal to the blade behavior degree, and the micro-aggregation image group data is updated; if the blade micro-aggregation degree is equal to the blade behavior degree, strip recognition points and circular recognition points are set, and according to the strip recognition points and the circular recognition points, the local abnormal range of the blades and the concentrated abnormal range of the blades in the micro-aggregation image sample are respectively obtained, and according to the local abnormal range of the blades and the total range of the micro-aggregation image samples, the blade coverage factor one is obtained; according to the concentrated abnormal range of the blades and the total range of the micro-aggregation strip image samples, the blade coverage factor two is obtained; according to the blade coverage factor one and the blade coverage factor two, the blade coverage factor is obtained.

[0009] Furthermore, the process of extracting the micro-aggregate image data in a targeted manner through the leaf texture factor to obtain the leaf fingerprint state data includes: The micro-aggregate image data is extracted in a targeted manner through the leaf texture factor. The fingerprint state one data of the leaf is obtained according to the leaf texture factor one and the micro-aggregate image data. The fingerprint state two data of the leaf is obtained according to the leaf texture factor two and the micro-aggregate image data. And the fingerprint state data of the leaf is obtained according to the fingerprint state one data and the fingerprint state two data.

[0010] Furthermore, the fingerprint status data is analyzed to obtain the blade fault damage value, and the process of wind turbine blade maintenance management is performed based on the blade fault damage value, including: The fingerprint status data is analyzed, and the blade fault one damage value is obtained according to the fingerprint status one data; the blade fault two damage value is obtained according to the fingerprint status two data; the environmental impact coefficient is set, and the blade fault damage value is obtained according to the environmental impact coefficient, the blade fault one damage value and the blade fault two damage value; and the fault damage of the wind turbine blade is located through the micro-aggregate image samples, and the fault damage location, micro-aggregate image samples and blade fault damage value are sent to the operation and maintenance personnel to perform wind turbine blade maintenance management.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: setting up a drone detection group to obtain blade motion data of wind turbine blades, and obtaining blade rhythm data based on the blade motion data, and making a rhythm judgment on the blade motion data based on the blade rhythm data to obtain the blade behavior degree, which can improve the detection accuracy and efficiency, timely discover blade failures, avoid major accidents, and improve the safety and stability of wind turbines; according to the blade behavior degree, sending personalized wind turbine blade instructions to the drone detection group, according to the personalized wind turbine blade instructions, the drone detection group obtains the unit blade image data of the wind turbine blades, which can reduce manual intervention, improve detection efficiency, improve the pertinence and accuracy of data acquisition, and reduce the collection of a large amount of unnecessary data; according to the unit Blade image data, obtain micro-aggregate image data and blade micro-aggregate degree, analyze the micro-aggregate image data according to the blade micro-aggregate degree, and obtain the blade coverage factor, which can improve the accuracy of image data, improve the anti-interference ability of image acquisition, and help to timely discover potential fault risks, improve the operating efficiency and safety of wind turbines, reduce maintenance costs, reduce downtime, and improve the overall operation and maintenance efficiency of wind farms; through the blade coverage factor, the micro-aggregate image data is extracted in a targeted manner to obtain the blade fingerprint status data, and the fingerprint status data is analyzed to obtain the blade fault damage value, and the blade fault damage value is used to manage the maintenance of wind turbine blades, which can formulate more accurate and efficient maintenance plans to avoid excessive maintenance or greater problems caused by ignoring faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a schematic diagram of a wind turbine blade fault detection method based on image data according to an embodiment of the present application. DETAILED DESCRIPTION

[0013] like Figure 1 As shown, a wind turbine blade fault detection method based on image data includes the following steps: Step S1: setting up a drone detection group to obtain blade motion data of wind turbine blades, and obtaining blade rhythm data based on the blade motion data, and performing rhythm judgment on the blade motion data based on the blade rhythm data to obtain blade behavior degree; Step S2: sending a personalized wind turbine blade instruction to the drone detection group according to the blade behavior degree, and the drone detection group acquires the wind turbine blade image data according to the personalized wind turbine blade instruction; Step S3: obtaining micro-aggregation image data and blade micro-aggregation degree according to the blade image data of the unit, analyzing the micro-aggregation image data according to the blade micro-aggregation degree, and obtaining the blade coverage factor; Step S4: extract the micro-aggregate image data in a targeted manner through the blade texture factor to obtain the blade fingerprint status data, analyze the fingerprint status data to obtain the blade fault damage value, and perform wind turbine blade maintenance management through the blade fault damage value.

[0014] It should be further explained that, in the specific implementation process, a drone detection group is set to obtain the blade motion data of the wind turbine blades, and the blade rhythm data is obtained based on the blade motion data. The blade motion data is judged based on the blade rhythm data, and the process of obtaining the blade behavior degree includes: The drone detection group includes a plurality of drone detection groups, each of which is provided with n detection drones; It should be further explained that, in the specific implementation process, the drone detection group is used to obtain the blade motion data of the wind turbine blades, and the detection drone is provided with a plurality of micro cameras, and m detection drones are provided to obtain the blade motion data of the same wind turbine, and m is less than n; The blade motion data includes blade motion image data and blade motion time; Set a detection cycle, and the drone detection collects blade motion data of all wind turbine blades according to the detection cycle; According to the blade motion data, the blade motion data is analyzed in state to obtain the blade rhythm data; It should be further explained that, in the specific implementation process, the process of state analysis of blade motion data according to blade motion data includes: using big data to identify the contours of wind turbine blades in blade motion image data, and initially marking the blades on the initially collected blade motion image data, recording them as blade A, blade B and blade C; and tracking the blade motion image data according to the blade initial marking and blade motion time to obtain blade vibration amplitude and blade vibration time; the blade tracking refers to the process of overlapping blade motion image data according to the blade motion time sequence and blade initial marking, and comparing the contours of blade motion image data of adjacent blade motion times; through the detection cycle, the number of times the blade vibration amplitude is the same in the detection cycle is counted and the ratio is calculated with the detection cycle to obtain the blade vibration frequency, and the blade vibration amplitude, blade vibration time, blade initial marking and blade vibration frequency are recorded as blade rhythm data; According to the blade rhythm data, the blade motion data is judged rhythmically to obtain the blade behavior degree; It needs to be further explained that, in the specific implementation process, the specific process of judging the rhythm of the blade motion data according to the blade rhythm data is as follows: according to the initial mark of the blades in the blade rhythm data, the wind turbine blades of each wind turbine are analyzed through the detection cycle, and within the same detection cycle, the blade vibration amplitude and blade vibration frequency of all wind turbine blades of the wind turbine are analyzed, the detection cycle is divided into m blade cycles, and the blade vibration amplitude and blade vibration frequency of adjacent wind turbine blades within the blade cycle are calculated to obtain the blade amplitude difference value and the blade frequency difference value; the blade amplitude difference value is calculated as a ratio with the blade frequency difference value to obtain the blade behavior degree.

[0015] It should be further explained that, in the specific implementation process, according to the blade behavior degree, a personalized wind turbine blade instruction is sent to the UAV detection group. According to the personalized wind turbine blade instruction, the process in which the UAV detection group obtains the unit blade image data of the wind turbine blade includes: Send personalized wind turbine blade instructions to the drone detection group based on blade behavior; It should be further explained that, in the specific implementation process, according to the blade behavior degree, the specific process of sending personalized wind turbine blade instructions to the drone detection group is: analyzing the blade motion data through deep learning, and setting the blade behavior parameter coefficient; if the blade behavior degree is less than or equal to the blade behavior parameter coefficient, it means that the wind turbine is operating normally; if the blade behavior degree is greater than the blade behavior parameter coefficient, it means that the wind turbine is abnormal, and the wind turbine positioning information is obtained, and all wind turbine blades of the wind turbine are randomly divided into regions, recorded as wind turbine blade regions, and wind turbine blade region blocks are obtained. Through the wind turbine positioning information and wind turbine blade region blocks, personalized wind turbine blade instructions are sent to the drone detection group, and the personalized wind turbine blade instructions refer to targeted limitation of the detection content of the drone detection group; According to the personalized wind turbine blade instructions, the drone detection group plans the driving path of the drone detection group according to the wind turbine positioning information, drives to the wind turbine corresponding to the wind turbine positioning, and plans the detection of the drone detection group according to the wind turbine blade area block, arranges the drone detection group to detect the wind turbine blades in turn through the initial blade marking, and arranges the drones in the drone detection group to perform partitioned and rotated detection according to the wind turbine blade area, obtains the unit blade image data of the wind turbine blade, and associates the blade initial marking with the unit blade image to obtain the blade initial association relationship, and the unit blade image data includes the image data of each wind turbine blade area and the blade initial association relationship.

[0016] It should be further explained that, in the specific implementation process, the process of obtaining the micro-aggregation image data and the blade micro-aggregation degree according to the unit blade image data, analyzing the micro-aggregation image data according to the blade micro-aggregation degree, and obtaining the blade coverage factor includes: According to the initial association relationship of the blades in the unit blade image data, the unit blade image data are grouped to obtain the unit blade image group data; the unit blade image group data are respectively identified as damage-free by deep learning to obtain the unit blade damage-free image group data; and the non-damaged area in the unit blade image group data is separated from the unit blade image group data by using a semantic segmentation algorithm to obtain micro-aggregate image data, and the micro-aggregate image data are grouped according to the initial association relationship of the blades to obtain micro-aggregate image group data; Set an extraction sample, the specification of the extraction sample is e*h, e is the length of the extraction sample, and h is the width of the extraction sample; According to the extracted samples, sample extraction is performed on the unit blade image group data and the micro-aggregate image group data respectively to obtain the unit blade image samples and the micro-aggregate image samples; and the number of extracted samples of the unit blade image group data and the micro-aggregate image group data is counted respectively to obtain the blade sample volume and the micro-aggregate sample volume respectively; the ratio of the micro-aggregate sample volume to the blade sample volume is calculated to obtain the blade micro-aggregate degree; The micro-aggregation image data is analyzed according to the leaf micro-aggregation degree to obtain the leaf covering factor; It should be further explained that, in the specific implementation process, the specific process of analyzing the micro-aggregation image data according to the blade micro-aggregation degree is as follows: if the blade micro-aggregation degree is not equal to the blade behavior degree, the unit blade lossless image group data is gray-scale processed, and the unit blade lossless image group data is respectively identified as damage-free by deep learning until the blade micro-aggregation degree is equal to the blade behavior degree, and the micro-aggregation image group data is updated; if the blade micro-aggregation degree is equal to the blade behavior degree, a strip recognition point and a circular recognition point are set, and the strip recognition point is used to identify the abnormal blade details in the micro-aggregation image data, and the circular recognition point is used to identify the abnormal blade details in the micro-aggregation image data. The identification points are used to identify concentrated abnormalities of leaves in micro-aggregate image data. The micro-aggregate image samples are identified by strip identification points and circular identification points, and the local abnormal range of leaves and the concentrated abnormal range of leaves in the micro-aggregate image samples are obtained respectively. The ratio of the local abnormal range of leaves to the total range of micro-aggregate image samples is calculated to obtain the leaf covering factor one; the ratio of the concentrated abnormal range of leaves to the total range of micro-aggregate strip image samples is calculated to obtain the leaf covering factor two; the leaf covering factor one and the leaf covering factor two are recorded as leaf covering factors; and the leaf covering factors are marked in the corresponding micro-aggregate image samples.

[0017] It should be further explained that, in the specific implementation process, the micro-aggregate image data is extracted specifically through the blade texture factor to obtain the blade fingerprint status data, and the fingerprint status data is analyzed to obtain the blade fault damage value, and the process of wind turbine blade maintenance management through the blade fault damage value includes: The leaf fingerprint status data is obtained by extracting the micro-aggregate image data in a targeted manner through the leaf texture factor; It should be further explained that, in the specific implementation process, the specific process of extracting the micro-aggregate image data in a targeted manner through the leaf texture factor is as follows: the micro-aggregate image data is locally amplified through the leaf texture factor 1 to obtain the fingerprint state 1 data of the leaf, and the fingerprint state 1 data includes the starting point, the extension direction, the length and the depth; the micro-aggregate image data is centrally observed through the leaf texture factor 2 to obtain the fingerprint state 2 data of the leaf, and the fingerprint state 2 data includes the area size and the distribution density; and the fingerprint state 1 data and the fingerprint state 2 data are recorded as the leaf fingerprint state data; Analyze the fingerprint status data to obtain the blade fault damage value; It should be further explained that, in the specific implementation process, the specific process of analyzing the fingerprint state data includes: multiplying and accumulating the length and depth of the fingerprint state data to obtain the severity of the fault damage at the local details of the wind turbine blade, which is recorded as the blade fault damage value; multiplying and accumulating the area size and distribution density of the fingerprint state data to obtain the severity of the fault damage at the concentrated part of the wind turbine blade, which is recorded as the blade fault damage value; setting the environmental impact coefficient, which is related to the environmental factors in which the wind turbine blade is located, multiplying the environmental impact coefficient, the blade fault damage value, and the blade fault damage value to obtain the severity of the blade fault damage of the wind turbine blade under the influence of environmental factors, which is recorded as the blade fault damage value; The fault damage of wind turbine blades is located through micro-aggregate image samples, and the fault damage location, micro-aggregate image samples and blade fault damage values ​​are sent to operation and maintenance personnel to carry out wind turbine blade maintenance management.

[0018] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A wind turbine blade fault detection method based on image data, characterized in that: The following steps are involved: Step S1: setting up a drone detection group to obtain blade motion data of wind turbine blades, and obtaining blade rhythm data based on the blade motion data, and performing rhythm judgment on the blade motion data based on the blade rhythm data to obtain blade behavior degree; Step S2: sending a personalized wind turbine blade instruction to the drone detection group according to the blade behavior degree, and the drone detection group acquires the wind turbine blade image data according to the personalized wind turbine blade instruction; Step S3: obtaining micro-aggregation image data and blade micro-aggregation degree according to the blade image data of the unit, analyzing the micro-aggregation image data according to the blade micro-aggregation degree, and obtaining the blade coverage factor; Step S4: extract the micro-aggregate image data in a targeted manner through the blade texture factor to obtain the blade fingerprint status data, analyze the fingerprint status data to obtain the blade fault damage value, and perform wind turbine blade maintenance management through the blade fault damage value.

2. The method for detecting wind turbine blade faults based on image data according to claim 1, characterized in that: The process of setting up a drone detection group to obtain blade motion data of wind turbine blades and obtaining blade rhythm data based on the blade motion data includes: The drone detection group includes several drone detection groups, each of which is provided with n detection drones; the blade motion data includes blade motion image data and blade motion time; a detection cycle is set, and the drone detection collects blade motion data of all wind turbine blades according to the detection cycle; According to the blade motion data, the blade initial mark is obtained; and according to the blade initial mark, the blade motion time and the blade motion image data, the blade vibration amplitude and the blade vibration time are obtained; according to the detection period and the blade vibration amplitude, the blade vibration frequency is obtained, and according to the blade vibration amplitude, the blade vibration time, the blade initial mark and the blade vibration frequency, the blade rhythm data is obtained.

3. The method for detecting wind turbine blade faults based on image data according to claim 2, characterized in that: According to the blade rhythm data, the blade motion data is judged rhythmically and the process of obtaining the blade behavior degree includes: According to the blade initial mark, detection period, blade vibration amplitude and blade vibration frequency of the blade rhythm data, the blade amplitude difference value and the blade frequency difference value are obtained; according to the blade amplitude difference value and the blade frequency difference value, the blade behavior degree is obtained.

4. The method for detecting wind turbine blade faults based on image data according to claim 3, characterized in that: According to the blade behavior degree, a personalized wind turbine blade instruction is sent to the UAV detection group. According to the personalized wind turbine blade instruction, the process in which the UAV detection group obtains the unit blade image data of the wind turbine blade includes: According to the blade motion data, the blade behavior parameter coefficient is set; according to the blade behavior degree and the blade behavior parameter coefficient, the wind turbine blade area block is obtained, and according to the wind turbine positioning information and the wind turbine blade area block, the personalized wind turbine blade instruction is sent to the UAV detection group; according to the personalized wind turbine blade instruction, the unit blade image data of the wind turbine blade is obtained, and according to the blade initial mark and the unit blade image, the blade initial association relationship is obtained, and the unit blade image data includes the image data of each wind turbine blade area and the blade initial association relationship.

5. The method for detecting wind turbine blade faults based on image data according to claim 4, characterized in that: The process of obtaining micro-aggregation image data and blade micro-aggregation degree according to the unit blade image data includes: According to the initial association relationship of the blades of the unit blade image data, the unit blade image group data is obtained; according to the unit blade image group data, the unit blade lossless image group data is obtained; and according to the unit blade image group data and the unit blade lossless image group data, the micro-aggregate image data is obtained, and according to the initial association relationship of the blades, the micro-aggregate image group data is obtained; the extraction sample is set; according to the extraction sample, the unit blade image group data and the micro-aggregate image group data, the unit blade image sample and the micro-aggregate image sample are obtained; and the blade sample amount and the micro-aggregate sample amount are obtained; according to the micro-aggregate sample amount and the blade sample amount, the blade micro-aggregate degree is obtained.

6. The method for detecting wind turbine blade faults based on image data according to claim 5, characterized in that: The process of analyzing the micro-aggregation image data according to the leaf micro-aggregation and obtaining the leaf texture factor includes: If the blade micro-aggregation degree is not equal to the blade behavior degree, the grayscale processing is performed on the lossless image group data of the unit blades until the blade micro-aggregation degree is equal to the blade behavior degree, and the micro-aggregation image group data is updated; if the blade micro-aggregation degree is equal to the blade behavior degree, strip recognition points and circular recognition points are set, and according to the strip recognition points and the circular recognition points, the local abnormal range of the blades and the concentrated abnormal range of the blades in the micro-aggregation image sample are respectively obtained, and according to the local abnormal range of the blades and the total range of the micro-aggregation image samples, the blade coverage factor one is obtained; according to the concentrated abnormal range of the blades and the total range of the micro-aggregation strip image samples, the blade coverage factor two is obtained; according to the blade coverage factor one and the blade coverage factor two, the blade coverage factor is obtained.

7. The method for detecting wind turbine blade faults based on image data according to claim 6, characterized in that: The process of extracting micro-aggregate image data in a targeted manner through leaf texture factors and obtaining leaf fingerprint status data includes: The micro-aggregate image data is extracted in a targeted manner through the leaf texture factor. The fingerprint state one data of the leaf is obtained according to the leaf texture factor one and the micro-aggregate image data. The fingerprint state two data of the leaf is obtained according to the leaf texture factor two and the micro-aggregate image data. And the fingerprint state data of the leaf is obtained according to the fingerprint state one data and the fingerprint state two data.

8. The method for detecting wind turbine blade faults based on image data according to claim 7, characterized in that: The process of analyzing the fingerprint status data, obtaining the blade fault damage value, and performing wind turbine blade maintenance management based on the blade fault damage value includes: The fingerprint status data is analyzed, and the blade fault one damage value is obtained according to the fingerprint status one data; the blade fault two damage value is obtained according to the fingerprint status two data; the environmental impact coefficient is set, and the blade fault damage value is obtained according to the environmental impact coefficient, the blade fault one damage value and the blade fault two damage value; and the fault damage of the wind turbine blade is located through the micro-aggregate image samples, and the fault damage location, micro-aggregate image samples and blade fault damage value are sent to the operation and maintenance personnel to perform wind turbine blade maintenance management.

Images