Intelligent inspection and diagnosis system, method, equipment and device for photovoltaic power station unmanned aerial vehicle

Abstract

The invention provides an intelligent inspection and diagnosis system, method and device for a photovoltaic power station unmanned aerial vehicle and a medium, and the system comprises a heterogeneous perception fusion module which is used for constructing digital mirror image mapping of a power station global physical field through multi-mode sensor space-time coding and electromagnetic fingerprint matching; the group intelligent decision module is used for realizing autonomous task negotiation and anti-fragile path evolution of a multi-unmanned aerial vehicle cluster based on a dynamic Bayesian game model; the embedded diagnostic kernel module is used for running a quantization feature extraction algorithm at an edge computing node to realize component-level defect subsurface-level diagnostic reasoning; the superbody self-evolution module is used for continuously reconstructing a diagnostic knowledge graph by means of a genetic programming framework to realize Darwin asymptotic optimization of a system cognitive architecture; the problems that in unmanned aerial vehicle inspection, fault type judgment and accurate geographic positioning cannot be conducted online, photovoltaic array arrangement changes and sudden shielding scenes cannot be self-adapted, and transmission bandwidth limitation and analysis lag are likely to happen in a 5G weak coverage area are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station inspection, and in particular to an intelligent inspection and diagnosis system, method, device and medium for an unmanned aerial vehicle of a photovoltaic power station. Background Art

[0002] With the accelerating transformation of the global energy structure towards cleaner energy, the installed capacity of photovoltaic power stations continues to grow, and the stability and power generation efficiency of photovoltaic modules are directly related to the benefits of the power station. Existing photovoltaic power stations are generally distributed in complex terrains and large-scale fields (such as hills, deserts, water surfaces), and it is necessary to regularly inspect thousands to hundreds of thousands of photovoltaic modules for hidden faults such as hot spot effects, PID (Potential Induced Degradation) degradation, hidden cracks, and dust accumulation. Traditional operation and maintenance mainly rely on manual handheld inspection with infrared thermal imagers or fixed monitoring devices, which have pain points such as long inspection cycles, low coverage rates, high risks of high-altitude operations, and fault identification relying on manual experience. Especially in ultra-large-scale power stations with a capacity of hundreds of megawatts, the traditional method requires 2-3 weeks for a single comprehensive inspection, and there are risks of missed inspections and misjudgments, which severely restricts the improvement of the full life cycle benefits of the power station.

[0003] Most existing unmanned aerial vehicle inspection systems are equipped with single visible light or infrared imaging devices, and manual remote control flight is required to transmit a large amount of raw data back to the background for analysis, resulting in three core defects: First, spectral sensing and AI real-time edge computing are not deeply integrated, and it is impossible to online discriminate fault types and accurately geolocate, relying on manual secondary screening, resulting in response delays; Second, traditional flight path planning relies on flying along preset coordinate points and cannot adapt to changes in the photovoltaic array layout and sudden occlusion scenarios, resulting in a high missed inspection rate; Third, video processing depends on the cloud center server, and in areas with weak 5G coverage, problems such as limited transmission bandwidth and analysis lag are likely to occur, making it difficult to meet the second-level emergency response requirements for emergencies such as wildfires and foreign object intrusion.

[0004] Therefore, there is an urgent need to propose an intelligent inspection and diagnosis system for an unmanned aerial vehicle of a photovoltaic power station to solve the technical problems of inability to online discriminate fault types and accurately geolocate, inability to adapt to changes in the photovoltaic array layout and sudden occlusion scenarios, and limited transmission bandwidth and analysis lag in areas with weak 5G coverage during unmanned aerial vehicle inspection. Summary of the Invention

[0005] To overcome the problems existing in the related art, the present disclosure provides an intelligent inspection and diagnosis system, method, device and medium for an unmanned aerial vehicle of a photovoltaic power station to solve the technical problems of inability to online discriminate fault types and accurately geolocate, inability to adapt to changes in the photovoltaic array layout and sudden occlusion scenarios, and limited transmission bandwidth and analysis lag in areas with weak 5G coverage during unmanned aerial vehicle inspection in the related art.

[0006] One or more embodiments of this specification provide an intelligent inspection and diagnosis system for a photovoltaic power station UAV, including a heterogeneous perception fusion module, a swarm intelligence decision-making module, an embedded diagnosis kernel module, and a hyperspace self-evolution module;

[0007] The heterogeneous perception fusion module is used to construct a digital mirror mapping of the power station's global physical field through multimodal sensor spatio-temporal encoding and electromagnetic fingerprint matching;

[0008] The swarm intelligence decision-making module is used to achieve autonomous task negotiation and anti-fragile path evolution of multiple UAV clusters based on a dynamic Bayesian game model;

[0009] The embedded diagnosis kernel module is used to run a quantization feature extraction algorithm on an edge computing node to achieve component-level defect sub-surface-level diagnostic reasoning;

[0010] The hyperspace self-evolution module is used to continuously reconstruct the diagnostic knowledge graph with the help of a genetic programming framework to achieve Darwinian asymptotic optimization of the system's cognitive architecture.

[0011] Preferably, the heterogeneous perception fusion module includes a multimodal spatio-temporal encoding unit, a sub-millimeter attitude control unit, an electromagnetic fingerprint map unit, a self-compensating optical channel unit, and a nano-level vibration sensing unit;

[0012] The multimodal spatio-temporal encoding unit is used to integrate spatio-temporal sequence data of visible light, thermal infrared, lidar, and electric field intensity sensors;

[0013] The sub-millimeter attitude control unit is used for quantum gyro + visual SLAM composite positioning;

[0014] The electromagnetic fingerprint map unit is used to establish a three-dimensional map of the power station's electromagnetic environment characteristics;

[0015] The self-compensating optical channel unit is used to adjust the photography parameter optical flow algorithm in real time according to atmospheric disturbances;

[0016] The nano-level vibration sensing unit is used to monitor component micro-deformation through laser Doppler vibrometry;

[0017] The multimodal spatio-temporal encoding unit triggers the sub-millimeter attitude control unit to perform quantum gyro + visual SLAM composite positioning calibration by integrating spatio-temporal sequence data of visible light, thermal infrared, lidar, and electric field intensity;

[0018] The sub-millimeter attitude control unit triggers the electromagnetic fingerprint map unit to dynamically construct a three-dimensional map of the power station's electromagnetic environment characteristics by providing a three-dimensional coordinate with an accuracy of 0.1 mm;

[0019] The electromagnetic fingerprint map unit triggers the self-compensating optical channel unit to correct the compensation coefficient of the atmospheric turbulence optical flow algorithm in real time by analyzing the radiation spectrum characteristics of the inverter;

[0020] The self-compensating optical channel unit triggers the nanoscale sensing unit to start the laser Doppler micro-deformation monitoring protocol by stabilizing the optical imaging quality.

[0021] Preferably, the swarm intelligence decision-making module includes a swarm task allocation unit, a counter-intuitive path planning unit, a resilient communication topology unit, a peer knowledge sharing unit, and a digital mirror mapping unit;

[0022] The swarm task allocation unit is used for multi-UAV dynamic task segmentation based on the consensus algorithm;

[0023] The counter-intuitive path planning unit is used for fusing the A* variant algorithm of the fault probability field;

[0024] The resilient communication topology unit is used for a communication reliability controller that supports multi-hop transmission in ad hoc networks;

[0025] The peer knowledge sharing unit is used for the incremental model parameter transfer mechanism between UAVs;

[0026] The digital mirror mapping unit is used for constructing a digital copy of the physical power station in real time;

[0027] The swarm task allocation unit triggers the counter-intuitive path planning unit to load the dynamic obstacle avoidance path of the fault probability field through the multi-aircraft cooperation scheme generated by the Byzantine fault-tolerant consensus algorithm;

[0028] The counter-intuitive path planning unit triggers the resilient communication topology unit to activate the dynamic QoS guarantee mechanism for multi-hop transmission in ad hoc networks through the generated non-Euclidean geometry path;

[0029] The resilient communication topology unit triggers the peer knowledge sharing unit to implement the neural network weight synchronization under the federated learning framework through the link quality prediction data;

[0030] The peer knowledge sharing unit triggers the digital mirror mapping unit to update the mapping relationship of the equipment deterioration state of the power station digital copy through the aggregated incremental model parameters.

[0031] Preferably, the embedded diagnosis kernel module includes a quantum noise analysis unit, a defect propagation simulation unit, a material fingerprint comparison unit, an electric potential field reconstruction engine unit, and an entropy value anomaly detection unit;

[0032] The quantum noise analysis unit is used for extracting the quantization characteristics of the original sensor signal;

[0033] The defect propagation simulation unit is used for fault evolution prediction based on the lattice Boltzmann method;

[0034] The material fingerprint comparison unit is used for matching digital specimens of the microstructure of photovoltaic materials;

[0035] The electric potential field reconstruction engine unit is used to reconstruct the three-dimensional electric potential distribution from discrete measurement values;

[0036] The entropy anomaly detection unit is used for monitoring the state mutation of a photovoltaic array based on Shannon entropy;

[0037] The quantum noise analysis unit triggers the defect propagation simulation unit to run the crack diffusion simulation of the lattice Boltzmann method by extracting the quantization projection of the sensor signal feature vector;

[0038] The defect propagation simulation unit dynamically triggers the material fingerprint comparison unit to perform topological matching verification of the microstructure atlas of photovoltaic modules through the predicted subsurface cracks;

[0039] The material fingerprint comparison unit triggers the electric potential field reconstruction unit to solve the three-dimensional Poisson equation under non-uniform medium conditions by identifying the material lattice defect characteristics;

[0040] The electric potential field reconstruction unit triggers the entropy anomaly detection unit to calculate the Shannon entropy change gradient matrix at the array level through the reconstructed carrier concentration distribution.

[0041] Preferably, the hyperbody self-evolution module includes a cross-generation feature distillation unit, an adversarial simulation sandbox unit, a cognitive loop optimization unit, an evolutionary constraint solving unit, and a legacy system adaptation unit;

[0042] The cross-generation feature distillation unit is used for the parameter migration channel between the old and new diagnostic models;

[0043] The adversarial simulation sandbox unit is used to generate an adversarial network to construct an extreme scenario simulator;

[0044] The cognitive loop optimization unit is used to adjust the weights of each module of the system through reinforcement learning;

[0045] The evolutionary constraint solving unit is used to ensure that the system update does not violate the physical law verification layer;

[0046] The legacy system adaptation unit is used to abstract the virtualization interface components of hardware differences;

[0047] The cross-generation feature distillation unit triggers the adversarial simulation sandbox unit to generate adversarial samples containing extreme states such as avalanche breakdown through the cross-generation constraints constructed by the parameter migration channel;

[0048] The adversarial simulation sandbox unit triggers the cognitive circuit optimization unit to perform meta-reinforcement learning of multi-objective weighted coefficients through the generated high-dimensional adversarial interference data;

[0049] The cognitive loop optimization unit triggers the evolutionary constraint solving unit to implement the Lagrange multiplier method for physical law compatibility verification through the updated module coupling weights;

[0050] The evolution constraint solving unit triggers the legacy system adaption unit to generate virtualized middleware to realize generalized access of multi-protocol devices through the verified system evolution strategy.

[0051] One or more embodiments of this specification provide a method for intelligent inspection and diagnosis of a photovoltaic power station drone, including the following steps:

[0052] Construct a digital mirror image of the entire physical field of the power station through multimodal sensor spatiotemporal encoding and electromagnetic fingerprint matching;

[0053] Autonomous task negotiation and anti-fragile path evolution of multi-UAV swarms based on dynamic Bayesian game model;

[0054] Run quantized feature extraction algorithms on edge computing nodes to implement sub-surface diagnostic reasoning for component-level defects;

[0055] The genetic programming framework is used to continuously reconstruct the diagnostic knowledge graph to achieve Darwinian asymptotic optimization of the system cognitive architecture.

[0056] Preferably, the autonomous task negotiation and anti-fragile path evolution of a multi-UAV cluster based on a dynamic Bayesian game model specifically includes the following steps:

[0057] The dynamic obstacle avoidance path of the multi-machine collaborative solution loading fault probability field generated by the Byzantine fault-tolerant consensus algorithm;

[0058] Activate the dynamic QoS guarantee mechanism of multi-hop transmission in ad hoc networks through the generated non-Euclidean geometric paths;

[0059] Implement neural network weight synchronization in a federated learning framework using link quality prediction data;

[0060] The equipment degradation status mapping relationship of the power plant digital replica is updated through the aggregated incremental model parameters.

[0061] Preferably, the step of running a quantized feature extraction algorithm on an edge computing node to implement sub-surface level diagnostic reasoning of component-level defects specifically includes the following steps:

[0062] Run crack propagation simulations using the lattice Boltzmann method by extracting quantized projections of the sensor signal eigenvectors;

[0063] Perform topological matching verification of the microstructure atlas of the photovoltaic module by predicting the dynamic of subsurface cracks;

[0064] Solve the three-dimensional Poisson equation under inhomogeneous medium conditions by identifying the characteristics of material lattice defects;

[0065] Calculate the change gradient matrix of array-level Shannon entropy through the reconstructed carrier concentration distribution.

[0066] One or more embodiments of this specification provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the intelligent inspection and diagnosis method of the photovoltaic power station drone as described above.

[0067] One or more embodiments of this specification provide a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps of the intelligent inspection and diagnosis method of the photovoltaic power station drone as described above.

[0068] An intelligent inspection and diagnosis system, method, device, and medium for a photovoltaic power station unmanned aerial vehicle provided by the present disclosure have the advantages that through a heterogeneous perception fusion module, it is used to construct a digital mirror mapping of the power station's entire physical field through multi-modal sensor spatio-temporal encoding and electromagnetic fingerprint matching. By means of multi-modal sensor spatio-temporal encoding and electromagnetic fingerprint matching, a digital mirror mapping of the power station's entire physical field is constructed, achieving an accurate modeling of the physical environment of the photovoltaic power station, being able to monitor the operation status of the power station in real time, timely discover potential problems, and improve the operation and maintenance efficiency of the power station; a swarm intelligence decision-making module, which is used to achieve autonomous task negotiation and anti-fragile path evolution of multiple unmanned aerial vehicle clusters based on a dynamic Bayesian game model. Based on the dynamic Bayesian game model, autonomous task negotiation of multiple unmanned aerial vehicle clusters is realized, improving the collaborative operation ability of the unmanned aerial vehicle clusters. Through anti-fragile path evolution, it ensures that the path planning of unmanned aerial vehicles in complex environments is robust, can handle emergencies, and improves the reliability and efficiency of inspections; an embedded diagnosis kernel module, which is used to implement component-level defect sub-surface-level diagnostic reasoning by running a quantization feature extraction algorithm on an edge computing node. By running a quantization feature extraction algorithm on an edge computing node, sub-surface-level diagnostic reasoning of photovoltaic component defects is realized, improving the accuracy and depth of defect detection. The deployment of edge computing nodes enables diagnostic reasoning to be quickly completed locally, reducing data transmission latency and improving the real-time performance of diagnosis; a hyperbody self-evolution module, which is used to continuously reconstruct a diagnostic knowledge graph by means of a genetic programming framework to achieve Darwinian asymptotic optimization of the system's cognitive architecture. By means of a genetic programming framework, the diagnostic knowledge graph is continuously reconstructed, enabling the system to continuously learn and evolve, adapt to new diagnostic requirements, and gradually improve the performance of the system's cognitive architecture through Darwinian asymptotic optimization, making the diagnostic system more intelligent and efficient. Through the comprehensive application of multi-modal sensors, dynamic Bayesian game models, quantization feature extraction algorithms, and genetic programming frameworks, high-precision, high-efficiency, and high-reliability intelligent inspection and diagnosis of photovoltaic power station unmanned aerial vehicles are achieved. The digital mirror mapping of the entire physical field provides accurate environmental modeling. Autonomous task negotiation and anti-fragile path evolution of multiple unmanned aerial vehicle clusters ensure the coordination and path optimization of inspection tasks. Component-level defect sub-surface-level diagnostic reasoning improves the accuracy and real-time performance of defect detection. The continuous reconstruction of the diagnostic knowledge graph and the optimization of the system's cognitive architecture enable the system to continuously evolve and improve, significantly enhancing the operation and maintenance efficiency and intelligent level of photovoltaic power stations. Description of the Drawings

[0069] In order to more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in this specification. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0070] Figure 1 Schematic structural diagram of an intelligent inspection and diagnosis system for a photovoltaic power station drone provided for one or more embodiments of this specification;

[0071] Figure 2 Schematic structural diagram of the units included in an intelligent inspection and diagnosis system for a photovoltaic power station drone provided for one or more embodiments of this specification;

[0072] Figure 3 Schematic flowchart of an intelligent inspection and diagnosis method for a photovoltaic power station drone provided for one or more embodiments of this specification;

[0073] Figure 4 Schematic structural diagram of a computer device provided for one or more embodiments of this specification. Specific embodiments

[0074] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will clearly and completely describe the technical solutions in one or more embodiments of this specification in conjunction with the accompanying drawings in one or more embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this invention document.

[0075] The following will make a detailed description of the present invention in conjunction with the specific embodiments and the accompanying drawings of the specification.

[0076] System embodiments

[0077] According to an embodiment of the present invention, an intelligent inspection and diagnosis system for a photovoltaic power station drone is provided. As Figure 1 shown, it is a schematic structural diagram of the intelligent inspection and diagnosis system for a photovoltaic power station drone provided for this embodiment. The intelligent inspection and diagnosis system for a photovoltaic power station drone according to an embodiment of the present invention includes a heterogeneous perception fusion module 11, a swarm intelligence decision-making module 12, an embedded diagnosis kernel module 13, and a superbody self-evolution module 14.

[0078] The heterogeneous perception fusion module 11 is used to construct a digital mirror mapping of the power station's global physical field through multi-modal sensor spatio-temporal coding and electromagnetic fingerprint matching;

[0079] The swarm intelligence decision-making module 12 is used to realize autonomous task negotiation and anti-fragile path evolution of multiple drone clusters based on a dynamic Bayesian game model;

[0080] An embedded diagnostic kernel module 13 is used to run a quantization feature extraction algorithm on an edge computing node to implement component-level defect sub-surface-level diagnostic reasoning;

[0081] A superbody self-evolution module 14 is used to continuously reconstruct a diagnostic knowledge graph by means of a genetic programming framework to achieve Darwinian asymptotic optimization of the system's cognitive architecture.

[0082] The system provided in this embodiment includes a heterogeneous perception fusion module 11, which is used to construct a digital mirror mapping of the power station's global physical field through multi-modal sensor spatio-temporal encoding and electromagnetic fingerprint matching, and construct a digital mirror mapping of the power station's global physical field through multi-modal sensor spatio-temporal encoding and electromagnetic fingerprint matching, so as to achieve an accurate modeling of the physical environment of the photovoltaic power station, be able to monitor the operation status of the power station in real time, timely discover potential problems, and improve the operation and maintenance efficiency of the power station; a swarm intelligence decision-making module 12, which is used to achieve autonomous task negotiation and anti-fragile path evolution of a multi-UAV cluster based on a dynamic Bayesian game model. Based on the dynamic Bayesian game model, autonomous task negotiation of the multi-UAV cluster is realized to improve the collaborative operation ability of the UAV cluster. Through anti-fragile path evolution, it is ensured that the path planning of the UAVs in a complex environment is robust, can cope with emergencies, and improve the reliability and efficiency of inspection; an embedded diagnostic kernel module 13 is used to run a quantization feature extraction algorithm on an edge computing node to implement component-level defect sub-surface-level diagnostic reasoning. By running a quantization feature extraction algorithm on an edge computing node, sub-surface-level diagnostic reasoning of photovoltaic component defects is realized, the accuracy and depth of defect detection are improved, and the deployment of edge computing nodes enables diagnostic reasoning to be quickly completed locally, reducing data transmission delay and improving the real-time performance of diagnosis; a superbody self-evolution module 14 is used to continuously reconstruct a diagnostic knowledge graph by means of a genetic programming framework to achieve Darwinian asymptotic optimization of the system's cognitive architecture. By means of a genetic programming framework, the diagnostic knowledge graph is continuously reconstructed, enabling the system to continuously learn and evolve to adapt to new diagnostic requirements. Through Darwinian asymptotic optimization, the performance of the system's cognitive architecture is gradually improved, making the diagnostic system more intelligent and efficient. Through the comprehensive application of multi-modal sensors, dynamic Bayesian game models, quantization feature extraction algorithms, and genetic programming frameworks, high-precision, high-efficiency, and high-reliability of intelligent UAV inspection and diagnosis in photovoltaic power stations are achieved. The digital mirror mapping of the global physical field provides accurate environmental modeling, autonomous task negotiation and anti-fragile path evolution of the multi-UAV cluster ensure the coordination of inspection tasks and path optimization, component-level defect sub-surface-level diagnostic reasoning improves the accuracy and real-time performance of defect detection, and the continuous reconstruction of the diagnostic knowledge graph and the optimization of the system's cognitive architecture enable the system to continuously evolve and improve, significantly improving the operation and maintenance efficiency and intelligent level of photovoltaic power stations.

[0083] As Figure 2As shown in the figure, it is a schematic structural diagram of the units included in an intelligent inspection and diagnosis system for a photovoltaic power station unmanned aerial vehicle provided in this embodiment.

[0084] In one embodiment, the heterogeneous perception fusion module 11 includes a multimodal spatio-temporal coding unit 111, a sub-millimeter attitude control unit 112, an electromagnetic fingerprint map unit 113, a self-compensating optical channel unit 114, and a nanoscale vibration perception unit 115.

[0085] The multimodal spatio-temporal coding unit 111 is used to integrate spatio-temporal sequence data of visible light, thermal infrared, lidar, and electric field intensity sensors. The sub-millimeter attitude control unit 112 is used for quantum gyro + visual SLAM composite positioning. The electromagnetic fingerprint map unit 113 is used to establish a three-dimensional map of the electromagnetic environment characteristics of the power station. The self-compensating optical channel unit 114 is used to adjust the photography parameters of the optical flow algorithm in real time according to atmospheric disturbances. The nanoscale vibration perception unit 115 is used to monitor the micro-deformation of components through laser Doppler vibrometry.

[0086] The multimodal spatio-temporal coding unit 111 triggers the sub-millimeter attitude control unit 112 to perform quantum gyro + visual SLAM (Simultaneous Localization and Mapping) composite positioning calibration by integrating spatio-temporal sequence data of visible light, thermal infrared, lidar, and electric field intensity. The sub-millimeter attitude control unit 112 triggers the electromagnetic fingerprint map unit 113 to dynamically construct a three-dimensional map of the electromagnetic environment characteristics of the power station by providing three-dimensional coordinates with an accuracy of 0.1 mm. The electromagnetic fingerprint map unit 113 triggers the self-compensating optical channel unit 114 to correct the compensation coefficient of the atmospheric turbulence optical flow algorithm in real time by analyzing the radiation spectrum characteristics of the inverter. The self-compensating optical channel unit 114 triggers the nanoscale perception unit 115 to start the laser Doppler micro-deformation monitoring protocol by stabilizing the optical imaging quality.

[0087] The system provided in this embodiment integrates the spatio-temporal data of visible light, thermal infrared, lidar, and electric field intensity sensors through an unmanned aerial vehicle cluster, constructs a digital mirror of the power station by combining the high-precision positioning of a quantum gyro and visual SLAM, and then collaboratively plans an anti-fragile inspection path based on a dynamic Bayesian game model to avoid fault hot spots in real time.

[0088] In one embodiment, the swarm intelligence decision-making module 12 includes a bee colony task allocation unit 121, a counter-intuitive path planning unit 122, a resilient communication topology unit 123, a peer knowledge sharing unit 124, and a digital mirror mapping unit 125.

[0089] The swarm task allocation unit 121 is used for the dynamic task segmentation of multiple unmanned aerial vehicles (UAVs) based on a consensus algorithm. The counter-intuitive path planning unit 122 is used for the A* variant algorithm that fuses the fault probability field. The resilient communication topology unit 123 is used for the communication reliability controller that supports multi-hop transmission in ad hoc networks. The peer knowledge sharing unit 124 is used for the incremental model parameter transfer mechanism between UAVs. The digital mirror mapping unit 125 is used for the construction of a digital copy that synchronizes the physical power station in real time.

[0090] Among them, the A* variant algorithm is as follows:

[0091] f(n) = g(n) + λ * h(n) * φ(p f (n));

[0092] Among them, g(n) represents the actual cost from the starting point to node n (including flight energy consumption and inspection task weight); h(n) represents the classical heuristic function (usually using the Euclidean distance); φ(p f (n)) = 1 + α(p f (n)) β represents the fault probability field modulation factor; (p f (n) ∈ [0, 1] represents the equipment fault probability at node n; λ represents the dynamic weight coefficient, which follows the time decay mechanism of (γ is the environmental disturbance intensity, t is the flight time).

[0093] The swarm task allocation unit 121 triggers the counter-intuitive path planning unit 122 to load the dynamic obstacle avoidance path of the fault probability field through the multi-aircraft cooperation scheme generated by the Byzantine fault-tolerant consensus algorithm. The counter-intuitive path planning unit 122 triggers the resilient communication topology unit 123 to activate the dynamic QoS (Quality of Service) guarantee mechanism for multi-hop transmission in ad hoc networks through the generated non-Euclidean geometric path. The resilient communication topology unit 123 triggers the peer knowledge sharing unit 124 to implement the neural network weight synchronization under the federated learning framework through the link quality prediction data. The peer knowledge sharing unit 124 triggers the digital mirror mapping unit 125 to update the device degradation state mapping relationship of the power station digital copy through the aggregated incremental model parameters.

[0094] The system provided in this embodiment realizes the full-process intelligent management of the UAV inspection task through the collaborative work of the swarm intelligence decision-making module 12. From task allocation to path planning, from communication guarantee to knowledge sharing, and then to digital mirror mapping, each unit closely cooperates to form an efficient, reliable, and adaptive intelligent inspection system. This solution significantly improves the efficiency, accuracy, and reliability of the UAV inspection of the photovoltaic power station, while reducing the operation and maintenance costs and risks, providing strong technical support for the intelligent operation and maintenance of the photovoltaic power station.

[0095] In one embodiment, the embedded diagnostic kernel module 13 includes a quantum noise analysis unit 131, a defect propagation simulation unit 132, a material fingerprint comparison unit 133, an electric potential field reconstruction engine unit 134, and an entropy anomaly detection unit 135.

[0096] The quantum noise analysis unit 131 is used to extract the quantization characteristics of the original sensor signal. The defect propagation simulation unit 132 is used to predict the fault evolution based on the lattice Boltzmann method. The material fingerprint comparison unit 133 is used to match the digital specimens of the microstructure of photovoltaic materials. The electric potential field reconstruction engine unit 134 is used to reconstruct the three-dimensional electric potential distribution from discrete measurement values. The entropy anomaly detection unit 135 is used to monitor the state mutation of the photovoltaic array based on Shannon entropy.

[0097] The quantum noise analysis unit 131 triggers the crack diffusion simulation of the lattice Boltzmann method by the quantization projection of the feature vector of the sensor signal. The defect propagation simulation unit 132 triggers the topological matching verification of the microstructure atlas of the photovoltaic module by the predicted subsurface crack dynamics. The material fingerprint comparison unit 133 triggers the electric potential field reconstruction unit 134 to solve the three-dimensional Poisson equation under the condition of inhomogeneous medium by the identified material lattice defect characteristics. The electric potential field reconstruction unit 134 triggers the entropy anomaly detection unit 135 to calculate the change gradient matrix of the array-level Shannon entropy by the reconstructed carrier concentration distribution.

[0098] During flight, the nanoscale vibration perception and electric potential field reconstruction technology can capture the micro-deformation and electric potential anomaly of the component in real time, while the edge computing node runs the quantization feature extraction algorithm for subsurface-level defect diagnosis, combined with the crack evolution trend of the lattice Boltzmann method.

[0099] The system provided in this embodiment realizes multi-level and high-precision diagnosis and prediction of photovoltaic module defects through the collaborative work of the embedded diagnostic kernel module 13. From signal analysis to defect evolution prediction, from microstructure matching to electric potential field reconstruction, and then to state mutation monitoring, each unit works closely together to form an efficient, precise, and adaptive intelligent diagnostic system. This solution significantly improves the accuracy, depth, and real-time performance of photovoltaic module defect detection, provides strong technical support for the preventive maintenance and fault warning of photovoltaic power stations, and reduces the operation and maintenance costs and risks at the same time.

[0100] In one embodiment, the superbody self-evolution module 14 includes a cross-generation feature distillation unit 141, an adversarial simulation sandbox unit 142, a cognitive loop optimization unit 143, an evolutionary constraint solving unit 144, and a legacy system adaptation unit 145.

[0101] The cross-generation feature distillation unit 141 is used for the parameter migration channel between the old and new diagnostic models. The adversarial simulation sandbox unit 142 is used to generate an adversarial network to construct an extreme scenario simulator. The cognitive loop optimization unit 143 is used to adjust the weights of each module of the system through reinforcement learning. The evolutionary constraint solving unit 144 is used to ensure that the system update does not violate the physical law verification layer. The legacy system adaptation unit 145 is used to abstract the virtualization interface components of hardware differences.

[0102] The cross-generation feature distillation unit 141 triggers the adversarial simulation sandbox unit 142 to generate adversarial samples containing extreme states such as avalanche breakdown through the cross-generation constraints constructed by the parameter migration channel. The adversarial simulation sandbox unit 142 triggers the cognitive loop optimization unit 143 to perform meta-reinforcement learning of multi-objective weighting coefficients through the generated high-dimensional adversarial interference data. The cognitive loop optimization unit 143 triggers the evolutionary constraint solving unit 144 to implement the Lagrange multiplier method for physical law compatibility verification through the updated module coupling weights. The evolutionary constraint solving unit 144 triggers the legacy system adaptation unit 145 to generate virtualization middleware to achieve generalized access to multi-protocol devices through the verified system evolution strategy.

[0103] The system continuously evolves through the genetic programming framework, uses the adversarial simulation sandbox to generate extreme scenario training data, realizes the dynamic optimization of the diagnostic knowledge graph, and is compatible with multiple devices through the virtualization interface, and finally forms a closed-loop intelligent operation and maintenance system with the capabilities of autonomous decision-making, precise positioning, subsurface defect diagnosis, and continuous evolution.

[0104] The system provided in this embodiment realizes the self-evolution, high adaptability, and seamless integration with the traditional system of the intelligent diagnosis system of the photovoltaic power station through the collaborative work of the hyperbody self-evolution module 14. From cross-generation model optimization to extreme scenario simulation, from dynamic weight adjustment to physical law compatibility verification, and then to generalized access to multi-protocol devices, each unit closely cooperates to form an efficient, reliable, and adaptive intelligent system. This solution significantly improves the intelligence level, robustness, and compatibility of the system, provides strong technical support for the long-term stable operation and continuous optimization of the photovoltaic power station, and at the same time reduces the complexity and cost of system upgrade and operation and maintenance.

[0105] Method Embodiment

[0106] According to an embodiment of the present invention, there is provided an intelligent inspection and diagnosis method for an unmanned aerial vehicle of a photovoltaic power station, as Figure 3 shown, which is a schematic flowchart of the intelligent inspection and diagnosis method for an unmanned aerial vehicle of a photovoltaic power station provided in this embodiment. The intelligent inspection and diagnosis method for an unmanned aerial vehicle of a photovoltaic power station according to an embodiment of the present invention includes the following steps:

[0107] S301. Construct a digital mirror mapping of the power station's global physical field through spatio-temporal encoding of multi-modal sensors and electromagnetic fingerprint matching.

[0108] S302. Achieve autonomous task negotiation and anti-fragile path evolution of multi-UAV clusters based on the dynamic Bayesian game model.

[0109] S303. Run a quantization feature extraction algorithm on edge computing nodes to achieve component-level defect subsurface-level diagnostic reasoning.

[0110] S304. Continuously reconstruct the diagnostic knowledge graph with the help of the genetic programming framework to achieve Darwinian asymptotic optimization of the system's cognitive architecture.

[0111] The method provided in this embodiment constructs a digital mirror mapping of the power station's global physical field through spatio-temporal encoding of multi-modal sensors and electromagnetic fingerprint matching. By spatio-temporal encoding of multi-modal sensors and electromagnetic fingerprint matching, a digital mirror mapping of the power station's global physical field is constructed, enabling accurate modeling of the physical environment of the photovoltaic power station, capable of real-time monitoring of the power station's operating status, timely detection of potential problems, and improvement of the power station's operation and maintenance efficiency. Achieve autonomous task negotiation and anti-fragile path evolution of multi-UAV clusters based on the dynamic Bayesian game model. Based on the dynamic Bayesian game model, autonomous task negotiation of multi-UAV clusters is achieved, improving the collaborative operation ability of the UAV cluster. Through anti-fragile path evolution, it ensures that the path planning of UAVs in complex environments is robust, capable of handling emergencies, and improving the reliability and efficiency of inspections. Run a quantization feature extraction algorithm on edge computing nodes to achieve component-level defect subsurface-level diagnostic reasoning. Run a quantization feature extraction algorithm on edge computing nodes to achieve subsurface-level diagnostic reasoning of photovoltaic module defects, improving the accuracy and depth of defect detection. The deployment of edge computing nodes enables diagnostic reasoning to be quickly completed locally, reducing data transmission latency and improving the real-time performance of diagnosis. Continuously reconstruct the diagnostic knowledge graph with the help of the genetic programming framework to achieve Darwinian asymptotic optimization of the system's cognitive architecture. With the help of the genetic programming framework, continuously reconstruct the diagnostic knowledge graph, enabling the system to continuously learn and evolve to adapt to new diagnostic requirements. Through Darwinian asymptotic optimization, gradually improve the performance of the system's cognitive architecture, making the diagnostic system more intelligent and efficient. Through the comprehensive application of multi-modal sensors, dynamic Bayesian game models, quantization feature extraction algorithms, and genetic programming frameworks, high-precision, high-efficiency, and high-reliability of intelligent UAV inspection and diagnosis of photovoltaic power stations are achieved. The digital mirror mapping of the global physical field provides accurate environmental modeling. Autonomous task negotiation and anti-fragile path evolution of multi-UAV clusters ensure the coordination of inspection tasks and path optimization. Component-level defect subsurface-level diagnostic reasoning improves the accuracy and real-time performance of defect detection. The continuous reconstruction of the diagnostic knowledge graph and the optimization of the system's cognitive architecture enable the system to continuously evolve and improve, significantly enhancing the operation and maintenance efficiency and intelligent level of photovoltaic power stations.

[0112] In one embodiment, the autonomous task negotiation and anti-fragile path evolution of multi-UAV clusters are realized based on a dynamic Bayesian game model, which specifically includes the following steps:

[0113] Load the dynamic obstacle avoidance path of the failure probability field with the multi-aircraft cooperation scheme generated by the Byzantine fault-tolerant consensus algorithm.

[0114] Activate the dynamic QoS guarantee mechanism of multi-hop transmission in the ad-hoc network through the generated non-Euclidean geometry path.

[0115] Implement the neural network weight synchronization under the federated learning framework through the link quality prediction data.

[0116] Update the device degradation state mapping relationship of the power station digital copy through the aggregated incremental model parameters.

[0117] The method provided in this embodiment realizes the full-process intelligent management of the UAV inspection task through collaborative work. From task assignment to path planning, from communication guarantee to knowledge sharing, and then to digital mirror mapping, each unit closely cooperates to form an efficient, reliable, and adaptive intelligent inspection system. This solution significantly improves the efficiency, accuracy, and reliability of UAV inspections in photovoltaic power stations, while reducing the operation and maintenance costs and risks, providing strong technical support for the intelligent operation and maintenance of photovoltaic power stations.

[0118] In one embodiment, the component-level defect sub-surface-level diagnostic reasoning is realized by running a quantization feature extraction algorithm on an edge computing node, which specifically includes the following steps:

[0119] Run the crack diffusion simulation of the lattice Boltzmann method through the quantization projection of the sensor signal feature vector extraction.

[0120] Perform the topological matching verification of the photovoltaic component microstructure atlas through the predicted sub-surface crack dynamics.

[0121] Solve the three-dimensional Poisson equation under non-uniform medium conditions through the identified material lattice defect characteristics.

[0122] Calculate the change gradient matrix of the array-level Shannon entropy through the reconstructed carrier concentration distribution.

[0123] The method provided in this embodiment realizes the multi-level and high-precision diagnosis and prediction of photovoltaic component defects through collaborative work. From signal analysis to defect evolution prediction, from microstructure matching to electric potential field reconstruction, and then to state mutation monitoring, each unit closely cooperates to form an efficient, accurate, and adaptive intelligent diagnosis system. This solution significantly improves the accuracy, depth, and real-time performance of photovoltaic component defect detection, provides strong technical support for the preventive maintenance and fault warning of photovoltaic power stations, while reducing the operation and maintenance costs and risks.

[0124] The embodiments of the present invention are corresponding device embodiments to the above method embodiments. The specific operations of each module processing step can be understood by referring to the description of the method embodiments, and will not be elaborated herein.

[0125] As Figure 4 shown, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the intelligent patrol diagnosis method in the above embodiments, or when the computer program is executed by a processor, it implements the intelligent patrol diagnosis method of the unmanned aerial vehicle in the photovoltaic power station in the above embodiments.

[0126] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0127] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the key point of each embodiment is to illustrate the differences from other embodiments. In particular, for the apparatus or system embodiments, since they are basically similar to the method embodiments, they are described relatively simply, and for the relevant parts, reference can be made to the corresponding descriptions in the method embodiments. The apparatus and system embodiments described above are only illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative work.

[0128] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and the content not described in detail in this specification of the present invention belongs to the well-known technology of those skilled in the art.

Claims

1. An intelligent inspection and diagnosis system for a photovoltaic power station drone, characterized in that, It includes a heterogeneous perception fusion module, a swarm intelligence decision-making module, an embedded diagnostic kernel module, and a superbody self-evolution module; The heterogeneous perception fusion module is used to construct a digital mirror mapping of the power station's global physical field through multi-modal sensor spatio-temporal encoding and electromagnetic fingerprint matching; The swarm intelligence decision-making module is used to achieve autonomous task negotiation and anti-fragile path evolution of multiple UAV clusters based on a dynamic Bayesian game model; The embedded diagnostic kernel module is used to run a quantization feature extraction algorithm at the edge computing node to achieve component-level defect sub-surface-level diagnostic reasoning; The superbody self-evolution module is used to continuously reconstruct the diagnostic knowledge graph with the help of a genetic programming framework to achieve Darwinian asymptotic optimization of the system's cognitive architecture.

2. The intelligent inspection and diagnosis system of the photovoltaic power station UAV according to claim 1, characterized in that The heterogeneous perception fusion module includes a multi-modal spatio-temporal encoding unit, a sub-millimeter attitude control unit, an electromagnetic fingerprint map unit, a self-compensating optical channel unit, and a nano-level vibration perception unit; The multi-modal spatio-temporal encoding unit is used to integrate spatio-temporal sequence data of visible light, thermal infrared, lidar, and electric field intensity sensors; The sub-millimeter attitude control unit is used for quantum gyro + visual SLAM composite positioning; The electromagnetic fingerprint map unit is used to establish a three-dimensional map of the power station's electromagnetic environment characteristics; The self-compensating optical channel unit is used to adjust the photography parameter optical flow algorithm in real time according to atmospheric disturbances; The nano-level vibration perception unit is used to monitor component micro-deformation through laser Doppler vibrometry; The multi-modal spatio-temporal encoding unit triggers the sub-millimeter attitude control unit to perform quantum gyro + visual SLAM composite positioning calibration by integrating spatio-temporal sequence data of visible light, thermal infrared, lidar, and electric field intensity; The sub-millimeter attitude control unit triggers the electromagnetic fingerprint map unit to dynamically construct a three-dimensional map of the power station's electromagnetic environment characteristics by providing three-dimensional coordinates with an accuracy of 0.1 mm; The electromagnetic fingerprint map unit triggers the self-compensating optical channel unit to real-time correct the compensation coefficient of the atmospheric turbulence optical flow algorithm by analyzing the radiation spectrum characteristics of the inverter; The self-compensating optical channel unit triggers the nano-level perception unit to start the laser Doppler micro-deformation monitoring protocol by stabilizing the optical imaging quality.

3. The intelligent inspection and diagnosis system of the photovoltaic power station UAV according to claim 1, characterized in that, The swarm intelligence decision-making module includes a swarm task allocation unit, a counter-intuitive path planning unit, a resilient communication topology unit, a peer knowledge sharing unit, and a digital mirror mapping unit; The swarm task allocation unit is used for multi-UAV dynamic task segmentation based on a consensus algorithm; The counter-intuitive path planning unit is used to fuse the A* variant algorithm of the fault probability field; The resilient communication topology unit is used as a communication reliability controller to support multi-hop transmission of self-organizing networks; The peer knowledge sharing unit is used for the incremental model parameter transfer mechanism between UAVs; The digital mirror mapping unit is used to construct a real-time synchronized digital copy of the physical power station; The swarm task allocation unit triggers the counter-intuitive path planning unit to load the dynamic obstacle avoidance path of the fault probability field through a multi-machine cooperation scheme generated by the Byzantine fault-tolerant consensus algorithm; The counter-intuitive path planning unit triggers the resilient communication topology unit to activate the dynamic QoS guarantee mechanism of multi-hop transmission of self-organizing networks by generating a non-Euclidean geometry path; The resilient communication topology unit triggers the peer knowledge sharing unit to implement neural network weight synchronization under the federated learning framework through link quality prediction data; The peer knowledge sharing unit triggers the digital mirror mapping unit to update the device degradation state mapping relationship of the power station digital copy through the aggregated incremental model parameters.

4. The intelligent inspection and diagnosis system for the photovoltaic power station UAV according to claim 1, characterized in that, The embedded diagnostic kernel module includes a quantum noise analysis unit, a defect propagation simulation unit, a material fingerprint comparison unit, an electric potential field reconstruction engine unit, and an entropy anomaly detection unit; The quantum noise analysis unit is used to extract the quantization characteristics of the original sensor signal; The defect propagation simulation unit is used to predict the fault evolution based on the lattice Boltzmann method; The material fingerprint comparison unit is used to match the digital specimens of the microstructure of photovoltaic materials; The electric potential field reconstruction engine unit is used to reconstruct the three-dimensional electric potential distribution from discrete measurement values; The entropy anomaly detection unit is used to monitor the state mutation of the photovoltaic array based on Shannon entropy; The quantum noise analysis unit triggers the defect propagation simulation unit to run the crack diffusion simulation of the lattice Boltzmann method by extracting the quantization projection of the sensor signal feature vector; The defect propagation simulation unit triggers the material fingerprint comparison unit to perform topological matching verification of the microstructure atlas of the photovoltaic module by predicting the dynamic subsurface crack; The material fingerprint comparison unit triggers the electric potential field reconstruction unit to solve the three-dimensional Poisson equation under non-uniform medium conditions by identifying the material lattice defect characteristics; The electric potential field reconstruction unit triggers the entropy anomaly detection unit to calculate the Shannon entropy change gradient matrix at the array level by reconstructing the carrier concentration distribution.

5. The intelligent inspection and diagnosis system of the photovoltaic power station drone according to claim 1, wherein, The hyperbody self-evolution module includes a cross-generation feature distillation unit, an adversarial simulation sandbox unit, a cognitive loop optimization unit, an evolutionary constraint solving unit, and a legacy system adaptation unit; The cross-generation feature distillation unit is used for the parameter migration channel between the old and new diagnostic models; The adversarial simulation sandbox unit is used to generate an adversarial network to construct an extreme scenario simulator; The cognitive loop optimization unit is used to adjust the weights of each module of the system through reinforcement learning; The evolutionary constraint solving unit is used to ensure that the system update does not violate the physical law verification layer; The legacy system adaptation unit is used to abstract the virtualization interface components of hardware differences; The cross-generation feature distillation unit triggers the adversarial simulation sandbox unit to generate adversarial samples containing extreme states such as avalanche breakdown through the cross-generation constraints constructed by the parameter migration channel; The adversarial simulation sandbox unit triggers the cognitive loop optimization unit to perform meta-reinforcement learning of multi-objective weighted coefficients through the generated high-dimensional adversarial interference data; The cognitive loop optimization unit triggers the evolutionary constraint solving unit to implement the Lagrange multiplier method for physical law compatibility verification through the updated module coupling weights; The evolutionary constraint solving unit triggers the legacy system adaptation unit to generate virtualization middleware to achieve generalized access to multi-protocol devices through the verified system evolution strategy.

6. An intelligent inspection and diagnosis method for a photovoltaic power station drone, characterized in that, It includes the following steps: Construct a digital mirror mapping of the power station's global physical field through multi-modal sensor spatio-temporal encoding and electromagnetic fingerprint matching; Autonomous task negotiation and anti-fragile path evolution of multi-UAV swarms based on dynamic Bayesian game model; Run quantized feature extraction algorithms on edge computing nodes to implement sub-surface diagnostic reasoning for component-level defects; The genetic programming framework is used to continuously reconstruct the diagnostic knowledge graph to achieve Darwinian asymptotic optimization of the system cognitive architecture.

7. The intelligent inspection and diagnosis method of the photovoltaic power station UAV according to claim 6, characterized in that, The autonomous task negotiation and anti-fragile path evolution of a multi-UAV cluster based on a dynamic Bayesian game model specifically includes the following steps: The dynamic obstacle avoidance path of the multi-machine collaborative solution loading fault probability field generated by the Byzantine fault-tolerant consensus algorithm; Activate the dynamic QoS guarantee mechanism of multi-hop transmission in ad hoc networks through the generated non-Euclidean geometric paths; Implement neural network weight synchronization in a federated learning framework using link quality prediction data; The equipment degradation status mapping relationship of the power plant digital replica is updated through the aggregated incremental model parameters.

8. The intelligent inspection and diagnosis method of the photovoltaic power station UAV according to claim 6, characterized in that, The implementation of the quantized feature extraction algorithm on the edge computing node to implement the sub-surface diagnostic reasoning of component-level defects specifically includes the following steps: Run crack propagation simulations using the lattice Boltzmann method by extracting quantized projections of the sensor signal eigenvectors; Topological matching validation of PV module microstructure maps performed through predicted subsurface crack dynamics; Solve the three-dimensional Poisson's equation under inhomogeneous medium conditions by identifying the material lattice defect characteristics; The array-level Shannon entropy change gradient matrix is calculated using the reconstructed carrier concentration distribution.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, the intelligent inspection and diagnosis method for the photovoltaic power station drone as described in any one of claims 6 to 8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, the steps of the intelligent inspection and diagnosis method of the photovoltaic power station drone as claimed in any one of claims 6 to 8 are implemented.

Images