Wind turbine cooperative inspection method and system in dynamic operation area

Abstract

This invention relates to the field of wind turbine operation and maintenance technology, and particularly to a collaborative inspection method and system for wind turbines in a dynamic operating domain. The method includes: acquiring real-time operating parameters of the wind turbine; determining the parts of the wind turbine to be inspected based on the correlation analysis between the real-time operating parameters and a damage evolution model; the damage evolution model employing a combination of a physical model and a neural network model; comparing the parts to be inspected with a pre-planned inspection range to obtain areas requiring supplementary inspection; allocating inspection tasks to the areas requiring supplementary inspection and the pre-planned inspection range based on the energy consumption constraints of the wall-climbing robot and the drone, generating collaborative operation instructions; and scheduling the wall-climbing robot and the drone to perform inspections according to the collaborative operation instructions. This invention achieves precise matching between wind turbine damage risks and inspection tasks, balancing inspection accuracy, coverage integrity, and equipment energy consumption under non-stop operating conditions, thereby improving the efficiency and reliability of collaborative inspections.

Description

Technical Field

[0001] This invention relates to the technical field of wind turbine operation and maintenance, and in particular to a collaborative inspection method and system for wind turbines in a dynamic operating domain. Background Technology

[0002] Wind turbines operate in complex environments with strong winds and significant temperature fluctuations for extended periods. Core load-bearing structures such as towers and blades are prone to surface cracks, coating peeling, and corrosion damage. Timely and accurate inspections are crucial for ensuring the safe and stable operation of wind turbines and reducing maintenance costs. Currently, wind turbine inspections commonly employ a collaborative approach using wall-climbing robots and rotorcraft drones. Wall-climbing robots enable close-range, detailed inspection of the tower surface, while rotorcraft drones provide extensive coverage inspection of high-altitude areas such as blades and nacelles. This collaborative approach has become the mainstream technology for non-stop, full-structure inspections of wind turbines.

[0003] However, existing collaborative inspection methods for wind turbines cannot accurately match the inspection scope and multi-device inspection tasks based on the actual damage risk of the wind turbines. This makes it difficult for inspection operations to simultaneously meet the comprehensive requirements of defect detection accuracy, detection coverage integrity and energy consumption control, and cannot adapt to the high-efficiency inspection needs of wind turbines under non-stop operation. Summary of the Invention

[0004] This invention provides a method and system for collaborative inspection of wind turbine units in a dynamic operating domain, which can effectively solve the problems in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The collaborative inspection method for wind turbine units in dynamic operating domains includes: Obtain real-time operating parameters of wind turbine units; Based on the correlation analysis between real-time operating parameters and the damage evolution model, the parts of the wind turbine to be reviewed are determined; the damage evolution model adopts a combination of physical model and neural network model. The areas to be reviewed are compared with the pre-planned testing range to determine the areas requiring supplementary testing. Based on the energy consumption constraints of the wall-climbing robot and the drone, inspection tasks are assigned to the areas requiring supplementary inspection and the pre-planned inspection range, and collaborative operation instructions are generated. Based on the collaborative operation instructions, the wall-climbing robot and the drone are scheduled to perform the inspection.

[0006] Furthermore, the real-time operating parameters include: The data is based on at least one of the following: strain distribution data acquired by a fiber optic sensor, vibration data acquired by an accelerometer, wind speed data acquired by an anemometer, and temperature data acquired by an infrared thermal imager.

[0007] Furthermore, the methods for determining the areas to be reviewed include: Input the real-time operating parameters into the physical model to obtain the theoretical damage hotspot area; By inputting real-time operating parameters into the neural network model, the damage probability at each spatial location is obtained; The theoretical damage hotspot area is fused with the damage probability, and the spatial location in the fusion result that exceeds the preset damage threshold is taken as the part to be reviewed.

[0008] Furthermore, the fusion of theoretical damage hotspots and damage probabilities adopts a weighted fusion method. The weights of the physical model are determined based on its theoretical fit, while the weights of the neural network model are determined based on the fitting accuracy of its training samples. After weighted fusion, a comprehensive damage score for each spatial location is obtained. Spatial locations with comprehensive damage scores exceeding a preset damage threshold are the areas to be reviewed.

[0009] Furthermore, the pre-planned inspection range is determined by dividing the surface of the wind turbine into multiple inspection blocks and setting an inspection period for each inspection block.

[0010] Furthermore, a method for dividing the surface of a wind turbine into multiple inspection areas includes: The tower is divided into several annular blocks along its height and several sections along the blade span.

[0011] Furthermore, the methods for assigning inspection tasks include: With the optimization objectives of minimizing total energy consumption and maximizing the coverage ratio of areas requiring supplementary inspection, the walking path of the wall-climbing robot and the flight path of the drone are determined through an optimization algorithm under energy consumption constraints; the optimization algorithm is either ant colony algorithm or genetic algorithm.

[0012] Furthermore, energy consumption constraints include the remaining power of the wall-climbing robot, the remaining power of the drone, and the maximum energy consumption quota for a single inspection task.

[0013] Furthermore, during the testing process, wind speed and remaining battery power of the equipment are monitored in real time; When the wind speed exceeds the safety threshold for drones, drone operations will be suspended, and some areas requiring additional inspection will be handled by wall-climbing robots. When the drone's battery level drops below the preset lower limit, the drone operation will be suspended, and some areas requiring additional inspection will be handled by the wall-climbing robot. When the wall-climbing robot's battery level drops below the preset lower limit, the drone will prioritize covering the remaining areas that need to be inspected.

[0014] On the other hand, the present invention also provides a wind turbine collaborative inspection system for dynamic operating domains, including: The parameter acquisition module is used to acquire the real-time operating parameters of the wind turbine. The damage analysis module is used to determine the parts of the wind turbine to be reviewed based on the correlation analysis between real-time operating parameters and the damage evolution model; the damage evolution model adopts a combination of physical model and neural network model. The area comparison module is used to compare the area to be reviewed with the pre-planned detection range to obtain the area that needs to be re-inspected; The instruction generation module is used to allocate inspection tasks to the areas requiring supplementary inspection and the pre-planned inspection range based on the energy consumption constraints of the wall-climbing robot and the drone, and generate collaborative operation instructions. The inspection and scheduling module is used to schedule wall-climbing robots and drones to perform inspections based on collaborative operation instructions.

[0015] The technical solution of this invention can achieve the following technical effects: By combining real-time operating parameters of wind turbines with physical and neural network models to form a damage evolution model, the system can accurately locate the parts to be inspected. By comparing the model with the pre-planned inspection range, the system can identify the areas requiring supplementary inspection, achieving a precise match between the inspection range and the actual damage risk, ensuring that inspection operations focus on high-risk areas. Based on the energy consumption constraints of wall-climbing robots and drones, the system uses ant colony algorithms or genetic algorithms to allocate inspection tasks and plan equipment operation paths and routes. This ensures comprehensive coverage of the areas requiring supplementary inspection and the pre-set inspection range while controlling total energy consumption, balancing defect detection accuracy with energy consumption control requirements. During the inspection process, the system monitors wind speed and remaining power in real time, dynamically adjusting the operating equipment according to wind speed thresholds and power limits to ensure continuous inspection operations without shutdown. This adapts to the efficient inspection needs of wind turbines in complex field environments, ensuring the safe and stable operation of the units and controlling maintenance costs.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the wind turbine collaborative inspection method for the dynamic operating domain of the present invention. Figure 2This is a schematic diagram of the wind turbine collaborative inspection system in the dynamic operation domain of the present invention. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] like Figure 1 As shown, the wind turbine collaborative inspection method for dynamic operating domains of the present invention specifically includes the following steps: Step S1: Obtain the real-time operating parameters of the wind turbine; Step S2: Based on the correlation analysis between real-time operating parameters and the damage evolution model, determine the parts of the wind turbine to be reviewed; the damage evolution model adopts a combination of physical model and neural network model. Step S3: Compare the area to be reviewed with the pre-planned testing range to obtain the area that needs to be re-inspected; Step S4: Based on the energy consumption constraints of the wall-climbing robot and the drone, assign inspection tasks to the areas requiring supplementary inspection and the pre-planned inspection range, and generate collaborative operation instructions; Step S5: Based on the collaborative operation instructions, schedule the wall-climbing robot and the drone to perform the inspection.

[0022] In this embodiment, by correlating the real-time operating parameters of the wind turbine with the damage evolution model and combining the energy consumption constraints of the wall-climbing robot and the drone for dynamic task allocation, a closed-loop inspection system from accurate risk perception to optimal resource allocation is constructed. Through the damage evolution model, high-risk areas to be reviewed can be identified and areas requiring supplementary inspection can be dynamically generated, reducing invalid inspection paths. At the same time, by utilizing the matching mechanism of equipment energy consumption constraints, priority coverage of key defect areas is ensured under limited energy conditions. Ultimately, a high-precision, full-coverage, and low-energy-consumption efficient inspection effect is achieved without shutting down the system, thereby improving the intelligence level, economy, and safety of wind turbine operation and maintenance.

[0023] In a specific implementation, as one example, given that the structural damage risk of wind turbines needs to be assessed through operating parameters, it is necessary to select operating parameters that are directly related to the structural damage of the wind turbines and can reflect potential damage risks. Through targeted sensor deployment, data acquisition, and preprocessing, real-time and reliable acquisition of these parameters can be achieved, supporting damage risk analysis and collaborative inspection task planning. This embodiment achieves the acquisition of real-time operating parameters of the wind turbines through sensor selection and deployment, parameter acquisition, and preprocessing, as detailed below: Step S11: Select sensors that meet the requirements of complex field environments, capable of withstanding strong winds, temperature fluctuations, and dust interference to ensure the stability and accuracy of data acquisition. The real-time operating parameters acquired by the selected sensors include at least one of strain distribution data, vibration data, wind speed data, and temperature data. Considering the differences in wind turbine inspection scenarios and the varying emphasis on parameters for different inspection needs, some scenarios do not require the collection of all parameters. Selecting one or more parameters corresponding to the current inspection focus is sufficient to meet the needs of damage risk assessment and inspection task planning. The sensor selection and arrangement methods for each parameter are as follows: a. Fiber Optic Sensors: Distributed fiber optic sensors are selected, which have a wide measurement range, high sensitivity, and strong anti-electromagnetic interference capability. They can be adapted to the long-distance, large-scale strain detection requirements of wind turbine towers and blades to acquire strain distribution data. The distributed fiber optic sensors are embedded in the inner wall of the tower along the height direction and embedded in the inside of the blade along the spanwise direction. During the arrangement, stress concentration areas such as tower welds and blade leading edges are avoided to prevent the sensors from affecting the structural strength of the unit. At the same time, it is ensured that the sensors are in close contact with the structural surface to ensure the accuracy of strain data transmission. b. Accelerometer: A piezoelectric accelerometer is selected, which has a fast response speed and high measurement accuracy. It can accurately capture vibration signals during the operation of the unit and obtain vibration data. The accelerometer is fixed at the bottom of the tower, the nacelle base and the blade root. These parts are the areas where the wind turbine vibrates most significantly during operation and can most directly reflect the vibration state of the unit structure. c. Anemometer: An ultrasonic anemometer is selected because it is not affected by mechanical rotating parts, has a wide measurement range, and can be adapted to wind speed measurement in strong wind environments in the field to obtain wind speed data. The ultrasonic anemometer is installed on the windward side of the top of the nacelle, higher than the outline of the nacelle, to avoid the nacelle from obstructing the wind speed measurement and to ensure that the wind speed data obtained can accurately reflect the wind environment in which the unit is located. d. Infrared thermal imager: A non-contact infrared thermal imager is selected, which can achieve long-distance, non-contact temperature measurement. It can acquire temperature distribution data without contacting the unit structure, avoiding interference with the unit's operation. The infrared thermal imager is installed on the bottom of the UAV fuselage and matched with the UAV's inspection path to ensure that the tower and blade surfaces can be fully covered during the UAV's flight. At the same time, the shooting angle and focal length of the infrared thermal imager are adjusted to ensure the clarity and accuracy of the temperature data acquisition. Step S12: Start the selected sensor to collect the corresponding real-time operating parameters, and at the same time perform targeted preprocessing on the collected raw data to remove interference signals and ensure data accuracy: Fiber optic sensors collect data via optical signal transmission. During the acquisition process, a fixed acquisition frequency is set to ensure that the dynamic changes in strain during unit operation can be captured. After the acquired optical signal is converted into an electrical signal, noise reduction processing is performed to remove noise signals caused by electromagnetic interference and dust interference in the field environment. Then, the strain distribution data is obtained through signal conversion to clarify the strain magnitude and distribution pattern at different spatial locations of the unit. The strain distribution data can directly reflect the stress state of the structure. The strain value in areas of abnormal stress in the structure will show significant deviation, which can serve as a direct basis for judging potential damage. Accelerometers collect data by sensing changes in acceleration caused by vibration. The acquisition frequency is consistent with that of fiber optic sensors to ensure data time synchronization. After converting the vibration signal into an electrical signal, the signal is filtered to remove interference signals caused by the sensor's own vibration and environmental vibration, thus obtaining key parameters such as vibration amplitude and frequency at different parts of the unit. Vibration data can reflect the operational stability of the structure. When the structure suffers damage such as cracks or loosening, the vibration amplitude and frequency will fluctuate abnormally, which can indirectly reflect the risk of damage. Ultrasonic anemometers transmit and receive ultrasonic signals, calculate and collect data based on the time difference of ultrasonic wave propagation, and set an appropriate acquisition frequency to ensure timely capture of dynamic changes in wind speed while avoiding data redundancy. Only outliers are removed from the collected wind speed data. Wind speed data directly affects the unit's operating load and structural stress. In high wind speed environments, the structural stress of the unit increases, raising the risk of damage. At the same time, wind speed data also provides a basis for the safety management of UAV inspection operations. Infrared thermal imagers capture infrared radiation signals from the surface of the unit structure and convert them into temperature distribution data. The acquisition frequency is matched with the flight speed of the drone to ensure that each area to be inspected can acquire the corresponding temperature data. The acquired temperature data is calibrated to eliminate the influence of ambient temperature and lighting conditions on the measurement results, so as to obtain the true temperature distribution of the unit surface. When the structure is damaged by corrosion, coating peeling or other damage, its surface heat dissipation characteristics will change, and abnormal hot spots or cold spots will appear in the temperature distribution, which can be used as a basis for judging potential damage.

[0024] In this embodiment, the occurrence of structural damage to the wind turbine is related to stress state, vibration state, wind environmental load, and temperature changes. At least one parameter from strain distribution data, vibration data, wind speed data, and temperature data is selected for acquisition. This allows for targeted reflection of the turbine's operating status and potential damage, avoiding potential omissions or misjudgments that might occur with single-parameter acquisition. Simultaneously, it considers actual inspection needs, eliminating the need to collect all parameters. Based on the core requirement of structural damage evolution, this embodiment does not employ a single sensor or non-targeted sensor layout. The sensor selection and placement for each parameter are combined with the parameter's purpose and the turbine's structural characteristics. Simultaneous acquisition and targeted preprocessing ensure the real-time nature, accuracy, and effectiveness of the parameters, achieving deep adaptation between parameter acquisition and subsequent damage risk analysis and collaborative inspection task planning. Sensor selection considers the complex field environment, enabling it to withstand strong winds, temperature fluctuations, and dust interference. Reasonable placement and targeted data preprocessing remove various interference signals, further ensuring parameter reliability for damage risk analysis and determination of areas requiring verification.

[0025] In some embodiments of the present invention, existing methods for determining the location to be reviewed often employ a single model, such as a physical model or a neural network model alone. A single physical model, derived from structural mechanics theory, cannot adapt to the random damage evolution characteristics of wind turbines in complex field environments, easily leading to a disconnect between theory and reality and resulting in deviations in the location to be reviewed. A single neural network model, trained on a large amount of historical data, lacks physical mechanism support, is prone to misjudgment when data is insufficient, and cannot explain the inherent logic of damage generation, making it difficult to meet the accuracy requirements of inspection. Consequently, collaborative inspection cannot accurately match the inspection range and multi-equipment inspection tasks based on actual damage risk, making it difficult to simultaneously meet the comprehensive requirements of defect detection accuracy, detection coverage integrity, and energy consumption control. This embodiment employs a damage evolution model combining a physical model and a neural network model, combined with real-time operating parameters, to achieve accurate determination of the location to be reviewed through precise correlation between parameters and the combined model. Specifically, the following operations are performed: Step S21: Construct a damage evolution combined model. This model is a combination of a physical model and a neural network model, which are trained independently and work together. The physical model is a finite element model based on the structural mechanics characteristics of wind turbines. Its construction is based on the material properties, geometric parameters, and stress characteristics of the wind turbine tower, blades, and other load-bearing structures. During model construction, the influence of environmental factors such as strong winds and temperature fluctuations on the structural stress is considered, and environmental loads are transformed into model input boundary conditions to ensure that the model accurately reflects the intrinsic relationship between structural stress and damage evolution. It can derive the theoretical correlation between real-time operating parameters and structural damage through structural mechanics theory, providing a physical basis for damage judgment. The neural network model is a convolutional neural network model, including an input layer, convolutional layers, pooling layers, and full-processing layers. The model consists of a connection layer and an output layer. The input layer corresponds to real-time operating parameters, while the output layer corresponds to the damage probability of each spatial location of the wind turbine. During model training, historical operating parameters of the wind turbine and corresponding structural damage detection data are selected as training samples. The samples cover the parameter-damage correspondence under different damage types and environmental conditions. During training, parameters such as the number of network layers and the size of convolutional kernels are adjusted to optimize the model fitting accuracy, ensuring that the model can accurately output the damage probability of each spatial location based on actual operating parameters. This approach can compensate for the limitations of physical models and adapt to the randomness and complexity of damage evolution. The physical model provides theoretical support, while the neural network model provides the ability to adapt to actual data. The output results of both are calculated independently, and then integrated through a fusion algorithm to avoid the defects of a single model. Step S22: Input the real-time running parameters into the physical model. Based on structural mechanics theory, the physical model analyzes and calculates the parameters: For the input strain distribution data, it calculates the stress concentration coefficient at each spatial location of the structure. Areas with stress concentration coefficients higher than a preset threshold are identified as theoretical damage hotspots. In such areas, the structure experiences concentrated stress and is prone to damage such as cracks. For the input vibration data, it calculates the vibration response amplitude at each part of the structure. Areas with abnormal vibration response amplitudes are included in the theoretical damage hotspots. In such areas, the structure has insufficient stability and is prone to loosening and crack propagation. For the input wind speed data, it calculates the stress distribution at each part of the structure under wind load. Areas with excessive stress are included in the theoretical damage hotspots. For the input temperature data, it calculates the thermal stress distribution on the structural surface. Areas with concentrated thermal stress are included in the theoretical damage hotspots. The theoretical damage hotspots output by the physical model clearly indicate the location information, damage type prediction, and theoretical basis of each area. Step S23: Input the real-time operating parameters into the neural network model. The neural network model extracts parameter features through convolutional layers, compresses feature dimensions through pooling layers, and maps features to damage probabilities through fully connected layers. It outputs the damage probability of each spatial location of the wind turbine. The damage probability ranges from 0 to 1. The higher the value, the greater the possibility of damage in the corresponding area. The output of the neural network model needs to be calibrated in conjunction with the confidence level of the training samples to ensure the reliability of the damage probability. If there are multiple input parameters, the model outputs the damage probability by integrating multi-dimensional features to improve the comprehensiveness of the judgment. If there is a single input parameter, the model is based on the damage type related to the feature corresponding to that parameter to ensure the specificity of the judgment. Step S24: Fusing the theoretical damage hotspots with the damage probabilities; the fusion adopts a weighted fusion method, and the weight allocation is based on the reliability of the physical model and the neural network model: the weight of the physical model is determined based on its theoretical fit, combined with the wind turbine structural design parameters and mechanical characteristics to ensure the reliability of theoretical support; the weight of the neural network model is determined based on the fitting accuracy of its training samples. The higher the fitting accuracy, the higher the weight ratio, ensuring the accuracy of actual data adaptation; the fusion process is as follows: assigning corresponding weights to the theoretical damage hotspots output by the physical model, assigning corresponding weights to the damage probabilities of each spatial location output by the neural network model, and then weighting and fusing the two to obtain the comprehensive damage score of each spatial location; Step S25: Spatial locations where the comprehensive damage score exceeds the preset damage threshold are designated as areas to be reviewed. The preset damage threshold is determined based on historical damage data of wind turbine units, operation and maintenance experience, and safe operation standards, combined with the importance of different structural parts. For critical parts such as blade roots and tower welds, the threshold is set relatively low to ensure no damage is missed. For non-critical parts, the threshold is set relatively high to reduce redundant reviews. At the same time, the location information, damage type prediction, comprehensive damage score, and judgment basis of the areas to be reviewed are marked to ensure the traceability and accuracy of the areas to be reviewed.

[0026] In this embodiment, the combination of physical and neural network models takes into account both physical mechanisms and actual data characteristics, avoiding misjudgments or omissions that may be caused by a single model. The detailed annotation of the parts to be reviewed provides a reliable reference for the delineation of supplementary inspection areas and the allocation of inspection tasks. This enables the matching of inspection scope with inspection tasks of multiple devices, achieving synergistic optimization of defect detection accuracy, detection coverage integrity and energy consumption control, and adapting to the high-efficiency inspection needs of wind turbine units under non-stop operation.

[0027] In practical implementation, as one example, existing methods often use fixed and unchanging regional divisions for pre-planned inspection ranges, without dynamically adjusting them based on the actual damage risk of the wind turbine. This results in some areas with potential damage not being included in the planning range, or the planning range not matching the damage risk, leading to problems such as inspections not being conducted when they are not needed, and repeated inspections when they are unnecessary. This embodiment constructs a pre-planned inspection range that adapts to the damage risk by dividing the area into blocks that fits the structural characteristics of the wind turbine. Then, by comparing the area to be reviewed with the planning range, the need for supplementary inspections is clarified, thus determining the supplementary inspection area. The specific implementation steps are as follows: Step S31: Construct a pre-planned inspection range. This range is determined by dividing the surface of the wind turbine into multiple inspection blocks and setting an inspection period for each block. The division of inspection blocks must conform to the structural stress characteristics of the wind turbine and the convenience of inspection. The core load-bearing structures of the wind turbine are the tower and blades, which have different structural characteristics; therefore, the division logic must be consistent with the evolution law of structural damage. The tower is a columnar structure, divided into several annular blocks along the height direction. The top, middle and bottom of the tower are the dividing points, and the blocks are divided from top to bottom. Each annular block corresponds to a clear spatial range, and the block number, height and corresponding structural part are marked. The inspection priority of each block is clearly defined, and the core stress area has a higher priority than the non-stress area. The blade is a long strip structure, divided into several sections along the span. Each section corresponds to a specific length range, and the section number and coverage area are marked. At the same time, the corresponding damage type is associated with the section. The tip section is susceptible to strong wind loads, so cracks are the focus of attention. The root section is under concentrated stress, so structural loosening is the focus of attention. The testing time period is set in combination with the operating conditions of the wind turbine and the operating status of the equipment. A suitable testing time period is assigned to each block to be inspected to avoid the impact of environmental factors such as high wind speed and extreme temperature difference on the testing accuracy. At the same time, it avoids conflicts between multiple equipment inspections. For example, during high wind speed periods, priority is given to the inspection of the bottom block of the tower, and during low wind speed periods, the blades and high-altitude areas are inspected to ensure the reliability of the testing data. Step S32: Compare the area to be reviewed with the pre-planned inspection range. Before comparison, ensure that the spatial coordinates and time periods of the area to be reviewed and the planned range are consistent. The spatial coordinates should use the same coordinate system, establishing a three-dimensional coordinate system with the unit base as the origin, and clarifying the coordinate parameters of each block and each area to be reviewed. The time period should be consistent, meaning that the inspection time period corresponding to the area to be reviewed is consistent with the preset inspection time period of that block in the planned range, to avoid comparison failure due to time period deviation. During the comparison, check each area to be reviewed one by one to determine whether it falls within any inspection block of the pre-planned inspection range and whether the inspection was completed within the preset inspection time period of that block. If the area to be reviewed falls within a certain planning block, and the inspection period of that block has already covered the inspection requirements for that area, it means that the area has been included in the planned inspection scope and no supplementary inspection is required. If the area to be reviewed falls within a certain planning block, but the inspection period of that block does not cover the damage evolution cycle of that area, then the block where that area is located is included in the area that needs to be re-inspected. If the area to be reviewed does not fall within any pre-planned inspection area, it is directly determined to be an area requiring supplementary inspection. If the area to be reviewed falls within the planned area, but the inspection of that area has not been completed, the area corresponding to that area will be included in the area requiring supplementary inspection. The area to be reviewed is located within the planned area. The inspection of this area has been completed, but the data is abnormal. The area corresponding to this area is included in the area that needs to be re-inspected. Step S33: The areas requiring supplementary inspection need to be categorized and labeled: For areas outside the planning scope that need to be reviewed, mark their specific locations, the corresponding damage types of the areas to be reviewed, and explain why the areas were not included in the planning. For areas within the planning scope that have not been tested, indicate the reasons for the incomplete testing and the planned time for supplementary testing; For areas within the planning scope that have been tested but have abnormal test data and require secondary verification, mark the reasons for the abnormalities and the key points of the verification. At the same time, based on the damage risk level of the parts to be reviewed, the areas to be inspected are ranked, so that areas with high damage risk and critical structures can be prioritized in the allocation of inspection tasks.

[0028] In this embodiment, the annular division of the tower conforms to the uniform stress distribution of its columnar structure, facilitating segmented inspection by the wall-climbing robot. The spanwise division of the blades conforms to the characteristics of its elongated structure, ensuring that the damage risk of each segment can be assessed individually. This also adapts to inspection needs; when only a single parameter is collected, supplementary inspections can be conducted on the corresponding segment, reducing redundant inspections. The pre-planned range and supplementary inspection area are interconnected. The pre-planned range serves as the basic inspection benchmark, while the supplementary inspection area provides damage risk-oriented supplementary inspections. This combination avoids the rigidity of a single plan and the blindness of supplementary inspections. Supplementary inspection requirements are dynamically determined based on the damage risk level of the area to be reviewed, rather than mechanically comparing positions, achieving a combination of risk orientation and range comparison. There is no fixed limit to the number of blocks; they can be adjusted according to unit specifications and inspection needs. The inspection period can be adjusted based on environmental factors, and the determination of the supplementary inspection area is unambiguous, adapting to different inspection scenarios.

[0029] In practical implementation, as one example, in the collaborative inspection of wind turbine units, the task allocation method often adopts a fixed allocation mode, failing to simultaneously consider the dual requirements of minimizing total energy consumption and maximizing the coverage of the area requiring supplementary inspection. Furthermore, the specific definition of energy consumption constraints is not clearly defined, and the selection of optimization algorithms lacks a reasonable basis, making it impossible to adapt to the collaborative inspection needs of the area requiring supplementary inspection and the pre-planned inspection range. This results in energy waste, incomplete coverage of the supplementary inspection area, or low inspection efficiency, failing to connect the area requiring supplementary inspection with equipment scheduling, and making it difficult to meet the comprehensive requirements of efficient, non-stop inspection of wind turbine units. Therefore, this embodiment combines the equipment characteristics and energy consumption constraints of wall-climbing robots and drones, guided by dual optimization objectives, and achieves accurate task allocation between the area requiring supplementary inspection and the pre-planned inspection range through the reasonable selection of optimization algorithms, generating directly executable collaborative operation instructions. The specific implementation steps are as follows: Step S41: Define energy consumption constraints, including the remaining battery power of the wall-climbing robot, the remaining battery power of the drone, and the maximum energy consumption quota for a single inspection task. The remaining battery power directly determines the duration and scope of a single inspection task, which is fundamental to whether the equipment can complete the task. The maximum energy consumption quota for a single inspection task is set in conjunction with the wind turbine maintenance cost and the equipment's endurance to avoid increased maintenance costs or equipment failure due to excessive energy consumption of a single device. The combination of these three factors comprehensively covers the energy consumption limitations for equipment operation and task execution, adapting to different inspection scenarios. The remaining battery power of the wall-climbing robot and drone is collected in real-time by the equipment's built-in power detection module. During the collection process, instantaneous power fluctuations are removed to ensure the accuracy and reliability of the power data. This directly reflects the equipment's current available energy reserves. The maximum energy consumption quota for a single inspection task is set based on the equipment's rated energy consumption, inspection task duration, and field maintenance and resupply conditions. It is suitable for different models and inspection scenarios. For example, small drones have weaker endurance and can be set with a lower maximum energy consumption quota, while large wall-climbing robots can have their quota appropriately increased to ensure that the quota setting matches the actual capabilities of the equipment and inspection needs. The definition standard for energy consumption constraints is: the remaining power of the wall-climbing robot and drone must not be lower than the preset minimum operating power to ensure that the equipment can complete the inspection task and return safely. The total energy consumption of a single inspection task must not exceed the maximum energy consumption quota. If the constraints are exceeded, the task allocation plan needs to be adjusted to avoid equipment shutdown or excessive energy consumption. Step S42: Integrate the areas requiring supplementary inspection with the pre-planned inspection range to clarify the total inspection area for this inspection. Based on the equipment characteristics of the wall-climbing robot and the drone, divide the basic inspection area. The wall-climbing robot is suitable for close-range fine inspection of the tower surface, and the tower area is given priority, including the pre-planned tower annular block and the tower area requiring supplementary inspection. The drone is suitable for large-scale coverage inspection of high-altitude areas such as blades and nacelles, and the blade area is given priority, including the pre-planned blade section and the blade area requiring supplementary inspection. Step S43: With the optimization objectives of minimizing total energy consumption and maximizing the coverage ratio of areas requiring supplementary inspection, the allocation scheme is adjusted through an optimization algorithm under energy consumption constraints. In the optimization objectives, minimum total energy consumption refers to the lowest total energy consumption of the wall-climbing robot and drone in performing this inspection task, which aligns with the energy-saving requirements of wind turbine operation and maintenance, reduces equipment operating costs, and extends equipment lifespan. Maximum coverage ratio of areas requiring supplementary inspection refers to the highest ratio of the actual coverage area of ​​the areas requiring supplementary inspection to the total area of ​​the areas requiring supplementary inspection after task allocation, which aligns with the inspection purpose, ensuring full coverage of the areas requiring supplementary inspection and avoiding omissions of potential damage. Since the purpose of the inspection is to identify potential damage, if only minimizing energy consumption results in the omission of areas requiring supplementary inspection, the inspection will lose its meaning. Therefore, maximizing the coverage ratio of areas requiring supplementary inspection takes priority over minimizing total energy consumption. Under the premise of ensuring full coverage of the areas requiring supplementary inspection, the goal is to minimize total energy consumption. If full coverage cannot be achieved, priority is given to covering areas with high damage risk. The optimization algorithm can be either ant colony optimization (ACO) or genetic algorithm. Ant colony optimization has strong pathfinding capabilities, accurately optimizing the walking path of the wall-climbing robot and the flight path of the drone, adapting to the dynamic planning requirements of inspection paths. Genetic algorithm has strong multi-objective optimization capabilities, simultaneously considering total energy consumption and coverage ratio, adapting to the needs of dual-objective collaborative optimization. If ant colony optimization is used, the inspection area is divided into several nodes, the movement paths of the wall-climbing robot and the drone are transformed into path search between nodes, energy consumption constraints are transformed into path search constraints, and the optimization objective is transformed into an evaluation function for path search. By adjusting the search parameters of the ant colony, the path search efficiency is optimized to ensure that the path with the lowest energy consumption and most comprehensive coverage is found. If genetic algorithm is used, the task allocation scheme is used as the genetic individual, the energy consumption constraint is used as the individual selection condition, and the optimization objective is used as the individual fitness evaluation standard. Through genetic selection, crossover, and mutation operations, the optimal task allocation scheme is selected. Step S44: Transform the inspection task allocation plan into collaborative operation instructions that can be recognized and executed by the wall-climbing robot and the drone, clarify the inspection tasks, paths, time nodes and operation requirements of the equipment, ensure that the two cooperate and avoid inspection conflicts.

[0030] In this embodiment, the task allocation scheme adapts to energy consumption constraints and inspection requirements, ensuring full coverage of the inspection area while reducing total energy consumption, improving inspection efficiency, avoiding energy waste and omission of potential damage, adapting to the high-efficiency inspection requirements of wind turbine units without shutting down, and providing reliable support for equipment scheduling by generating standardized and executable instructions, thus ensuring the collaborative inspection effect of the wall-climbing robot and the drone.

[0031] In a specific implementation, as one example, given that wind turbine inspection operations need to be carried out continuously and safely in complex field environments, and that significant fluctuations in wind conditions can easily affect drone flight safety, and that changes in equipment power levels may lead to inspection interruptions or missed areas for supplementary inspection, it is necessary to monitor key influencing parameters in real time to cope with emergencies and ensure the integrity of the inspection. This embodiment achieves collaborative operation between the wall-climbing robot and the drone by parsing collaborative operation instructions, completing equipment initialization, real-time monitoring of key parameters, dynamic adjustment of inspection tasks, and final summary execution. The specific implementation steps are as follows: Step S51: Parse the collaborative operation instructions to complete the initialization preparation of the wall-climbing robot and the drone; by extracting various key information from the instructions, clarify the inspection area, path, detection sequence, energy consumption control threshold, collaborative requirements, and anomaly handling principles of the two devices; the parsing process adopts a standardized parsing method that can be recognized by the devices to avoid omission of instruction information or parsing deviation; the initialization preparation includes: checking whether the walking mechanism, detection module, and power detection module of the wall-climbing robot are working properly, adjusting the walking speed and detection accuracy to match the fine detection requirements of the tower area; checking whether the flight mechanism, navigation module, and detection module of the drone are working properly, calibrating the flight path and shooting angle to match the large-area coverage detection requirements of high-altitude areas such as blades and nacelles; synchronizing the time base of the two devices to clarify the time nodes for collaborative cooperation, avoiding overlapping or omission of inspection areas, and ensuring that the initialization status of the devices meets the instruction requirements; after initialization, generate a device readiness feedback signal to confirm that the detection operation can be started; Step S52: Based on the parsed collaborative operation instructions, schedule the wall-climbing robot to perform close-range fine-grained inspection along the preset walking path. During the inspection, maintain a matching walking speed with inspection accuracy, record inspection data in real time, and upload it synchronously to the control terminal. Schedule the drone to perform wide-area coverage inspection along the preset flight route. During the inspection, maintain a stable flight altitude and shooting angle to ensure clear inspection data, and upload it synchronously to the control terminal. Real-time communication between the two devices is achieved through the control terminal, and their respective operating status is synchronously fed back. When one device malfunctions, the other device suspends its current inspection of non-core areas, prepares for emergency cooperation, and ensures that the inspection operation proceeds in an orderly manner. Step S53: During the inspection process, wind speed and remaining battery power of the equipment are monitored in real time. Wind speed directly affects the flight safety of the drone. The wind environment in the wild fluctuates greatly, and excessive wind speed can easily cause the drone to lose control and crash, thus interrupting the inspection operation. Real-time data is collected by an ultrasonic anemometer, covering the drone's flight area and the wall-climbing robot's inspection area. During the data collection process, interference from instantaneous wind speed fluctuations is removed to ensure that the wind speed data accurately reflects the current wind environment. The remaining battery power of the equipment determines the continuous operating capability of the two devices. Insufficient battery power will prevent the devices from completing the assigned inspection tasks, resulting in missed areas for supplementary inspection. Real-time data is collected through their respective battery detection modules, with the collection frequency matched to the equipment's operating status to ensure timely capture of battery changes and directly reflect the equipment's current available endurance. Step S54: Based on the monitoring results, determine the abnormal situation and execute the corresponding task adjustment operation; the abnormal situation is divided into three categories, and targeted adjustment operations are performed for each category: a) Wind speed exceeds the drone's safety threshold: This threshold is set based on the drone's flight performance and wind resistance level, adapting to the characteristics of outdoor wind environments. For example, small drones have weaker wind resistance, so the threshold is set relatively low. The judgment is based on the real-time collected wind speed data continuously exceeding the preset threshold for a preset duration. At this time, the drone operation is immediately suspended, and the drone is controlled to return to a safe docking point to prevent equipment loss of control. At the same time, the areas that the drone has not completed the inspection are reviewed, and areas that can be reached by the wall-climbing robot are selected. These areas are then assigned to the wall-climbing robot for execution, while the unadjusted areas are temporarily stored until the wind speed drops below the safety threshold, at which point the drone is reassigned to complete the inspection. b. Drone battery level below preset lower limit: This preset lower limit is set in conjunction with the drone's return trip energy consumption and emergency operation energy consumption to ensure that the drone can still safely return to the docking point after the battery level drops below this lower limit, avoiding equipment crashes or being stranded at high altitudes; the judgment is based on the real-time collected data of the drone's remaining battery level continuously being below the preset lower limit. At this time, the drone is immediately controlled to suspend inspection operations and start the return trip procedure to ensure the equipment returns safely; at the same time, the unfinished inspection areas of the drone are sorted out, and the parts of the areas that need to be inspected that can be covered by the wall-climbing robot are prioritized for inspection by the wall-climbing robot. The remaining areas that cannot be covered will be rescheduled for inspection after the drone replenishes its battery level. c. The wall-climbing robot's battery level is below the preset lower limit: This lower limit is set in conjunction with the wall-climbing robot's return trip energy consumption and emergency operation energy consumption to ensure that the equipment can still return safely after the battery level drops below this limit, avoiding equipment delays. The judgment is based on the real-time collected data on the remaining battery level of the wall-climbing robot, which is consistently below the preset lower limit. At this time, the wall-climbing robot is immediately controlled to stop the current inspection operation and start the return trip program to ensure the equipment returns safely. At the same time, the unfinished inspection areas of the wall-climbing robot are sorted out and adjusted to be covered by drones first. The inspection of areas that need to be inspected is performed first, and then the pre-planned non-core areas are inspected to ensure that no areas need to be inspected are missed and to avoid inspection interruptions. During abnormal adjustments, the inspection task information is updated synchronously, uploaded to the control terminal, and the reasons for the adjustment, the content of the adjustment, and the time of the adjustment are recorded to ensure that the adjustment process is traceable. Step S55: After completing the inspection, dispatch the wall-climbing robot and drone back to the preset docking point, shut down the inspection module, walking mechanism or flying mechanism, and perform equipment status inspection and maintenance; at the same time, check the coverage of the inspection area, confirm that all areas that need to be inspected have been inspected, and if there are any incomplete areas, formulate a plan for the next inspection based on the equipment status and environmental conditions to ensure that the inspection task is fully covered and without omissions.

[0032] In this embodiment, by monitoring key parameters in real time during the detection process, clarifying the anomaly judgment criteria and dynamic adjustment logic, flexible task switching is achieved instead of fixed instruction execution. Anomaly adjustments prioritize coverage of the supplementary inspection area, taking into account both equipment safety and inspection needs. It can promptly respond to emergencies such as sudden wind speed changes and insufficient power, avoid equipment failure and inspection interruption, adapt to the high-efficiency inspection needs of wind turbine units without shutting down, and form a complete inspection closed loop.

[0033] Based on the same inventive concept as the wind turbine collaborative inspection method for a dynamic operating domain described in the foregoing embodiments, this invention also provides a wind turbine collaborative inspection system for a dynamic operating domain, such as... Figure 2 As shown, the system includes: The parameter acquisition module is used to acquire the real-time operating parameters of the wind turbine. The damage analysis module is used to determine the parts of the wind turbine to be reviewed based on the correlation analysis between real-time operating parameters and the damage evolution model; the damage evolution model adopts a combination of physical model and neural network model. The area comparison module is used to compare the area to be reviewed with the pre-planned detection range to obtain the area that needs to be re-inspected; The instruction generation module is used to allocate inspection tasks to the areas requiring supplementary inspection and the pre-planned inspection range based on the energy consumption constraints of the wall-climbing robot and the drone, and generate collaborative operation instructions. The inspection and scheduling module is used to schedule wall-climbing robots and drones to perform inspections based on collaborative operation instructions.

[0034] The system described above in this invention can effectively realize the collaborative inspection method of wind turbine units in dynamic operating domains, and the technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0035] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for dynamic operation field wind turbine cooperative inspection, characterized in that, include: Obtain real-time operating parameters of wind turbine units; Based on the correlation analysis between the real-time operating parameters and the damage evolution model, the parts of the wind turbine to be reviewed are determined. The damage evolution model is a combination of a physical model and a neural network model; The area to be reviewed is compared with the pre-planned inspection range to obtain the area that needs to be re-inspected; Based on the energy consumption constraints of the wall-climbing robot and the drone, inspection tasks are assigned to the areas requiring supplementary inspection and the pre-planned inspection range, and collaborative operation instructions are generated. Based on the collaborative operation instructions, the wall-climbing robot and the drone are scheduled to perform the inspection.

2. The wind turbine collaborative inspection method for dynamic operating domains according to claim 1, characterized in that, The real-time operating parameters include: The data is based on at least one of the following: strain distribution data acquired by a fiber optic sensor, vibration data acquired by an accelerometer, wind speed data acquired by an anemometer, and temperature data acquired by an infrared thermal imager.

3. The method for collaborative inspection of wind turbine units in a dynamic operating domain according to claim 2, characterized in that, The method for determining the area to be reviewed includes: The real-time operating parameters are input into the physical model to obtain the theoretical damage hotspot region; The real-time operating parameters are input into the neural network model to obtain the damage probability at each spatial location; The theoretical damage hotspot region is fused with the damage probability, and the spatial location in the fusion result that exceeds the preset damage threshold is taken as the part to be reviewed.

4. The wind turbine collaborative inspection method for dynamic operating domains according to claim 3, characterized in that, The fusion of theoretical damage hotspots and damage probabilities is achieved through weighted fusion. The weights of the physical model are determined based on its theoretical fit, while the weights of the neural network model are determined based on the fitting accuracy of its training samples. After weighted fusion, a comprehensive damage score is obtained for each spatial location. Spatial locations with comprehensive damage scores exceeding a preset damage threshold are designated as areas to be reviewed.

5. The method for collaborative inspection of wind turbine units in a dynamic operating domain according to claim 1, characterized in that, The pre-planned inspection range is determined by dividing the surface of the wind turbine into multiple inspection blocks and setting an inspection period for each inspection block.

6. The wind turbine collaborative inspection method for dynamic operating domains according to claim 5, characterized in that, The method for dividing the surface of a wind turbine into multiple inspection blocks includes: The tower is divided into several annular blocks along its height and several sections along the blade span.

7. The method for collaborative inspection of wind turbine units in a dynamic operating domain according to claim 1, characterized in that, The method for assigning inspection tasks includes: With the optimization objectives of minimizing total energy consumption and maximizing the coverage ratio of areas requiring supplementary inspection, the walking path of the wall-climbing robot and the flight path of the drone are determined by an optimization algorithm under the energy consumption constraint; the optimization algorithm is either an ant colony algorithm or a genetic algorithm.

8. The method for collaborative inspection of wind turbine units in a dynamic operating domain according to claim 7, characterized in that, The energy consumption constraints include the remaining power of the wall-climbing robot, the remaining power of the drone, and the maximum energy consumption quota for a single inspection task.

9. The method for collaborative inspection of wind turbine units in a dynamic operating domain according to claim 1, characterized in that, During the testing process, wind speed and remaining battery power of the equipment are monitored in real time. When the wind speed exceeds the safety threshold for drones, drone operations will be suspended, and some areas requiring additional inspection will be handled by wall-climbing robots. When the drone's battery level drops below the preset lower limit, the drone operation will be suspended, and some areas requiring additional inspection will be handled by the wall-climbing robot. When the wall-climbing robot's battery level drops below the preset lower limit, the drone will prioritize covering the remaining areas that need to be inspected.

10. A collaborative inspection system for wind turbine generators in a dynamic operating domain, characterized in that, include: The parameter acquisition module is used to acquire the real-time operating parameters of the wind turbine. The damage analysis module is used to determine the parts of the wind turbine to be reviewed based on the correlation analysis between the real-time operating parameters and the damage evolution model; the damage evolution model is a combination of a physical model and a neural network model. The area comparison module is used to compare the area to be reviewed with the pre-planned detection range to obtain the area that needs to be re-inspected; The instruction generation module is used to allocate inspection tasks to the area requiring supplementary inspection and the pre-planned inspection range based on the energy consumption constraints of the wall-climbing robot and the drone, and generate collaborative operation instructions. The inspection and scheduling module is used to schedule the wall-climbing robot and the drone to perform inspections according to the collaborative operation instructions.

Images