Power grid operation safety patrol risk prediction method fusing artificial intelligence technology

Abstract

The application discloses a power grid operation safety patrol risk prediction method fusing artificial intelligence technology and relates to the technical field of power grid operation safety patrol, and specifically comprises the following steps: performing operation logic analysis on all identified parallel operation tasks, screening all operation tasks with operation conflicts, and forming operation interference task clusters; obtaining parallel operation interference feature information in the operation interference task clusters, analyzing the same, and evaluating the task interference degree between the operation tasks in the operation interference task clusters; and according to the evaluation result, performing dynamic regulation and control on the operation task input data set for risk prediction. The application solves the task conflict problem caused by the operation demand and safety constraint difference of different operation tasks during the power grid operation patrol process, and achieves the purpose of improving the accuracy of risk prediction by dynamically regulating and controlling the operation task input data set and accurately evaluating the task interference degree.

Description

Technical Field

[0001] This invention relates to the field of power grid operation safety inspection technology, specifically to a method for predicting power grid operation safety inspection risks by integrating artificial intelligence technology. Background Technology

[0002] Power grid operation safety inspection refers to the systematic inspection and recording of equipment operating status, environmental safety factors, and potential fault hazards during the operation and maintenance of power systems, especially at key nodes such as high-voltage transmission lines, substation equipment, and distribution facilities. This is done manually or with intelligent equipment (such as inspection robots and drones). The aim is to identify and eliminate safety hazards, ensuring the safety of personnel and the stable operation of the power grid system. However, traditional inspection methods suffer from problems such as information silos, delayed response, and reliance on experience for risk identification, making it difficult to effectively predict future risks. Therefore, a power grid operation safety inspection risk prediction method integrating artificial intelligence technology has emerged. Based on multi-source data (such as historical inspection records, environmental parameters, and image / video information), it utilizes AI algorithms such as machine learning and deep learning to build risk identification and prediction models. This enables real-time analysis, trend judgment, and early warning of potential safety risks, helping personnel take preventative measures before risks occur, effectively improving inspection efficiency and reducing personal and equipment risks. By deeply embedding artificial intelligence technology into power grid operation scenarios, this method not only enhances the intelligence and automation capabilities of safety management and control, but also lays the technical foundation for building a more forward-looking and responsive power grid safety assurance system.

[0003] Existing power grid operation safety inspection risk prediction technologies that integrate artificial intelligence mainly predict potential risks during power grid operations by constructing data-driven intelligent analysis models. The risk prediction process typically includes the following key stages: First, in the data acquisition stage, the system acquires multimodal data such as images, videos, environmental parameters (e.g., temperature, humidity, wind speed, electromagnetic intensity), and worker behavior trajectories at the work site through the deployment of various sensors, drones, inspection robots, and video surveillance. Second, in the data preprocessing and fusion stage, the system cleans, denoises, and extracts features from the collected data, and integrates structured and unstructured information to form a complete operational safety profile. Subsequently, in the risk modeling stage, the system utilizes… Artificial intelligence algorithms, such as convolutional neural networks (CNNs) for image recognition, long short-term memory networks (LSTMs) for processing time-series data, or ensemble learning models for comprehensive evaluation, learn potential risk characteristic patterns from historical accident samples. Then, in the risk prediction stage, the model analyzes the deviations and trends between inspection data and historical data in real time, predicting potential high-risk events such as equipment failures, operational violations, and sudden environmental changes. Finally, through a risk grading and early warning module, the prediction results are transformed into specific risk level assessments, and combined with a geographic information system (GIS) or dispatch platform, visualized early warning prompts are generated and fed back to work dispatchers or on-site operators in real time, thereby achieving closed-loop management of early detection, proactive intervention, and intelligent decision-making. Overall, this technology, with artificial intelligence at its core, runs through the entire process from data perception to intelligent decision-making, significantly improving the scientific and intelligent level of power grid operation safety management.

[0004] The existing technology has the following shortcomings:

[0005] During power grid operation and inspection, when multiple tasks are executed concurrently, such as clearing obstacles and equipment maintenance on the same line segment, conflicts can easily arise due to the different operational requirements and safety constraints of each task. For example, maintenance typically requires power outages, while clearing obstacles relies on line power supply to ensure equipment operation. If the data from both tasks are indiscriminately input into an AI risk prediction model, the model may fail to recognize the task exclusion relationships, confusing the task context logic. This can lead to low-risk tasks being misjudged as high-risk, or high-risk tasks being unrecognized. Existing AI-integrated power grid operation safety inspection risk prediction technologies cannot dynamically adjust the input dataset for risk prediction based on the degree of task interference in situations with operational conflicts. Consequently, the model cannot decouple operational logic and determine conflicts under parallel task conditions, potentially leading to prediction deviations, incorrect scheduling strategies, work plan conflicts, and even personal and equipment safety accidents.

[0006] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide a method for predicting the risks of power grid operation safety inspections that integrates artificial intelligence technology, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for predicting the risk of power grid operation safety inspection integrating artificial intelligence technology, specifically including the following steps:

[0009] During power grid operation and inspection, the received power grid inspection and dispatch data is analyzed to identify all operation tasks currently being executed in parallel within the same power equipment area;

[0010] Perform operational logic analysis on all identified parallel execution tasks, filter out all tasks with operational conflicts, and form a cluster of tasks with operational interference.

[0011] Obtain the interference characteristics of parallel jobs in the inter-operational interference task cluster, analyze them, and evaluate the degree of task interference between each job in the inter-operational interference task cluster.

[0012] Based on the evaluation results, dynamic adjustments are made to the task input dataset used for risk prediction.

[0013] Risk prediction is performed based on the regulated input data, the risk results for each task are output, and the risk results and corresponding task conditions are recorded to optimize the subsequent risk prediction process.

[0014] Preferably, operational logic analysis is performed on all identified parallel execution tasks to filter out all tasks with operational conflicts and form a cluster of tasks with operational interference, specifically:

[0015] Based on all the identified parallel-executed job tasks, extract the operation steps, equipment target information and safety dependency rules for each job task.

[0016] The execution order relationship between tasks is analyzed based on operation step information; the device access relationship between tasks is analyzed based on device target information; and the control condition relationship between tasks is analyzed based on security dependency rules.

[0017] Based on the analysis results, all job tasks with operational conflicts were identified and grouped into a cluster of interfering job tasks.

[0018] Preferably, the method involves acquiring and analyzing the parallel job interference characteristics of the interfering task cluster, and evaluating the degree of task interference between each job in the interfering task cluster. This specifically includes the following steps:

[0019] Obtain the interference feature information of parallel jobs in the interfering task cluster and perform preprocessing after acquisition;

[0020] Resource interference analysis information and step interference evaluation information are extracted from the preprocessed parallel operation interference feature information, and then analyzed to generate resource conflict intensity coefficient and operation step conflict index, respectively.

[0021] Based on the generated resource conflict intensity coefficient and operation step conflict index, a task interference index is generated by weighted summation.

[0022] Determine the pre-set threshold range for the task interference index, and compare it with the generated task interference index after determination. Based on the comparison, evaluate the degree of task interference between each task in the task cluster.

[0023] Preferably, the logic for obtaining the resource conflict intensity coefficient is as follows:

[0024] Resource interference analysis information is extracted from the preprocessed parallel job interference feature information. Specifically, this includes the proportion of total equipment resources allocated to each job in the operational interference task cluster, the number of times equipment resources are requested, and the number of times resource access conflicts occur with other tasks. These are then labeled as A. i R i and C i A i R represents the proportion of the total equipment resources allocated to the i-th job in the interfering task cluster. i C represents the number of times the i-th job in the interfering task cluster requests device resources. i This represents the number of resource access conflicts that occur between the i-th job and other tasks in the interfering task cluster, where i = 1, 2, 3, ..., g, and g is a positive integer;

[0025] The specific formula for calculating the resource conflict intensity coefficient is as follows:

[0026]

[0027] In the formula, RCIC is the resource conflict intensity coefficient.

[0028] Preferably, the logic for obtaining the conflict index of the operation steps is as follows:

[0029] Step interference evaluation information is extracted from the preprocessed parallel job interference feature information. Specifically, this includes the number of steps executed by each job in the operation interference task cluster, the number of conflicts with other tasks in the execution steps, and the proportion of equipment resources occupied during the execution of steps, and these are respectively labeled as S. i P i and U i S i P represents the number of steps executed by the i-th job in the cluster of interfering operations. i U represents the number of conflicts in the execution steps between the i-th job and other tasks in the interfering task cluster. i This represents the proportion of device resources occupied by the i-th job in the interfering task cluster during the execution of steps, where i = 1, 2, 3, ..., g, and g is a positive integer;

[0030] The specific formula for calculating the operational step conflict index is as follows:

[0031]

[0032] In the formula, OSCI is the Operational Conflict Index.

[0033] Preferably, based on the generated Resource Conflict Intensity Index (RCIC) and Operational Step Conflict Index (OSCI), a task interference index is generated by weighted summation. The specific calculation formula is as follows:

[0034] TII=ω1*RCIC+ω2*OSCI

[0035] In the formula, TII is the task interference index, ω1 and ω2 are the non-zero weight coefficients of the resource conflict intensity coefficient (RCIC) and the operation step conflict index (OSCI), respectively, and ω1+ω2=1.

[0036] Preferably, a pre-defined threshold range for the task interference index [TII] is determined. min TII max After determination, it is compared with the generated Task Interference Index (TII). Based on the comparison, the degree of task interference between each job task in the operational interference task cluster is evaluated. The specific comparison analysis is as follows:

[0037] If TII <TII min The degree of task interference between tasks in the inter-task cluster is low.

[0038] If TII min ≤TII≤TII max The degree of task interference between tasks in the inter-task cluster is moderate.

[0039] If TII>TII maxThe degree of task interference between tasks in an operational interference task cluster is defined as the severity.

[0040] Preferably, based on the evaluation results, dynamic adjustments are made to the task input dataset used for risk prediction, specifically as follows:

[0041] If the assessment result is low, the original state of the task input dataset used for risk prediction is maintained, and no adjustments are made to the data to ensure that the data is used for risk prediction within the pre-set normal range.

[0042] If the assessment result is moderate, the weighted adjustment of the task input dataset used for risk prediction is performed, increasing the weight of tasks involving high conflict in the task data;

[0043] If the assessment result is severity, the input dataset for the job tasks used for risk prediction is filtered and reordered, prioritizing data from tasks with high interference.

[0044] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0045] 1. This invention effectively solves the problem of task conflicts caused by different operational requirements and safety constraints among tasks during multi-task parallel execution. Traditional methods treat multiple tasks as a single batch for processing, failing to identify the exclusionary relationships between tasks. This leads to low-risk tasks being misclassified as high-risk, or high-risk tasks not being identified. By analyzing the degree of task interference and dynamically adjusting the task input dataset under different interference levels, this solution can accurately assess and distinguish task conflicts in a multi-task parallel environment, effectively avoiding conflict confusion and task misjudgment in traditional methods.

[0046] 2. This invention combines important indicators such as resource conflict intensity coefficient and operational step conflict index, generates a task interference index through weighted summation, and classifies task interference according to a set threshold range. This mechanism ensures that highly disruptive tasks receive priority during task data processing, thereby ensuring that these tasks receive appropriate attention in risk prediction. The strategy of dynamically adjusting the input dataset of job tasks guarantees the rational allocation of resources during task scheduling and ensures that the risk assessment results for each task are more accurate and reliable.

[0047] 3. This invention ensures that task conflicts are effectively resolved during risk prediction by filtering, weighting, and prioritizing data for high-conflict tasks. Under severe interference, the system can reorder and filter input data based on the degree of conflict between tasks, prioritizing the assessment of high-risk tasks. This not only improves the prediction accuracy of power grid operation safety inspections but also provides more reliable decision support for the stable operation of the power grid. Through these precise interference assessments and data adjustments, this technical solution significantly improves the safety of power grid operations and reduces scheduling errors and potential safety accident risks caused by task conflicts. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0049] Figure 1 This is a flowchart illustrating the power grid operation safety inspection risk prediction method that integrates artificial intelligence technology, as described in this invention. Detailed Implementation

[0050] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0051] This invention provides, for example Figure 1 The power grid operation safety inspection risk prediction method shown, which integrates artificial intelligence technology, specifically includes the following steps:

[0052] During power grid operation and inspection, the received power grid inspection and dispatch data is analyzed to identify all operation tasks currently being executed in parallel within the same power equipment area;

[0053] Parallel relationship analysis of power grid operation tasks can be achieved by constructing a job scheduling data parsing algorithm. This algorithm takes structured job information received from the scheduling platform as input, including fields such as job number, scheduling time, job type, target equipment number, power grid topology number, and geographical coordinates, and maps this information to a unified time and space indexing system. It identifies temporal overlap relationships between job tasks by setting time window sliding matching rules, and simultaneously uses a spatial matching algorithm based on equipment affiliation to determine whether tasks operate within the same power equipment area. Tasks with temporal overlap and consistent spatial affiliation are identified as parallel job tasks and further grouped into a candidate task set for subsequent analysis and processing.

[0054] When identifying all parallel tasks currently executing within the same power equipment area, the planned start and end times of each task are first standardized to generate a unified time interval index. Simultaneously, the device number and topology region identifier are extracted to establish a device affiliation mapping relationship for the tasks. Then, time interval comparisons are used to determine the temporal intersection between tasks, and the spatial correlation is determined through the device mapping relationship. When two or more tasks overlap in time and act on the same power equipment or its adjacent branches, they are considered to have a parallel execution relationship. Finally, all tasks meeting these conditions are merged into the same set of parallel tasks.

[0055] Identifying parallel tasks is crucial for pre-modeling the logical relationships within the task data before risk prediction. The temporal and spatial overlap of multiple tasks often indicates potential interference between their operations. Indiscriminately treating these tasks as a single input batch in an AI risk prediction model will make it difficult for the model to accurately identify operational dependencies and mutual exclusions, leading to biased risk assessments. Proactively identifying parallel tasks during data preprocessing facilitates subsequent classification and interference assessment of potentially conflicting tasks, thereby improving the reliability and accuracy of overall risk prediction.

[0056] Perform operational logic analysis on all identified parallel execution tasks, filter out all tasks with operational conflicts, and form a cluster of tasks with operational interference.

[0057] In this embodiment, operational logic analysis is performed on all identified parallel execution tasks to filter out all tasks with operational conflicts and form a cluster of tasks with operational interference, specifically:

[0058] Based on all the identified parallel-executed job tasks, extract the operation steps, equipment target information and safety dependency rules for each job task.

[0059] A structured parsing algorithm for job task information can be constructed to extract operation step information, equipment target information, and safety dependency rules. First, in the received task data from the scheduling platform, metadata entries for each job task are identified through field mapping rules, including job number, job type, scheduling instructions, job instruction list, power grid equipment number, and job safety control requirements. Then, a rule engine is used to perform semantic parsing and sequence analysis on the job instruction list, extracting the job execution order, key operation nodes, and dependency chains to form operation step information. The equipment number is matched with the power grid equipment ledger database to determine the physical equipment and topology location corresponding to the job target, extracting this as equipment target information. Finally, based on the job scheduling requirements and the control strategy requirements attached to the safety technical specifications, constraint logic on job conditions is extracted, such as whether power outage is required, whether warning lines need to be laid, and whether job interval duration is required, and these are uniformly modeled as safety dependency rules. The entire process is driven at the software layer by configurable field binding and parsing rules, completing the extraction from raw task data to structured logical information for subsequent operational logic analysis.

[0060] The execution order relationship between tasks is analyzed based on operation step information; the device access relationship between tasks is analyzed based on device target information; and the control condition relationship between tasks is analyzed based on security dependency rules.

[0061] An algorithm for analyzing the logical relationships between tasks can be constructed to analyze the execution order, equipment access, and control conditions among tasks. First, for operation step information, a directed graph structure is built for the instruction sequence of each task, extracting the temporal logic and execution dependency chains of key operation nodes. Then, the temporal logic of nodes between different tasks is cross-compared to identify task pairs with pre-dependencies, overlapping operations, or conflicting instructions, thus determining their execution order. Second, for equipment target information, the target equipment number associated with each task is mapped to the power grid equipment topology graph to determine whether tasks point to the same equipment, the same electrical circuit, or adjacent nodes. If resource overlap exists, it is marked as a device access conflict. Finally, for safety dependency rules, the operational constraints of each task are modeled as a set of Boolean logic expressions. Logical solving and rule matching are used to analyze whether there are mutually exclusive or interfering control conditions between different tasks, such as one task requiring power outage while another requires continuous power supply. The entire analysis process uses software to perform task comparison, graph analysis, and rule calculation on an intermediate data structure, automatically generating an operational logic relationship matrix between tasks as the basis for determining operational conflicts.

[0062] Based on the analysis results, all job tasks with operational conflicts were identified and grouped into a cluster of interfering job tasks.

[0063] Selecting and grouping conflicting tasks into clusters can be achieved using a task clustering algorithm based on a logical conflict graph. This algorithm first reads the execution order, device access, and control condition relationships between tasks analyzed in the previous step, and then constructs an operational logic conflict matrix, where each conflict relationship is represented as an edge connection between task nodes. By establishing this conflict graph structure, the system runs an aggregation traversal algorithm on the graph to identify connections between all tasks with direct or indirect conflict paths, and automatically groups densely interacting tasks into a preliminary task set based on task relevance.

[0064] When screening conflicting tasks, the system determines conflict paths for each pair of tasks based on edge weights and conflict types in the conflict graph. If any type of valid conflict exists between tasks (e.g., inconsistent instruction order, overlapping equipment resources, conflicting control conditions), the task pair is marked as conflicting. Subsequently, through connectivity analysis in the graph, all tasks with conflicting paths are grouped into the same group, forming disjoint task subgraphs. Each subgraph represents a cluster of mutually conflicting tasks. This process is completed by traversing all task nodes in the graph and recording their conflict connections, ensuring that tasks within each cluster have verifiable logical conflict links, thus achieving structured aggregation of task conflicts.

[0065] By grouping conflicting tasks into the same cluster of interfering tasks, we can avoid misinterpreting conflicting information as independent task logic in subsequent processing. This also provides downstream risk prediction models with clear task context boundaries, facilitating the model's identification of local conflict patterns and high-risk combinations. Especially in densely populated areas or during batch task execution, ignoring operational interference between tasks can lead to prediction errors or scheduling conflicts. Therefore, task screening and merging based on analysis results is a crucial step in improving the accuracy of AI risk prediction and achieving proactive management, significantly enhancing system controllability and the security of task scheduling.

[0066] Obtain the interference characteristics of parallel jobs in the inter-operational interference task cluster, analyze them, and evaluate the degree of task interference between each job in the inter-operational interference task cluster.

[0067] In this embodiment, the interference characteristic information of parallel jobs in the interfering task cluster is obtained and analyzed to evaluate the degree of task interference between each job in the interfering task cluster. Specifically, this includes the following steps:

[0068] Obtain the interference feature information of parallel jobs in the interfering task cluster and perform preprocessing after acquisition;

[0069] Obtaining the interference characteristics of parallel operations within a cluster of interfering tasks can be achieved by accessing the task data recorded in the power grid operation scheduling platform. First, the system needs to extract raw data related to the tasks from the scheduling platform, including the frequency of resource requests, resource allocation, the type and number of task execution steps, and the number of resource access conflicts. This data is typically stored in a database in structured form. The software retrieves this data by calling database interfaces or APIs and converts it into a processable format. Furthermore, the system can supplement this with additional information related to equipment operating status and environmental conditions through real-time data collected by sensors and monitoring equipment, further enriching the interference characteristics of parallel operations.

[0070] The purpose of preprocessing is to ensure that the acquired operational interference characteristics provide high-quality data support for subsequent analysis and modeling. Preprocessing steps include data cleaning, denoising, and standardization. First, invalid or erroneous data, such as missing values, outliers, or data that does not conform to the specified format, needs to be removed. Second, data from different tasks with different units or ranges is standardized to ensure that all types of data have the same dimensions and comparability. Then, potential noise is smoothed, for example, by using filtering algorithms to remove extremely fluctuating instantaneous data, ensuring data stability. Finally, the system categorizes and integrates the cleaned and standardized data according to task type and resource type for easier subsequent analysis. Preprocessed data can more accurately reflect the interference characteristics between various operational tasks, providing reliable input for generating interference evaluation coefficients.

[0071] Resource interference analysis information and step interference evaluation information are extracted from the preprocessed parallel operation interference feature information, and then analyzed to generate resource conflict intensity coefficient and operation step conflict index, respectively.

[0072] Extracting resource interference analysis information and step interference assessment information can be achieved through data analysis and feature extraction algorithms. First, in the preprocessed interference feature information, the system identifies and distinguishes between resource-related features (such as resource request frequency, resource allocation amount, conflict frequency, etc.) and operation step-related features (such as the number of operation steps, conflict frequency between steps, etc.). Extracting resource interference analysis information mainly relies on statistically analyzing and comparing the resources occupied by tasks, including calculating the resource occupancy ratio of each task and its impact on resource access by other tasks. Step interference assessment information focuses on analyzing the cross-conflicts of task execution steps, especially when multiple tasks are running in parallel, to determine whether step conflicts or operational dependencies exist. The system classifies and clusters standardized data, uses statistical analysis or correlation analysis to extract these key features, and constructs two types of interference analysis information. This process can be automated using machine learning models or data mining techniques to extract and classify this information, ensuring efficiency and accuracy.

[0073] Based on the generated resource conflict intensity coefficient and operation step conflict index, a task interference index is generated by weighted summation.

[0074] Determine the pre-set threshold range for the task interference index, and compare it with the generated task interference index after determination. Based on the comparison, evaluate the degree of task interference between each task in the task cluster.

[0075] Determining a pre-defined threshold range for task interference indices can be achieved through data analysis and statistical methods using historical operational data. First, the system needs to collect and analyze past power grid operational data, particularly interference indices between different tasks (e.g., resource conflicts and operational step conflicts between tasks), and perform statistical analysis on these indices, such as calculating their average, standard deviation, maximum, and minimum values. Based on this statistical data, the system can set threshold ranges for the interference indices. Typically, these threshold ranges vary depending on the task type and operating environment. The system can adjust the thresholds based on actual needs and experience, for example: low interference (minor interference), medium interference (moderate interference), and high interference (high interference). Furthermore, the system can use machine learning algorithms to automatically optimize the threshold ranges based on patterns in historical data, further improving the accuracy of interference index assessment. This process can be achieved through data mining techniques, cluster analysis, or regression analysis to ensure that the threshold settings match the interference situations encountered in actual operations.

[0076] In this embodiment, the logic for obtaining the resource conflict intensity coefficient is as follows:

[0077] Resource interference analysis information is extracted from the preprocessed parallel job interference feature information. Specifically, this includes the proportion of total equipment resources allocated to each job in the operational interference task cluster, the number of times equipment resources are requested, and the number of times resource access conflicts occur with other tasks. These are then labeled as A. i R i and C i A i R represents the proportion of the total equipment resources allocated to the i-th job in the interfering task cluster. i C represents the number of times the i-th job in the interfering task cluster requests device resources. i This represents the number of resource access conflicts that occur between the i-th job and other tasks in the interfering task cluster, where i = 1, 2, 3, ..., g, and g is a positive integer;

[0078] Data such as the proportion of total equipment resources allocated to each task in the interfering task cluster, the number of times equipment resources are requested, and the number of resource access conflicts with other tasks can be obtained through the resource management module, task logs, and data interfaces of the monitoring system in the power grid operation scheduling system. First, the proportion of total equipment resources A... i This information can be obtained by accessing the device allocation data in the resource management module. The system records the type and quantity of device resources allocated to each job task, typically stored as key-value pairs using device ID and task ID. After obtaining the task's resource allocation information, the system can calculate the proportion of resources used by the task, usually the ratio of the task's resource usage to the total resource quantity.

[0079] Secondly, the number of times R requests device resources i This can be obtained by accessing the task scheduling records in the scheduling system. Whenever a job requestes access to device resources, the system records the operation and accumulates the number of requests. The software can extract the resource request records for each job from the job scheduling table or operation log, and obtain the frequency of device resource requests by counting the number of times the task requests access to device resources.

[0080] Finally, the number of resource access conflicts C with other tasks iThis requires analyzing concurrent resource access patterns. Whenever multiple job tasks execute concurrently and attempt to access the same device resource, the system records a conflict event. By analyzing the device resource access logs, the software can identify overlapping accesses to the same device resource across different tasks and calculate the number of resource access conflicts between each job task and other tasks. Typically, a conflict is determined when multiple job tasks request the same device resource at the same time or during overlapping periods. The software can match and count these requests using scheduling records and device usage logs to determine the number of conflicts. This data is collected and processed in real-time by the software system and stored in a relevant database, ensuring that the resource usage, requests, and conflicts for each task can be accurately tracked and calculated.

[0081] The specific formula for calculating the resource conflict intensity coefficient is as follows:

[0082]

[0083] In the formula, RCIC is the resource conflict intensity coefficient.

[0084] The formula used to calculate the Resource Conflict Intensity Coefficient (RCIC) reasonably quantifies the degree of resource interference between tasks by combining resource usage, request frequency, and conflict count. First, A in the formula... i This represents the proportion of total equipment resources allocated to each task, reflecting the intensity of resource consumption by the task. (This is achieved through interaction with R...) i Multiplication also takes into account the frequency of resource requests for tasks, indicating that tasks with more frequent resource requests will have a greater impact on device resource demands and potential conflicts. Secondly, C i The sigmoid function represents the number of resource access conflicts that occur between a task and other tasks. The number of conflicts is smoothed. Through the smoothing effect of an exponential function, the interference between tasks is smaller when the number of conflicts is low, while the impact increases rapidly when the number of conflicts is high. This non-linear approach ensures that the weight of high-conflict tasks is significantly increased in the overall interference coefficient. Finally, the entire formula, through weighted summation, integrates task resource consumption, request frequency, and conflict severity to derive a comprehensive resource conflict intensity coefficient, which effectively reflects the intensity of interference between tasks at the resource level.

[0085] The Resource Conflict Intensity Coefficient (RCIC) is directly related to the degree of interference between tasks within an evaluation cluster of interfering tasks, showing a positive correlation. Specifically, a higher RCIC value indicates higher resource consumption and more frequent resource requests by tasks, along with more frequent resource access conflicts, suggesting a higher degree of interference. In this case, task interference becomes more severe, potentially leading to scheduling conflicts or excessive equipment resource consumption, thus affecting the efficiency and safety of power grid operations. Conversely, a lower RCIC value indicates lower resource demands, fewer resource requests, and fewer conflicts, resulting in less interference and better coordination and stability of power grid operations. RCIC evaluation quantifies the degree of resource interference between tasks, helping the system determine whether scheduling adjustments or optimization of task execution order are necessary to avoid potential conflicts and safety hazards.

[0086] In this embodiment, the logic for obtaining the operation step conflict index is as follows:

[0087] Step interference evaluation information is extracted from the preprocessed parallel job interference feature information. Specifically, this includes the number of steps executed by each job in the operation interference task cluster, the number of conflicts with other tasks in the execution steps, and the proportion of equipment resources occupied during the execution of steps, and these are respectively labeled as S. i P i and U i S i P represents the number of steps executed by the i-th job in the cluster of interfering operations. i U represents the number of conflicts in the execution steps between the i-th job and other tasks in the interfering task cluster. i This represents the proportion of device resources occupied by the i-th job in the interfering task cluster during the execution of steps, where i = 1, 2, 3, ..., g, and g is a positive integer;

[0088] Data such as the number of steps executed by each job task in an operational interference task cluster, the number of conflicts with other tasks in execution steps, and the proportion of equipment resources occupied during execution steps can be obtained through the data interface of the integrated job scheduling system, task execution monitoring system, and equipment resource management module. First, the number of steps (S) i This information can be obtained by extracting the operational steps of each job from the task execution log. Each job typically executes multiple steps according to a predetermined process, and the system can analyze the task execution logs to count the total number of steps for each task. This data is generally stored in a structured format in the task management database, and the software can obtain it by calling an interface.

[0089] Secondly, the number of step conflicts (P)i Resource usage can be monitored in real time by a monitoring system when multiple tasks are executed concurrently. Conflicts occur when two or more tasks access the same device resources during execution. The system can determine the number of times each task conflicts with other tasks during execution by comparing task execution time windows and resource usage status. When a conflict occurs, the system records the conflict event and performs statistical analysis for subsequent purposes.

[0090] Finally, the resource usage ratio of each step (U) i This refers to the proportion of equipment resources used by each job task during its execution steps. The actual equipment resource usage during task execution can be obtained through the resource management module. The system tracks the types and amounts of resources used by each task in each step, and then calculates the proportion of equipment resources used by each task during its execution steps. This data is calculated by monitoring equipment resource usage in real time and combining it with task allocation information from the job scheduling system to obtain the resource usage proportion for each task step. Through this method, the system can accurately and in real-time obtain this data, providing effective support for subsequent interference assessment.

[0091] The specific formula for calculating the operational step conflict index is as follows:

[0092]

[0093] In the formula, OSCI is the Operational Conflict Index.

[0094] The Operational Conflict Index (OSCI) is calculated by combining the number of steps in a task, the frequency of step conflicts, and the proportion of resources used to reasonably assess the intensity of step interference during task execution. First, e Ui Using exponential operations to process the step resource usage ratio U i This is because, during task execution, an increase in resource consumption can have a non-linear amplification effect on other tasks. That is, when a task consumes more resources during its execution, its interference with other tasks increases dramatically. Modeling this dramatic increase using an exponential function better reflects the actual resource consumption interference. Next, P... i This represents the number of conflicts between a task and other tasks in their execution steps. It directly reflects the competition and conflict between tasks during the execution process. The higher the number of conflicts, the greater the interference intensity. Therefore, when multiplied by other factors, it can amplify the impact of conflicts between tasks. Finally, ln(1+S) i The number of steps S iIntroducing logarithmic calculations effectively smooths out the impact of the number of steps, preventing excessive amplification of the interference index when there are too many steps. The logarithmic function ensures that the increase in interference intensity with the increase in the number of steps gradually decreases, thus ensuring a more accurate reflection of the interference effect of each step. By combining these factors, the final Operational Step Conflict Index (OSCI) can comprehensively evaluate the interference intensity between various tasks during task execution and accurately reflect the complex interference relationships between tasks through a nonlinear mathematical model.

[0095] The Operation Conflict Index (OSCI) is positively correlated with the degree of task interference among tasks in the evaluated operation interference task cluster. Specifically, a higher OSCI value means that the task consumes a larger proportion of resources during the execution of its steps (U...). i The value is relatively high, and the frequency of step conflicts between tasks (P) is also high. i The number of steps in the task (S) is relatively high. i A high OSCI (Operational Conflict Index) indicates a high degree of mutual interference between tasks during execution, potentially leading to resource contention, operational delays, and other problems, thus increasing the overall interference within the task cluster. Conversely, a low OSCI value indicates that task execution steps consume fewer resources, have fewer conflicts, and fewer steps, suggesting lower interference between tasks and better coordination among job tasks. Therefore, the Operational Conflict Index effectively reflects the intensity of interference between job tasks at the step execution level, providing an important reference for task scheduling and resource allocation, helping to reduce unnecessary conflicts and optimize the execution order of job tasks.

[0096] In this embodiment, based on the generated Resource Conflict Intensity Index (RCIC) and Operation Step Conflict Index (OSCI), a task interference index is generated by weighted summation. The specific calculation formula is as follows:

[0097] TII=ω1*RCIC+ω2*OSCI

[0098] In the formula, TII is the task interference index, ω1 and ω2 are the non-zero weight coefficients of the resource conflict intensity coefficient (RCIC) and the operation step conflict index (OSCI), respectively, and ω1+ω2=1.

[0099] To calculate the Task Interference Index (TII), a weighted sum is first needed based on the already generated Resource Conflict Intensity Index (RCIC) and Operational Step Conflict Index (OSCI). After extracting and calculating these two conflict indices from the task data, the system assigns appropriate non-zero weight coefficients (ω1 and ω2) to them according to the specific requirements of the task. These two weight coefficients reflect the relative importance of resource conflict and operational step conflict in the task interference assessment. Typically, these weight coefficients are determined through historical data or experimental analysis and adjusted based on the nature of the task and the actual working environment. To ensure the rationality of these two weight coefficients, ω1 + ω2 = 1, meaning their sum is 1, ensuring a balanced overall interference assessment. By adjusting the weight coefficients, the system can flexibly and precisely adjust the impact of resource conflict and operational step conflict on the task interference index according to the characteristics and working conditions of different tasks, ultimately outputting an accurate Task Interference Index (TII). These calculations can be automated through software, ensuring dynamic adaptability and efficiency under different tasks and environments.

[0100] In this embodiment, a pre-defined task interference index threshold range [TII] is determined. min TII max After determination, it is compared with the generated Task Interference Index (TII). Based on the comparison, the degree of task interference between each job task in the operational interference task cluster is evaluated. The specific comparison analysis is as follows:

[0101] If TII <TII min The degree of task interference between tasks in the inter-task cluster is low.

[0102] This indicates that there are few resource and step conflicts among the tasks in the interfering task cluster, and the task execution processes are relatively independent and well-coordinated. Low interference means that the resource consumption and execution steps of the tasks have little mutual impact, the tasks can be performed efficiently, and the task scheduling system does not need frequent adjustments. This situation usually indicates that the system is running stably, the tasks can be executed smoothly as planned, and resource utilization is relatively efficient.

[0103] If TII min ≤TII≤TII max The degree of task interference between tasks in the inter-task cluster is moderate.

[0104] This situation indicates that there is a certain degree of resource conflict and step interference between the tasks, but it has not yet reached the level of serious interference. This suggests that some resource contention or step conflicts may occur during task execution, requiring some system intervention and optimization, such as appropriately adjusting the task execution order and reallocating resources to reduce interference. Although system operation may be affected to some extent, it can generally still maintain basic operational efficiency and stability.

[0105] If TII>TII max The degree of task interference between tasks in an operational interference task cluster is defined as the severity.

[0106] This situation indicates significant resource and step conflicts between tasks, which may have a substantial impact on task execution. Severe interference signifies poor coordination between tasks and intense resource contention, potentially leading to task delays, equipment overload, or operational failures. In such cases, the system needs immediate intervention, which may include rescheduling tasks, adjusting task priorities, increasing resource allocation, or suspending the execution of some tasks to avoid affecting the normal operation and safety of the entire power grid.

[0107] Based on the evaluation results, dynamic adjustments are made to the task input dataset used for risk prediction.

[0108] In this embodiment, dynamic adjustment is performed on the task input dataset used for risk prediction based on the evaluation results, specifically as follows:

[0109] If the assessment result is low, the original state of the task input dataset used for risk prediction is maintained, and no adjustments are made to the data to ensure that the data is used for risk prediction within the pre-set normal range.

[0110] When the assessment result is low, the original state of the task input dataset used for risk prediction is maintained without adjustment, and risk prediction analysis can be achieved by directly utilizing the original data. This process first relies on the system automatically identifying the degree of interference between tasks. If the assessment result shows that the task interference is small, the system can confirm that the original data is stable enough and suitable for subsequent risk prediction analysis. In this case, the software system will automatically extract the recorded original task data by accessing the task information in the database. This data includes key features such as the frequency of resource requests and execution steps of the task, and this data does not exceed the preset normal range. Since the interference is small, the original data can provide sufficient prediction accuracy, so no additional adjustment or weighting is required. In this way, unnecessary calculations or processing can be avoided when the interference is small, ensuring efficient use of system resources, while maintaining the integrity and consistency of task data, and avoiding errors or inaccurate prediction results that may be caused by excessive intervention.

[0111] If the assessment result is moderate, the input dataset of the task for risk prediction is weighted and adjusted to increase the weight of tasks involving high-conflict tasks in the task data, so that the impact of these tasks on risk prediction is more significant.

[0112] When the assessment result is moderate, weighting the input dataset of job tasks can effectively improve the influence of high-conflict tasks in risk prediction. This process first requires the system to analyze the conflict between job tasks and identify those with high conflict levels. Typically, the system calculates the conflict level of each task based on the frequency of resource requests, overlap in execution steps, and other conflict factors. When a task has a high conflict level, the system automatically assigns it a higher weight. This weighting process can amplify the relevant data of conflicting tasks proportionally, such as applying weighting coefficients to data on the number of resource requests and conflict frequency, thereby increasing the influence of high-conflict tasks. In implementing this process, the software can set dynamic weights based on task conflict, and the weighting coefficients can be adjusted based on historical data or real-time analysis results, ensuring that tasks with higher conflict levels receive more attention in subsequent risk prediction. This method helps ensure that, even with moderate task interference, the prediction model can identify and prioritize high-risk tasks, preventing them from potentially impacting job safety. This approach improves the accuracy and reliability of predictions while making task scheduling more scientific and rational.

[0113] If the assessment result is severity, the task input dataset used for risk prediction is filtered and reordered, prioritizing data from high-interference tasks to ensure that key tasks are given priority in risk prediction.

[0114] When the assessment result is severity, the process of filtering and reordering the input dataset of job tasks aims to ensure that high-interference tasks are prioritized in risk prediction. The software identifies tasks with high interference risk by analyzing conflicts and interference between tasks. These tasks typically consume significant equipment resources, have high resource request frequency, or involve frequent step conflicts. First, the system analyzes all job task data and assigns an interference weight to each task based on its interference level; higher weights indicate greater interference. Then, the software filters tasks based on these interference weights, prioritizing tasks with higher interference levels and removing or delaying tasks with lower interference levels. Next, the software reorders the task data according to their interference weights, ensuring that high-interference tasks are prioritized for evaluation. This ensures that these tasks receive more attention during risk prediction, preventing the potential threat of high-interference tasks to power grid operations from being overlooked. In this way, the software optimizes task scheduling and resource allocation, reduces potential risks caused by high-interference tasks, and ensures job safety and system stability.

[0115] Risk prediction is performed based on the regulated input data, the risk results for each task are output, and the risk results and corresponding task conditions are recorded to optimize the subsequent risk prediction process.

[0116] The process of risk prediction based on regulated input data first relies on the regulation and optimization of task data. When the regulated input dataset is used for risk prediction by an algorithmic model (such as a machine learning or deep learning model), the system analyzes the combined impact of various factors such as task conflicts, resource consumption, and operational steps. The software system accesses the regulated data and inputs it into the risk assessment model. The model evaluates each task individually and outputs the risk result for each task, typically including a risk level or specific risk score. Through risk prediction of tasks, the system can promptly identify high-risk tasks, providing data support for scheduling and resource allocation.

[0117] Furthermore, the system records the risk outcomes and corresponding operational details for each task, generating a detailed risk assessment report. These records include information such as task execution time, resource allocation, interference levels, and predicted risk levels. The purpose of this recording is to optimize subsequent risk prediction processes by accumulating historical data. The software analyzes these historical risk assessment results, identifies patterns and regularities, and adjusts and optimizes the parameters and algorithms of the risk prediction model, making future risk predictions more accurate. This optimization process based on historical data improves the model's accuracy and reliability, ensuring the safety of power grid operations.

[0118] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0119] 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 wired or wireless (e.g., infrared, wireless, microwave, etc.) means. A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes 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.

[0120] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0121] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0122] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units 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 an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0123] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0124] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0125] 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.

Claims

1. A method for predicting safety inspection risks in power grid operations by integrating artificial intelligence technology, characterized in that: Specifically, the following steps are included: During power grid operation and inspection, the received power grid inspection and dispatch data is analyzed to identify all operation tasks currently being executed in parallel within the same power equipment area; Perform operational logic analysis on all identified parallel execution tasks, filter out all tasks with operational conflicts, and form a cluster of tasks with operational interference. Obtain the interference characteristics of parallel jobs in the inter-operational interference task cluster, analyze them, and evaluate the degree of task interference between each job in the inter-operational interference task cluster. Specifically, the following steps are included: Obtain the interference feature information of parallel jobs in the interfering task cluster and perform preprocessing after acquisition; Resource interference analysis information and step interference evaluation information are extracted from the preprocessed parallel operation interference feature information, and then analyzed to generate resource conflict intensity coefficient and operation step conflict index, respectively. The logic for obtaining the resource conflict intensity coefficient is as follows: Resource interference analysis information is extracted from the preprocessed parallel job interference feature information. Specifically, this includes the proportion of total device resources allocated to each job in the interfering task cluster, the number of times device resources are requested, and the number of resource access conflicts with other tasks. These are then labeled as follows: , and , Indicates the first in the cluster of tasks that interfere with each other. The proportion of total equipment resources allocated to each task. Indicates the first in the cluster of tasks that interfere with each other. The number of times a task requests device resources. Indicates the first in the cluster of tasks that interfere with each other. The number of times a task encounters resource access conflicts with other tasks. , It is a positive integer; The specific formula for calculating the resource conflict intensity coefficient is as follows: In the formula, This is the resource conflict intensity coefficient; The logic for obtaining the conflict index in the operation steps is as follows: Step interference evaluation information is extracted from the preprocessed parallel job interference feature information. Specifically, this includes the number of steps executed by each job in the operational interference task cluster, the number of conflicts with other tasks in step execution, and the proportion of equipment resources occupied during step execution. These are then labeled as follows: , and , Indicates the first in the cluster of tasks that interfere with each other. The number of steps executed in each job task Indicates the first in the cluster of tasks that interfere with each other. The number of times a task conflicts with other tasks in terms of execution steps. Indicates the first in the cluster of tasks that interfere with each other. The proportion of equipment resources used by each task during its execution steps. , It is a positive integer; The calculation formula for the operation step conflict index is as follows: In the formula, The conflict index is the number of steps involved in the operation. Based on the generated resource conflict intensity coefficient and operation step conflict index, a task interference index is generated by weighted summation. Determine the pre-set threshold range for the task interference index, and compare it with the generated task interference index after determination. Based on the comparison, evaluate the degree of task interference between each task in the task cluster. Based on the evaluation results, dynamic adjustments are made to the task input dataset used for risk prediction. Risk prediction is performed based on the regulated input data, the risk results for each task are output, and the risk results and corresponding task conditions are recorded to optimize the subsequent risk prediction process.

2. The power grid operation safety inspection risk prediction method integrating artificial intelligence technology according to claim 1, characterized in that, All identified parallel execution tasks are subjected to operational logic analysis to filter out all tasks with operational conflicts and form a cluster of tasks with operational interference, specifically: Based on all the identified parallel-executed job tasks, extract the operation steps, equipment target information and safety dependency rules for each job task. The execution order relationship between tasks is analyzed based on operation step information; the device access relationship between tasks is analyzed based on device target information; and the control condition relationship between tasks is analyzed based on security dependency rules. Based on the analysis results, all job tasks with operational conflicts were identified and grouped into a cluster of interfering job tasks.

3. The power grid operation safety inspection risk prediction method integrating artificial intelligence technology according to claim 2, characterized in that, Based on the generated resource conflict intensity coefficient Conflict index with operating procedures The task interference index is generated by weighted summation, and the specific calculation formula is as follows: In the formula, This represents the task interference index. and Resource conflict intensity coefficients Conflict index with operating procedures The non-zero weight coefficients, and .

4. The power grid operation safety inspection risk prediction method integrating artificial intelligence technology according to claim 3, characterized in that, Determine the pre-defined threshold range for task interference index. And after determination, it is compared with the generated task interference index. A comparison was performed, and the degree of task interference between each job task in the operational interference task cluster was evaluated based on the comparison. The specific comparison analysis is as follows: like The degree of task interference between tasks in the inter-task cluster is low. like The degree of task interference between tasks in the inter-task cluster is moderate. like The degree of task interference between tasks in an operational interference task cluster is defined as the severity.

5. The power grid operation safety inspection risk prediction method integrating artificial intelligence technology according to claim 4, characterized in that, Based on the evaluation results, dynamic adjustments are performed on the task input dataset used for risk prediction, specifically: If the assessment result is low, the original state of the task input dataset used for risk prediction is maintained, and no data is adjusted to ensure that the data is used for risk prediction within the pre-set normal range. If the assessment result is moderate, the weighted adjustment of the task input dataset used for risk prediction is performed, increasing the weight of tasks involving high conflict in the task data; If the assessment result is severity, the input dataset for the job tasks used for risk prediction is filtered and reordered, prioritizing data from tasks with high interference.

Images