Intelligent monitoring system for power transmission lines based on 5G network slicing

Abstract

This invention relates to the field of power transmission line monitoring technology and provides an intelligent monitoring system for power transmission lines based on 5G network slicing. The system includes: a slice configuration module for establishing a 5G slice-power transmission line monitoring model; a state perception module for obtaining a spatiotemporal perception model of equipment state and environmental impact; an anomaly discrimination module for obtaining an optimized anomaly discrimination model for power transmission lines based on the 5G slice-power transmission line monitoring model; a strategy generation module for obtaining the spatiotemporal distribution characteristics of equipment state based on the equipment state-environmental impact spatiotemporal perception model, performing strategy analysis based on the equipment state spatiotemporal distribution characteristics and the optimized anomaly discrimination model for power transmission lines, and determining the power transmission line monitoring strategy parameters; and a feedback optimization module for optimizing the power transmission line monitoring strategy parameters. This invention ensures the timeliness and integrity of data transmission, improves the accuracy and adaptability of power transmission line monitoring, and effectively guarantees the stable operation of power transmission lines.

Description

Technical Field

[0001] This invention relates to the field of power transmission line monitoring technology, specifically to an intelligent monitoring system for power transmission lines based on 5G network slicing. Background Technology

[0002] With the continuous expansion of power systems, transmission lines, as the key carriers of power transmission, directly affect the continuity of power supply through their operational stability. Currently, transmission lines are often distributed in areas with complex geographical environments, encompassing various terrains such as mountains, hills, and deserts. They face multiple challenges, including extreme weather, geological disasters, and equipment aging. Traditional monitoring methods are no longer sufficient to meet the real-time, accuracy, and reliability requirements of modern power systems.

[0003] Traditional power transmission line monitoring relies heavily on a combination of manual inspections and single-sensor monitoring. Manual inspections are not only resource-intensive but also limited by inspection cycles and geographical conditions, making it difficult to achieve 24 / 7, comprehensive monitoring of the lines. This is especially true in severe weather or remote areas, where inspections face significant challenges. Single-sensor monitoring, on the other hand, suffers from limited data acquisition dimensions and severe information silos. Data from different types of sensors is difficult to correlate effectively, resulting in a one-sided understanding of the power transmission line's operational status.

[0004] In terms of data transmission, traditional communication networks often suffer from insufficient bandwidth and excessive latency when dealing with massive amounts of monitoring data. Real-time data generated by power transmission line monitoring points includes equipment operating parameters, ambient temperature and humidity, wind speed and direction, etc. This data needs to be transmitted to the monitoring center for processing in a timely manner. However, the existing network architecture lacks support for differentiated transmission of different types of data. Important data may be delayed due to network congestion, affecting the timely response to abnormal situations.

[0005] Existing monitoring systems have significant shortcomings in data analysis capabilities. Most systems can only perform simple analysis on single types of data, lacking in-depth fusion and spatiotemporal correlation analysis of multi-source data. There are complex dynamic relationships between equipment operating status and environmental factors. For example, high temperatures may accelerate equipment aging, and strong winds may increase the amplitude of line sway. Traditional analysis methods struggle to capture these spatiotemporal interactions, resulting in low accuracy in identifying abnormal equipment states and a high likelihood of misjudgments or omissions.

[0006] Existing anomaly detection models are often based on fixed feature parameters, making it difficult to adapt to the dynamic changes in the transmission line operating environment. When the external environment or equipment operating status changes, the model's accuracy drops significantly, requiring manual parameter readjustment, which not only increases maintenance costs but also affects the timeliness of anomaly identification. In the strategy generation stage, existing systems rely heavily on empirical decisions, lacking in-depth analysis of the spatiotemporal distribution patterns of equipment status. This results in insufficient targeting of the generated monitoring strategies, making it difficult to dynamically adjust them according to actual conditions.

[0007] With the development of 5G technology, its low latency and high bandwidth characteristics have provided new possibilities for power transmission line monitoring. However, how to deeply integrate 5G network slicing technology with power transmission line monitoring to achieve dynamic resource allocation and efficient data transmission remains an unsolved problem. In existing technologies, the configuration of 5G network resources lacks precise matching with power transmission line monitoring data, making it difficult to meet the differentiated needs of different monitoring scenarios and hindering the improvement of intelligent monitoring levels for power transmission lines. Summary of the Invention

[0008] The purpose of this invention is to provide an intelligent monitoring system for power transmission lines based on 5G network slicing, thereby solving the problems mentioned in the background section. This invention adopts the following technical solution:

[0009] The intelligent monitoring system for power transmission lines based on 5G network slicing includes:

[0010] The slice configuration module is used to acquire 5G network parameters and power transmission line monitoring data, define slice service attributes, configure slice resources based on the slice service attributes for the 5G network parameters and power transmission line monitoring data, and establish a 5G slice-power transmission line monitoring model.

[0011] The state perception module is used to obtain equipment operation dataset and environmental monitoring dataset by association through multi-source data fusion technology, and to perform state perception modeling on the equipment operation dataset and environmental monitoring dataset using spatiotemporal analysis to obtain a spatiotemporal perception model of equipment state-environmental impact.

[0012] The anomaly detection module is used to initialize feature extraction parameters based on the 5G slice-transmission line monitoring model, optimize the feature extraction parameters for transmission line anomaly detection according to the anomaly identification target, and iteratively obtain the optimized transmission line anomaly detection model.

[0013] The strategy generation module is used to obtain the spatiotemporal distribution characteristic information of equipment status based on the equipment status-environmental impact spatiotemporal perception model, perform strategy analysis based on the equipment status spatiotemporal distribution characteristic information and the transmission line anomaly optimization discrimination model, and determine the transmission line monitoring strategy parameters.

[0014] The feedback optimization module is used to simulate and verify the parameters of the transmission line monitoring strategy, obtain the implementation effect of the monitoring strategy, and optimize the parameters of the transmission line monitoring strategy based on the implementation effect of the monitoring strategy.

[0015] Preferably, the establishment of the 5G slice-transmission line monitoring model includes:

[0016] Determine the slice resource attributes and resource connection edge attributes based on the slice service attributes;

[0017] Based on the slice resource attributes, resource identification and attribute labeling are performed on the 5G network parameters and power transmission line monitoring data to obtain a slice resource set and a slice resource attribute set.

[0018] The connection relationships of the slice resource set are analyzed using the 5G network parameters and power transmission line monitoring data to construct a resource proximity connection set.

[0019] Based on the resource connection edge attributes, the resource proximity connection set is sliced ​​and abstracted to obtain the sliced ​​resource topology adjacency matrix;

[0020] Based on the slice resource topology adjacency matrix and the slice resource attribute set, the slice resource set is connected by slice identification to establish the 5G slice-transmission line monitoring model.

[0021] Preferably, the method for obtaining the spatiotemporal perception model of device status and environmental impact includes:

[0022] Spatiotemporal analysis was used to perform spatiotemporal arrangement and characteristic analysis on the equipment operation dataset and environmental monitoring dataset to obtain the equipment status characteristic dataset and environmental impact characteristic dataset.

[0023] The equipment state characteristic dataset and the environmental impact characteristic dataset are trained using random forest to obtain the equipment state perception model and the environmental impact perception model, respectively.

[0024] The equipment state perception model and the environmental impact perception model are fused in parallel to generate an initial equipment state spatiotemporal perception model.

[0025] The initial equipment state spatiotemporal perception model is verified and adjusted using a model calibrator to obtain the equipment state-environment impact spatiotemporal perception model.

[0026] Preferably, the iterative acquisition of the transmission line anomaly optimization discrimination model includes:

[0027] Anomaly evaluation indicators are extracted from the anomaly identification target to obtain an anomaly evaluation indicator set, and an anomaly discrimination effect fitness function is constructed based on the anomaly evaluation indicator set.

[0028] Based on the feature extraction parameters, determine the individual gene and individual mutation rate of the parameter, and use the anomaly discrimination effect fitness function to evaluate the fitness of the feature extraction parameters to obtain a feature extraction fitness set;

[0029] Based on the feature extraction fitness set, the parameter individual genes and parameter individual mutation rates are iteratively evolved until a preset termination condition is met, and the optimal parameter individual with the largest fitness is determined. The optimal parameter individual includes transmission line anomaly discrimination features.

[0030] The 5G slice-transmission line monitoring model is optimized and updated based on the anomaly discrimination features of the transmission line to obtain the optimized anomaly discrimination model of the transmission line.

[0031] Preferably, determining the transmission line monitoring strategy parameters includes:

[0032] Based on the transmission line anomaly optimization discrimination model, the equipment state distribution impact analysis is performed to obtain the equipment state distribution impact parameters, which include line operation impact parameters and monitoring response impact parameters.

[0033] A monitoring strategy library is constructed. The spatiotemporal distribution characteristics of the equipment status are combined with the influence parameters of the equipment status distribution. The monitoring strategy library is matched and analyzed to obtain the monitoring and matching control strategy for the transmission line.

[0034] Based on the transmission line monitoring and matching control strategy, the spatiotemporal distribution characteristics of the equipment status and the parameters affecting the equipment status distribution are divided into selection thresholds to obtain the monitoring strategy parameter selection thresholds.

[0035] The anomaly detection effect fitness function is used to perform global optimization within the threshold value of the monitoring strategy parameters to determine the monitoring strategy parameters of the transmission line.

[0036] Preferably, the step of using the anomaly detection effect fitness function to perform global optimization within the threshold value for selecting the monitoring strategy parameters to determine the transmission line monitoring strategy parameters includes:

[0037] Multiple strategy parameters are randomly selected within the threshold range of the monitoring strategy parameters, and the multiple strategy parameters are evaluated using the anomaly discrimination effect fitness function to obtain the fitness of multiple parameters;

[0038] Based on the fitness of the multiple parameters, the threshold for selecting the monitoring strategy parameters is approximated to determine the local optimization region of the strategy parameters.

[0039] Set the parameter search step size according to the local optimization region of the strategy parameters;

[0040] The strategy parameters are searched and evaluated within the local optimization region of the strategy parameters according to the parameter search step size, and the optimization region is iteratively approximated based on the parameter search evaluation results until the preset number of iterations is reached. The transmission line monitoring strategy parameters are then determined by parameter fitness comparison.

[0041] Preferably, the step of optimizing the transmission line monitoring strategy parameters based on the implementation effect of the monitoring strategy includes:

[0042] Based on the implementation effect of the monitoring strategy, the optimization direction of the transmission line monitoring strategy parameters is analyzed to obtain parameter variation optimization rules.

[0043] The parameters of the transmission line monitoring strategy are mutated and expanded according to the parameter mutation optimization rules to obtain a monitoring strategy parameter cluster;

[0044] Within the monitoring strategy parameter cluster, parameter comparison and optimization are performed to obtain monitoring and control optimization strategy parameters, and the transmission line monitoring and optimization control is performed through the monitoring and control optimization strategy parameters.

[0045] Preferably, the step of using spatiotemporal analysis to perform spatiotemporal arrangement and characteristic analysis on the equipment operation dataset and the environmental monitoring dataset includes:

[0046] The equipment operation dataset is divided into time series to obtain equipment status time series classified by hour, day, and month.

[0047] The environmental monitoring dataset is spatially divided to obtain the spatial regions of environmental impact classified by tower, section, and line.

[0048] Based on the time series of equipment status and the spatial region of environmental impact, the equipment operation dataset and the environmental monitoring dataset are jointly arranged to construct a spatiotemporal grid data matrix.

[0049] Feature values ​​are extracted from the spatiotemporal grid data matrix to obtain equipment state fluctuation characteristics and environmental impact gradient characteristics, forming the equipment state characteristic dataset and the environmental impact characteristic dataset.

[0050] Preferably, the step of parsing the connection relationships of the slice resource set using the 5G network parameters and power transmission line monitoring data to construct a resource proximity connection relationship set includes:

[0051] Slice bandwidth analysis is performed on the 5G network parameters to obtain slice bandwidth allocation information;

[0052] The monitoring data of the transmission line is used to locate the monitoring nodes and obtain the location information of the monitoring nodes;

[0053] Based on the slice bandwidth allocation information and monitoring node location information, the connection strength of the slice resource set is calculated to determine the resource connection strength threshold.

[0054] Select slice resource pairs with connection strength greater than the resource connection strength threshold, and construct the resource proximity connection relationship set.

[0055] Preferably, the step of parallel fusion of the device state perception model and the environmental impact perception model to generate an initial device state spatiotemporal perception model includes:

[0056] Set the operating state weight coefficients for the device state perception model;

[0057] Set environmental factor weight coefficients for the environmental impact perception model;

[0058] The outputs of the equipment state perception model and the environmental impact perception model are weighted and summed based on the weight coefficients of the operating state and the weight coefficients of the environmental factors to generate the initial equipment state spatiotemporal perception model.

[0059] Compared with the prior art, the beneficial effects of the present invention are:

[0060] The slicing configuration module enables precise matching of 5G network parameters and power transmission line monitoring data. Based on the slice service attributes, slice resources are configured and a 5G slice-power transmission line monitoring model is established. This allows network resources to be dynamically adjusted according to monitoring needs, avoiding data transmission delays or congestion caused by unreasonable resource allocation in traditional networks. It ensures that different types of power transmission line monitoring data, such as equipment operation data and environmental monitoring data, can be transmitted efficiently in dedicated network slices, ensuring the timeliness and integrity of data transmission.

[0061] By employing multi-source data fusion technology through a state-aware module to obtain equipment operation datasets and environmental monitoring datasets, and then establishing a spatiotemporal perception of equipment state and environmental impact through spatiotemporal analysis, this approach overcomes the limitations of traditional single-data monitoring. This method can capture the complex temporal and spatial relationships between equipment operating status and environmental factors. For example, it can detect the gradual impact of continuous rainfall in a certain area on the insulation performance of surrounding equipment, or the dynamic relationship between temperature changes and equipment load over different time periods, thereby providing a more comprehensive understanding of the overall operating status of transmission lines.

[0062] The anomaly detection module initializes feature extraction parameters based on the established 5G slice-transmission line monitoring model and optimizes the anomaly detection based on the anomaly identification target. The resulting optimized anomaly detection model can better adapt to the complex operating environment of transmission lines. Traditional anomaly detection methods often struggle to cope with changing anomalies due to fixed parameters, while the anomaly detection module, through continuous parameter optimization, can improve the ability to identify various anomalies, whether they are potential faults in the equipment itself or transmission line anomalies caused by the external environment, enabling more accurate identification.

[0063] By combining the spatiotemporal distribution characteristics of equipment status with the transmission line anomaly optimization discrimination model through the strategy generation module, the determined transmission line monitoring strategy parameters are more targeted and adaptable. Based on the status distribution of equipment in different times and spaces, as well as the identified anomalies, the generated transmission line monitoring strategy parameters can cover all key nodes and potential risk areas of the transmission line, changing the situation of traditional monitoring strategies that are highly general but lack specificity, thus enabling monitoring work to be more targeted.

[0064] The feedback optimization module simulates and verifies the parameters of the transmission line monitoring strategy and optimizes them based on the implementation effect of the monitoring strategy, forming a closed-loop optimization mechanism. Through simulation verification, potential deficiencies in the transmission line monitoring strategy can be identified in advance. Adjustments can then be made based on the actual implementation effect of the monitoring strategy, enabling the transmission line monitoring strategy to dynamically optimize with changes in the operating status and environment of the transmission line. This ensures that the entire monitoring system maintains a consistently high level of efficiency and better guarantees the stable operation of the transmission line. Attached Figure Description

[0065] Figure 1 A schematic diagram illustrating the working principle of the intelligent monitoring system for power transmission lines based on 5G network slicing provided by this invention;

[0066] Figure 2 This is a flowchart of the 5G slice-transmission line monitoring model established in this invention;

[0067] Figure 3 This is a flowchart of the spatiotemporal perception model for obtaining the device status-environmental influence in this invention;

[0068] Figure 4 This is a flowchart of the iterative acquisition of the transmission line anomaly optimization discrimination model in this invention;

[0069] Figure 5 This is a flowchart for determining the parameters of the transmission line monitoring strategy in this invention. Detailed Implementation

[0070] 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.

[0071] Example 1: Please refer to Figures 1-5 The intelligent monitoring system for power transmission lines based on 5G network slicing provided in this embodiment includes a slice configuration module, a status perception module, an anomaly detection module, a strategy generation module, and a feedback optimization module. The specific functions and interaction methods of each module are as follows:

[0072] The slice configuration module is used to acquire 5G network parameters and power transmission line monitoring data, define slice service attributes, and configure slice resources based on these attributes to establish a 5G slice-power transmission line monitoring model. The 5G network parameters include performance indicators such as network bandwidth, latency, and jitter, while the power transmission line monitoring data includes operational data such as line temperature, current, voltage, and tower tilt. Slice service attributes are determined according to monitoring requirements, such as real-time requirements and data transmission priority.

[0073] The state awareness module is used to obtain equipment operation datasets and environmental monitoring datasets through multi-source data fusion technology. It then uses spatiotemporal analysis to perform state awareness modeling on these datasets, resulting in a spatiotemporal awareness model of equipment state-environmental impact. The multi-source data fusion technology integrates data from various sources such as sensors, cameras, and weather stations. The equipment operation dataset covers real-time operating parameters of various components of the transmission line, while the environmental monitoring dataset includes environmental factor data such as wind speed, rainfall, and temperature.

[0074] The anomaly detection module is used to initialize feature extraction parameters based on the 5G slice-transmission line monitoring model. It then optimizes these parameters for transmission line anomaly detection according to the anomaly identification targets, iteratively obtaining an optimized anomaly detection model. Feature extraction parameters include data fluctuation thresholds and anomaly feature dimensions. Anomaly identification targets include fault types such as line short circuits, equipment aging, and icing.

[0075] The strategy generation module is used to obtain the spatiotemporal distribution characteristics of equipment status based on the equipment status-environment impact spatiotemporal perception model. Based on this information and the transmission line anomaly optimization discrimination model, it performs strategy analysis to determine the transmission line monitoring strategy parameters. The equipment status spatiotemporal distribution characteristics reflect the operational status patterns of equipment at different times and locations. The transmission line monitoring strategy parameters include monitoring frequency, early warning thresholds, and operation and maintenance scheduling schemes.

[0076] The feedback optimization module is used to simulate and verify the parameters of the transmission line monitoring strategy, obtain the implementation effect of the monitoring strategy, and optimize the transmission line monitoring strategy parameters based on the implementation effect. The simulation verification is achieved by simulating the strategy execution process under different operating conditions. The implementation effect of the monitoring strategy includes evaluation indicators such as fault detection rate, false alarm rate, and response time.

[0077] Example 2: Based on the above, the establishment of the 5G slicing-transmission line monitoring model includes:

[0078] The slice resource attributes and resource connection edge attributes are determined based on the slice service attributes.

[0079] The slice service attributes are defined based on the actual monitoring needs of transmission lines, covering real-time data transmission requirements, minimum transmission bandwidth standards, data transmission priority allocation, and reliability indicators. Based on these slice service attributes, slice resource attributes and resource connection edge attributes are further clarified. Slice resource attributes include resource type (e.g., computing resources, storage resources, transmission resources); resource capacity (e.g., processing power of computing resources, storage space size of storage resources, bandwidth limit of transmission resources); resource availability duration (how long a resource can be used by the system); and resource latency characteristics (the time delay in data processing or transmission). Resource connection edge attributes include connection bandwidth (the upper limit of data transmission rate between two resources); connection stability (measured by the number of connection interruptions per unit time); connection transmission latency (the time taken to transmit data from one resource to another); and connection encryption level (used to ensure data security during transmission, with different levels corresponding to different encryption algorithms and key management methods).

[0080] Based on slice resource attributes, resource identification and attribute labeling are performed on 5G network parameters and power transmission line monitoring data to obtain slice resource sets and slice resource attribute sets.

[0081] The 5G network parameters include information such as base station coverage, signal strength, available bandwidth, network latency, and packet loss rate; the power transmission line monitoring data includes data such as line current, voltage, temperature, tower tilt angle, conductor sag, and insulator pollution level. During resource identification, based on the type characteristics in the slice resource attributes, various transmission and computing resources are identified from the 5G network parameters, and sensor acquisition and data storage resources are identified from the power transmission line monitoring data. Subsequently, each identified resource is labeled with attributes, associating its capacity, latency characteristics, and other attributes with the corresponding resource. This process forms a slice resource set, which is a summary of all identified resources; simultaneously, a slice resource attribute set is formed, recording the attribute information corresponding to each resource.

[0082] By analyzing the connectivity relationships of the slice resource set using 5G network parameters and power transmission line monitoring data, a set of resource proximity connectivity relationships is constructed.

[0083] First, the 5G network parameters are analyzed to clarify the connection methods between various network resources, such as the connection between base stations and the core network, the connection between the core network and the data processing center, and the specific paths and dependencies of these connections. For power transmission line monitoring data, the communication relationships between monitoring devices are analyzed, such as the connections between various sensors and data aggregation nodes, and the connections between data aggregation nodes and remote monitoring centers. Based on these analyses, the connection relationships between various resources in the slice resource set are parsed to determine which resources have direct data interaction and which resources are indirectly connected through intermediate resources. For resource pairs with connection relationships, they are recorded to form a resource proximity connection relationship set, which details the proximity connections between resources.

[0084] Based on the resource connection edge attributes, the set of resource proximity connections is sliced ​​and abstracted to obtain the sliced ​​resource topology adjacency matrix.

[0085] In this set of resource proximity connections, each connection corresponds to a feature in the resource connection edge attributes, such as connection bandwidth and transmission latency. During the slicing abstraction process, the connections are quantized based on these attributes; for example, connection bandwidth is converted into specific numerical values ​​according to certain rules, and transmission latency is also converted into corresponding quantized values. Then, using the resources in the slice resource set as the rows and columns of a matrix, the element values ​​are determined based on the quantized results of the connection between the corresponding two resources. If a connection exists between two resources, the element value is the quantized value of that connection; otherwise, the element value is 0. Through this processing, the set of resource proximity connections is transformed into a slice resource topological adjacency matrix, which clearly shows the topological connections between each resource and the quantized characteristics of those connections.

[0086] Based on the slice resource topology adjacency matrix and slice resource attribute set, slice identification and connection are performed on the slice resource set to establish a 5G slice-transmission line monitoring model.

[0087] Each resource in the slice resource set has a unique identifier. Combined with the resource connection relationships presented by the slice resource topology adjacency matrix, each resource is identified and connected according to its connection status. During the connection process, the slice resource attribute set is also referenced, ensuring that each connection reflects not only the topological relationship between resources but also the attribute characteristics of the resources. For example, when two transmission resources are connected through a certain connection relationship, their bandwidth, latency, and other attribute information are also incorporated into the connection description. In this way, the scattered resources, their attributes, and connection relationships are integrated to form a complete 5G slice-transmission line monitoring model. The 5G slice-transmission line monitoring model can comprehensively reflect the correlation between 5G network slice resources and transmission line monitoring data, providing basic support for subsequent status awareness modules, anomaly detection modules, etc., and realizing effective management and scheduling of transmission line monitoring resources.

[0088] Example 3: Based on the above, a spatiotemporal perception model of equipment status and environmental impact is obtained, including:

[0089] Spatiotemporal analysis was used to perform spatiotemporal arrangement and characteristic analysis on the equipment operation dataset and the environmental monitoring dataset to obtain the equipment status characteristic dataset and the environmental impact characteristic dataset.

[0090] The equipment operation dataset contains real-time operating parameters of various components of the transmission line, such as conductor temperature, vibration frequency, and current carrying capacity; tower verticality and mechanical stress at connections; and insulator leakage current and surface temperature. This data is collected in real-time by sensors deployed along the line and stored with timestamps. The environmental monitoring dataset covers various parameters of the environment in which the transmission line is located, including wind speed, wind direction, precipitation, air humidity, ambient temperature, air pressure, light intensity, icing thickness, and vegetation height. Data sources include weather stations, drone inspection equipment, and infrared imagers.

[0091] In practice, spatiotemporal analysis is used to perform spatiotemporal arrangement and characteristic analysis of the equipment operation dataset and the environmental monitoring dataset, including:

[0092] The equipment operation dataset is divided into time series segments, constructing equipment status time series hierarchically by hour, day, and month. Specifically, the hourly series records the average, maximum, minimum, and fluctuation range of various parameters within that hour; the daily series summarizes the changing trends of the hourly data for the day, marking the peak and stable periods of each day; and the monthly series integrates the characteristics of the daily data for the month to analyze the operational patterns within the month.

[0093] The environmental monitoring dataset is spatially divided, and the spatial areas of environmental impact are determined according to towers, sections, and lines. Specifically, an individual tower is used as the smallest unit to record environmental parameters within a 50-meter radius around the tower; several adjacent towers are divided into sections, and the differences in the distribution of environmental parameters within the section are statistically analyzed; the entire transmission line is used as the unit, and the environmental data of all sections are integrated to form the line-level environmental distribution characteristics.

[0094] Based on the time series of equipment status and the spatial regions of environmental impact, a spatiotemporal grid data matrix is ​​constructed by jointly arranging the equipment operation dataset and the environmental monitoring dataset. The row dimensions of the spatiotemporal grid data matrix correspond to different spatial regions (towers, sections, lines), and the column dimensions correspond to different time nodes (hours, days, months). Each element in the spatiotemporal grid data matrix contains combined information on equipment operation parameters and environmental parameters at the corresponding spatiotemporal location. This arrangement allows for a clear visualization of the correspondence between equipment status and environmental conditions in a specific region at a given moment, as well as the status changes of the same region at different times and the differences between different regions at the same time.

[0095] Feature value extraction was performed on the spatiotemporal grid data matrix to obtain equipment status fluctuation characteristics and environmental impact gradient characteristics. Equipment status fluctuation characteristics include the rate of parameter change, the frequency of consecutive outliers, and the degree of deviation from historical data for the same period, such as the rate of conductor temperature increase of 5°C within one hour or the number of times tower verticality exceeded the normal range for three consecutive days. Environmental impact gradient characteristics include the spatial rate of change of environmental parameters and the degree of correlation between different environmental factors, such as the wind speed increase rate from the start to the end of the line and the correlation coefficient between precipitation and air humidity. These features were then organized to form equipment status characteristic datasets and environmental impact characteristic datasets.

[0096] The equipment state characteristic dataset and the environmental impact characteristic dataset were trained using random forest to obtain the equipment state perception model and the environmental impact perception model, respectively.

[0097] The random forest consists of multiple decision trees. During training, samples and features are randomly drawn from the equipment status characteristic dataset to construct the training set for each decision tree. Each node of the decision tree splits according to a threshold value of the feature value. For example, using a change rate of 0.5℃ / minute in the equipment status fluctuation feature as the threshold, samples are divided into two classes until the sample classes at the leaf nodes tend to be consistent. After training, the equipment status awareness model can output the current operating status assessment result of the equipment based on the input equipment status characteristic data. Similarly, the environmental impact awareness model is trained using the environmental impact characteristic dataset. By learning the correlation between the gradient features of environmental parameters and equipment status changes, the environmental impact awareness model can assess the potential impact of different environmental conditions on the equipment.

[0098] The equipment status perception model and the environmental impact perception model are fused in parallel to generate an initial equipment status spatiotemporal perception model. Specifically, the process includes: setting operating state weight coefficients for the equipment state perception model, which are determined based on the importance of equipment state characteristic data in historical records. For example, the weight of current-carrying parameters, which have a greater impact on line safety, is higher than that of vibration frequency parameters; setting environmental factor weight coefficients for the environmental impact perception model, which are determined based on the actual impact range of environmental parameters on equipment operation. For example, the weight of icing thickness is higher than that of light intensity; and weighting and summing the outputs of the equipment state perception model and the environmental impact perception model based on the operating state weight coefficients and environmental factor weight coefficients to generate an initial equipment state spatiotemporal perception model. In other words, during fusion, the outputs of the equipment state perception model and the environmental impact perception model are weighted and summed according to their respective weights. For example, if the state score output by the equipment state perception model is 80 points (out of 100) with a weight of 0.6, and the impact score output by the environmental impact perception model is 70 points with a weight of 0.4, then the fused result is 80 × 0.6 + 70 × 0.4 = 76 points, thus generating the initial equipment state spatiotemporal perception model.

[0099] The initial equipment state spatiotemporal perception model was validated and adjusted using a model calibrator to obtain the equipment state-environment impact spatiotemporal perception model.

[0100] The model calibrator uses historical data not used in training and recent actual monitoring data as a validation set. This data is input into the initial equipment state spatiotemporal perception model to obtain prediction results. The prediction results are compared with the actual equipment state to calculate the error value. If the error exceeds the allowable range, the parameters of the initial equipment state spatiotemporal perception model are adjusted. Adjustments include the number of decision trees in the random forest, the depth of each tree, the node splitting threshold, and the weight coefficients of the equipment state perception model and the environmental impact perception model. For example, when the equipment state prediction error for a certain segment is large, the splitting priority of the corresponding feature in the decision tree for that segment is increased, or the weight coefficients of environmental factors are adjusted. After multiple rounds of validation and adjustment, the deviation between the model's prediction results and the actual situation is brought within a reasonable range, ultimately forming the equipment state-environmental impact spatiotemporal perception model. This model comprehensively considers temporal changes and spatial differences, accurately reflecting the dynamic relationship between equipment state and environmental impact.

[0101] Example 4: Based on the above, an iterative optimization discrimination model for transmission line anomalies is obtained, including:

[0102] Anomaly evaluation indicators are extracted from the anomaly identification target to obtain an anomaly evaluation indicator set, and an anomaly discrimination effect fitness function is constructed based on the anomaly evaluation indicator set.

[0103] The anomaly identification targets cover various abnormal situations that may occur in transmission lines, such as excessively thick conductor icing, excessive tower tilt, aging and damage to insulators, line short circuits, and foreign object entanglement. For these anomaly identification targets, the extracted anomaly evaluation indicators include: anomaly detection rate (the proportion of actual anomalies correctly identified by the model); missed detection rate (the proportion of actual anomalies not identified by the model); false detection rate (the proportion of anomalies identified by the model but actually normal); identification delay time (the time interval between the occurrence of an anomaly and its identification by the model); and feature extraction time (the time required for the model to process data and extract anomaly features).

[0104] When constructing the fitness function for anomaly detection, the influence of the aforementioned anomaly evaluation indicators must be considered. The expression for the fitness function for anomaly detection is:

[0105]

[0106] in, This represents the fitness function value. Indicates the abnormality detection rate. Indicates the false negative rate. Indicates the false positive rate. Indicates the recognition delay time. Indicates the time spent on feature extraction; , , , , These are all weighting coefficients, used to adjust the influence of each anomaly assessment indicator in the fitness function of anomaly discrimination. The values ​​of the weighting coefficients are determined according to the importance of different anomaly types. For example, for severe anomalies such as line short circuits, The value of can be appropriately increased to improve the proportion of abnormality detection rate in fitness assessment.

[0107] Based on the feature extraction parameters, the individual gene and individual mutation rate of the parameter are determined. The fitness of the feature extraction parameters is evaluated using the anomaly discrimination effect fitness function to obtain the feature extraction fitness set.

[0108] The feature extraction parameters include data sampling frequency, feature window size, anomaly threshold range, number of feature dimensions, and filtering coefficients. Individual parameter genes encode these feature extraction parameters into a sequence of binary or decimal numbers, with each number corresponding to a specific parameter value. For example, an 8-bit binary number can represent the level of data sampling frequency. The individual parameter mutation rate controls the probability of mutation during the iteration process. Its value is typically between 0.01 and 0.1, with the specific value determined based on the parameter's sensitivity. For parameters that significantly impact anomaly detection (such as the anomaly threshold range), a lower mutation rate can be set to maintain stability. For parameters with less impact (such as filtering coefficients), the mutation rate can be appropriately increased to broaden the search range.

[0109] When evaluating the fitness of feature extraction parameters, each set of feature extraction parameters is input into the 5G slice-transmission line monitoring model. The model calculates various evaluation indicators based on the identification results of abnormal samples, and then substitutes these indicators into the anomaly discrimination effect fitness function to obtain the corresponding fitness value. All combinations of feature extraction parameters and their corresponding fitness values ​​are summarized to form a feature extraction fitness set.

[0110] Based on the feature extraction fitness set, the parameter individual genes and parameter individual mutation rate are iteratively evolved until the preset termination condition is met, and the optimal parameter individual with the largest fitness is determined. The optimal parameter individual includes the transmission line anomaly discrimination features.

[0111] The iterative evolutionary process begins with an initial parameter population, which consists of multiple randomly generated sets of feature extraction parameters. In each iteration, parameter individuals are selected based on their fitness values; those with higher fitness values ​​are more likely to be selected. These selected individuals serve as parents in the crossover operation. The crossover operation generates new offspring parameter individuals by exchanging partial gene segments between two parent parameter individuals. For example, combining the first half of the gene from parent A with the second half of the gene from parent B forms a new gene sequence.

[0112] After the crossover operation, the genes of some offspring individuals are mutated according to a preset individual mutation rate, that is, the values ​​at certain positions in the gene sequence are randomly changed to generate new parameter combinations. Mutation helps increase population diversity and avoids the iteration process getting stuck in local optima. The iteration process continues until a preset termination condition is met. The preset termination condition can be set as the number of iterations reaching a preset maximum (e.g., 100 times), or the change in the maximum fitness value in multiple consecutive iterations being less than a set threshold (e.g., 0.001). When the preset termination condition is met, the individual with the highest fitness value is selected from the current population. The features contained in this optimal individual (such as specific feature dimensions, anomaly thresholds, window size, etc.) are the transmission line anomaly detection features.

[0113] The 5G slice-transmission line monitoring model is optimized and updated based on the anomaly discrimination features of transmission lines to obtain an optimized anomaly discrimination model for transmission lines.

[0114] During the optimization and update process, the anomaly detection features of transmission lines were integrated into the feature extraction module of the 5G slice-transmission line monitoring model. The model's data processing methods were adjusted; for example, data segments were re-divided based on the newly determined feature window size, and the judgment criteria for the degree of data anomaly were adjusted according to the anomaly threshold range. Simultaneously, the computational resources of the 5G slice-transmission line monitoring model were reallocated based on the number of feature dimensions, allocating more resources to the extraction of key features to improve processing efficiency. Through these adjustments, the optimized transmission line anomaly detection model can more accurately capture the anomaly features of transmission lines, maintaining the ability to identify various anomalies while reducing the false negative and false positive rates and shortening the identification latency, thereby achieving efficient identification of abnormal states of transmission lines.

[0115] Example 5: Based on the above, determine the transmission line monitoring strategy parameters, including:

[0116] Based on the transmission line anomaly optimization discrimination model, the influence of equipment state distribution is analyzed to obtain the equipment state distribution influence parameters, which include line operation influence parameters and monitoring response influence parameters.

[0117] Among them, the line operation impact parameters reflect the effect of equipment status on the overall operation of the transmission line. Examples include the degree to which conductor temperature exceeds a certain range limits the line's current carrying capacity; the change in line mechanical tension caused by each degree increase in tower tilt angle; the decline in insulation performance due to insulator aging; and the impact of increased icing thickness on conductor sag. The monitoring response impact parameters reflect the effect of abnormal equipment status on the monitoring system's response mechanism. Examples include the factor by which the monitoring system needs to increase the data sampling frequency after three consecutive minor anomalies in a certain section of equipment; the delay time for the monitoring center to issue early warning signals after a specific type of anomaly (such as excessive conductor vibration); and the impact of the distance between the anomaly location and the nearest maintenance station on the arrival time of repair personnel.

[0118] A monitoring strategy library is constructed, and the spatiotemporal distribution characteristics of equipment status are combined with the parameters affecting the equipment status distribution. The results are then matched and analyzed with the monitoring strategy library to obtain the matching control strategy for transmission line monitoring.

[0119] The monitoring strategy library contains various preset strategies, categorized into three types based on monitoring intensity: routine monitoring, enhanced monitoring, and emergency monitoring. Routine monitoring is suitable for situations where equipment status is stable and environmental impact is minimal, such as collecting tower tilt data daily and recording conductor temperature hourly. Enhanced monitoring is used in scenarios where equipment status fluctuates or environmental conditions are complex, such as adjusting the wind speed sensor's data sampling frequency from once every 10 minutes to once every 2 minutes during windy weather, while also increasing the frequency of drone inspections. Emergency monitoring is for situations where obvious anomalies have occurred, such as when the thickness of ice on the conductor exceeds a set value, initiating real-time video monitoring and transmitting icing data to the monitoring center every 5 minutes.

[0120] The spatiotemporal distribution characteristics of equipment status show that during the high-temperature period in summer (12:00 to 16:00 daily), the conductor temperature in the middle 30-kilometer section of a 100-kilometer transmission line is generally 5-8°C higher than other sections. Furthermore, this area has experienced two temporary current-carrying capacity limitations due to excessive temperature within the past three months. Based on the equipment status distribution impact parameters (for every 1°C increase in conductor temperature in this area, the current-carrying capacity limitation value decreases by 2%), an enhanced monitoring strategy was matched from the monitoring strategy library. Specifically, from 10:00 to 18:00 daily, the conductor temperature sampling frequency for this middle 30-kilometer section will be increased to once every 5 minutes, while the number of times infrared imagers photograph conductor joints will be increased, and a temperature change trend report will be generated hourly.

[0121] Based on the transmission line monitoring and matching control strategy, the selection thresholds for equipment status spatiotemporal distribution characteristics and equipment status distribution influence parameters are divided to obtain the monitoring strategy parameter selection thresholds.

[0122] Taking conductor temperature monitoring as an example, based on the matched enhanced monitoring strategy and considering the temperature range (35℃ to 48℃) during the high-temperature period of summer and the influence parameters of equipment status distribution (current carrying capacity begins to be limited when the temperature exceeds 40℃), the temperature monitoring threshold is divided into three intervals: 35℃ to 40℃ is the attention interval, 40℃ to 45℃ is the warning interval, and above 45℃ is the emergency interval. The sampling frequency thresholds for each interval are once every 5 minutes, once every 2 minutes, and real-time sampling, respectively; the warning signal thresholds are yellow warning (interval one), orange warning (interval two), and red warning (interval three).

[0123] The anomaly detection effect fitness function is used to perform global optimization within the threshold of the monitoring strategy parameters to determine the monitoring strategy parameters for transmission lines.

[0124] Based on the above examples, the effects of different sampling frequencies and combinations of warning signals were tested within three temperature monitoring threshold ranges. In the monitoring range (35℃ to 40℃), the monitoring effects were tested at sampling frequencies of every 5 minutes, every 6 minutes, and every 4 minutes. The anomaly identification efficiency at each frequency was evaluated using an anomaly detection fitness function. In the warning range (40℃ to 45℃), the combined effects of sampling frequencies of every 2 minutes, every 3 minutes, and every 1 minute with the timing of orange warning signal issuance (e.g., when the temperature reaches 42℃ or exceeds 40℃ three times consecutively) were tested. In the emergency range (above 45℃), different schemes of real-time sampling and delay times for red warning signal issuance (e.g., immediate issuance or issuance after 30 seconds of confirmation) were tested. Through evaluation of all parameter combinations, the monitoring strategy parameters for the 30-kilometer section in the middle section during the summer high-temperature period were finally determined: the sampling frequency of the monitoring section is once every 5 minutes, and a yellow warning is issued when the temperature reaches 40℃; the sampling frequency of the warning section is once every 2 minutes, and an orange warning is issued when the temperature exceeds 42℃; the emergency section is sampled in real time, and a red warning is immediately issued when the temperature reaches 45℃, and the linkage command of the flow restriction is automatically triggered.

[0125] In another scenario, towers 50 to 60 of a power transmission line in a mountainous area are located in a windy region. Information on the spatiotemporal distribution of equipment status shows that in this area, wind speeds exceeding 10 m / s account for 60% of the days during winter (December to February of the following year), and the tower tilt exhibits significant fluctuations when wind speeds exceed 12 m / s. The equipment status distribution impact parameters indicate that for every 2 m / s increase in wind speed, the monitoring error of tower tilt may increase by 1.5%. The enhanced monitoring strategy matched from the monitoring strategy library requires improving the monitoring accuracy of tower tilt when wind speeds exceed 10 m / s. Through threshold division, wind speed monitoring thresholds are set to three ranges: 10 m / s to 12 m / s, 12 m / s to 15 m / s, and above 15 m / s. The corresponding tower tilt sampling accuracy is improved to 0.1 degrees, 0.05 degrees, and 0.02 degrees, respectively. After global optimization, the monitoring strategy parameters were determined as follows: when the wind speed reaches 10 m / s, the sampling interval for tower tilt is shortened from once per hour to once every 20 minutes; when the wind speed exceeds 12 m / s, dual-sensor cross-validation is initiated (i.e., two sensors at different locations are used simultaneously to monitor the tilt); when the wind speed reaches 15 m / s, an emergency drone inspection is triggered, transmitting a real-time image of the top of the tower every 30 minutes.

[0126] Example 6: Based on the above, the parameters of the transmission line monitoring strategy are optimized by feedback based on the implementation effect of the monitoring strategy, including:

[0127] Based on the implementation effect of the monitoring strategy, the optimization direction of the transmission line monitoring strategy parameters is analyzed to obtain the parameter variation optimization rules.

[0128] The monitoring strategy implementation effect is formed by tracking and recording the entire strategy execution process, covering the actual performance of various parameters during strategy execution, such as whether the data collection frequency is consistent with the setting, whether the triggering of early warning signals is in line with the actual situation, and the execution efficiency of operation and maintenance scheduling instructions. It also includes various problems that occur during strategy implementation. For example, in the monitoring of a certain transmission line, according to the established strategy parameters, when the ambient temperature reaches 38℃, conductor temperature sampling should be initiated every 10 minutes. However, in actual implementation, it was found that due to sensor response delay, the first sampling was only completed 2-3 minutes after the temperature exceeded the threshold on several occasions. The icing thickness early warning threshold set for a certain section is 10mm. However, in actual monitoring, due to the terrain, significant conductor sag changes occurred when the icing reached 8mm, resulting in a delay in the issuance of early warning signals.

[0129] Optimization directions were analyzed to address these actual performance issues and problems, clarifying which parameters needed adjustment and the general direction of adjustment. For sampling frequency-related parameters, when sampling delay occurred, the analysis revealed the optimization direction of shortening the sampling interval; for warning threshold parameters, when warning lag occurred, the analysis revealed the optimization direction of lowering the threshold. Based on these optimization directions, parameter variation optimization rules were formulated, specifying the range and method of parameter adjustment. For example, for the conductor temperature sampling frequency, the original parameter was 10 minutes / time, and the optimization rule set it to be adjustable within the range of 8-12 minutes, with each adjustment not exceeding 2 minutes; for the icing thickness warning threshold, the original parameter was 10 mm, and the optimization rule set it to be adjustable within the range of 7-10 mm, with each adjustment not exceeding 1 mm.

[0130] Based on the parameter mutation optimization rules, the parameters of the transmission line monitoring strategy are mutated and expanded to obtain a monitoring strategy parameter cluster.

[0131] Taking a power transmission line traversing a mountainous area as an example, its original monitoring strategy parameters included: a wind speed monitoring threshold of 12 m / s (enhanced monitoring is activated when this value is exceeded), a tower tilt sampling frequency of 30 minutes / time, and a conductor vibration amplitude warning threshold of 5 mm. Following parameter variation optimization rules, these parameters were expanded through variation. The wind speed monitoring threshold, within the range of 10-14 m / s, was adjusted in 1 m / s increments, resulting in five variation parameters: 10 m / s, 11 m / s, 12 m / s, 13 m / s, and 14 m / s. The tower tilt sampling frequency, within the range of 20-40 minutes, was adjusted in 5-minute increments, resulting in five variation parameters: 20 minutes / time, 25 minutes / time, 30 minutes / time, 35 minutes / time, and 40 minutes / time. The conductor vibration amplitude warning threshold, within the range of 4-6 mm, was adjusted in 0.5 mm increments, resulting in five variation parameters: 4 mm, 4.5 mm, 5 mm, 5.5 mm, and 6 mm. These mutated parameters are combined to form a monitoring strategy parameter cluster containing 5×5×5=125 parameter combinations, each combination representing a possible monitoring strategy parameter setting.

[0132] Parameters are compared and optimized within the monitoring strategy parameter cluster to obtain monitoring and control optimization strategy parameters, and then the monitoring and control optimization strategy parameters are used to perform transmission line monitoring and optimization control.

[0133] Using the above example, we simulated each of the 125 parameter combinations in the monitoring strategy parameter cluster. During the simulation, we input historical environmental data and equipment operation data for the mountain transmission line, and observed the performance of the monitoring system under each parameter combination, including whether it could promptly capture abnormal states and whether there was excessive monitoring leading to resource waste. The simulation revealed that when the wind speed monitoring threshold was 11 m / s, the tower tilt sampling frequency was 25 minutes / time, and the conductor vibration amplitude warning threshold was 4.5 mm, this parameter combination could respond promptly to the variable wind conditions in the mountainous area, strengthening monitoring by initiating the threshold 1 m / s in advance to avoid missing anomalies due to sudden increases in wind speed; it could also more promptly grasp changes in tower tilt by shortening the sampling interval by 5 minutes; and at the same time, lowering the vibration warning threshold by 0.5 mm allowed for earlier detection of abnormal conductor vibration.

[0134] The best-performing parameter combination was comprehensively compared with other parameter combinations to evaluate its adaptability and effectiveness in various scenarios, ultimately determining it as the optimal monitoring and control strategy parameter. Subsequently, the optimized parameters were applied to the actual monitoring of the transmission line in the mountainous area, and their effectiveness was further verified through actual operation, achieving optimized control of the transmission line monitoring. This process is not a one-time event; as the operating environment of the transmission line changes and the equipment ages, the above-mentioned variation, expansion, comparison, and optimization steps can be repeated periodically to continuously update the monitoring strategy parameters to adapt to ever-changing monitoring needs.

[0135] Example 7: Based on the above, the process of determining the transmission line monitoring strategy parameters by using the anomaly detection effect fitness function to perform global optimization within the threshold range of the monitoring strategy parameters is as follows:

[0136] Multiple strategy parameters are randomly selected within the threshold range for monitoring strategy parameters, and the fitness function of anomaly detection effect is used to evaluate these parameters, obtaining the fitness of multiple parameters. The threshold range for monitoring strategy parameter selection is derived by dividing the parameters into thresholds based on the spatiotemporal distribution characteristics of equipment status and the parameters influencing equipment status distribution in the transmission line monitoring and matching control strategy. This covers various adjustable parameter ranges related to transmission line monitoring, such as data acquisition frequency, anomaly response time, and monitoring equipment power. Random selection avoids parameter selection bias caused by human factors, ensuring that the selected strategy parameters have a certain breadth and representativeness. For example, regarding data acquisition frequency, different frequency parameters such as every 10 seconds, every 15 seconds, and every 20 seconds can be randomly selected within the threshold range; regarding anomaly response time, parameters such as 0.5 seconds, 1 second, and 1.5 seconds can be randomly selected. These randomly selected strategy parameters are substituted into the anomaly detection fitness function. The anomaly detection fitness function calculates the accuracy, timeliness and other indicators of anomaly detection based on the strategy parameters in the simulated monitoring scenario, thereby obtaining the parameter fitness for each strategy parameter. The level of parameter fitness reflects the effectiveness of the strategy parameter in anomaly detection.

[0137] The optimization region for monitoring strategy parameter selection thresholds is approximated based on the fitness of multiple parameters to determine the local optimization region for the strategy parameters. Specifically, after obtaining multiple parameter fitness values, these fitness values ​​are analyzed to identify the range of strategy parameters with high fitness. Since strategy parameters with high fitness tend to concentrate within a specific sub-interval, the range is gradually narrowed by continuously narrowing the range towards this sub-interval, ultimately determining a smaller local optimization region for the strategy parameters. For example, assuming the initial monitoring strategy parameter selection threshold has a data acquisition frequency range of 5 to 30 seconds, fitness analysis of randomly selected frequency parameters reveals that the frequency parameters with high fitness are mainly concentrated between 10 and 20 seconds. Therefore, this 10 to 20 second interval can be determined as the local optimization region for the data acquisition frequency strategy parameters. This optimization region approximation method reduces the scope of subsequent parameter searches, improves optimization efficiency, and allows subsequent parameter searches to focus more on areas that may produce better strategy parameters.

[0138] Based on the local optimization region of the strategy parameters, set the parameter search step size. The setting of the parameter search step size needs to comprehensively consider the size of the local optimization region of the strategy parameters and the required parameter accuracy. If the local optimization region of the strategy parameters is small, but the required parameter accuracy is high, then a smaller search step size can be set; if the local optimization region of the strategy parameters is large, but the required accuracy is relatively low, then a larger search step size can be set. For example, within a local optimization region of the strategy parameters of the data acquisition frequency of 10 to 20 seconds, if a parameter accuracy of 1 second is desired, then the search step size can be set to 1 second; if the required accuracy is 2 seconds, then the search step size can be set to 2 seconds. A reasonable parameter search step size can ensure optimization accuracy while avoiding excessive computation due to an excessively small step size, or missing the optimal parameter due to an excessively large step size.

[0139] The strategy parameters are searched and evaluated within the local optimization region according to the parameter search step size. Based on the evaluation results, the optimization region is iteratively approximated until a preset number of iterations is reached. The transmission line monitoring strategy parameters are then determined through parameter fitness comparison. Specifically, after determining the local optimization region and parameter search step size, strategy parameters are selected sequentially from the starting point of the local optimization region according to the set parameter search step size. Each selected strategy parameter is substituted into the anomaly discrimination effect fitness function for evaluation to obtain the corresponding parameter fitness. After each search and evaluation within the current local optimization region, the evaluation results are analyzed to find the smaller interval containing the strategy parameter with the highest fitness. This interval is used as the new local optimization region for the next search and evaluation, thus achieving iterative approximation of the optimization region.

[0140] For example, in the local optimization region of strategy parameters within a data acquisition frequency of 10 to 20 seconds, with a step size of 1 second, the strategy parameters for 10 seconds, 11 seconds, and so on, are evaluated sequentially. If the parameter fitness corresponding to 12 seconds is found to be the highest, then the next local optimization region can be narrowed down to 11 to 13 seconds, with the step size still 1 second, and the search and evaluation are performed again. This iterative process is repeated continuously, each time narrowing the optimization region to a range more likely to contain the optimal strategy parameters. When the number of iterations reaches a preset number, such as 10 iterations, in the final local optimization region of the strategy parameters, the strategy parameter with the highest fitness is selected as the final determined transmission line monitoring strategy parameter by comparing the fitness of all strategy parameters. Through this process, the optimal monitoring strategy parameter can be gradually and accurately located, ensuring that the strategy parameter can play a good role in transmission line monitoring and meet the system's requirements for transmission line anomaly detection and monitoring.

[0141] Example 8: Based on the above, the process of constructing a resource proximity connection set by parsing the connection relationships of the slice resource set using 5G network parameters and power transmission line monitoring data is as follows:

[0142] Slice bandwidth analysis is performed on 5G network parameters to obtain slice bandwidth allocation information. These parameters include frequency bands, transmission rates, latency, jitter, and other data. Slice bandwidth analysis needs to be combined with the service requirements of power transmission line monitoring scenarios, matching these parameters with the bandwidth requirements of different monitoring services. For example, for high-definition video monitoring of power transmission lines, which has high bandwidth requirements, it is necessary to analyze the maximum bandwidth and stable bandwidth range that the 5G network can provide in the corresponding frequency band. For sensor data acquisition services, the bandwidth requirements are relatively lower, but the latency requirements are more stringent; therefore, the focus is on analyzing the bandwidth allocation under the condition of meeting latency requirements. Through this analysis, the bandwidth allocation values, dynamic bandwidth adjustment range, and bandwidth guarantee mechanisms corresponding to different slices are clarified, forming complete slice bandwidth allocation information.

[0143] Monitoring node location is determined by analyzing transmission line monitoring data. Specifically, the monitoring data originates from various monitoring devices distributed along the transmission line, such as temperature sensors, humidity sensors, and image acquisition equipment on towers, as well as current transformers and voltage transformers along the line. Monitoring node location requires combining the geographical information data of the transmission line with methods such as GPS positioning and tower number mapping to determine the specific geographical coordinates of each monitoring device. Simultaneously, the positional relationship of each monitoring node within the transmission line topology must be clarified, such as which tower a monitoring node is located on and which line segment it belongs to, thus forming monitoring node location information containing the latitude and longitude, tower number, and section of each monitoring node.

[0144] The connection strength of a slice resource set is calculated based on slice bandwidth allocation information and monitoring node location information to determine the resource connection strength threshold. The slice resource set includes network resources such as base stations and core network elements in the 5G network, as well as monitoring resources such as monitoring equipment and data processing units in power transmission lines. The connection strength calculation needs to comprehensively consider two factors: bandwidth support capability and spatial distance. From the perspective of bandwidth support capability, if the slice bandwidth allocation of a network resource can meet the transmission needs of the data generated by a monitoring node, and the bandwidth stability is high, the connection strength between the two is relatively high; conversely, if the bandwidth is insufficient or fluctuates greatly, the connection strength is low. From the perspective of spatial distance, the closer the network resource is to the monitoring node, the lower the signal transmission loss, the higher the connection stability, and the greater the connection strength; the farther the distance, the more severe the signal attenuation, and the lower the connection strength. By setting a reasonable calculation model, bandwidth support capability and spatial distance are converted into quantifiable values, and then the connection strength value between each pair of slice resources is calculated.

[0145] Slice resource pairs with connection strengths greater than a resource connection strength threshold are selected to construct a resource proximity connection set. The resource connection strength threshold is set with reference to the operational requirements of the transmission line monitoring system and the performance indicators of the 5G network. Through analysis of historical data and system simulation testing, a minimum connection strength value that can guarantee data transmission quality and system stability is determined. After calculating the connection strength of all slice resource pairs, resource pairs with connection strengths greater than the resource connection strength threshold are selected. These resource pairs possess stable and reliable connection conditions, meeting the data transmission and service interaction requirements of the intelligent transmission line monitoring system. These selected resource pairs, along with their corresponding connection strengths, connection methods, and other information, are then organized to form a resource proximity connection set.

[0146] Through the above process, the connection relationships of the slice resource set are analyzed, and the constructed resource proximity connection relationship set provides an important foundation for the subsequent establishment of the 5G slice-transmission line monitoring model. This enables 5G network slice resources and transmission line monitoring resources to work collaboratively according to reasonable connection relationships, ensuring the effective operation of the intelligent transmission line monitoring system. In practical applications, as the 5G network status changes and the transmission line monitoring requirements are adjusted, slice bandwidth allocation information and monitoring node location information may change. In this case, it is necessary to recalculate the connection strength and reconstruct the resource proximity connection relationship set to ensure the accuracy and adaptability of the connection relationships. For example, when a new monitoring node is added, the monitoring node needs to be relocated, and the connection strength between the new node and each network resource needs to be calculated. Resource pairs that meet the conditions are added to the resource proximity connection relationship set. When the 5G network bandwidth is adjusted, the slice bandwidth allocation information also needs to be re-analyzed, and the connection strength recalculated to update the resource proximity connection relationship set. This dynamic adjustment mechanism ensures that the resource proximity connection relationship set always matches the actual operating state of the system, thus providing strong support for the stable and efficient operation of the entire intelligent monitoring system.

[0147] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart monitoring system for power transmission lines based on 5G network slicing, characterized in that, include: The slice configuration module is used to acquire 5G network parameters and power transmission line monitoring data, define slice service attributes, configure slice resources based on the slice service attributes for the 5G network parameters and power transmission line monitoring data, and establish a 5G slice-power transmission line monitoring model. The state perception module is used to obtain equipment operation dataset and environmental monitoring dataset by association through multi-source data fusion technology, and to perform state perception modeling on the equipment operation dataset and environmental monitoring dataset using spatiotemporal analysis to obtain a spatiotemporal perception model of equipment state-environmental impact. The anomaly detection module is used to initialize feature extraction parameters based on the 5G slice-transmission line monitoring model, optimize the feature extraction parameters for transmission line anomaly detection according to the anomaly identification target, and iteratively obtain the optimized transmission line anomaly detection model. The strategy generation module is used to obtain the spatiotemporal distribution characteristic information of equipment status based on the equipment status-environmental impact spatiotemporal perception model, perform strategy analysis based on the equipment status spatiotemporal distribution characteristic information and the transmission line anomaly optimization discrimination model, and determine the transmission line monitoring strategy parameters. The feedback optimization module is used to simulate and verify the parameters of the transmission line monitoring strategy, obtain the implementation effect of the monitoring strategy, and optimize the parameters of the transmission line monitoring strategy based on the implementation effect of the monitoring strategy.

2. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 1, characterized in that, The establishment of the 5G slice-transmission line monitoring model includes: Determine the slice resource attributes and resource connection edge attributes based on the slice service attributes; Based on the slice resource attributes, resource identification and attribute labeling are performed on the 5G network parameters and power transmission line monitoring data to obtain a slice resource set and a slice resource attribute set. The connection relationships of the slice resource set are analyzed using the 5G network parameters and power transmission line monitoring data to construct a resource proximity connection set. Based on the resource connection edge attributes, the resource proximity connection set is sliced ​​and abstracted to obtain the sliced ​​resource topology adjacency matrix; Based on the slice resource topology adjacency matrix and the slice resource attribute set, the slice resource set is connected by slice identification to establish the 5G slice-transmission line monitoring model.

3. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 1, characterized in that, The obtained device status-environmental impact spatiotemporal perception model includes: Spatiotemporal analysis was used to perform spatiotemporal arrangement and characteristic analysis on the equipment operation dataset and environmental monitoring dataset to obtain the equipment status characteristic dataset and environmental impact characteristic dataset. The equipment state characteristic dataset and the environmental impact characteristic dataset are trained using random forest to obtain the equipment state perception model and the environmental impact perception model, respectively. The equipment state perception model and the environmental impact perception model are fused in parallel to generate an initial equipment state spatiotemporal perception model. The initial equipment state spatiotemporal perception model is verified and adjusted using a model calibrator to obtain the equipment state-environment impact spatiotemporal perception model.

4. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 1, characterized in that, The iterative acquisition of the optimized discriminant model for transmission line anomalies includes: Anomaly evaluation indicators are extracted from the anomaly identification target to obtain an anomaly evaluation indicator set, and an anomaly discrimination effect fitness function is constructed based on the anomaly evaluation indicator set. Based on the feature extraction parameters, determine the individual gene and individual mutation rate of the parameter, and use the anomaly discrimination effect fitness function to evaluate the fitness of the feature extraction parameters to obtain a feature extraction fitness set; Based on the feature extraction fitness set, the parameter individual genes and parameter individual mutation rates are iteratively evolved until a preset termination condition is met, and the optimal parameter individual with the largest fitness is determined. The optimal parameter individual includes transmission line anomaly discrimination features. The 5G slice-transmission line monitoring model is optimized and updated based on the anomaly discrimination features of the transmission line to obtain the optimized anomaly discrimination model of the transmission line.

5. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 4, characterized in that, The determination of transmission line monitoring strategy parameters includes: Based on the transmission line anomaly optimization discrimination model, the equipment state distribution impact analysis is performed to obtain the equipment state distribution impact parameters, which include line operation impact parameters and monitoring response impact parameters. A monitoring strategy library is constructed. The spatiotemporal distribution characteristics of the equipment status are combined with the influence parameters of the equipment status distribution. The monitoring strategy library is matched and analyzed to obtain the monitoring and matching control strategy for the transmission line. Based on the transmission line monitoring and matching control strategy, the spatiotemporal distribution characteristics of the equipment status and the parameters affecting the equipment status distribution are divided into selection thresholds to obtain the monitoring strategy parameter selection thresholds. The anomaly detection effect fitness function is used to perform global optimization within the threshold value of the monitoring strategy parameters to determine the monitoring strategy parameters of the transmission line.

6. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 5, characterized in that, The step of using the anomaly detection effect fitness function to perform global optimization within the selected threshold of the monitoring strategy parameters to determine the transmission line monitoring strategy parameters includes: Multiple strategy parameters are randomly selected within the threshold range of the monitoring strategy parameters, and the multiple strategy parameters are evaluated using the anomaly discrimination effect fitness function to obtain the fitness of multiple parameters; Based on the fitness of the multiple parameters, the threshold for selecting the monitoring strategy parameters is approximated to determine the local optimization region of the strategy parameters. Set the parameter search step size according to the local optimization region of the strategy parameters; The strategy parameters are searched and evaluated within the local optimization region of the strategy parameters according to the parameter search step size, and the optimization region is iteratively approximated based on the parameter search evaluation results until the preset number of iterations is reached. The transmission line monitoring strategy parameters are then determined by parameter fitness comparison.

7. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 1, characterized in that, The step of optimizing the transmission line monitoring strategy parameters based on the implementation effect of the monitoring strategy includes: Based on the implementation effect of the monitoring strategy, the optimization direction of the transmission line monitoring strategy parameters is analyzed to obtain parameter variation optimization rules. The parameters of the transmission line monitoring strategy are mutated and expanded according to the parameter mutation optimization rules to obtain a monitoring strategy parameter cluster; Within the monitoring strategy parameter cluster, parameter comparison and optimization are performed to obtain monitoring and control optimization strategy parameters, and the transmission line monitoring and optimization control is performed through the monitoring and control optimization strategy parameters.

8. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 3, characterized in that, The process of using spatiotemporal analysis to perform spatiotemporal arrangement and characteristic analysis on the equipment operation dataset and environmental monitoring dataset includes: The equipment operation dataset is divided into time series to obtain equipment status time series classified by hour, day, and month. The environmental monitoring dataset is spatially divided to obtain the spatial regions of environmental impact classified by tower, section, and line. Based on the time series of equipment status and the spatial region of environmental impact, the equipment operation dataset and the environmental monitoring dataset are jointly arranged to construct a spatiotemporal grid data matrix. Feature values ​​are extracted from the spatiotemporal grid data matrix to obtain equipment state fluctuation characteristics and environmental impact gradient characteristics, forming the equipment state characteristic dataset and the environmental impact characteristic dataset.

9. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 2, characterized in that, The step of parsing the connection relationships of the slice resource set using the 5G network parameters and power transmission line monitoring data to construct a resource proximity connection relationship set includes: Slice bandwidth analysis is performed on the 5G network parameters to obtain slice bandwidth allocation information; The monitoring data of the transmission line is used to locate the monitoring nodes and obtain the location information of the monitoring nodes; Based on the slice bandwidth allocation information and monitoring node location information, the connection strength of the slice resource set is calculated to determine the resource connection strength threshold. Select slice resource pairs with connection strength greater than the resource connection strength threshold, and construct the resource proximity connection relationship set.

10. The intelligent monitoring system for power transmission lines based on 5G network slicing as described in claim 3, characterized in that, The step of merging the device state perception model and the environmental impact perception model in parallel to generate an initial device state spatiotemporal perception model includes: Set the operating state weight coefficients for the device state perception model; Set environmental factor weight coefficients for the environmental impact perception model; The outputs of the equipment state perception model and the environmental impact perception model are weighted and summed based on the weight coefficients of the operating state and the weight coefficients of the environmental factors to generate the initial equipment state spatiotemporal perception model.

Images