Power station operation and maintenance method, system and equipment based on unmanned aerial vehicle inspection data and medium

Abstract

The invention relates to the technical field of power station operation and maintenance, and discloses a power station operation and maintenance method, system and device based on unmanned aerial vehicle inspection data and a medium. Accurate positioning, category division and grade evaluation of defect targets of the power station equipment are realized, the efficiency and accuracy of defect target identification are improved, and timely discovery of early fine defects of the power station equipment is guaranteed; meanwhile, by combining skill labels, real-time geographic positions and actual workloads of the operation and maintenance personnel, the optimal operation and maintenance personnel are matched through a task allocation algorithm, and operation and maintenance work orders are automatically constructed and distributed, so that reasonable configuration of operation and maintenance resources is realized, the problem of defect disposal lag caused by unreasonable dispatching is avoided, and the operation and maintenance efficiency is improved. And operation and maintenance personnel are ensured to process each defect target in time, and reliable technical guarantee is provided for safe and stable operation of power station equipment.

Description

Technical Field

[0001] This invention relates to the field of power plant operation and maintenance technology, specifically to a power plant operation and maintenance method, system, equipment, and medium based on UAV inspection data. Background Technology

[0002] With the rapid development of the new energy industry and power infrastructure construction, the scale and complexity of power plant equipment are constantly increasing. Equipment inspection and operation and maintenance of various power plants, such as photovoltaic power plants, wind power plants, and substations, have become core links in ensuring the safe and stable operation of power plants. Unmanned aerial vehicle (UAV) inspection technology, with its advantages of high efficiency, wide coverage, strong adaptability to operating environments, and avoidance of the safety risks of manual inspection, can quickly collect massive amounts of power plant equipment inspection image data, providing rich visual evidence for power plant defect investigation. It has gradually replaced traditional manual inspection and become the mainstream technical means for power plant operation and maintenance inspection, and is widely used in the investigation of appearance defects and operational anomalies in power plant equipment.

[0003] The power plant operation and maintenance methods disclosed in related technologies rely on manual screening and identification of defects in inspection image data collected by drones, followed by the issuance of corresponding maintenance work orders. However, the identification results of these methods are significantly affected by the professional experience and working conditions of the maintenance personnel, leading to potential omissions, misjudgments, and inconsistencies in defect classification and grading standards. Furthermore, these methods are time-consuming. In conclusion, the power plant operation and maintenance methods in these related technologies are insufficient to meet the current operation and maintenance requirements of power plants. Summary of the Invention

[0004] This invention provides a power plant operation and maintenance method, system, equipment, and medium based on UAV inspection data, in order to solve the problem that power plant operation and maintenance methods in related technologies are difficult to meet the operation and maintenance requirements of existing power plants.

[0005] In a first aspect, the present invention provides a power plant operation and maintenance method based on unmanned aerial vehicle (UAV) inspection data, the method comprising: Acquire inspection image data collected by the UAV, wherein the inspection image data includes physical information of the image acquisition location; Based on the inspection image data, the trained target detection algorithm is used to filter and obtain image data of multiple defective targets. Based on the image data of each defective target, the trained deep learning model is used to identify the defect, and the defect category and defect level of the corresponding defective target are obtained. For each defect target whose defect level is higher than the preset level, the optimal maintenance personnel are matched using a task allocation algorithm by taking into account each maintenance personnel's skill tags, real-time geographical location and actual workload, and corresponding maintenance work orders are generated and sent to the corresponding user terminals.

[0006] Through the above implementation methods, based on the inspection image data collected by UAVs, and relying on the trained target detection algorithm and deep learning model, the system achieves accurate positioning, classification, and level assessment of defective targets in power plant equipment. This improves the efficiency and accuracy of defective target identification, ensuring the timely detection of early and subtle defects in power plant equipment. Simultaneously, by combining the skill tags, real-time geographical location, and actual workload of maintenance personnel, the system uses a task allocation algorithm to match the optimal maintenance personnel and automatically construct and dispatch maintenance work orders. This achieves rational allocation of maintenance resources, avoids delays in defect handling due to unreasonable work assignments, and ensures that maintenance personnel can promptly address each defective target. This provides reliable technical support for the safe and stable operation of power plant equipment and overcomes the shortcomings of existing power plant maintenance methods that are difficult to meet the current maintenance requirements of power plants.

[0007] In one optional implementation, the deep learning model includes a deep convolutional neural network and a multi-factor evaluation model. The step of using the trained deep learning model to identify each defective target based on its image data, thereby obtaining the defect category and defect level of the corresponding defective target, includes: Based on the image data of each defective target, a convolutional neural network model is used for identification to obtain the defect type of the corresponding defective target; Based on the physical size, location, and type characteristics of defects in the image data of each defective target, a multi-factor evaluation model is used to obtain the defect level of each defective target. The defect type and defect level of each defect target are associated, and the result is output as the defect category and defect level of the corresponding defect target.

[0008] Through the above implementation methods, the feature information of defect images is extracted using a trained convolutional neural network model, and the defect type of the corresponding defect target is determined, so as to achieve efficient and accurate determination of the defect type of power plant equipment and avoid the type classification bias caused by human subjective judgment. At the same time, by using a multi-factor evaluation model, multiple influencing factors such as defect physical size, equipment location, and type characteristics are comprehensively analyzed to achieve scientific and objective evaluation of defect level. This ensures that the final defect level has a quantifiable basis for judgment, avoids the problem of misjudgment of level caused by single-dimensional evaluation, and lays an accurate data foundation for subsequent processing of high-level defects.

[0009] In one alternative implementation, it further includes: Receive maintenance data of defect targets uploaded by user terminals and update the status of the corresponding maintenance work orders.

[0010] Through the above implementation methods, based on the maintenance data uploaded in real time by operation and maintenance personnel through user terminals during the defect handling process, the corresponding work order process status is automatically updated, replacing the inefficient traditional method of manual reporting and manual work order updates. This effectively improves the timeliness and accuracy of work order status management, allowing operation and maintenance management personnel to keep track of the handling progress of each defect in real time, and realizes the effective correlation between defect data, work order data and maintenance data. It breaks down the information silos in each link of the power plant operation and maintenance process, and provides complete and continuous data support for the traceability and analysis of subsequent operation and maintenance data and the health status assessment of power plant equipment, further improving the digital closed-loop management of power plant operation and maintenance work.

[0011] In one optional implementation, receiving maintenance data of defective targets uploaded by the user terminal and updating the status of the corresponding maintenance work order includes: Receive maintenance data of defective targets uploaded by user terminals; the maintenance data includes records of on-site defect handling process, defect handling results information, on-site operation evidence images, spare parts replacement details, operation and maintenance personnel's operation time and maintenance operation remarks; The received maintenance data is structured and validated, and the compliant data is classified and organized according to preset dimensions to establish a mapping between maintenance data and corresponding defect target association information; the defect target association information includes defect number, defect category, defect level and defect target collection location; By combining the maintenance data, the results of previous defect identification, and the mapping between maintenance data and the corresponding defect target information, the on-site operation evidence images and the image data of the defect targets are compared and displayed, and the status of the maintenance work order is updated.

[0012] Through the above implementation methods, maintenance data is structured, validated, and accurately mapped to information associated with defect targets. This effectively eliminates invalid and redundant data, ensuring a one-to-one correspondence between maintenance data and original defect data and maintenance work orders. This avoids data confusion and misalignment, improving the accuracy of maintenance data management. Simultaneously, by comparing and displaying on-site work evidence images with original defect images, the effectiveness of defect handling is presented intuitively, facilitating maintenance management personnel to quickly verify the rectification status of defect targets. By automatically updating work order status based on verification results and maintenance data, the traditional manual verification and update method is replaced, significantly improving the efficiency and accuracy of work order status management.

[0013] In one alternative implementation, it further includes: In response to the report generation command provided by the user, based on the category and level of each target defect, and combined with the preset report framework, the system uses a trained feature extraction model to fill in the data and obtain the inspection report of the target power plant.

[0014] Through the above implementation methods, relying on the trained feature extraction model and combined with the preset report framework, the data can be accurately filled based on the identification results such as the category and level of the target defect. The report can be directly generated based on the structured data of defect identification, realizing the seamless connection between inspection data and defect identification results and report output, which facilitates timely and reliable data support for the decision-making of power plant operation and maintenance management personnel.

[0015] In one alternative implementation, it further includes: Based on the inspection image data collected by UAV, the data is processed using data preprocessing methods to obtain preprocessed inspection image data. The preprocessed inspection image data is used to screen defective targets through target detection algorithms.

[0016] Through the above implementation method, by preprocessing the raw inspection image data collected by the UAV, interference factors such as noise and blurring in the raw images are effectively eliminated. At the same time, the inspection image data is rectified and optimized, improving the clarity and effectiveness of the inspection image data. This avoids interference from invalid and messy raw data on the target detection algorithm, significantly reduces the probability of false detection and false negative detection of the target detection algorithm, and significantly improves the accuracy and efficiency of defect target screening.

[0017] In one optional implementation, the data preprocessing method includes: Image denoising and enhancement, data integrity verification, data cleaning, or data format standardization.

[0018] Through the above implementation methods, image denoising and enhancement methods effectively eliminate noise, blur, and other interference in inspection image data, improving the clarity and defect feature identification of the inspection image data, and ensuring that the target detection algorithm can more accurately capture defect details. Data integrity verification and data cleaning methods remove missing, invalid, and redundant inspection image data, avoiding invalid data from consuming computing resources and interfering with algorithm judgment, thus improving the overall efficiency of inspection image data processing. The data format standardization method achieves unified image data format under different UAV equipment and different acquisition scenarios, solving the problem of incompatibility between multi-source data formats, and improving the adaptability and robustness of the target detection algorithm to inspection image data acquired by different UAVs.

[0019] Secondly, the present invention provides a power plant operation and maintenance system based on UAV inspection data, the system comprising: The image acquisition module is used to acquire inspection image data collected by the UAV, and the inspection image data includes physical information of the acquisition location; The defect screening module is used to screen multiple defect targets based on the inspection image data using a trained target detection algorithm. The defect identification module is used to identify each defective target based on its image data using a trained deep learning model, thereby obtaining the defect category and defect level of the corresponding defective target. The task allocation module is used to match the optimal maintenance personnel for each defect target with a defect level greater than the preset level by taking into account each maintenance personnel's skill tags, real-time geographical location and actual workload, and to generate corresponding maintenance work orders and send them to the corresponding user terminals.

[0020] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the power plant operation and maintenance method based on UAV inspection data described in the first aspect or any corresponding embodiment.

[0021] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the power plant operation and maintenance method based on UAV inspection data described in the first aspect or any corresponding embodiment. Attached Figure Description

[0022] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the first process of a power plant operation and maintenance method based on UAV inspection data according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the second process of the power plant operation and maintenance method based on UAV inspection data according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the third process of the power plant operation and maintenance method based on UAV inspection data according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the third process of the power plant operation and maintenance method based on UAV inspection data according to an embodiment of the present invention; Figure 5 This is a structural block diagram of a power plant operation and maintenance system based on UAV inspection data according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0026] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0027] The power plant operation and maintenance methods based on inspection image data collected by drones disclosed in related technologies have the following defects: 1. The massive amounts of image data collected by drones require manual screening, which is time-consuming and prone to missing details. Data from a single inspection mission may require analysts to spend days or even weeks to complete the initial screening. This results in a significant lag in inspection results, making it impossible to provide real-time feedback on equipment status. Furthermore, prolonged exposure to highly similar images can easily lead to visual fatigue and decreased attention, causing users to miss subtle or early-stage defects (such as small cracks or faint heat spots), which are often precursors to major safety hazards.

[0028] 2. Relying on human experience to determine defect types and severity levels is inconsistent and accuracy is difficult to guarantee. For the same defect (such as insulator damage or hot spots on components), different analysts may give different type judgments and severity level assessments (such as "minor", "moderate", or "severe"). Even the same analyst may make inaccurate judgments on the same defect at different times.

[0029] 3. Identified defects are difficult to translate into maintenance actions in a timely manner, lacking a complete closed loop from problem discovery to problem resolution. The ultimate goal of inspection is to eliminate defects and ensure safety. The current model results in a serious disconnect between "inspection" and "repair".

[0030] 4. Inspection data, identification results, and maintenance actions are scattered across different systems, making collaboration difficult. The original images are in one system (such as a drone management platform), the identification results are in another system (such as an AI analysis platform), and the maintenance work orders are in a third system (such as an asset management system). Maintenance personnel need to switch between different systems repeatedly to verify information, which greatly reduces collaboration efficiency and increases communication costs.

[0031] In summary, to address the aforementioned shortcomings, this invention provides a power plant operation and maintenance method based on UAV inspection data. By acquiring inspection image data collected by UAVs and relying on trained target detection algorithms and deep learning models, it achieves precise location, classification, and grading of defective targets in power plant equipment, improving the efficiency and accuracy of defective target identification and ensuring the timely detection of early and subtle defects in power plant equipment. Simultaneously, by combining the skill tags, real-time geographical location, and actual workload of operation and maintenance personnel, the optimal operation and maintenance personnel are matched through a task allocation algorithm, and operation and maintenance work orders are automatically constructed and dispatched. This achieves rational allocation of operation and maintenance resources, avoids delays in defect handling due to unreasonable work assignments, and ensures that operation and maintenance personnel can promptly address each defective target. This provides reliable technical support for the safe and stable operation of power plant equipment and overcomes the shortcomings of related power plant operation and maintenance methods that are difficult to meet the current operation and maintenance requirements of power plants.

[0032] According to an embodiment of the present invention, a power plant operation and maintenance method based on UAV inspection data is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0033] This embodiment provides a power plant operation and maintenance method based on UAV inspection data, which can be used in power plant operation and maintenance server terminals. Figure 1 This is a flowchart of a power plant operation and maintenance method based on UAV inspection data according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: S101, acquire inspection image data collected by the UAV; the inspection image data includes physical information of the image acquisition location.

[0034] The inspection image data collected by the drone is transmitted in real time via 4G / 5G network through the data upload unit integrated into the drone flight control App or SDK during the inspection operation, or the original images and metadata (such as location, attitude, timestamp) are automatically uploaded in batches after the operation is completed.

[0035] Meanwhile, to further facilitate the use of target detection algorithms to filter defective targets in inspection image data, the method provided in this embodiment also includes: Based on the inspection image data collected by UAV, the data is processed using data preprocessing methods to obtain preprocessed inspection image data; the preprocessed inspection image data is used to screen defective targets through target detection algorithms.

[0036] For example, data preprocessing methods include: Image denoising and enhancement, data integrity verification, data cleaning, or data format standardization.

[0037] Image denoising and enhancement methods effectively eliminate noise, blur, and other interference in inspection image data, improving the clarity and defect feature identification of the data, ensuring that the target detection algorithm can more accurately capture defect details. Data integrity verification and data cleaning methods remove missing, invalid, and redundant inspection image data, preventing invalid data from consuming computing resources and interfering with algorithm judgment, thus improving the overall efficiency of inspection image data processing. Data format standardization methods unify the format of image data from different UAVs and different acquisition scenarios, solving the problem of incompatibility between multi-source data formats and improving the adaptability and robustness of the target detection algorithm to inspection image data acquired by different UAVs.

[0038] By preprocessing the raw inspection image data collected by UAVs, interference factors such as noise and blurring in the raw images are effectively eliminated. At the same time, the inspection image data is regularized and optimized, improving the clarity and effectiveness of the inspection image data. This avoids interference from invalid and messy raw data on the target detection algorithm, significantly reducing the probability of false detection and false negative detection of the target detection algorithm, and significantly improving the accuracy and efficiency of defect target screening.

[0039] S102, based on the inspection image data, the trained target detection algorithm is used to filter and obtain image data of multiple defective targets.

[0040] Object detection algorithms include YOLO (You Only Look Once, an object detection algorithm) and Faster R-CNN (Region-based Convolutional Neural Network, another object detection algorithm), which are used to automatically and accurately locate and select potential defective objects in images.

[0041] S103, based on the image data of each defective target, the trained deep learning model is used to identify the defect, and the defect category and defect level of the corresponding defective target are obtained.

[0042] The deep learning model includes a deep convolutional neural network model and a multi-factor evaluation model, which are used to intelligently identify and evaluate the level of defects in power plants after being trained and optimized with large-scale power plant industry defect image data.

[0043] Deep convolutional neural network models are a type of deep learning model specifically designed for image data processing. They can automatically extract visual features (such as the shape, texture, color, and outline of defects) from defect images, enabling accurate determination of defect types. They are the core model for identifying defect categories.

[0044] For example, the convolutional neural network model provided in this application includes: a multi-level feature extraction layer, a spatial pyramid pooling enhancement layer, a path aggregation network bidirectional fusion layer, and a defect type classification and confidence output layer.

[0045] A multi-layered feature extraction layer is used to receive pre-processed defect target image data (size uniformly adjusted to...). The received defective target image data is fed into the input layer of the network, and based on the CSPDarknet53 (convolutional neural network model architecture) backbone network, the received defective target image data is output as feature maps of three different depths, including: Shallow features ( ): Preserves rich texture details and edge information, suitable for detecting minute defects (such as microcracks and pinholes).

[0046] Mid-layer features ( It balances semantic and spatial information and is suitable for detecting medium-sized defects.

[0047] Deep features ( It has the strongest semantic abstraction capability and is suitable for detecting large-sized or complex-shaped defects (such as large-area inclusions and severe deformation).

[0048] The CSPDarknet53 backbone network is composed of multiple Resblock_body residual modules stacked together, which can reduce the amount of computation while retaining gradient information by utilizing cross-stage partial connection structure.

[0049] The spatial pyramid pooling enhancement layer receives deep features output from multi-level feature extraction layers, processes them, and performs... , , The max pooling operation captures global contextual information under different receptive fields and concatenates the pooling result with the original feature map, thereby increasing the receptive field of neurons and enabling the model to distinguish the overall morphological features of defects, avoiding misjudging background noise as defects.

[0050] The path aggregation network's bidirectional fusion layer utilizes the constructed feature pyramid to deeply fuse deep semantic features with shallow detailed features through upsampling, downsampling, and channel concatenation operations. This generates three feature layers with rich semantics and precise localization (corresponding to...). , , (Resolution), ensuring that the model can identify both large defect types and accurately classify small defects.

[0051] The defect type classification and confidence output layer is used to receive the fused feature map and perform final decoding using three independent monitoring heads.

[0052] First, each detection head goes through several convolutional layers to predict the bounding box coordinates, object confidence, and defect category probability distribution within each grid cell, based on its corresponding feature scale.

[0053] For the preset Each defect type (e.g., cracks, porosity, scratches, inclusions, etc.) outputs a string of length [length missing]. probability vector , in This indicates that the defective target belongs to the first... The probability of a class of defects; The output probability vector is processed using the Softmax activation function to sum to 1, forming a standard probability distribution. The maximum value in the probability vector is then selected. and its corresponding category index Using a preset confidence threshold Evaluation: like Then the type of the defective target is determined to be the first. kind; like If the defect is identified as an "unknown defect" or background noise, it will not be output or will be marked as pending manual review.

[0054] The multi-factor assessment model is a quantitative assessment model built based on the actual needs of power plant operation and maintenance. It can comprehensively analyze multiple key factors such as the physical size of the defect, its location in the equipment, and the characteristics of the defect type to objectively assess the severity of the defect. It is the core model for determining the defect level.

[0055] The multi-factor evaluation model can be implemented as a random forest model trained using historical defect data with defect level labels calibrated by experts. By taking the physical size features, location features, and type features of the defect target acquired in real time as input parameters, it directly outputs the defect level label of the corresponding defect target.

[0056] The input parameters for the multi-factor evaluation model are: The physical size characteristics of the defective target, such as the actual physical area converted from the defective pixel area ( ), maximum chord length, aspect ratio, and perimeter; Location features, such as calculating the Euclidean distance from the defect center point to the stress concentration area of ​​the workpiece, or used to represent the regional coding of different risk areas of the workpiece; Type features: The defect type output by the convolutional neural network model is encoded using one-hot encoding or assigned a basic risk weight coefficient based on the severity of the defect (e.g., crack coefficient = 1.0, porosity coefficient = 0.6).

[0057] The output parameter of the multi-factor evaluation model is: the defect level label of the defect target.

[0058] Defect category refers to the standardized classification of defects based on the type characteristics of power station equipment defects. It is the recognition output of a deep convolutional neural network model and corresponds to the specific defect type of the power station equipment, such as hot spots on photovoltaic panels, spontaneous explosion of insulators, foreign objects in conductors, equipment cracks, and component damage.

[0059] Defect level refers to the degree of impact of a target defect on the safe operation of power plant equipment. The standardized classification of defects is the evaluation output of a multi-factor assessment model. It is usually classified into emergency, major, and general levels according to the degree of impact from high to low, which is used to facilitate the subsequent dispatch of operation and maintenance work orders.

[0060] Specifically, S103 above includes: Based on the image data of each defective target, a convolutional neural network model is used for identification to obtain the defect type of the corresponding defective target; Based on the physical size, location, and type characteristics of defects in the image data of each defective target, a multi-factor evaluation model is used to obtain the defect level of each defective target. The defect type and defect level of each defect target are associated, and the result is output as the defect category and defect level of the corresponding defect target.

[0061] By using a trained convolutional neural network model to extract feature information from defect images and determine the defect type of the corresponding defect target, we can achieve efficient and accurate determination of the defect type of power plant equipment and avoid the type classification bias caused by human subjective judgment. At the same time, by using a multi-factor evaluation model, we can conduct quantitative analysis on multiple influencing factors such as defect physical size, equipment location, and type characteristics to achieve scientific and objective evaluation of defect level. This ensures that the final defect level has a quantifiable basis for judgment and avoids the problem of misjudgment of level caused by single-dimensional evaluation, laying an accurate data foundation for subsequent processing of high-level defects.

[0062] S104: For each defect target with a defect level greater than the preset level, the system combines the skill tags, real-time geographical location and actual workload of each maintenance personnel, uses a task allocation algorithm to match the optimal maintenance personnel, and constructs corresponding maintenance work orders to send to the corresponding user terminals.

[0063] The preset level is a threshold for handling defects that the power plant sets in advance based on the equipment operation and maintenance management specifications and the degree of impact of defects on the safe operation of the equipment. It is usually the dividing point between "requiring immediate handling" and "routine inspection attention" in the defect level system (for example, if the "major" level is set as the preset level, defects with a level higher than this value are "urgent" and are targets that need to be dispatched for handling). It is the core basis for determining whether a defect needs to initiate operation and maintenance dispatch.

[0064] Skill tags refer to standardized capability identifiers established for operation and maintenance personnel that match their ability to handle defects in power plant equipment. These tags are set based on the professional skills, certifications, and work experience of operation and maintenance personnel (such as tags for "hot spot handling of photovoltaic panels," "insulator replacement," and "foreign object removal from conductors"). They are the core dimension for matching defect handling needs.

[0065] Real-time geolocation is the actual location information of the operations and maintenance personnel. It is collected in real time through user terminals and used to calculate the distance between the operations and maintenance personnel and the defect target collection location, ensuring the spatial rationality of work assignment and improving the response efficiency of defect handling.

[0066] Actual workload refers to the work status information such as the number of maintenance work orders currently undertaken by maintenance personnel, the progress of work order completion, and the duration of work. This information is used to avoid overloading maintenance personnel and to achieve a balanced allocation of maintenance resources.

[0067] The task allocation algorithm is a quantitative matching algorithm based on multi-dimensional dispatching factors. By assigning weights and quantifying the matching degree of skill tags, real-time geographical distance, and actual workload, it outputs the matching degree score between each maintenance personnel and the defect target, thereby achieving the scientific selection of the optimal maintenance personnel.

[0068] The maintenance work order is a standardized maintenance task document automatically constructed based on defect target information and work assignment results. It contains core information such as the category, level, collection location, and image data of the defect target, as well as the operation requirements and task time limits of the maintenance personnel. It is the core basis for maintenance personnel to carry out on-site operations.

[0069] User terminals are mobile work terminals (such as mobile phones, tablets, industrial PDAs, etc.) used by maintenance personnel. They support functions such as work order reception, on-site work recording, and maintenance data feedback. They serve as the carrier for work order dispatch and maintenance data interaction.

[0070] By automatically triggering and generating maintenance work orders based on preset rules (such as defect level being "urgent" or above), the work orders automatically associate key information such as defect location, images, and descriptions. Based on factors such as the skill tags of maintenance personnel, real-time geographical location, and current workload, the task allocation algorithm optimizes the matching and automatic dispatch of maintenance work orders, improving dispatch efficiency and rationality.

[0071] The power plant operation and maintenance method based on UAV inspection data provided in this embodiment, based on the inspection image data collected by UAVs, and relying on the trained target detection algorithm and deep learning model, achieves accurate positioning, classification, and level assessment of defect targets in power plant equipment, improving the efficiency and accuracy of defect target identification and ensuring the timely detection of early and subtle defects in power plant equipment. At the same time, by combining the skill tags, real-time geographical location, and actual workload of operation and maintenance personnel, the optimal operation and maintenance personnel are matched through a task allocation algorithm, and operation and maintenance work orders are automatically generated and dispatched, realizing the rational allocation of operation and maintenance resources, avoiding the problem of delayed defect handling due to unreasonable work assignment, and ensuring that operation and maintenance personnel can handle each defect target in a timely manner. This provides a reliable technical guarantee for the safe and stable operation of power plant equipment and overcomes the shortcomings of power plant operation and maintenance methods in related technologies that are difficult to meet the current operation and maintenance requirements of power plants.

[0072] This embodiment provides a power plant operation and maintenance method based on UAV inspection data, which can be used in power plant operation and maintenance server terminals. Figure 2 This is a flowchart of a power plant operation and maintenance method based on UAV inspection data according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps: S201, acquire inspection image data collected by the drone, wherein the inspection image data includes physical information of the image acquisition location. For details, please refer to [link to details]. Figure 1 S101 of the illustrated embodiment will not be described again here.

[0073] S202, based on the inspection image data, a trained target detection algorithm is used to filter and obtain image data of multiple defective targets. For details, please refer to [link to relevant documentation]. Figure 1S102 of the illustrated embodiment will not be described again here.

[0074] S203, based on the image data of each defective target, uses a trained deep learning model to identify the defect, obtaining the defect category and defect level of the corresponding defective target. For details, please refer to [link to relevant documentation]. Figure 1 S103 of the illustrated embodiment will not be described again here.

[0075] S204: For each defect target with a defect level exceeding a preset level, the system comprehensively considers each operations and maintenance (O&M) personnel's skill tags, real-time geographical location, and actual workload. Using a task allocation algorithm, it matches the optimal O&M personnel and generates corresponding O&M work orders, which are then sent to the corresponding user terminals. For details, please refer to [link to details]. Figure 1 S104 of the illustrated embodiment will not be described again here.

[0076] S205 receives maintenance data of defect targets uploaded by user terminals and updates the status of the corresponding maintenance work orders.

[0077] The maintenance data for defect targets are the full-dimensional on-site operation data related to the handling of the defect target, uploaded by operation and maintenance personnel through the user terminal after the defect handling is completed. This includes the defect handling process record, defect handling result information, on-site operation supporting images, spare parts replacement details, operation and maintenance personnel operation time and maintenance operation notes.

[0078] The status of an operation and maintenance work order refers to the standardized status identifiers at different stages throughout the entire lifecycle of the work order, from its creation to its final completion. These identifiers indicate the processing progress of the work order. The statuses of operation and maintenance work orders include: created, dispatched, in progress, completed, pending acceptance, and accepted. The core of updating the work order status in this step is to synchronize the "in progress" work order to the subsequent statuses such as "completed" or "pending acceptance" based on the maintenance data, thereby achieving real-time tracking of the work order progress.

[0079] Based on maintenance data uploaded in real time by operation and maintenance personnel through user terminals during the defect handling process, the corresponding work order process status is automatically updated, replacing the inefficient traditional method of manual reporting and manual work order updates. This effectively improves the timeliness and accuracy of work order status management, allowing operation and maintenance management personnel to keep track of the handling progress of each defect in real time. It also enables the effective correlation between defect data, work order data, and maintenance data, breaking down information silos in various stages of power plant operation and maintenance. This provides complete and continuous data support for subsequent traceability and analysis of operation and maintenance data and assessment of the health status of power plant equipment, further improving the digital closed-loop management of power plant operation and maintenance.

[0080] This embodiment provides a power plant operation and maintenance method based on UAV inspection data, which can be used in power plant operation and maintenance server terminals. Figure 3This is a flowchart of a power plant operation and maintenance method based on UAV inspection data according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps: S301, acquire inspection image data collected by the drone, wherein the inspection image data includes physical information of the image acquisition location. For details, please refer to [link to details]. Figure 1 S101 of the illustrated embodiment will not be described again here.

[0081] S302, based on the inspection image data, a trained target detection algorithm is used to filter and obtain image data of multiple defective targets. For details, please refer to [link to relevant documentation]. Figure 1 S102 of the illustrated embodiment will not be described again here.

[0082] S303, based on the image data of each defective target, uses a trained deep learning model to identify the defect, obtaining the defect category and defect level of the corresponding defective target. For details, please refer to [link to details]. Figure 1 S103 of the illustrated embodiment will not be described again here.

[0083] S304: For each defect target with a defect level exceeding a preset level, the system comprehensively considers each operations and maintenance (O&M) personnel's skill tags, real-time geographical location, and actual workload. Using a task allocation algorithm, it matches the optimal O&M personnel and generates corresponding O&M work orders, which are then sent to the corresponding user terminals. For details, please refer to [link to details]. Figure 1 S104 of the illustrated embodiment will not be described again here.

[0084] S305 receives maintenance data of defect targets uploaded by user terminals and updates the status of the corresponding maintenance work orders.

[0085] Specifically, the aforementioned S305 includes: S3051, Receive maintenance data of defective targets uploaded by user terminals; the maintenance data includes defect on-site handling process records, defect handling result information, on-site operation supporting images, spare parts replacement details, operation and maintenance personnel operation time and maintenance operation remarks. S3052, perform structured parsing and validity verification on the received maintenance data, classify and organize the compliant data according to preset dimensions, and establish a mapping between maintenance data and corresponding defect target association information; the defect target association information includes defect number, defect category, defect level, and defect target collection location; S3053, combining the maintenance data, the previous defect identification results, and the mapping between the maintenance data and the corresponding defect target association information, the on-site operation evidence images and the image data of the defect targets are compared and displayed, and the status of the maintenance work order is updated.

[0086] By performing structured parsing, validity verification, and precise mapping of maintenance data to defect target information, invalid and redundant data is effectively eliminated, ensuring a one-to-one correspondence between maintenance data and original defect data and operation and maintenance work orders. This avoids data chaos and misalignment, improving the accuracy of operation and maintenance data management. At the same time, by comparing and displaying on-site operation evidence images with original defect images, the effect of defect handling is presented intuitively, facilitating operation and maintenance management personnel to quickly verify the rectification status of defect targets. By combining the verification results with maintenance data to automatically update the work order status, the traditional manual verification and update mode is replaced, significantly improving the efficiency and accuracy of work order status management.

[0087] This embodiment provides a power plant operation and maintenance method based on UAV inspection data, which can be used in power plant operation and maintenance server terminals. Figure 4 This is a flowchart of a power plant operation and maintenance method based on UAV inspection data according to an embodiment of the present invention, such as... Figure 4 As shown, the process includes the following steps: S401, acquire inspection image data collected by the drone, wherein the inspection image data includes physical information of the image acquisition location. For details, please refer to [link to details]. Figure 1 S101 of the illustrated embodiment will not be described again here.

[0088] S402, based on the inspection image data, the trained target detection algorithm is used to filter and obtain image data of multiple defective targets. For details, please refer to [link to relevant documentation]. Figure 1 S102 of the illustrated embodiment will not be described again here.

[0089] S403: Based on the image data of each defective target, a trained deep learning model is used for identification to obtain the defect category and defect level of the corresponding defective target. For details, please refer to [link to details]. Figure 1 S103 of the illustrated embodiment will not be described again here.

[0090] S404, for each defect target with a defect level exceeding a preset level, combines each operations and maintenance (O&M) personnel's skill tags, real-time geographical location, and actual workload, using a task allocation algorithm to match the optimal O&M personnel, and generates corresponding O&M work orders which are then sent to the corresponding user terminals. For details, please refer to [link to details]. Figure 1 S104 of the illustrated embodiment will not be described again here.

[0091] S405, in response to the report generation command provided by the user, uses the trained feature extraction model to fill in the data based on the category and level of each target defect, combined with the preset report framework, to obtain the inspection report of the target power plant.

[0092] By relying on the trained feature extraction model and combining it with the preset report framework, the system can accurately fill in the data based on the identification results such as the category and level of the target defect. It can directly generate reports based on the structured data of defect identification, realizing the seamless connection between inspection data and defect identification results and report output, and providing timely and reliable data support for the decision-making of power plant operation and maintenance management personnel.

[0093] This embodiment also provides a power plant operation and maintenance system based on UAV inspection data. This system is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that performs a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0094] Reference Figure 5 This embodiment provides a power plant operation and maintenance system based on UAV inspection data, the system comprising: The image acquisition module 510 is used to acquire inspection image data collected by the UAV, and the inspection image data includes physical information of the acquisition location; The defect screening module 520 is used to screen multiple defect targets based on the inspection image data using a trained target detection algorithm. The defect identification module 530 is used to identify each defect target based on its image data using a trained deep learning model, thereby obtaining the defect category and defect level of the corresponding defect target. The task allocation module 540 is used to match the optimal maintenance personnel for each defect target with a defect level greater than the preset level by combining each maintenance personnel's skill tags, real-time geographical location and actual workload, and to generate corresponding maintenance work orders and send them to the corresponding user terminals.

[0095] In some alternative implementations, the defect identification module 530 includes: The type recognition unit is used to identify the defect type of each defect target based on the image data of each defect target using a convolutional neural network model. The rating assessment unit is used to obtain the defect rating of each defective target based on the physical size, location and type characteristics of the defects in the image data of each defective target, using a multi-factor assessment model. The result output unit is used to establish a correlation between the defect type and defect level of each defect target, and output the defect category and defect level of the corresponding defect target.

[0096] In some alternative implementations, it also includes: The work order update module is used to receive maintenance data of defect targets uploaded by user terminals and update the status of the corresponding maintenance work orders.

[0097] In some optional implementations, the work order update module includes: The data receiving unit is used to receive maintenance data of defective targets uploaded by user terminals; the maintenance data includes records of on-site defect handling process, defect handling result information, on-site operation evidence images, spare parts replacement details, operation and maintenance personnel's operation time and maintenance operation remarks. The data mapping unit is used to perform structured parsing and validity verification on the received maintenance data, classify and organize the compliant data according to preset dimensions, and establish a mapping between maintenance data and corresponding defect target association information; the defect target association information includes defect number, defect category, defect level, and defect target collection location; The work order update unit is used to combine the maintenance data, the previous defect identification results, and the mapping between the maintenance data and the corresponding defect target association information to compare and display the on-site operation evidence images and the image data of the defect targets, and to continue to update the status of the maintenance work order.

[0098] In some alternative implementations, it also includes: The report generation module is used to respond to the report generation command provided by the user, and based on the category and level of the target defect, combined with the preset report framework, use the trained feature extraction model to fill in the data and obtain the inspection report of the target power station.

[0099] In some alternative implementations, it also includes: The data preprocessing module is used to process the inspection image data collected by the UAV using data preprocessing methods to obtain preprocessed inspection image data; the preprocessed inspection image data is used to screen defective targets through target detection algorithms.

[0100] In some optional implementations, the data preprocessing method of the data preprocessing module includes: Image denoising and enhancement, data integrity verification, data cleaning, or data format standardization.

[0101] The power plant operation and maintenance system based on UAV inspection data provided in this embodiment of the invention can execute the power plant operation and maintenance method based on UAV inspection data provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the above modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0102] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0103] The following is a detailed reference. Figure 6 This diagram illustrates a suitable structural design for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 601, which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) 602 or a program loaded from memory 608 into random access memory (RAM) 603. RAM 603 also stores various programs and data required for the operation of the electronic device. The processor 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0104] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0105] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a memory 608, or installed from a ROM 602. When the computer program is executed by the processor 601, it performs the functions defined in the power plant operation and maintenance method based on UAV inspection data according to embodiments of the present invention.

[0106] Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0107] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the power plant operation and maintenance method based on UAV inspection data shown in the above embodiments is implemented.

[0108] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0109] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A power plant operation and maintenance method based on UAV inspection data, characterized in that, The method includes: Acquire inspection image data collected by the UAV, wherein the inspection image data includes physical information of the image acquisition location; Based on the inspection image data, the trained target detection algorithm is used to filter and obtain image data of multiple defective targets. Based on the image data of each defective target, the trained deep learning model is used to identify the defect, and the defect category and defect level of the corresponding defective target are obtained. For each defect target whose defect level is higher than the preset level, the optimal maintenance personnel are matched using a task allocation algorithm by taking into account each maintenance personnel's skill tags, real-time geographical location and actual workload, and corresponding maintenance work orders are generated and sent to the corresponding user terminals.

2. The method according to claim 1, characterized in that, The deep learning model includes a deep convolutional neural network and a multi-factor evaluation model. Based on the image data of each defective target, the trained deep learning model is used for identification to obtain the defect category and defect level of the corresponding defective target, including: Based on the image data of each defective target, a convolutional neural network model is used for identification to obtain the defect type of the corresponding defective target; Based on the physical size, location, and type characteristics of defects in the image data of each defective target, a multi-factor evaluation model is used to obtain the defect level of each defective target. The defect type and defect level of each defect target are associated, and the result is output as the defect category and defect level of the corresponding defect target.

3. The method according to claim 1, characterized in that, Also includes: Receive maintenance data of defect targets uploaded by user terminals and update the status of the corresponding maintenance work orders.

4. The method according to claim 3, characterized in that, The step of receiving maintenance data of defective targets uploaded by user terminals and updating the status of corresponding maintenance work orders includes: Receive maintenance data of defective targets uploaded by user terminals; the maintenance data includes records of on-site defect handling process, defect handling results information, on-site operation evidence images, spare parts replacement details, operation and maintenance personnel's operation time and maintenance operation remarks; The received maintenance data is structured and validated, and the compliant data is classified and organized according to preset dimensions to establish a mapping between maintenance data and corresponding defect target association information; the defect target association information includes defect number, defect category, defect level and defect target collection location; By combining the maintenance data, the results of previous defect identification, and the mapping between maintenance data and the corresponding defect target information, the on-site operation evidence images and the image data of the defect targets are compared and displayed, and the status of the maintenance work order is continuously updated.

5. The method according to claim 1, characterized in that, Also includes: In response to the report generation command provided by the user, based on the category and level of each target defect, and combined with the preset report framework, the system uses a trained feature extraction model to fill in the data and obtain the inspection report of the target power plant.

6. The method according to claim 1, characterized in that, Also includes: Based on the inspection image data collected by UAV, the data is processed using data preprocessing methods to obtain preprocessed inspection image data. The preprocessed inspection image data is used to screen defective targets through target detection algorithms.

7. The method according to claim 6, characterized in that, The data preprocessing method includes: Image denoising and enhancement, data integrity verification, data cleaning, or data format standardization.

8. A power plant operation and maintenance system based on UAV inspection data, characterized in that, The system includes: The image acquisition module is used to acquire inspection image data collected by the UAV, and the inspection image data includes physical information of the acquisition location; The defect screening module is used to screen multiple defect targets based on the inspection image data using a trained target detection algorithm. The defect identification module is used to identify each defective target based on its image data using a trained deep learning model, thereby obtaining the defect category and defect level of the corresponding defective target. The task allocation module is used to match the optimal maintenance personnel for each defect target with a defect level greater than the preset level by taking into account each maintenance personnel's skill tags, real-time geographical location and actual workload, and to generate corresponding maintenance work orders and send them to the corresponding user terminals.

9. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the power plant operation and maintenance method based on UAV inspection data as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, which are used to cause the computer to execute the power plant operation and maintenance method based on UAV inspection data as described in any one of claims 1 to 7.

Images