A method, device and system for unmanned operation and maintenance of a photovoltaic power station

Abstract

The application discloses an unmanned operation and maintenance method, device and system for photovoltaic power stations, and particularly relates to the technical field of automatic control, and is used for solving the problem of collaborative failure of the existing unmanned operation and maintenance system for photovoltaic power stations due to the conflict between centralized optimization and local autonomous decision-making of equipment; the global resource graph is constructed synchronously by acquiring the resource request instruction sequence in real time, the resource request instruction sequence is matched with the global resource graph to determine whether there is a resource allocation circular waiting chain, when there is, the execution path intersection density of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain on the physical space and the logical sequence is quantitatively evaluated, the interference effect is analyzed based on the execution path intersection density and the behavior randomness characteristics of each heterogeneous operation and maintenance equipment in path selection to trace back the root cause of the conflict, the key conflict task is determined based on the root cause of the conflict, and the execution equipment assignment is dynamically reconstructed to remove the resource allocation circular waiting chain.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, and in particular to a method, device and system for unmanned operation and maintenance of photovoltaic power plants. Background Technology

[0002] Currently, the level of automation in the operation and maintenance of photovoltaic power plants is gradually improving. Centralized control and dispatching systems can be used to manage tasks for heterogeneous equipment within the plant. The aim is to achieve unified dispatch of operation and maintenance tasks and equipment collaboration through a central system, thereby improving operation and maintenance efficiency. At the same time, to enhance the system's responsiveness to changes in the local environment, some equipment has also been given a certain degree of autonomous decision-making authority, such as planning detour paths when encountering obstacles. In existing technologies, centralized global optimization and equipment local autonomous decision-making are two typical control modes that are both applied in unmanned operation and maintenance systems for photovoltaic power plants.

[0003] However, when existing technologies combine centralized optimization with simple local autonomy, the two control modes have inherent differences in decision-making objectives and timeliness, which can easily lead to objective conflicts within the system. In dynamic, multi-task parallel operation and maintenance scenarios, such conflicts can cause system-level collaborative failures, manifested as repeated oscillations in task execution or deadlock in resource allocation, which seriously restricts the overall efficiency and operational stability of unmanned operation and maintenance systems for photovoltaic power plants. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing a method, device, and system for unmanned operation and maintenance of photovoltaic power plants.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] A method for unmanned operation and maintenance of photovoltaic power plants includes:

[0007] S1. Real-time acquisition of resource request command sequences issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station, and simultaneous construction of a global resource map reflecting the real-time occupancy status of shared resources across the entire station;

[0008] S2. Match the resource request instruction sequence with the global resource graph to determine if there is a resource allocation loop waiting chain that causes the task to stall;

[0009] S3. When a resource allocation circular waiting chain exists, quantitatively evaluate the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in the physical space and logical sequence.

[0010] S4. Based on the intersection density of execution paths, and combined with the randomness of the behavior of each heterogeneous operation and maintenance device in path selection, analyze the interference effect between the two to trace the root cause of the conflict.

[0011] S5. Key conflict tasks that lead to the formation of a resource allocation cycle waiting chain based on the determination of the root cause of the conflict;

[0012] S6. Based on the determination results of critical conflict tasks, dynamically reconstruct the execution equipment assignment of critical conflict tasks to eliminate the resource allocation circular waiting chain.

[0013] Furthermore, the system acquires in real-time the sequence of resource request commands issued by various heterogeneous operation and maintenance devices within the photovoltaic power station, and simultaneously constructs a global resource map reflecting the real-time occupancy status of shared resources across the entire station, including:

[0014] Real-time acquisition of resource request commands issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station during task execution, and organization into a resource request command sequence according to the time sequence;

[0015] Simultaneously monitor the real-time occupancy status of shared resources across the entire site, and construct a global resource map represented in a graph structure based on the relationship between resource type and physical location;

[0016] In the global resource graph, nodes correspond to shared resources, and edges correspond to the occupancy or dependency relationships between resources.

[0017] Furthermore, the resource request instruction sequence is matched against the global resource graph to determine whether there is a resource allocation circular wait chain that causes the task to stall, including:

[0018] Parse the resource request command sequence to extract the request information of each heterogeneous operation and maintenance device for shared resources;

[0019] The extracted request information is compared with the current occupancy status of resources in the global resource map;

[0020] Based on the comparison results, search the global resource graph for a closed-loop path formed by resource requests from multiple heterogeneous operation and maintenance devices.

[0021] When a closed-loop path exists, it is determined that a resource allocation circular waiting chain exists.

[0022] Furthermore, when a resource allocation circular waiting chain exists, the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in physical space and logical sequence is quantitatively evaluated, including:

[0023] Based on the historical execution trajectory of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain, spatial conflict hotspots formed in the physical space movement path are identified. Spatial conflict hotspots are determined by calculating the crossover frequency and average dwell time of different task paths within a preset time period.

[0024] Synchronously analyze the operation flow of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in the logical sequence, identify the logical competition bottleneck nodes formed during the execution of the logical sequence, and determine the logical competition bottleneck nodes by statistically analyzing the request order and waiting time of different tasks for the same shared resources.

[0025] By integrating the distribution characteristics of spatial conflict hotspots with the occurrence patterns of logical competition bottleneck nodes, a multi-dimensional quantitative evaluation result of the intersection density of execution paths is generated.

[0026] Furthermore, based on the intersection density of execution paths and the randomness of the behavior exhibited by various heterogeneous operation and maintenance devices in path selection, the interference effect between the two is analyzed to trace the root cause of the conflict, including:

[0027] Identify high-density conflict regions in physical space and logical sequence based on execution path intersection density;

[0028] Extract the stochastic characteristics of the behavior of each heterogeneous operation and maintenance device in the path selection process, including the decision difference of the device in choosing an alternative path when encountering path obstacles and the path selection variation coefficient of the device when repeatedly performing the same task;

[0029] Spatiotemporal correlation mapping is performed between high-density conflict areas and behavioral randomness characteristics to establish a distribution map of random equipment behavior in conflict areas;

[0030] By analyzing the spatial overlap and logical coupling between high-density conflict regions and devices with highly random behavior in the distribution map of behavioral randomness, the systemic interference regions that lead to the formation of resource allocation cycle waiting chains can be identified.

[0031] Based on the distribution characteristics and intensity of the systemic interference region, the structural contradiction between equipment autonomous decision-making and system resource allocation is traced back to the root cause of the conflict.

[0032] Furthermore, based on the determination of the root causes of the conflict, the key conflict tasks leading to the formation of a resource allocation cycle waiting chain include:

[0033] Based on the distribution characteristics of systematic interference regions in the root causes of conflict, candidate conflict tasks that are simultaneously located in high interference intensity regions in both physical space and logical sequence are identified.

[0034] Analyze the topological position of each candidate conflicting task in the resource allocation loop waiting chain and its contribution to loop formation;

[0035] By combining the randomness of equipment behavior and the intensity of task dependence of candidate conflict tasks, the conflict intensification index of each candidate conflict task is calculated.

[0036] Candidate conflict tasks with the highest conflict escalation index and the greatest contribution to the formation of the resource allocation cycle waiting chain are identified as critical conflict tasks.

[0037] Furthermore, the contribution is determined by the amount of resources held by the candidate conflicting task that are waiting for by other tasks and its hub status in the loop.

[0038] Furthermore, based on the determination results of critical conflict tasks, the execution equipment assignments for critical conflict tasks are dynamically restructured to eliminate resource allocation circular waiting chains, including:

[0039] Based on the determination results of critical conflict tasks, identify the original heterogeneous operation and maintenance equipment currently executing critical conflict tasks and its resource occupancy status in the resource allocation loop waiting chain;

[0040] Based on the real-time workload and task execution capability characteristics of the available heterogeneous operation and maintenance equipment group in the photovoltaic power plant, a candidate set of alternative heterogeneous operation and maintenance equipment that is qualified to perform critical conflict tasks is selected.

[0041] Evaluate the potential path intersection density and behavioral randomness interference risk of each device in the candidate set of alternative heterogeneous operation and maintenance equipment when performing critical conflicting tasks and other operation and maintenance tasks in the resource allocation circular waiting chain;

[0042] Select the alternative heterogeneous operation and maintenance equipment with the lowest potential path intersection density and the lowest risk of interference from behavioral randomness, and dynamically reassign critical conflict tasks from the original heterogeneous operation and maintenance equipment to the selected alternative heterogeneous operation and maintenance equipment.

[0043] After the execution equipment that has completed the critical conflict task is reassigned, verify whether the resource allocation circular wait chain has been resolved.

[0044] On the other hand, the present invention provides an unmanned operation and maintenance system for photovoltaic power plants, comprising:

[0045] The information determination module is used to acquire the sequence of resource request instructions issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station in real time, and simultaneously construct a global resource map reflecting the real-time occupancy status of shared resources across the entire station.

[0046] The loop determination module is used to match the sequence of resource request instructions with the global resource graph to determine whether there is a resource allocation loop waiting chain that causes the task to stall.

[0047] The density quantification module is used to quantify and evaluate the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in the physical space and logical sequence when a resource allocation circular waiting chain exists.

[0048] The root cause tracing module is used to analyze the interference effect between execution path intersection density and the randomness of the behavior of various heterogeneous operation and maintenance devices in path selection in order to trace the root cause of the conflict.

[0049] The conflict determination module is used to determine the key conflict tasks that lead to the formation of a resource allocation loop waiting chain based on the root cause of the conflict.

[0050] The loop release module is used to dynamically reconstruct the execution device assignment of critical conflict tasks based on the judgment results of critical conflict tasks, so as to release the resource allocation loop wait chain.

[0051] On the other hand, the present invention provides an unmanned operation and maintenance device for a photovoltaic power plant. The device includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, an unmanned operation and maintenance method for a photovoltaic power plant is implemented.

[0052] The beneficial effects of this invention are:

[0053] 1. By acquiring the resource request command sequence issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station in real time, and simultaneously constructing a global resource map reflecting the real-time occupancy status of shared resources across the entire station, an efficient system status monitoring and coordination mechanism is established. This enables the central control system to fully perceive the changing trends of equipment resource demands and perform dynamic analysis in conjunction with the global resource status, thereby effectively capturing potential conflicts in resource allocation. By accurately matching the resource request command sequence with the global resource map, it can quickly determine whether there is a resource allocation loop waiting chain that causes task stagnation, and provide early warnings and interventions before problems occur. This significantly reduces the risk of collaborative failure caused by inconsistencies between equipment autonomous decision-making and centralized scheduling objectives, ensuring the coordination and continuity of various operation and maintenance tasks during parallel execution. At the same time, through real-time data synchronization and closed-loop feedback, the system's adaptability to dynamic environments is improved, reducing oscillations and delays during task execution, thereby optimizing the response efficiency and stability of the overall operation and maintenance process.

[0054] 2. By quantitatively evaluating the intersection density of execution paths of heterogeneous operation and maintenance tasks in the resource allocation circular waiting chain in both physical space and logical sequence, and combining this with the randomness of the behavior exhibited by the equipment in path selection, the root causes of conflicts were analyzed in depth. This enabled the accurate identification of key conflicting tasks, and the dynamic reconstruction of task assignment strategies based on the root causes of conflicts. By intelligently reassigning key conflicting tasks from the original equipment to more suitable alternative equipment, the system effectively eliminated the resource allocation circular waiting chain and deadlock. This not only resolved immediate conflicts but also prevented the recurrence of similar problems by optimizing resource allocation. It significantly improved the overall coordination, reliability, and long-term operating efficiency of the unmanned operation and maintenance system for photovoltaic power plants, while reducing dependence on external intervention and enhancing the system's adaptability in complex and ever-changing environments. Attached Figure Description

[0055] Figure 1 This is a flowchart of an unmanned operation and maintenance method for a photovoltaic power station according to the present invention;

[0056] Figure 2 This is a schematic diagram of the structure of an unmanned operation and maintenance system for a photovoltaic power station according to the present invention. Detailed Implementation

[0057] 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, and 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.

[0058] Example 1: Figure 1 This invention provides a method for unmanned operation and maintenance of photovoltaic power plants, comprising:

[0059] S1. Real-time acquisition of resource request command sequences issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station, and simultaneous construction of a global resource map reflecting the real-time occupancy status of shared resources across the entire station;

[0060] S2. Match the resource request instruction sequence with the global resource graph to determine if there is a resource allocation loop waiting chain that causes the task to stall;

[0061] S3. When a resource allocation circular waiting chain exists, quantitatively evaluate the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in the physical space and logical sequence.

[0062] S4. Based on the intersection density of execution paths, and combined with the randomness of the behavior of each heterogeneous operation and maintenance device in path selection, analyze the interference effect between the two to trace the root cause of the conflict.

[0063] S5. Key conflict tasks that lead to the formation of a resource allocation cycle waiting chain based on the determination of the root cause of the conflict;

[0064] S6. Based on the determination results of critical conflict tasks, dynamically reconstruct the execution equipment assignment of critical conflict tasks to eliminate the resource allocation circular waiting chain.

[0065] S1. Real-time acquisition of resource request command sequences issued by various heterogeneous operation and maintenance devices within the photovoltaic power station, and simultaneous construction of a global resource map reflecting the real-time occupancy status of shared resources across the entire station. Specifically, this is implemented as follows:

[0066] The system collects resource request commands issued by various heterogeneous operation and maintenance (O&M) devices within the photovoltaic (PV) power plant during task execution in real time, and organizes them into a chronological sequence. Specifically, this is achieved through a real-time communication link established between the PV power plant's central control system and each heterogeneous O&M device. This continuously receives request signals from each device regarding shared resources during O&M tasks. These request signals include a device identifier, a resource type identifier, and a request timestamp. The device identifier uniquely identifies the requesting heterogeneous O&M device, the resource type identifier specifies the type of shared resource requested (e.g., charging pile or tool storage point), and the request timestamp is generated based on the system's global clock to ensure timing accuracy. Upon receiving the request signal... After the initial call, the central control system sorts these signals according to the order of the request timestamps and stores them in the instruction buffer, forming a resource request instruction sequence. The instruction buffer is managed using a first-in-first-out queue data structure to ensure the timing consistency of the instructions. At the same time, the system periodically checks the integrity of the instruction sequence by verifying the continuity of timestamps and device status reports to avoid data loss or duplication. For example, it detects anomalies by comparing whether the difference between adjacent timestamps is within a preset time interval threshold. The preset time interval threshold is set according to the system sampling frequency. For example, if the system samples once per second, the preset time interval threshold is set to 1 second, thereby ensuring that the resource request instruction sequence can truly reflect the changes in resource requirements of various heterogeneous maintenance devices in a dynamic task environment.

[0067] The system synchronously monitors the real-time occupancy status of all shared resources across the entire station and constructs a global resource map represented by a graph structure based on the association between resource type and physical location. Specifically, this is achieved through a sensor network and resource status monitoring modules deployed within the photovoltaic power station. This data is used to collect real-time occupancy information for each shared resource, including a resource identifier, occupancy status, occupying device identifier, and physical coordinates. The resource identifier uniquely identifies each shared resource, the occupancy status indicates whether the resource is occupied or idle, the occupying device identifier indicates the device currently occupying the resource, and the physical coordinates are obtained through a global positioning system or an in-station positioning system. After acquiring the occupancy information, the system constructs a global resource map based on the association between resource type and physical location. The map uses a graph data structure, with nodes corresponding to shared resources. Each node stores a resource identifier, resource type, and physical coordinates. The system considers real-time occupancy status and the corresponding occupancy or dependency relationships between resources. Occupancy relationships represent the connections formed when a resource is occupied by a specific device, while dependency relationships represent the logical associations between resources during task execution, such as the sequential dependency of adjacent resource points in a device's movement path. When constructing the graph, the system first initializes the node set, classifying resources based on their type, such as charging piles or tool storage points. Then, it calculates the connecting edges between resources based on a physical location distance threshold. This physical location distance threshold is set by analyzing the layout of the photovoltaic power station and historical device movement path data. For example, it is calculated based on the average interaction distance between resources plus a standard deviation, with a typical value set at 5 meters, thus forming edge connections. Simultaneously, the system dynamically updates the graph, adjusting the edge relationships in real time when resource occupancy status changes to ensure that the global resource graph accurately reflects the real-time occupancy status and spatial logic of shared resources across the entire station.

[0068] In the global resource graph, nodes correspond to shared resources, and edges correspond to the occupancy or dependency relationships between resources. Specifically, during graph construction, node attributes include a unique resource identifier, a resource type enumeration value, physical coordinates, and a real-time status flag. The resource type enumeration value defines the category of the shared resource, such as a mobile charging station or a fixed tool station. The physical coordinates are used for spatial positioning, and the real-time status flag indicates whether the resource is available. Edge attributes include edge type and weight value. Edge types are divided into occupied edges and dependent edges. Occupied edges represent the relationship of a resource being occupied by a device, while dependent edges represent the sequential relationship of resources in the task flow. The weight value is used to quantify the strength of the relationship. For example, the weight of a dependent edge is calculated based on historical task execution data, set by statistically analyzing the frequency of dependencies between resources and normalizing it to a range of 0 to 1. The normalization method uses minimum-maximum scaling, where the minimum and maximum frequencies are extracted from historical data, thus providing a quantitative basis for subsequent analysis. The system maintains the connectivity and consistency of the graph by traversing nodes and edges. For example, when a change in resource occupancy status is detected, the relevant edges are automatically updated to ensure that the graph is always synchronized with the actual resource status of the photovoltaic power station, thereby providing accurate basic data for subsequent resource allocation analysis.

[0069] S2. Match the resource request instruction sequence with the global resource graph to determine if there is a resource allocation circular wait chain that causes the task to stall. Specifically, this is implemented as follows:

[0070] The system parses resource request command sequences to extract request information for shared resources from various heterogeneous maintenance devices. Specifically, it reads command entries one by one from the sequence. Each command entry contains a device identifier and a resource type identifier. The device identifier uniquely identifies the requesting heterogeneous maintenance device, while the resource type identifier specifies the type of shared resource requested, such as a charging pile or tool storage point. The parsing process iterates through all elements in the resource request command sequence, extracting the device identifier and resource type identifier from each element. This information is then organized into a request information list, where each record contains a correspondence between the device identifier and the resource type identifier. Simultaneously, the system performs data integrity checks, such as verifying whether the device identifier exists in a predefined device registry and whether the resource type identifier belongs to a predefined resource type enumeration value. This ensures that the extracted request information accurately reflects the real-time resource needs of each heterogeneous maintenance device.

[0071] The extracted request information is compared with the current occupancy status of resources in the global resource graph. Specifically, this involves obtaining the real-time occupancy status information of each shared resource node from the global resource graph, including whether the resource is occupied and the identifier of the occupying device. The comparison process involves traversing each request record in the request information list, searching for the node with the corresponding resource type identifier in the global resource graph, and comparing the occupancy status of that node with the device identifier in the request record. For example, if a request record shows that device A is requesting resource X, but the global resource graph shows that resource X is currently occupied by device B, a resource conflict event is recorded. The system uses an iterative comparison algorithm to process all request records, ensuring that each request matches the current status of the global resource graph, and recording the comparison results, including the number of conflicts, the conflicting device, and resource details, thus providing complete input data for subsequent search steps.

[0072] Based on the comparison results, the system searches the global resource graph for closed-loop paths formed by resource requests from multiple heterogeneous operation and maintenance devices. Specifically, it constructs a temporary dependency graph based on the conflict events recorded in the comparison results. Nodes in the temporary dependency graph represent heterogeneous operation and maintenance devices, and edges represent resource waiting relationships between devices. For example, if device A requests resource X but resource X is occupied by device B, an edge from device A to device B is added to the temporary dependency graph. When searching for closed-loop paths, the system uses a graph traversal method, starting from each device node in the temporary dependency graph and performing a depth-first traversal along the edge directions to check for loops. The system checks if a path exists that returns to the starting node, thus forming a closed loop. During traversal, the system maintains an access status marker array to track visited nodes, avoiding repeated traversals. A maximum search depth threshold is set to limit the traversal range. The maximum search depth threshold is set through statistical analysis based on the total number of devices in the photovoltaic power plant and the average length of historical task dependency chains. For example, it is obtained by calculating the average length of historical task dependency chains plus twice the standard deviation, with a typical value set to 10 layers. If a closed-loop path is found, the system records the path's detailed information, including the sequence of devices involved and resource dependencies, thus completing the search process.

[0073] When a closed-loop path exists, a resource allocation circular waiting chain is identified. This is achieved by verifying, based on the searched closed-loop path details, whether the path satisfies the circular waiting condition: each device in the path is waiting for resources occupied by the next device, and the first and last devices in the path are connected to form a closed loop. The determination process checks whether the resource waiting relationships of each edge in the closed-loop path are consistent. For example, if device A waits for device B, device B waits for device C, and device C waits for device A, it confirms that the resource occupancy relationship forms an uninterrupted cycle. The system outputs the determination result, including the identifier of the resource allocation circular waiting chain and a list of involved devices, and triggers subsequent processing mechanisms, such as recording the determination result in the system log or sending it to the monitoring interface, thereby ensuring that the system can promptly identify and respond to resource allocation problems.

[0074] S3. When a resource allocation circular waiting chain exists, quantitatively evaluate the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in both physical space and logical sequence. Specifically, this is implemented as follows:

[0075] Based on the historical execution trajectories of heterogeneous operation and maintenance tasks in the resource allocation cyclic waiting chain, spatial conflict hotspots formed in the physical space movement paths are identified. Specifically, this involves extracting historical execution trajectory data of each heterogeneous operation and maintenance task involved in the resource allocation cyclic waiting chain from the system database within a preset time period. The historical execution trajectory data records the sequence of movement path points for each task in the physical space, with each path point containing a timestamp and spatial coordinates. The identification process first divides the physical space into multiple grid cells. The size of the grid cell is set through statistical analysis based on the photovoltaic power station layout and the typical equipment movement range. For example, the grid size is determined by calculating the coverage area of ​​the equipment movement paths and the safety distance requirements between equipment; a typical value is set to 2 meters by 2 meters. Then, the intersection frequency and average dwell time of different task paths within the preset time period are calculated. The intersection frequency is obtained by counting the number of times different task paths intersect within each grid cell; for example, this is done within a 1-hour preset time period. The number of intersections between the paths of Task A and Task B within a specific grid cell, and the average dwell time are obtained by calculating the average dwell time of each task within the grid cell. The dwell time is calculated by accumulating the timestamp differences of consecutive trajectory points. The preset time period is set through historical data analysis based on the task execution cycle and system sampling frequency, for example, by analyzing the distribution of task execution time to determine typical statistical time periods. Spatial conflict hotspots are identified by setting intersection frequency thresholds and average dwell time thresholds. The intersection frequency threshold is set based on historical conflict data statistics, for example, taking the 75th percentile of the intersection frequency values ​​of all grid cells. The average dwell time threshold is set based on the normal operating time of the equipment through historical data analysis, for example, taking the average of historical dwell time data plus 1 standard deviation. The system traverses all grid cells. When the intersection frequency and average dwell time of a certain cell both exceed the corresponding thresholds, the cell is marked as a spatial conflict hotspot, and the area coordinates and conflict intensity index are recorded.

[0076] This analysis synchronously examines the operational flow of heterogeneous operation and maintenance tasks within the resource allocation circular waiting chain, identifying logical competition bottlenecks that arise during execution. Specifically, this involves extracting the logical operation sequences of each heterogeneous operation and maintenance task from the resource request instruction sequence and the global resource graph. These logical operation sequences record the request order and execution steps of each task for shared resources. The analysis first constructs a task-resource dependency graph, where nodes represent shared resources and edges represent the request relationships between tasks. Then, it statistically analyzes the request order and waiting time of different tasks for the same shared resource. The request order is determined by comparing the timestamps of the task resource requests; for example, if task A requests resource X earlier than task B, the order is recorded. The waiting time is obtained by calculating the time interval from when a task issues a resource request to when it receives a resource response. The time interval is extracted from the resource request instruction sequence and resource status log. Logical contention bottleneck nodes are identified by setting request order conflict thresholds and waiting time thresholds. The request order conflict threshold is set based on the frequency of resource requests through statistical analysis, such as statistically analyzing the distribution of the number of times the same resource is requested consecutively by multiple tasks, and taking the 80th percentile as the threshold. The waiting time threshold is set based on the normal response time of the task through historical data analysis, such as taking the 90th percentile of historical waiting times. The system traverses all shared resource nodes. When the request order conflict and waiting time of a resource both exceed the corresponding thresholds, the resource node is marked as a logical contention bottleneck node, and the node identifier and contention intensity index are recorded.

[0077] This method integrates the distribution characteristics of spatial conflict hotspots with the occurrence patterns of logical competition bottleneck nodes to generate a multi-dimensional quantitative evaluation result of execution path intersection density. Specifically, it maps the distribution characteristics of spatial conflict hotspots to the occurrence patterns of logical competition bottleneck nodes. Distribution characteristics include the number, location coordinates, and conflict intensity of spatial conflict hotspots, while occurrence patterns include the number, resource type, and competition intensity of logical competition bottleneck nodes. The integration process first aligns the spatial conflict hotspots and logical competition bottleneck nodes spatiotemporally, matching spatial regions with logical nodes based on the correspondence between resource physical coordinates and logical resource identifiers. Then, it calculates the execution path intersection density. The system generates a comprehensive assessment of spatial conflict intensity and logical competition intensity, for example, using a weighted summation method. The weights for spatial conflict intensity and logical competition intensity are set based on historical conflict impact analysis. The weight values ​​are allocated by statistically analyzing the contribution of spatial and logical conflicts to task delays, with a typical weight ratio set at 0.6 to 0.4. The multidimensional quantitative assessment results include numerical scores and level classifications. The numerical scores are obtained through normalization, with the normalization range set from 0 to 100. The level classifications are based on the scores, for example, 0 to 30 is low density, 31 to 70 is medium density, and 71 to 100 is high density. The system outputs assessment results, including the intersection density value of each resource allocation cycle waiting chain and related metadata, thus providing a quantitative basis for subsequent conflict analysis.

[0078] S4. Based on the intersection density of execution paths, and combined with the randomness of the behavior of various heterogeneous operation and maintenance devices in path selection, analyze the interference effect between the two to trace the root cause of the conflict. The specific implementation is as follows:

[0079] The system identifies high-density conflict areas in physical space and logical sequences based on execution path intersection density. Specifically, this involves extracting numerical scores and classification information from the multidimensional quantitative evaluation results of execution path intersection density. These results are generated by the aforementioned steps and include spatial conflict intensity and logical competition intensity indicators. The identification process first sets a high-density conflict area judgment threshold. This threshold is set statistically based on historical conflict data analysis, for example, by calculating the 80th percentile of the distribution of execution path intersection density numerical scores as the threshold benchmark. This threshold is then adjusted according to the actual operational needs of the photovoltaic power station, with the adjustment range dynamically determined based on task priority and the number of devices. The system traverses all evaluation areas. When the execution path intersection density numerical score of a certain area exceeds the high-density conflict area judgment threshold and is classified as high-density, the area is marked as a high-density conflict area, and its spatial coordinates and logical resource identifier are recorded. Simultaneously, the system distinguishes between physical space high-density conflict areas and logical sequence high-density conflict areas. Physical space high-density conflict areas are located using a geographic information system based on spatial coordinate information, while logical sequence high-density conflict areas are located using a resource dependency graph based on logical resource identifiers, thus completing the identification of high-density conflict areas.

[0080] This study extracts the stochastic characteristics of heterogeneous operation and maintenance equipment during path selection, including the decision-making variability when equipment encounters obstacles and the path selection variation coefficient when repeatedly performing the same task. Specifically, it extracts path selection records from the equipment's historical operation logs. These records contain information on alternative path selections when encountering obstacles and path trajectories when repeatedly performing the same task. The decision-making variability is calculated by statistically analyzing the frequency with which equipment selects different alternative paths when encountering the same type of obstacle. For example, it calculates the difference in the proportion of equipment choosing to detour to the left versus to the right when encountering a fixed obstacle. The decision difference value is quantified using the information entropy method, which constructs a probability distribution based on the frequency of occurrence of different path selection options and then calculates the degree of uncertainty of the probability distribution. The path selection variation coefficient is calculated by analyzing the path trajectory data when the device repeatedly performs the same task, such as comparing the ratio of the standard deviation to the mean of the path length when the same task is performed multiple times. The path trajectory data is obtained from the device navigation system and includes path point sequences and movement distance information, with the movement distance in meters. The system calculates the decision difference and path selection variation coefficient for each heterogeneous operation and maintenance device and stores them as a behavioral randomness feature vector, which is used for subsequent correlation analysis.

[0081] A spatiotemporal correlation mapping is performed between high-density conflict areas and behavioral randomness characteristics to establish a distribution map of equipment behavior randomness in conflict areas. Specifically, this involves matching the spatial coordinates and logical resource identifiers of high-density conflict areas with the behavioral randomness feature vectors of various heterogeneous maintenance equipment. The spatiotemporal correlation mapping process first establishes the correspondence between equipment locations and conflict areas. Based on real-time equipment location data and the spatial boundaries of high-density conflict areas, it calculates whether an equipment is located within a conflict area. Real-time equipment location data is obtained from the equipment positioning system and represented by latitude and longitude coordinates. Then, logical resource identifiers are correlated with equipment operation records, and the operational position of the equipment in the logical sequence is determined based on the resource request instruction sequence. The distribution map of equipment behavior randomness in conflict areas is represented by a graph structure. Nodes in the graph represent high-density conflict areas, and edges represent the propagation paths of equipment behavior randomness characteristics. Node attributes include region identifiers and region types, while edge attributes include decision difference values ​​and path selection variation coefficient values. During the map construction process, the system performs data alignment to ensure the consistency of timestamps and spatial coordinates. The timestamp alignment accuracy is set to 1 second, and the spatial coordinate alignment accuracy is set to 0.5 meters, thereby establishing a complete distribution map.

[0082] By analyzing the spatial overlap and logical coupling between high-density conflict areas and highly random behavior devices in the behavioral randomness distribution map, systemic interference areas leading to resource allocation loop waiting chains are identified. Specifically, this involves extracting high-density conflict area nodes and highly random behavior device edges from the behavioral randomness distribution map of conflict area devices. Highly random behavior devices are identified by setting a behavioral randomness threshold, which is set through statistical analysis based on the historical distribution of decision difference and path selection variation coefficient, for example, using the 85th percentile of the decision difference value and the 85th percentile of the path selection variation coefficient value as a joint threshold. Spatial overlap is obtained by calculating the ratio of the overlapping area between the spatial range of high-density conflict areas and the activity area of ​​highly random behavior devices. This overlapping area ratio is calculated using a geometric cross algorithm, for example, using the ratio of the intersection area of ​​polygons divided by the total area, implemented based on the vector cross product method. Logical coupling is determined by analyzing the logical resources and highly random behavior of high-density conflict areas. The system calculates the correlation strength of resource requests from machine-related devices. The correlation strength is obtained through a weighted summation of resource request frequency and waiting time, with weights set according to the importance of the resource type. The importance of the resource type is assessed through task criticality and resource utilization. Systematic interference regions are identified by setting spatial overlap and logical coupling thresholds. These thresholds are set based on historical interference event analysis, for example, by statistically analyzing typical values ​​from cases where resource allocation loops were successfully broken. The typical value for the spatial overlap threshold is set to 0.7, and the typical value for the logical coupling threshold is set to 0.6. The identification process includes calculating the spatial overlap and logical coupling of each candidate region and comparing these calculated values ​​with the corresponding thresholds. When the spatial overlap value of a region is greater than or equal to the spatial overlap threshold and the logical coupling value is greater than or equal to the logical coupling threshold, the region is marked as a systematic interference region. The system outputs a list of systematic interference regions, including region identifiers and interference strength indicators.

[0083] Based on the distribution characteristics and interference intensity of systematic interference regions, the structural contradiction between equipment autonomous decision-making and system resource layout is traced to the root cause of the conflict. Specifically, this involves analyzing the distribution patterns of systematic interference regions, including spatial clustering and logical correlation characteristics. Spatial clustering is obtained by calculating the spatial density of systematic interference regions, based on the number of interference regions per unit area. Logical correlation is determined by analyzing the topological position of systematic interference regions in the resource dependency graph. Interference intensity is extracted from the list of systematic interference regions, and the interference intensity index is calculated based on spatial overlap and logical coupling through normalization. The normalization method uses minimum and maximum scaling to a range of 0 to 100. The tracing process first identifies the degree of matching between equipment autonomous decision-making patterns and system resource layout. Equipment autonomous decision-making patterns are extracted from behavioral randomness characteristics, while system resource layout is obtained from the global resource map. Structural contradictions are identified by comparing the coordination between equipment path selection preferences and resource spatial distribution; for example, when highly randomized equipment frequently accesses resource-dense areas, it is marked as a layout conflict. The system generates a root cause report of the conflict, including a specific description of the contradiction and quantitative evidence, thus providing a basis for subsequent conflict resolution.

[0084] S5. The key conflict task that leads to the formation of a resource allocation circular waiting chain based on the determination of the root cause of the conflict is specifically implemented as follows:

[0085] Based on the distribution characteristics of systematic interference regions in the root causes of conflicts, candidate conflict tasks that are simultaneously located within high interference intensity regions in both physical space and logical sequence are identified. Specifically, this involves extracting distribution characteristic data from a list of systematic interference regions. This data includes the spatial coordinate range, logical resource coverage, and interference intensity value of the systematic interference regions. The identification process first sets a high interference intensity region determination threshold, which is determined based on statistical analysis of historical conflict data. For example, the 75th percentile of the distribution of interference intensity values ​​in systematic interference regions is used as a baseline threshold. This threshold is then adjusted based on factors related to the photovoltaic power plant's operating environment, with the adjustment range based on equipment density. The system determines the task complexity, traverses all systematic interference regions, and marks a region as a high-interference-intensity region when its interference intensity exceeds the threshold for high-interference-intensity region determination. Then, it identifies candidate conflicting tasks within this region. These tasks must simultaneously satisfy the following conditions: physically located within the coordinate range of the high-interference-intensity region and logically operating on logical resources covered by the high-interference-intensity region in their logical sequence. Physical location is verified using real-time positioning data from the task execution device, with a required accuracy of 0.5 meters. Logical sequence location is matched against the global resource map using task resource request records. The system outputs a list of candidate conflicting tasks, including task identifiers and their respective interference region information.

[0086] This analysis examines the topological position of each candidate conflicting task within the resource allocation cyclic waiting chain and its contribution to the formation of the loop. The contribution is determined by the number of resources held by a candidate conflicting task that are being waited for by other tasks and its hub status within the loop. Specifically, this involves extracting task resource dependencies from the resource allocation cyclic waiting chain structure and constructing a task resource waiting graph. Topological position analysis is achieved by calculating the centrality indices of candidate conflicting tasks within the task resource waiting graph. These centrality indices include degree centrality and betweenness centrality. Degree centrality is calculated by counting the number of edges directly connecting task nodes, while betweenness centrality is obtained by calculating the frequency with which task nodes appear in all shortest paths. The contribution calculation first involves counting the resources held by each candidate conflicting task. The number of tasks waiting for other tasks is determined by traversing the task resource waiting graph and counting the edges pointing to that task, with each edge representing a waiting relationship. Then, the hub status of candidate conflicting tasks in the loop is calculated. Hub status is determined by analyzing the frequency with which a task is on the critical path in the resource allocation loop waiting chain. The frequency is based on historical loop data statistics, and the statistical period is set according to the system's operating cycle. The final contribution value is obtained by weighted merging of the number of resources waiting and the hub status. The weights are set according to the resource type and task priority; for example, the weight for the number of resources waiting is set to 0.6, and the weight for the hub status is set to 0.4. The weight values ​​are determined through expert evaluation and historical data analysis. The system calculates and records the contribution score for each candidate conflicting task.

[0087] Combining the stochastic characteristics of equipment behavior and the intensity of task dependencies in candidate conflict tasks, a conflict intensification index is calculated for each candidate conflict task. Specifically, this involves extracting stochastic behavior characteristic data of the corresponding equipment from a database of equipment behavior stochastic characteristics. This data includes decision variability and path selection variation coefficients. Task dependency intensity is calculated by analyzing the resource dependencies of candidate conflict tasks in the resource allocation loop waiting chain. Resource dependencies include direct and indirect dependencies. The dependency intensity value is determined by a weighted sum based on the dependency chain length and resource criticality. Resource criticality is graded according to the importance of resources in task completion. The conflict intensification index is calculated using multi-dimensional feature fusion. The method first normalizes the randomness of equipment behavior and the intensity of task dependence, setting the normalization range to 0 to 1, and using minimum-maximum scaling. Then, it calculates the initial value of the conflict intensification index through weighted summation, with weights allocated according to feature importance, obtained through historical conflict data analysis. For example, the weight of randomness in equipment behavior is set to 0.5, and the weight of task dependence intensity is also set to 0.5. Finally, a time decay factor is introduced to adjust the conflict intensification index. The time decay factor is set based on the task duration; the longer the duration, the smaller the decay factor. The decay factor is calculated using an exponential decay formula, with a base of 0.9. The system outputs the conflict intensification index value for each candidate conflict task.

[0088] Candidate conflict tasks with the highest conflict escalation index and the greatest contribution to the formation of the resource allocation loop waiting chain are selected as critical conflict tasks. This is achieved by establishing a candidate conflict task evaluation matrix, which includes the conflict escalation index value and contribution score for each candidate task. The selection process first sets conflict escalation index thresholds and contribution thresholds. The conflict escalation index threshold is set to the 80th percentile based on historical data distribution, and the contribution threshold is set to the 70th percentile. The threshold update cycle is dynamically adjusted based on the system's operating status. Then, the system iterates through all candidate conflict tasks, comparing the conflict escalation index value of each task with the conflict escalation index threshold, and simultaneously comparing the contribution score with the contribution threshold, thus selecting the critical conflict escalation index value. Tasks with a conflict escalation index and a contribution score greater than or equal to the threshold are selected as the initial candidate set. A multi-objective decision-making method is then used to screen tasks. This method calculates a comprehensive score for each initial candidate task, obtained by a weighted product of the conflict escalation index and the contribution score. The weights are dynamically adjusted based on the system's optimization objectives; for example, when prioritizing conflict resolution, the conflict escalation index weight is set to 0.7, and the contribution score weight is set to 0.3. The system selects the task with the highest comprehensive score from the initial candidate set and verifies that both the conflict escalation index and contribution score exceed the corresponding thresholds. Finally, the task is determined to be a critical conflict task, and a critical conflict task report is generated, including detailed task information and processing recommendations.

[0089] S6. Based on the determination results of critical conflict tasks, dynamically reconstruct the execution device assignment of critical conflict tasks to break the resource allocation circular waiting chain. The specific implementation is as follows:

[0090] Based on the determination results of critical conflict tasks, the system identifies the original heterogeneous maintenance equipment currently executing the critical conflict task and its resource occupancy status in the resource allocation loop waiting chain. Specifically, this involves extracting task identifiers and device association information from the critical conflict task report. The critical conflict task report, generated by step S5, includes the task execution device identifier and a list of task resource requirements. The identification process first queries the task execution record database to obtain the original heterogeneous maintenance equipment identifier currently executing the critical conflict task. This identifier uniquely identifies the device instance responsible for executing the task. Then, the system extracts the resource occupancy status of this original heterogeneous maintenance equipment in the resource allocation loop waiting chain from the global resource graph. The resource occupancy status includes a list of shared resources currently occupied by the device and the occupancy duration for each resource. The occupancy duration is calculated by comparing the resource occupancy start time with the current system time, obtained from the resource status log. The system also records real-time changes in resource occupancy status by periodically scanning the global resource graph to update the status information. The scanning cycle is set according to system response requirements, for example, every 5 seconds, to ensure the accuracy of the status information and provide basic data for subsequent device selection.

[0091] Based on the real-time workload and task execution capability characteristics of the available heterogeneous O&M equipment group within the photovoltaic power plant, a candidate set of alternative heterogeneous O&M equipment qualified to perform critical conflict tasks is screened. Specifically, this involves obtaining real-time status data of all available heterogeneous O&M equipment within the photovoltaic power plant from the equipment management system. This real-time status data includes equipment workload indicators and task execution capability parameters. Workload indicators are obtained by calculating the number of tasks currently being executed by the equipment and the equipment resource utilization rate. The number of tasks currently being executed is counted from the task scheduling queue. Equipment resource utilization includes processing unit utilization and memory usage rate. This data is collected in real-time from the equipment monitoring system, with the collection frequency set according to the equipment type; for example, mobile equipment is collected once per second, and fixed equipment is collected once every 10 seconds. Task execution capability characteristics include equipment movement speed, task processing accuracy, and equipment functional compatibility. These characteristics are extracted from the equipment specification library and historical performance data. The historical performance data is based on the equipment specifications library. The system analyzes task completion records from the past 30 days. The screening process first sets workload and capability matching thresholds. The workload threshold is set through statistical analysis based on equipment type and photovoltaic power plant operation specifications; for example, the 70th percentile of the workload index of all available equipment is used as a benchmark. The capability matching threshold is calculated based on the similarity between the requirements of critical conflict tasks and the equipment capabilities. The similarity is determined by comparing the matching ratio between the task resource requirements and the equipment function list. The system iterates through all available heterogeneous maintenance equipment. When a device's workload is below the workload threshold and the similarity value of its task execution capabilities is greater than or equal to the capability matching threshold, the device is added to the candidate set of alternative heterogeneous maintenance equipment. The candidate set includes equipment identifiers and matching scores. The matching score is obtained by weighted calculation of workload and capability characteristics. The weights are dynamically adjusted according to the urgency of the task; for example, in high-urgency tasks, the workload weight is set to 0.7, and the capability characteristic weight is set to 0.3.

[0092] This study assesses the potential path intersection density and behavioral randomness interference risk between each device in the candidate set of alternative heterogeneous O&M equipment and other O&M tasks in the resource allocation cyclic waiting chain when performing critical conflict tasks. Specifically, this involves extracting path planning data and behavioral characteristic data for each device from the candidate set; assessing the potential path intersection density by simulating the movement paths of devices performing critical conflict tasks, with the movement paths generated based on task location and device navigation algorithms. The device navigation algorithm uses the A* path planning method, considering obstacles and traffic rules within the photovoltaic power plant; and calculating the path intersection density using the execution path intersection density assessment method in step S3, obtained by comparing the overlap between the simulated path and the planned paths of other tasks in the resource allocation cyclic waiting chain. The overlap degree is determined by... The ratio of path segment intersections to parallel distances is quantified, and the ratio is obtained through geometric calculations, such as calculating the number of intersections of two paths in the same grid cell divided by the total number of path points; the behavioral randomness interference risk assessment is based on the equipment behavioral randomness characteristic data, including decision variability and path selection variation coefficient, which are extracted from the equipment behavior database established in step S4; the interference risk value is calculated by analyzing the behavioral volatility exhibited by the equipment in similar task environments, and the volatility is measured by the standard deviation of historical behavioral data, which is calculated based on the past 50 task execution records; the system generates a potential path intersection density score and a behavioral randomness interference risk score for each candidate device, with the score range normalized to between 0 and 1, and the normalization method using min-max scaling.

[0093] The system selects alternative heterogeneous maintenance equipment with the lowest potential path intersection density and the lowest risk of behavioral randomness interference. Critical conflict tasks are dynamically reassigned from the original heterogeneous maintenance equipment to the selected alternative. This is achieved by establishing a candidate equipment evaluation matrix, which includes the potential path intersection density value and the behavioral randomness interference risk value for each candidate equipment. The selection process first sets optimal thresholds for path intersection density and interference risk. The optimal path intersection density threshold is set based on statistical analysis of historical task execution data, for example, taking the 20th percentile of the path intersection density values ​​of all candidate equipment. The optimal interference risk threshold is taken as the 20th percentile of the behavioral randomness interference risk value. These thresholds are updated dynamically based on system operating status, for example, recalculated every 10 minutes. Then, the system iterates through all candidate equipment, comparing the potential path intersection density value of each equipment with the optimal path intersection density threshold, and simultaneously comparing the behavioral randomness interference risk value with the optimal interference risk threshold. Equipment with a potential path intersection density value less than or equal to the optimal path intersection density threshold and exhibiting random behavior is selected. Devices with an interference risk value less than or equal to the interference risk preference threshold are selected as the initial candidate set. Then, a multi-objective optimization method is used for device selection. This method calculates the comprehensive priority index for each initial candidate device. The comprehensive priority index is obtained by weighted harmonic mean of path intersection density and interference risk. Path intersection density and interference risk are used as input parameters in the weighted harmonic mean calculation, with weights set according to system security and efficiency requirements. For example, the path intersection density weight is set to 0.6, and the interference risk weight is set to 0.4. The weight values ​​are determined through expert evaluation and historical conflict resolution effect analysis. The system selects the candidate device with the highest comprehensive priority index from the initial candidate set as the replacement heterogeneous maintenance device and performs a task reassignment operation. Task reassignment is achieved by updating the task allocation record and the device work queue. The task allocation record is modified in the central database, and the device work queue is synchronized through a message queue system to ensure resource status synchronization during the task handover process. Resource status synchronization includes releasing resources occupied by the original heterogeneous maintenance device and allocating new resources to the replacement heterogeneous maintenance device.

[0094] After the execution equipment for the critical conflict task is reassigned, it is verified whether the resource allocation circular wait chain has been resolved. This is achieved by rerunning the resource allocation circular wait chain detection process in step S2. This process searches for closed-loop paths based on the resource request command sequence and the global resource graph. The verification process first collects the reassigned resource request command sequence, extracted from the real-time request logs of each heterogeneous maintenance device. The extraction frequency matches the system sampling frequency, for example, once per second. Then, the global resource graph is updated to reflect the changes in resource occupancy after the task reassignment. The update operations include releasing resources occupied by the original heterogeneous maintenance device and allocating new resources to the replacement heterogeneous maintenance device. Device and resource release and allocation ensure data consistency through transactional operations. The system executes a closed-loop path search algorithm, which uses a depth-first traversal method to check for cyclic paths in the resource dependency graph. The depth-first traversal starts from each resource node and traverses along occupied and dependent edges, with a maximum search depth set to prevent infinite loops. The maximum search depth is set according to the number of resources, for example, twice the total number of resources. If no closed-loop path is detected, it is determined that the resource allocation circular waiting chain has been resolved, and the verification result is recorded in the system log. If a closed-loop path is still detected, a new round of conflict handling process is triggered, including re-evaluating key conflict tasks and adjusting device assignment strategies, and re-evaluating execution based on the latest resource status data.

[0095] Example 2: Figure 2 A schematic diagram of the structure of an unmanned operation and maintenance system for a photovoltaic power station according to the present invention is provided. The unmanned operation and maintenance system for a photovoltaic power station includes:

[0096] The information determination module is used to acquire the sequence of resource request instructions issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station in real time, and simultaneously construct a global resource map reflecting the real-time occupancy status of shared resources across the entire station.

[0097] The loop determination module is used to match the sequence of resource request instructions with the global resource graph to determine whether there is a resource allocation loop waiting chain that causes the task to stall.

[0098] The density quantification module is used to quantify and evaluate the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in the physical space and logical sequence when a resource allocation circular waiting chain exists.

[0099] The root cause tracing module is used to analyze the interference effect between execution path intersection density and the randomness of the behavior of various heterogeneous operation and maintenance devices in path selection in order to trace the root cause of the conflict.

[0100] The conflict determination module is used to determine the key conflict tasks that lead to the formation of a resource allocation loop waiting chain based on the root cause of the conflict.

[0101] The loop release module is used to dynamically reconstruct the execution device assignment of critical conflict tasks based on the judgment results of critical conflict tasks, so as to release the resource allocation loop wait chain.

[0102] Example 3: An unmanned operation and maintenance device for a photovoltaic power station. The device includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, an unmanned operation and maintenance method for a photovoltaic power station is implemented.

[0103] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0104] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. Computer-readable storage media can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0106] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0107] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0108] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0109] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0110] If a function is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0111] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0112] In conclusion, the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for unmanned operation and maintenance of a photovoltaic power station, characterized in that, include: S1. Real-time acquisition of resource request command sequences issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station, and simultaneous construction of a global resource map reflecting the real-time occupancy status of shared resources across the entire station; S2. Match the resource request instruction sequence with the global resource graph to determine if there is a resource allocation circular wait chain that causes the task to stall, including: Parse the resource request command sequence to extract the request information of each heterogeneous operation and maintenance device for shared resources; The extracted request information is compared with the current occupancy status of resources in the global resource map; Based on the comparison results, search the global resource graph for a closed-loop path formed by resource requests from multiple heterogeneous operation and maintenance devices. When a closed-loop path exists, it is determined that a resource allocation circular waiting chain exists; S3. When a resource allocation circular waiting chain exists, quantitatively evaluate the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in the physical space and logical sequence. S4. Based on the intersection density of execution paths, and combined with the randomness of the behavior of various heterogeneous operation and maintenance devices in path selection, analyze the interference effect between the two to trace the root cause of the conflict, including: Identify high-density conflict regions in physical space and logical sequence based on execution path intersection density; Extract the stochastic characteristics of the behavior of each heterogeneous operation and maintenance device in the path selection process, including the decision difference of the device in choosing an alternative path when encountering path obstacles and the path selection variation coefficient of the device when repeatedly performing the same task; Spatiotemporal correlation mapping is performed between high-density conflict areas and behavioral randomness characteristics to establish a distribution map of random equipment behavior in conflict areas; By analyzing the spatial overlap and logical coupling between high-density conflict regions and devices with highly random behavior in the distribution map of behavioral randomness, the systemic interference regions that lead to the formation of resource allocation cycle waiting chains can be identified. Based on the distribution characteristics and interference intensity of the systemic interference region, the structural contradiction between equipment autonomous decision-making and system resource layout is traced to the root cause of the conflict. S5. Key conflict tasks that lead to the formation of a resource allocation circular waiting chain based on the determination of the root cause of the conflict, including: Based on the distribution characteristics of systematic interference regions in the root causes of conflict, candidate conflict tasks that are simultaneously located in high interference intensity regions in both physical space and logical sequence are identified. Analyze the topological position of each candidate conflicting task in the resource allocation loop waiting chain and its contribution to loop formation; By combining the randomness of equipment behavior and the intensity of task dependence of candidate conflict tasks, the conflict intensification index of each candidate conflict task is calculated. Candidate conflict tasks with the highest conflict escalation index and the greatest contribution to the formation of the resource allocation cycle waiting chain are identified as key conflict tasks. S6. Based on the determination results of critical conflict tasks, dynamically reconstruct the execution equipment assignment of critical conflict tasks to eliminate the resource allocation circular waiting chain.

2. The unmanned operation and maintenance method for a photovoltaic power station according to claim 1, characterized in that, Real-time acquisition of resource request command sequences issued by various heterogeneous operation and maintenance devices within the photovoltaic power station, and simultaneous construction of a global resource map reflecting the real-time occupancy status of shared resources across the entire station, including: Real-time acquisition of resource request commands issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station during task execution, and organization into a resource request command sequence according to the time sequence; Simultaneously monitor the real-time occupancy status of shared resources across the entire site, and construct a global resource map represented in a graph structure based on the relationship between resource type and physical location; In the global resource graph, nodes correspond to shared resources, and edges correspond to the occupancy or dependency relationships between resources.

3. The unmanned operation and maintenance method for a photovoltaic power station according to claim 1, characterized in that, When a resource allocation circular waiting chain exists, quantitatively evaluate the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in both physical space and logical sequence, including: Based on the historical execution trajectory of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain, spatial conflict hotspots formed in the physical space movement path are identified. Spatial conflict hotspots are determined by calculating the crossover frequency and average dwell time of different task paths within a preset time period. Synchronously analyze the operation flow of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in the logical sequence, identify the logical competition bottleneck nodes formed during the execution of the logical sequence, and determine the logical competition bottleneck nodes by statistically analyzing the request order and waiting time of different tasks for the same shared resources. By integrating the distribution characteristics of spatial conflict hotspots with the occurrence patterns of logical competition bottleneck nodes, a multi-dimensional quantitative evaluation result of the intersection density of execution paths is generated.

4. The unmanned operation and maintenance method for a photovoltaic power station according to claim 1, characterized in that, The contribution is determined by the amount of resources held by the candidate conflicting task that are waiting for by other tasks and its hub status in the loop.

5. The unmanned operation and maintenance method for a photovoltaic power station according to claim 1, characterized in that, Based on the determination of critical conflict tasks, the execution equipment assignments for critical conflict tasks are dynamically restructured to break the resource allocation circular wait chain, including: Based on the determination results of critical conflict tasks, identify the original heterogeneous operation and maintenance equipment currently executing critical conflict tasks and its resource occupancy status in the resource allocation loop waiting chain; Based on the real-time workload and task execution capability characteristics of the available heterogeneous operation and maintenance equipment group in the photovoltaic power plant, a candidate set of alternative heterogeneous operation and maintenance equipment that is qualified to perform critical conflict tasks is selected. Evaluate the potential path intersection density and behavioral randomness interference risk between each device in the candidate set of alternative heterogeneous operation and maintenance equipment and other operation and maintenance tasks in the resource allocation circular waiting chain when each device performs a critical conflict task; Select the alternative heterogeneous operation and maintenance equipment with the lowest potential path intersection density and the lowest risk of interference from behavioral randomness, and dynamically reassign critical conflict tasks from the original heterogeneous operation and maintenance equipment to the selected alternative heterogeneous operation and maintenance equipment. After the execution equipment that has completed the critical conflict task is reassigned, verify whether the resource allocation circular wait chain has been resolved.

6. A photovoltaic power plant unmanned operation and maintenance system, used to implement the unmanned operation and maintenance method for a photovoltaic power plant as described in any one of claims 1-5, characterized in that, include: The information determination module is used to acquire the sequence of resource request instructions issued by various heterogeneous operation and maintenance equipment in the photovoltaic power station in real time, and simultaneously construct a global resource map reflecting the real-time occupancy status of shared resources across the entire station. The loop determination module is used to match the sequence of resource request instructions with the global resource graph to determine whether there is a resource allocation loop waiting chain that causes the task to stall. The density quantification module is used to quantify and evaluate the intersection density of execution paths of each heterogeneous operation and maintenance task in the resource allocation circular waiting chain in the physical space and logical sequence when a resource allocation circular waiting chain exists. The root cause tracing module is used to analyze the interference effect between execution path intersection density and the randomness of the behavior of various heterogeneous operation and maintenance devices in path selection in order to trace the root cause of the conflict. The conflict determination module is used to determine the key conflict tasks that lead to the formation of a resource allocation loop waiting chain based on the root cause of the conflict. The loop release module is used to dynamically reconstruct the execution device assignment of critical conflict tasks based on the judgment results of critical conflict tasks, so as to release the resource allocation loop wait chain.

7. An unmanned operation and maintenance device for a photovoltaic power station, characterized in that, The device includes: A processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein when the program or instructions are executed by the processor, they implement the unmanned operation and maintenance method for a photovoltaic power station as described in any one of claims 1-5.

Images