Wind power plant variable pitch system unattended intelligent operation and maintenance device based on AI and unmanned aerial vehicle cluster cooperation

Abstract

The invention provides a wind power plant pitch system unattended intelligent operation and maintenance device based on AI and unmanned aerial vehicle cluster cooperation, and relates to the technical field of intelligent operation and maintenance, and the device comprises a data collection module which is used for carrying out the data collection based on a preset high-precision sensor, carrying out the cooperative inspection based on an unmanned aerial vehicle cluster, and comprehensively obtaining the initial collection data; the processing fusion module is used for performing data processing on the initial collection data and performing data fusion to obtain target fusion data; the extraction and recognition module is used for carrying out feature extraction based on data features of operation of the variable pitch system and carrying out feature classification based on a machine learning algorithm to obtain classified data; and the diagnosis and maintenance module is used for carrying out fault diagnosis on the classified data and carrying out maintenance prediction on the target variable pitch system so as to carry out comprehensive intelligent maintenance according to a fault diagnosis result and a prediction maintenance result. The intelligent and unmanned operation and maintenance of the wind power plant variable pitch system can be realized, and the operation and maintenance efficiency and the equipment reliability are obviously improved.

Description

Technical Field

[0001] The present invention relates to the field of risk warning technology, and in particular to an unmanned intelligent operation and maintenance device for a wind farm pitch control system based on collaboration between AI and drone clusters. Background Art

[0002] Currently, in wind farm operations, the pitch system, as a core control component, frequently fails, accounting for as much as 60%-70% of total wind turbine failures, becoming a key factor affecting the power generation efficiency and operation and maintenance costs of wind farms.

[0003] However, traditional manual inspection methods are inefficient and struggle to respond to sudden faults in real time, resulting in delayed O&M responses. Furthermore, existing AI diagnostic technologies rely heavily on SCADA data and lack the multi-dimensional data fusion of high-precision sensors and drone near-field detection, limiting the accuracy and comprehensiveness of fault diagnosis. Furthermore, the use of drones in wind farms is largely limited to blade inspections, failing to form an effective closed-loop linkage with pitch control system O&M, hindering the full potential of drones in intelligent O&M.

[0004] Therefore, the present invention provides an unmanned intelligent operation and maintenance device for a wind farm pitch control system based on the collaboration of AI and drone clusters. Summary of the Invention

[0005] The present invention provides an unmanned intelligent operation and maintenance device for a wind farm pitch control system based on the collaboration of AI and drone clusters. The device is used to collect data through preset high-precision sensors and drone clusters, extract key features, and then use machine learning algorithms to classify and diagnose faults and predict maintenance needs. Finally, the drone cluster is used to carry out comprehensive intelligent maintenance. It can efficiently obtain comprehensive and accurate data, quickly and accurately diagnose faults, thereby ensuring the stable operation of the pitch control system and improving overall operation and maintenance efficiency.

[0006] The present invention provides an unmanned intelligent operation and maintenance device for a wind farm pitch control system based on collaboration between AI and a drone cluster, comprising: The data acquisition module is used to collect data based on preset high-precision sensors and conduct collaborative inspections based on drone clusters to comprehensively obtain initial collected data; The processing and fusion module is used to clean and standardize the initial collected data, and perform data fusion based on the processed initial collected data to obtain the target fusion data; An extraction and recognition module is used to extract features from the target fusion data based on the data features of the variable pitch system operation to obtain target extraction features, and to classify the target extraction features based on a machine learning algorithm to obtain classified data; The diagnostic maintenance module is used to diagnose faults on classified data and predict maintenance for the target variable pitch system, thereby performing comprehensive intelligent maintenance in combination with the UAV cluster based on the fault diagnosis results and predicted maintenance results.

[0007] The data acquisition module provided by the present invention includes: An operation acquisition unit is used to obtain real-time operation data and real-time maintenance data of the target pitch system of the wind farm based on preset high-precision sensors; The inspection and collection unit is used to inspect the target variable pitch system based on the preset inspection path by the drone cluster to obtain real-time inspection data; The data integration unit is used to integrate the real-time operation data, real-time maintenance data and real-time inspection data to obtain the initial acquisition data of the target pitch control system of the wind farm.

[0008] The inspection and collection unit provided by the present invention includes: The inspection planning subunit is used to combine the AI platform of the intelligent management terminal with the wind farm layout information and real-time wind farm environmental data and the drone equipment information of the drone cluster to plan the inspection route and obtain the preset inspection path; Cluster inspection subunit, used to equip the drone cluster with preset sensors and collect inspection data of the target variable pitch system based on the preset inspection path; Among them, the inspection data includes real-time blade images, real-time temperature data, real-time equipment operation vibration data, etc. Multiple data transmission subunits are used to upload inspection data to the AI platform of the intelligent management terminal in real time to obtain real-time inspection data.

[0009] The inspection planning subunit provided by the present invention includes: The first information acquisition block is used to obtain the electric field environment information of the target wind farm and the equipment distribution of the target pitch control system, and comprehensively obtain the wind farm layout information; The second information acquisition block is used to obtain real-time environmental data of the target wind farm and drone equipment information of the drone cluster; The initial path planning block is used to obtain the health status of each target device in the target pitch system and the real-time environmental data of the corresponding area, comprehensively determine the inspection priority of the current target device, and then determine the initial inspection planning path by combining the UAV device information and wind farm layout information; The optimization model determination block is used to determine the multi-objective optimization model of the drone cluster based on the comprehensive inspection objectives of the target wind farm; The inspection path optimization block is used to optimize the initial inspection planning path by combining the multi-objective optimization model with real-time environmental data to determine the preset inspection path.

[0010] The processing fusion module provided by the present invention includes: A data processing unit is used to perform data cleaning and data standardization on the initial collected data to obtain initial processed data; an algorithm setting unit, configured to set a data fusion algorithm based on data characteristics of a target pitch system to obtain a preset data fusion algorithm; The data fusion unit is used to perform data fusion on the initial processing data based on a preset data fusion algorithm to obtain target fusion data of the target pitch control system.

[0011] The extraction and identification module provided by the present invention includes: a feature extraction unit, configured to determine a comprehensive feature extraction scheme based on data features of the pitch system operation, and perform feature extraction on the target fusion data based on the comprehensive feature extraction scheme to obtain target extraction features; The feature classification unit is used to classify the target extracted features based on a preset machine learning algorithm to obtain classified data.

[0012] The diagnostic maintenance module provided by the present invention includes: A matching analysis unit, configured to perform rule matching on the classified data based on a preset expert database and identify the fault type of the classified data; A correlation analysis unit, used to perform correlation analysis on the classified data based on a preset neural network to determine the fault mode; A first prediction unit is configured to predict the service life of the target pitch system based on historical fault data in combination with a time series analysis method to obtain a first prediction result; a risk assessment unit, configured to perform a risk assessment on a failure risk level of a target pitch system based on a failure type, a failure mode, and a first prediction result of each device; The integrated maintenance unit is used to determine the maintenance strategy based on the risk assessment results and the drone cluster, and perform integrated intelligent maintenance on the target variable pitch system.

[0013] The integrated maintenance unit provided by the present invention comprises: A position synthesis subunit is used to obtain device position information of each target device of the target pitch system whose risk assessment result exceeds the preset standard assessment result, and obtain the comprehensive device position information of the target pitch system; The cluster response subunit is used to plan the inspection path based on the comprehensive equipment location information of the target variable pitch system, the UAV equipment information in the UAV cluster, and the real-time environmental data corresponding to each target equipment in the comprehensive equipment location information, to obtain the initial inspection planning path; The path judgment subunit is used to judge whether there is any overlap between the initial inspection plan path and the preset inspection path; A path extraction subunit is used to extract a portion of the initial inspection plan path that does not overlap with the preset inspection path to obtain a first inspection plan path; A path optimization subunit is used to obtain a second inspection planning path for the UAV cluster based on the starting point of the UAV cluster and the shortest path planning principle in combination with the first inspection planning path; A second judgment subunit is used to judge the path feasibility of the second inspection plan path; If there is an infeasible path, extract the infeasible path segment and replace it with the preset inspection path, so as to obtain the third inspection planning path of the drone cluster by combining the remaining inspection planning paths; Otherwise, the second inspection planning path is used as the third inspection planning path of the UAV cluster.

[0014] A depth inspection subunit is used to perform a depth inspection on the target variable pitch system using a UAV cluster combined with a third inspection planning path to obtain a depth inspection result; An inspection determination subunit, configured to determine a first impact weight of the in-depth inspection result and a second impact weight of the risk assessment result based on the historical impact levels of the in-depth inspection result and the risk assessment result; A fault synthesis subunit is used to combine the first impact weight with the in-depth inspection result, and the second impact weight with the risk assessment result to obtain a comprehensive fault condition of the target pitch system; A maintenance decision subunit is used to compare the comprehensive fault conditions of the target pitch system with the preset fault-maintenance database one by one, thereby determining the system maintenance strategy for the target pitch system; The decision-making execution subunit is used to realize automatic adjustment and control of each target device of the target pitch system based on the system maintenance strategy, thereby realizing comprehensive intelligent maintenance.

[0015] The unmanned intelligent operation and maintenance device for the wind farm pitch control system based on the collaboration of AI and drone clusters provided by the present invention collects data through preset high-precision sensors and drone clusters, extracts key features, and then uses machine learning algorithms to classify and diagnose faults and predict maintenance needs. Finally, it uses drone clusters to carry out comprehensive intelligent maintenance. It can efficiently obtain comprehensive and accurate data, quickly and accurately diagnose faults, thereby ensuring the stable operation of the pitch control system and improving overall operation and maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 This is a structural diagram of an unmanned intelligent operation and maintenance device for a wind farm pitch control system based on collaboration between AI and a drone cluster, provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0019] Example 1: The embodiment of the present invention provides an unmanned intelligent operation and maintenance device for a wind farm pitch control system based on AI and drone cluster collaboration, such as Figure 1 Shown, including: The data acquisition module is used to collect data based on preset high-precision sensors and conduct collaborative inspections based on drone clusters to comprehensively obtain initial collected data; The processing and fusion module is used to clean and standardize the initial collected data, and perform data fusion based on the processed initial collected data to obtain the target fusion data; An extraction and recognition module is used to extract features from the target fusion data based on the data features of the variable pitch system operation to obtain target extraction features, and to classify the target extraction features based on a machine learning algorithm to obtain classified data; The diagnostic maintenance module is used to diagnose faults on classified data and predict maintenance for the target variable pitch system, thereby performing comprehensive intelligent maintenance in combination with the UAV cluster based on the fault diagnosis results and predicted maintenance results.

[0020] In this embodiment, the preset high-precision sensor is a precision measuring device whose parameters and functions are set in advance according to specific needs. It can accurately capture various physical quantities such as temperature, pressure, vibration amplitude, etc., and provide a high-quality and reliable raw data basis for data analysis.

[0021] In this embodiment, the drone cluster is composed of several drones with collaborative working capabilities. Through advanced communication and coordination mechanisms, different drones in the drone cluster can jointly perform inspection, data collection and other tasks like a team, achieving efficient and comprehensive information acquisition.

[0022] In this embodiment, the initial collected data is measured by using preset high-precision sensors to measure each target object in the variable pitch system, and by using a drone cluster to collect information from different angles and heights. The resulting raw data set without any processing includes the original measurement values, images, videos, etc. of various physical quantities.

[0023] In this embodiment, data cleaning is to conduct a comprehensive check on the initially collected data to identify and eliminate noise interference, erroneous data, duplicate information, and incomplete data items, so as to ensure the accuracy and reliability of the data.

[0024] In this embodiment, data standardization processing is to convert data from different data sources with different formats and dimensions into a unified format and range through specific algorithms and rules, eliminate the differences and scale effects between the data, and enable different data to be compared and analyzed under the same standard.

[0025] In this embodiment, data fusion is to organically integrate information about the same target obtained from multiple sensors or different data sources to make up for the shortcomings of a single data.

[0026] In this embodiment, the target fusion data is a data set obtained after data cleaning, standardization, and data fusion and information integration, and has the characteristics of high quality, comprehensive information, and unified format.

[0027] In this embodiment, the pitch system is a vital component of the wind turbine generator set, and is mainly responsible for adjusting the angle of the blades to adapt to different wind speed conditions, so that the wind turbine can always maintain the best operating state and efficiently convert wind energy into electrical energy.

[0028] In this embodiment, the target extraction feature is to extract key information closely related to the operating status of the variable pitch system from the target fusion data using preset specific methods and algorithms, such as the vibration frequency of the blades, temperature changes of the transmission components, pressure fluctuations of the lubrication system, etc., which can reflect the operating characteristics and health status of the variable pitch system.

[0029] In this embodiment, the machine learning algorithm constructs mathematical models and algorithms to allow the computer to automatically learn patterns and rules from large amounts of data, and use the learned knowledge to make predictions and decisions. For example, in fault diagnosis, it can determine whether a device has failed, as well as the type and severity of the failure.

[0030] In this embodiment, the classification data is to input the target extracted features into the machine learning algorithm. After processing and analysis by the algorithm, each feature is divided into different categories, such as "normal", "minor fault", "serious fault", etc.

[0031] In this embodiment, fault diagnosis is based on the classified data, combined with professional knowledge and experience, to determine whether there is a fault in the pitch system, and to determine the specific type, location and severity of the fault.

[0032] In this embodiment, maintenance prediction is based on the historical operating data, current status information and fault diagnosis results of the variable pitch system. Prediction models and algorithms are used to predict possible failures and maintenance requirements of the variable pitch system in advance, thereby avoiding downtime losses and production interruptions caused by sudden failures.

[0033] In this embodiment, comprehensive intelligent maintenance is to comprehensively analyze the fault diagnosis results and maintenance prediction results, formulate the optimal maintenance strategy, and use drone clusters to perform specific maintenance tasks, such as fault location, component replacement, lubrication and maintenance, etc., to achieve automation, intelligence and efficiency of the maintenance process, improve maintenance quality and efficiency, and reduce maintenance costs.

[0034] In this embodiment, for example, the high-frequency vibration sensor and current harmonic detection chip built into the pitch system can collect mechanical vibration and motor operation data in real time, transmit it to the edge gateway via the LPWAN network, and provide a high-precision data source for AI diagnosis. AI-driven fault prediction includes: the ST-CNN model integrates sensor data and drone inspection data, and uses spatiotemporal feature analysis to identify faults such as pitch bearing wear and gearbox jamming in advance, with a prediction accuracy of 97%. Drone cluster collaborative inspection refers to the drone nest dynamically dispatching drones based on the AI platform's fault warnings, equipped with infrared thermal imagers, microphone arrays, and LiDAR to perform near-field inspections of the pitch cabinet, generating a three-dimensional point cloud model to locate the fault point. In automated decision-making and repair, for software faults, the AI platform remotely adjusts the pitch controller parameters; for hardware faults, the drone's robotic arm performs emergency tightening or component replacement, realizing a "detection-analysis-repair" closed loop.

[0035] The beneficial effects of the above technical solution are: by pre-setting high-precision sensors and drone clusters to collaboratively collect data, extract key features, and then use machine learning algorithms to classify and diagnose faults and predict maintenance needs, and finally use drone clusters to carry out comprehensive intelligent maintenance, it can efficiently obtain comprehensive and accurate data, quickly and accurately diagnose faults, thereby ensuring the stable operation of the variable pitch system and improving overall operation and maintenance efficiency.

[0036] Example 2: Based on Example 1, the data acquisition module includes: An operation acquisition unit is used to obtain real-time operation data and real-time maintenance data of the target pitch system of the wind farm based on preset high-precision sensors; The inspection and collection unit is used to inspect the target variable pitch system based on the preset inspection path by the drone cluster to obtain real-time inspection data; The data integration unit is used to integrate the real-time operation data, real-time maintenance data and real-time inspection data to obtain the initial acquisition data of the target pitch control system of the wind farm.

[0037] In this embodiment, the preset high-precision sensor is a precision measuring device whose parameters and functions are set in advance according to specific needs. It can accurately capture various physical quantities such as temperature, pressure, vibration amplitude, etc., and provide a high-quality and reliable raw data basis for data analysis.

[0038] In this embodiment, the drone cluster is composed of several drones with collaborative working capabilities. Through advanced communication and coordination mechanisms, different drones in the drone cluster can jointly perform inspection, data collection and other tasks like a team, achieving efficient and comprehensive information acquisition.

[0039] In this embodiment, the initial collected data is measured by using preset high-precision sensors to measure each target object in the variable pitch system, and by using a drone cluster to collect information from different angles and heights. The resulting raw data set without any processing includes the original measurement values, images, videos, etc. of various physical quantities.

[0040] The beneficial effect of the above technical solution is that by integrating multi-source data, it can more comprehensively and accurately reflect the status of the target variable pitch system, thereby providing a reliable basis for data processing fusion and equipment diagnosis and maintenance, and improving operation and maintenance efficiency and quality.

[0041] Example 3: Based on Example 2, the inspection and collection unit includes: The inspection planning subunit is used to combine the AI platform of the intelligent management terminal with the wind farm layout information and real-time wind farm environmental data and the drone equipment information of the drone cluster to plan the inspection route and obtain the preset inspection path; Cluster inspection subunit, used to equip the drone cluster with preset sensors and collect inspection data of the target variable pitch system based on the preset inspection path; Among them, the inspection data includes real-time blade images, real-time temperature data, real-time equipment operation vibration data, etc. Multiple data transmission subunits are used to upload inspection data to the AI platform of the intelligent management terminal in real time to obtain real-time inspection data.

[0042] In this embodiment, the AI platform refers to a system platform equipped with artificial intelligence technology and equipped with the intelligent management terminal.

[0043] In this embodiment, the wind farm layout information is detailed information about the distribution location, arrangement, spatial relationship, etc. of various wind turbine generator sets, substations, transmission lines and other facilities within the wind farm.

[0044] In this embodiment, the real-time wind farm environmental data is data on environmental conditions collected in real time during the operation of the wind farm, such as meteorological information such as wind speed, wind direction, temperature, humidity, and air pressure.

[0045] In this embodiment, the drone equipment information is detailed information about the performance parameters, functional characteristics, endurance, load capacity, etc. of each drone in the drone cluster.

[0046] In this embodiment, the preset inspection path is a drone flight route planned by using the AI platform of the intelligent management terminal based on wind farm layout information, real-time wind farm environmental data, and drone equipment information.

[0047] In this embodiment, the inspection data refers to various information about the target pitch system collected by preset sensors during the inspection process, including real-time blade images, real-time temperature data, and real-time equipment operation vibration data.

[0048] In this embodiment, the real-time inspection data is the inspection data uploaded to the AI platform of the intelligent management terminal in real time.

[0049] The beneficial effects of the above technical solution are: by using AI intelligent planning to achieve efficient and accurate inspection routes, the drone cluster is equipped with preset sensors to comprehensively collect data and transmit it to the platform in real time. It can timely grasp the status of the variable pitch system, quickly respond to faults, improve inspection efficiency and accuracy, and ensure the stable operation of the wind farm.

[0050] Example 4: Based on Example 3, the inspection planning subunit includes: The first information acquisition block is used to obtain the electric field environment information of the target wind farm and the equipment distribution of the target pitch control system, and comprehensively obtain the wind farm layout information; The second information acquisition block is used to obtain real-time environmental data of the target wind farm and drone equipment information of the drone cluster; The initial path planning block is used to obtain the health status of each target device in the target pitch system and the real-time environmental data of the corresponding area, comprehensively determine the inspection priority of the current target device, and then determine the initial inspection planning path by combining the UAV device information and wind farm layout information; The optimization model determination block is used to determine the multi-objective optimization model of the drone cluster based on the comprehensive inspection objectives of the target wind farm; The inspection path optimization block is used to optimize the initial inspection planning path by combining the multi-objective optimization model with real-time environmental data to determine the preset inspection path.

[0051] In this embodiment, the electric field environment information is environmental data related to the geography, climate, surrounding obstacles, etc. of the area where the wind farm is located.

[0052] In this embodiment, the target pitch system equipment distribution is the specific spatial arrangement of pitch system equipment within the wind farm.

[0053] In this embodiment, the wind farm layout information is the overall planning of the wind farm and the equipment arrangement details obtained by integrating the electric field environment information and the equipment distribution.

[0054] In this embodiment, the real-time environmental data is the data of meteorological factors such as wind speed, wind direction, temperature, etc. of the wind farm collected at a specific time, as well as environmental factors affecting inspection.

[0055] In this embodiment, the drone equipment information is the technical parameters and performance indicators of each drone in the drone cluster, such as model, endurance, etc.

[0056] In this embodiment, the device health status is the operating status and performance index evaluation result of the target pitch system device.

[0057] In this embodiment, the real-time environmental data is real-time environmental condition data of the area where each target pitch system device is located.

[0058] In this embodiment, the inspection priority is the order of equipment inspection determined according to the health status of the equipment and the real-time environmental data of the corresponding area.

[0059] In this embodiment, the initial inspection planning path is a drone flight route preliminarily planned based on drone equipment information, wind farm layout information, and inspection priority.

[0060] In this embodiment, the comprehensive inspection goal is to achieve the overall goal of wind farm inspection, such as comprehensive coverage of equipment and ensuring safe and stable operation.

[0061] In this embodiment, the multi-objective optimization model is a mathematical model for finding an optimal inspection path solution based on comprehensive inspection objectives and taking multiple factors into consideration.

[0062] In this embodiment, path optimization is a process of adjusting and improving the initial inspection planning path by combining a multi-objective optimization model with real-time environmental data.

[0063] In this embodiment, the preset inspection path is the final drone flight route determined after path optimization.

[0064] The beneficial effects of the above technical solution are: by using AI intelligent planning to achieve efficient and accurate inspection routes, the drone cluster is equipped with preset sensors to comprehensively collect data and transmit it to the platform in real time. It can timely grasp the status of the variable pitch system, quickly respond to faults, and improve inspection efficiency and accuracy.

[0065] Example 5: Based on Example 2, the processing fusion module includes: A data processing unit is used to perform data cleaning and data standardization on the initial collected data to obtain initial processed data; an algorithm setting unit, configured to set a data fusion algorithm based on data characteristics of a target pitch system to obtain a preset data fusion algorithm; The data fusion unit is used to perform data fusion on the initial processing data based on a preset data fusion algorithm to obtain target fusion data of the target pitch control system.

[0066] In this embodiment, data cleaning is to process the initially collected data to remove noise, correct erroneous data, fill in missing values, and other operations.

[0067] In this embodiment, data normalization involves converting cleaned data into a uniform format and range to make them comparable. Common normalization methods include scaling the data to the interval [0, 1] or setting the mean to 0 and the variance to 1. For example, measurement data from different sensors may have different dimensions and value ranges.

[0068] In this embodiment, the initially processed data is data that has been processed by data cleaning and data standardization.

[0069] In this embodiment, data characteristics refer to the characteristics and regularities of the data generated and collected by the target pitch system. For example, the angle data of the pitch system usually has the characteristics of periodic changes and has a certain correlation with data such as wind speed and rotation speed.

[0070] In this embodiment, the data fusion algorithm is set by selecting or designing an appropriate data fusion algorithm based on the data characteristics of the target pitch system. The data fusion algorithm is used to comprehensively process information from different sensors or data sources to improve data accuracy and reliability. For example, the angle data and speed data of the pitch system can be fused using algorithms such as weighted averaging and Kalman filtering.

[0071] In this embodiment, data fusion is a process of comprehensively processing the initially processed data using a preset data fusion algorithm, for example, fusing the data of multiple angle sensors of the pitch system to obtain a more accurate angle measurement value.

[0072] In this embodiment, the target fusion data is data of the target pitch system obtained after data fusion, which integrates information from multiple data sources.

[0073] The beneficial effect of the above technical solution is that by utilizing the fusion algorithm set by data characteristics, multi-source information can be integrated to obtain more accurate and reliable target fusion data, providing a solid foundation for the diagnosis and maintenance of the target pitch system.

[0074] Example 6: Based on Example 5, the identification module is extracted, including: a feature extraction unit, configured to determine a comprehensive feature extraction scheme based on data features of the pitch system operation, and perform feature extraction on the target fusion data based on the comprehensive feature extraction scheme to obtain target extraction features; The feature classification unit is used to classify the target extracted features based on a preset machine learning algorithm to obtain classified data.

[0075] In this embodiment, the data features are characteristics exhibited by the data generated during the operation of the pitch control system, such as the fluctuation range, change trend, periodicity, and correlation of the data.

[0076] In this embodiment, the comprehensive feature extraction scheme is a method and strategy for extracting effective features from target fusion data, which is formulated based on the data characteristics of the variable pitch system operation, including selecting appropriate feature extraction algorithms and parameter settings.

[0077] In this embodiment, the extracted features are based on a comprehensive feature extraction scheme, and are extracted from the target fusion data to obtain key features that can reflect the operating status and performance of the variable pitch system, such as the change pattern of specific parameters, the distribution of characteristic values, etc.

[0078] In this embodiment, the classified data is the result data obtained after classifying the target extracted features using a preset machine learning algorithm.

[0079] The beneficial effect of the above technical solution is: by accurately extracting key features based on the data characteristics of the variable pitch system, and then using machine learning algorithm classification, the operating status of the variable pitch system can be identified more quickly and accurately, thereby providing timely and accurate support for fault diagnosis and performance evaluation.

[0080] Example 7: Based on Example 6, the diagnosis and maintenance module includes: A matching analysis unit, configured to perform rule matching on the classified data based on a preset expert database and identify the fault type of the classified data; A correlation analysis unit, used to perform correlation analysis on the classified data based on a preset neural network to determine the fault mode; A first prediction unit is configured to predict the service life of the target pitch system based on historical fault data in combination with a time series analysis method to obtain a first prediction result; a risk assessment unit, configured to perform a risk assessment on a failure risk level of a target pitch system based on a failure type, a failure mode, and a first prediction result of each device; The integrated maintenance unit is used to determine the maintenance strategy based on the risk assessment results and the drone cluster, and perform integrated intelligent maintenance on the target variable pitch system.

[0081] In this embodiment, the preset expert database is a database that is pre-built and stores a large amount of professional knowledge and experience, covering information such as various fault types, characteristics, causes, and corresponding solutions that may occur in the pitch system.

[0082] In this embodiment, rule matching is a process of comparing and contrasting the classified data with the rules in the preset expert database.

[0083] In this embodiment, the preset neural network is a pre-designed and trained neural network model that can learn and capture complex relationships and patterns between data.

[0084] In this embodiment, correlation analysis is the process of processing and analyzing categorized data using a preset neural network to identify the inherent connections and relationships between the data. In pitch system fault analysis, correlation analysis can help identify the correlations between different features and fault modes, thereby determining the potential mode that led to the fault.

[0085] In this embodiment, the historical fault data is data related to various faults that occurred in the target pitch system during past operation, including information such as the time, location, type, mode, and duration of the fault.

[0086] In this embodiment, the time series analysis method is a method for analyzing a data series arranged in chronological order.

[0087] In this embodiment, the first prediction result is obtained by predicting the service life of the target pitch system based on historical fault data combined with a time series analysis method.

[0088] In this embodiment, the fault type is a specific category of fault occurring in the pitch system, such as a mechanical fault, an electrical fault, and the like.

[0089] In this embodiment, the failure mode is the specific characteristics and regularities exhibited when a pitch system failure occurs, reflecting the nature and occurrence mechanism of the failure.

[0090] In this embodiment, the fault risk level is a classification obtained by comprehensively considering factors such as the fault type, fault mode, and first prediction result of each device, and evaluating the possibility and severity of failure of the target pitch system.

[0091] In this embodiment, the maintenance strategy is a specific plan and measure for maintaining and servicing the target variable pitch system, formulated based on the risk assessment results and combined with the characteristics and capabilities of the drone cluster.

[0092] In this embodiment, comprehensive intelligent maintenance is an intelligent maintenance process that performs fault identification, pattern analysis, risk assessment, and maintenance decision-making on the target pitch control system.

[0093] The beneficial effects of the above technical solution are: by more accurately identifying the fault types and modes of the variable pitch system, accurately predicting the service life, and thus evaluating the risk level, and formulating maintenance strategies accordingly, efficient and intelligent maintenance can be achieved with the help of drone clusters, which can improve system reliability and maintenance efficiency.

[0094] Example 8: Based on Example 7, the comprehensive maintenance unit includes: A position synthesis subunit is used to obtain device position information of each target device of the target pitch system whose risk assessment result exceeds the preset standard assessment result, and obtain the comprehensive device position information of the target pitch system; The cluster response subunit is used to plan the inspection path based on the comprehensive equipment location information of the target variable pitch system, the UAV equipment information in the UAV cluster, and the real-time environmental data corresponding to each target equipment in the comprehensive equipment location information, to obtain the initial inspection planning path; The path judgment subunit is used to judge whether there is any overlap between the initial inspection plan path and the preset inspection path; A path extraction subunit is used to extract a portion of the initial inspection plan path that does not overlap with the preset inspection path to obtain a first inspection plan path; A path optimization subunit is used to obtain a second inspection planning path for the UAV cluster based on the starting point of the UAV cluster and the shortest path planning principle in combination with the first inspection planning path; A second judgment subunit is used to judge the path feasibility of the second inspection plan path; If there is an infeasible path, extract the infeasible path segment and replace it with the preset inspection path, so as to obtain the third inspection planning path of the drone cluster by combining the remaining inspection planning paths; Otherwise, the second inspection planning path is used as the third inspection planning path of the UAV cluster.

[0095] A depth inspection subunit is used to perform a depth inspection on the target variable pitch system using a UAV cluster combined with a third inspection planning path to obtain a depth inspection result; An inspection determination subunit, configured to determine a first impact weight of the in-depth inspection result and a second impact weight of the risk assessment result based on the historical impact levels of the in-depth inspection result and the risk assessment result; A fault synthesis subunit is used to combine the first impact weight with the in-depth inspection result, and the second impact weight with the risk assessment result to obtain a comprehensive fault condition of the target pitch system; A maintenance decision subunit is used to compare the comprehensive fault conditions of the target pitch system with the preset fault-maintenance database one by one, thereby determining the system maintenance strategy for the target pitch system; The decision-making execution subunit is used to realize automatic adjustment and control of each target device of the target pitch system based on the system maintenance strategy, thereby realizing comprehensive intelligent maintenance.

[0096] In this embodiment, the preset standard assessment result is a pre-set reference value used to measure whether the risk assessment result of the target pitch system is within a normal or acceptable range.

[0097] In this embodiment, the comprehensive device location information is information obtained by integrating the device location information of each target device in the target pitch system whose risk assessment results exceed the preset standard, and includes all device locations that require special attention.

[0098] In this embodiment, inspection path planning is the process of planning a flight path for the drone cluster that can complete the inspection task based on the comprehensive equipment location information of the target variable pitch system, drone equipment information and real-time environmental data.

[0099] In this embodiment, the initial inspection planned path is a preliminary planned path obtained after inspection path planning.

[0100] In this embodiment, the first inspection plan path is a path obtained by extracting a partial path that does not overlap with the preset inspection path from the initial inspection plan path, and is used for subsequent shortest path planning.

[0101] In this embodiment, the shortest path planning principle is a principle followed in path planning to make the UAV flight path distance as short as possible, so as to improve inspection efficiency and reduce energy consumption.

[0102] In this embodiment, the second inspection planned path is a planned path obtained based on the starting point of the drone cluster and the shortest path planning principle, combined with the first inspection planned path.

[0103] In this embodiment, path feasibility is to determine whether the drone cluster can complete the inspection task safely and smoothly when flying according to the second inspection plan path. It is a path validity judgment after considering the actual conditions such as the drone's flight capability, environmental factors, obstacles, etc.

[0104] In this embodiment, the infeasible path segment is a partial path that is judged to be infeasible in the second inspection planning path, and is a path segment that cannot be actually flown due to reasons such as drone flight restrictions and environmental obstacles.

[0105] In this embodiment, the third inspection planning path is the final planning path obtained by extracting the infeasible path segment and replacing the path with the preset inspection path when there is an infeasible path in the second inspection planning path, and then combining it with the remaining inspection planning paths; if the second inspection planning path is feasible, it is directly used as the third inspection planning path.

[0106] In this embodiment, the in-depth inspection result is obtained after using the drone cluster in combination with the third inspection planning path to conduct an in-depth inspection of the target variable pitch system, such as the specific fault location and fault degree of the equipment.

[0107] In this embodiment, the first impact weight is based on the historical impact of the deep inspection result, and is the weight assigned to the deep inspection result in the comprehensive evaluation.

[0108] In this embodiment, the second impact weight is based on the historical impact degree and is the weight assigned to the risk assessment result in the comprehensive assessment.

[0109] In this embodiment, the comprehensive fault condition is the overall fault state of the target pitch system obtained by combining the first impact weight with the in-depth inspection result and the second impact weight with the risk assessment result, which includes information such as the severity of the fault and the scope of impact.

[0110] In this embodiment, the preset fault-maintenance database is a pre-built database that stores the relationship between various fault conditions and corresponding maintenance strategies.

[0111] The beneficial effect of the above technical solution is that by formulating maintenance strategies more accurately, efficient and intelligent maintenance can be achieved with the help of drone clusters, which can improve system reliability and maintenance efficiency.

[0112] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An unmanned intelligent operation and maintenance device for wind farm pitch control systems based on AI and drone swarm collaboration, characterized by: include: The data acquisition module is used to collect data based on preset high-precision sensors and conduct collaborative inspections based on drone clusters to comprehensively obtain initial collected data; The processing and fusion module is used to clean and standardize the initial collected data, and perform data fusion based on the processed initial collected data to obtain the target fusion data; An extraction and recognition module is used to extract features from the target fusion data based on the data features of the variable pitch system operation to obtain target extraction features, and to classify the target extraction features based on a machine learning algorithm to obtain classified data; The diagnostic maintenance module is used to diagnose faults on classified data and predict maintenance for the target variable pitch system, thereby performing comprehensive intelligent maintenance in combination with the UAV cluster based on the fault diagnosis results and predicted maintenance results.

2. The unmanned intelligent operation and maintenance device for wind farm pitch control system based on AI and drone cluster collaboration according to claim 1 is characterized in that: Data acquisition module, including: An operation acquisition unit is used to obtain real-time operation data and real-time maintenance data of the target pitch system of the wind farm based on preset high-precision sensors; The inspection and collection unit is used to inspect the target variable pitch system based on the preset inspection path by the drone cluster to obtain real-time inspection data; The data integration unit is used to integrate the real-time operation data, real-time maintenance data and real-time inspection data to obtain the initial acquisition data of the target pitch control system of the wind farm.

3. The unmanned intelligent operation and maintenance device for wind farm pitch control system based on AI and drone cluster collaboration according to claim 2 is characterized in that: Inspection and collection unit, including: The inspection planning subunit is used to combine the AI platform of the intelligent management terminal with the wind farm layout information and real-time wind farm environmental data and the drone equipment information of the drone cluster to plan the inspection route and obtain the preset inspection path; Cluster inspection subunit, used to equip the drone cluster with preset sensors and collect inspection data of the target variable pitch system based on the preset inspection path; Among them, the inspection data includes real-time blade images, real-time temperature data, real-time equipment operation vibration data, etc. Multiple data transmission subunits are used to upload inspection data to the AI platform of the intelligent management terminal in real time to obtain real-time inspection data.

4. The unmanned intelligent operation and maintenance device for wind farm pitch control systems coordinated by local AI and drone clusters according to claim 3 is characterized in that: Inspection planning subunit, including: The first information acquisition block is used to obtain the electric field environment information of the target wind farm and the equipment distribution of the target pitch control system, and comprehensively obtain the wind farm layout information; The second information acquisition block is used to obtain real-time environmental data of the target wind farm and drone equipment information of the drone cluster; The initial path planning block is used to obtain the health status of each target device in the target pitch system and the real-time environmental data of the corresponding area, comprehensively determine the inspection priority of the current target device, and then determine the initial inspection planning path by combining the UAV device information and wind farm layout information; The optimization model determination block is used to determine the multi-objective optimization model of the drone cluster based on the comprehensive inspection objectives of the target wind farm; The inspection path optimization block is used to optimize the initial inspection planning path by combining the multi-objective optimization model with real-time environmental data to determine the preset inspection path.

5. The unmanned intelligent operation and maintenance device for wind farm pitch control system based on AI and drone cluster collaboration according to claim 2 is characterized in that: Processing fusion module, including: A data processing unit is used to perform data cleaning and data standardization on the initial collected data to obtain initial processed data; an algorithm setting unit, configured to set a data fusion algorithm based on data characteristics of a target pitch system to obtain a preset data fusion algorithm; The data fusion unit is used to perform data fusion on the initial processing data based on a preset data fusion algorithm to obtain target fusion data of the target pitch control system.

6. The unmanned intelligent operation and maintenance device for wind farm pitch control system based on AI and drone cluster collaboration according to claim 4 is characterized in that: Extraction and recognition modules, including: a feature extraction unit, configured to determine a comprehensive feature extraction scheme based on data features of the pitch system operation, and perform feature extraction on the target fusion data based on the comprehensive feature extraction scheme to obtain target extraction features; The feature classification unit is used to classify the target extracted features based on a preset machine learning algorithm to obtain classified data.

7. The unmanned intelligent operation and maintenance device for wind farm pitch control system based on AI and drone cluster collaboration according to claim 5 is characterized in that: Diagnostic maintenance module, including: A matching analysis unit, configured to perform rule matching on the classified data based on a preset expert database and identify the fault type of the classified data; A correlation analysis unit, used to perform correlation analysis on the classified data based on a preset neural network to determine the fault mode; A first prediction unit is configured to predict the service life of the target pitch system based on historical fault data in combination with a time series analysis method to obtain a first prediction result; a risk assessment unit, configured to perform a risk assessment on a failure risk level of a target pitch system based on a failure type, a failure mode, and a first prediction result of each device; The integrated maintenance unit is used to determine the maintenance strategy based on the risk assessment results and the drone cluster, and perform integrated intelligent maintenance on the target variable pitch system.

8. The unmanned intelligent operation and maintenance device for wind farm pitch control system based on AI and drone cluster collaboration according to claim 7 is characterized in that: Comprehensive maintenance unit, including: A position synthesis subunit is used to obtain device position information of each target device of the target pitch system whose risk assessment result exceeds the preset standard assessment result, and obtain the comprehensive device position information of the target pitch system; The cluster response subunit is used to plan the inspection path based on the comprehensive equipment location information of the target variable pitch system, the UAV equipment information in the UAV cluster, and the real-time environmental data corresponding to each target equipment in the comprehensive equipment location information, to obtain the initial inspection planning path; The path judgment subunit is used to judge whether there is any overlap between the initial inspection plan path and the preset inspection path; A path extraction subunit is used to extract a portion of the initial inspection plan path that does not overlap with the preset inspection path to obtain a first inspection plan path; A path optimization subunit is used to obtain a second inspection planning path for the UAV cluster based on the starting point of the UAV cluster and the shortest path planning principle in combination with the first inspection planning path; A second judgment subunit is used to judge the path feasibility of the second inspection plan path; If there is an infeasible path, extract the infeasible path segment and replace it with the preset inspection path, so as to obtain the third inspection planning path of the drone cluster by combining the remaining inspection planning paths; Otherwise, the second inspection planning path is used as the third inspection planning path of the UAV cluster; A depth inspection subunit is used to perform a depth inspection on the target variable pitch system using a UAV cluster combined with a third inspection planning path to obtain a depth inspection result; An inspection determination subunit, configured to determine a first impact weight of the in-depth inspection result and a second impact weight of the risk assessment result based on the historical impact levels of the in-depth inspection result and the risk assessment result; A fault synthesis subunit is used to combine the first impact weight with the in-depth inspection result, and the second impact weight with the risk assessment result to obtain a comprehensive fault condition of the target pitch system; A maintenance decision subunit is used to compare the comprehensive fault conditions of the target pitch system with the preset fault-maintenance database one by one, thereby determining the system maintenance strategy for the target pitch system; The decision-making execution subunit is used to realize automatic adjustment and control of each target device of the target pitch system based on the system maintenance strategy, thereby realizing comprehensive intelligent maintenance.

Images