Power transmission line multi-dimensional state monitoring method based on broadband low-power-consumption ad hoc network technology

Abstract

The invention provides a power transmission line multi-dimensional state monitoring method based on a broadband low-power-consumption ad hoc network technology. The method comprises the following steps: configuring a plurality of monitoring nodes along a power transmission line; collecting power transmission line monitoring data from the plurality of monitoring nodes; acquiring a historical false alarm rate, a real-time signal-to-noise ratio and an environmental interference parameter of each monitoring node; calculating a fusion feature vector of each monitoring node according to the historical false alarm rate, the real-time signal-to-noise ratio and the environmental interference parameter; analyzing the fusion feature vector and the power transmission line monitoring data acquired by each monitoring node by applying an anomaly judgment algorithm to generate a preliminary anomaly judgment result; recursive fusion is carried out by using a D-S evidence theory, and a final fusion confidence coefficient is output; and comparing the final fusion confidence coefficient with a preset security threshold, and triggering alarm information when the final fusion confidence coefficient is greater than the preset security threshold, thereby realizing effective processing of conflicts and uncertainty among multi-source data by using a broadband low-power-consumption ad hoc network technology.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power transmission line monitoring, and particularly relates to a power transmission line multi-dimensional state monitoring method based on a wideband low-power ad hoc network technology. BACKGROUND

[0002] With the development of smart grids and digital power systems, the safe and stable operation of power transmission lines, as a core component of the power grid, is of key significance to the entire power system. In recent years, with the rapid development of emerging technologies such as the Internet of Things, edge computing and deep learning, the use of multiple sensors for power transmission line state monitoring has become a research hotspot.

[0003] At present, there are many technical solutions and products for power transmission line monitoring at home and abroad, mainly using a single or a small number of types of sensors to collect data, such as image monitoring, temperature detection, weather data collection, etc., which lack. SUMMARY

[0004] Therefore, it is necessary to provide a power transmission line multi-dimensional state monitoring method based on a wideband low-power ad hoc network technology, which can overcome at least one of the above defects.

[0005] In a first aspect, the embodiments of the present application provide a power transmission line multi-dimensional state monitoring method based on a wideband low-power ad hoc network technology, applied to monitoring a power transmission line, the method comprising:

[0006] configuring a plurality of monitoring nodes along the power transmission line;

[0007] applying a wideband low-power ad hoc network technology to build an ad hoc network between the plurality of monitoring nodes;

[0008] collecting power transmission line monitoring data from the plurality of monitoring nodes through the ad hoc network;

[0009] obtaining historical false alarm rates, real-time signal-to-noise ratios and environmental interference parameters of each monitoring node;

[0010] calculating a fusion feature vector of each monitoring node according to the historical false alarm rates, the real-time signal-to-noise ratios and the environmental interference parameters;

[0011] applying an anomaly judgment algorithm to analyze the fusion feature vector and the power transmission line monitoring data obtained by each monitoring node to generate a preliminary anomaly judgment result;

[0012] converting the power transmission line monitoring data and the preliminary anomaly judgment result into basic probability assignments according to a mapping function;

[0013] recursively fusing a plurality of basic probability assignments using D-S evidence theory, and outputting a final fusion confidence;

[0014] The final fusion confidence level is compared with a preset security threshold. When the final fusion confidence level is greater than the preset security threshold, an alarm message is triggered.

[0015] In one embodiment, obtaining the historical false alarm rate, real-time signal-to-noise ratio, and environmental interference parameters of each monitoring node includes:

[0016] At each monitoring node, a comparison log of sensor alarm events and actual fault events is continuously recorded to calculate the historical false alarm rate of each monitoring node.

[0017] The power ratio of the raw signal to the background noise is collected in real time at each monitoring node to obtain the real-time signal-to-noise ratio of each monitoring node;

[0018] Environmental interference parameters are obtained by collecting data from each monitoring node within each preset period and using the standard deviation evaluation method.

[0019] In one embodiment, calculating the fusion feature vector of each monitoring node based on the historical false alarm rate, the real-time signal-to-noise ratio, and the environmental interference parameters includes:

[0020] The reliability score is calculated using the following formula:

[0021] r i =α·SNR i +β·(1-FR i )+γ·(1-NI i )

[0022] Where, r i The reliability score for the i-th monitoring node is given by α, β, γ, which are empirical coefficients and α + β + γ = 1. SNR i Let FR be the real-time signal-to-noise ratio of the i-th monitoring node. i Let NI be the historical false alarm rate of the i-th monitoring node. i Environmental interference parameters;

[0023] The scores of all monitoring nodes are normalized based on the reliability score to obtain weights, and the normalization formula is as follows:

[0024]

[0025] Among them, w i The normalized fusion weights are defined by j, which is an index variable indicating that all monitoring nodes are traversed during summation, and r... j This represents the reliability score for each monitoring node;

[0026] The original feature vector extracted by the i-th node is fused according to a weight to obtain a fused feature vector, and a calculation formula of the fusion process is as follows:

[0027]

[0028] wherein, F i is the original feature vector.

[0029] In an embodiment, the application of the anomaly judgment algorithm analyzes the fused feature vector and the transmission line monitoring data obtained by each monitoring node to generate a preliminary anomaly judgment result, including:

[0030] The original feature vector is input into a target detection submodule to generate a candidate abnormal area;

[0031] The region features and historical trajectories of the candidate abnormal area are input into a tracking submodule to predict a target motion state;

[0032] The target motion state and the region features are input into a CNN cascade submodule to output a preliminary confidence score, and the preliminary confidence score includes an abnormal confidence score and a normal confidence score.

[0033] In an embodiment, the transmission line monitoring data and the preliminary anomaly judgment result are converted into basic probability assignments according to a mapping function, including:

[0034] The abnormal confidence score and the normal confidence score are obtained according to the transmission line monitoring data and the preliminary anomaly judgment result;

[0035] A temperature coefficient is set to adjust the smoothness of the mapping function;

[0036] The mapping function is applied to convert the confidence score into a basic probability assignment, and the specific calculation formula is as follows:

[0037]

[0038] wherein, A represents an abnormal hypothesis, F k represents an abnormal confidence score of the k-th information source, represents a normal confidence score of the k-th monitoring node, T k is a temperature coefficient.

[0039] In an embodiment, the D-S evidence theory is applied to recursively fuse a plurality of basic probability assignments and output a final fused confidence, including:

[0040] A conflict factor is calculated for any two information sources, and the specific calculation formula is as follows:

[0041]

[0042] where K ij represents the conflict factor, i, j represent two different information sources, K ij represents the conflict factor, B, C are hypothesis sets, m i (B) represents the basic probability assignment of the ith information source to the hypothesis set B, m j (C) represents the basic probability assignment of the jth information source to the hypothesis set C;

[0043] The modified rule with the conflict relaxation coefficient is used for fusion, and the specific fusion formula is as follows:

[0044]

[0045] where A is a hypothesis, m i (A) represents the basic probability assignment of the ith information source to the hypothesis A, m j (A) represents the basic probability assignment of the jth information source to the hypothesis A, and λ is a conflict relaxation coefficient for adjusting the influence of conflict, m ij (A) represents the basic probability assignment of the fusion of the information sources i and j to the hypothesis A, min{m j (A), m j (A)} is a conflict relaxation term.

[0046] The basic probability assignment of the fusion of the information sources i and j to the hypothesis A is recursively fused with a third information source until all information sources are combined, and the final fusion confidence is output.

[0047] In an embodiment, the method further comprises:

[0048] According to the reliability score between the information sources, the specific condition formula is as follows:

[0049]

[0050] where η is an adjustment parameter, and τ is a reliability threshold.

[0051] In an embodiment, the method further comprises:

[0052] When the conflict relaxation coefficient is greater than 0, the conflict relaxation coefficient is adjusted to improve the rationality of the fusion result in the case of high conflict.

[0053] In an embodiment, the method further comprises:

[0054] After each fusion, the basic probability assignment of the current fusion result is verified.

[0055] If the basic probability assignment does not meet the preset rationality condition, the conflict buffer coefficient is readjusted and the weight of the information source participating in fusion is reevaluated.

[0056] In an embodiment, each of the monitoring nodes comprises at least one of a visible light image sensor, an infrared thermal imaging sensor, an ultrasonic weather sensor, a conductor tension sensor, a tower inclination sensor, and a non-contact fault location sensor.

[0057] The power transmission line multi-dimensional state monitoring method based on the wideband low-power self-organizing network technology provided by the embodiment of the application can utilize the wideband low-power self-organizing network technology to realize automatic networking, dynamic routing and low-power data transmission between the field sensors, the edge computing platform and the cloud data center, thereby ensuring the real-time performance of data collection and the stability of transmission. Meanwhile, the multi-source feature value analysis method is used to fuse the features extracted by different sensors, and the Dempster-Shafer evidence theory is combined to fuse the outputs from the target detection, tracking and classification modules, and finally the closed-loop feedback mechanism is used to realize self-correction and continuous optimization of the system. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 The flowchart of the power transmission line multi-dimensional state monitoring method based on the wideband low-power self-organizing network technology provided by the embodiment of the application is shown.

[0059] Figure 2 The schematic diagram of the electronic device provided by the embodiment of the application is shown.

[0060] Figure 3 The image recognition schematic diagram provided by the embodiment of the application is shown.

[0061] Explanation of main element symbols

[0062] Electronic device 20

[0063] Processor 21

[0064] Memory 22

[0065] Method steps S100-S900 DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are some of the embodiments of the application, but not all the embodiments of the application.

[0067] It should be noted that the "at least one" in the embodiments of the present application refers to one or more, and more refers to two or more than two. Unless otherwise defined, all technical and scientific terms used in the present application have the same meaning as understood by those skilled in the art belonging to the technical field in the present application. The terms used in the specification of the present application are only for the purpose of describing the specific embodiments and are not intended to limit the present application.

[0068] It should be noted that in the embodiments of the present application, the terms "first", "second", etc. are used only for the purpose of distinguishing the description, and cannot be understood as indicating or implying relative importance, nor can it be understood as indicating or implying order. The features limited by "first", "second" can explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the terms "exemplary" or "for example" are used to represent as an example, illustration or description. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the use of "exemplary" or "for example" and the like is intended to present the relevant concept in a specific manner.

[0069] Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application.

[0070] With the development of smart grid and digital power system, the safe and stable operation of transmission line as the core component of power grid is of key significance to the entire power system. In recent years, with the rapid development of emerging technologies such as Internet of Things, edge computing and deep learning, the use of various sensors for transmission line state monitoring has become a research hotspot. At present, there are many technical solutions and products for transmission line monitoring at home and abroad, mainly using single or a few types of sensors to collect data, such as image monitoring, temperature detection, weather data collection, etc. However, the traditional method has the following shortcomings:

[0071] Data acquisition limitations: single sensor or single monitoring technology is difficult to fully reflect the operation state of the transmission line. There are information islands between various sensor data, which is difficult to form a global and three-dimensional understanding of the line state, especially in the early warning of abnormal events (such as equipment aging, local faults, external damage, etc.).

[0072] Insufficient data fusion and intelligent analysis: most of the existing monitoring systems focus on data collection and local analysis, lack effective multi-source data fusion and intelligent decision-making means. Each monitoring module often runs independently, and fails to fully utilize advanced algorithms such as deep learning and Dempster-Shafer evidence theory to integrate uncertain information, resulting in insufficient accuracy of abnormal detection and real-time response capability.

[0073] Network transmission challenges: Power transmission lines are often located in remote areas with complex network environments and limited power supply conditions. Traditional monitoring systems are difficult to balance the requirements of large bandwidth and low power consumption in data transmission, leading to unstable data transmission and difficulty in realizing real-time monitoring and remote management.

[0074] Insufficient closed-loop feedback and self-optimization: Existing systems lack effective feedback mechanisms, and modules cannot form closed-loop control, making it difficult for the system to automatically adjust and self-optimize in the face of dynamic environmental changes and long-term operation, affecting overall monitoring effectiveness.

[0075] Therefore, the present application proposes a power transmission line multi-dimensional state monitoring method based on wideband low-power ad hoc network technology. This method makes full use of wideband low-power ad hoc network technology to realize automatic networking, dynamic routing and low-power data transmission between on-site sensors, edge computing platforms and cloud data centers, thereby ensuring the real-time data collection and stability of data transmission. At the same time, a multi-source feature value analysis method is used to fuse the features extracted by different sensors, and the outputs from the target detection, tracking and classification modules are combined using Dempster-Shafer evidence theory to fuse uncertain information, and finally a closed-loop feedback mechanism is used to realize self-correction and continuous optimization of the system.

[0076] Figure 1 is the flowchart of the power transmission line multi-dimensional state monitoring method based on wideband low-power ad hoc network technology provided by an embodiment of the present application, as shown in Figure 1 The power transmission line multi-dimensional state monitoring method based on wideband low-power ad hoc network technology includes at least the following steps: S100: configuring multiple monitoring nodes along the power transmission line; S200: applying wideband low-power ad hoc network technology to build an ad hoc network between the multiple monitoring nodes; S300: collecting power transmission line monitoring data from the multiple monitoring nodes through the ad hoc network; S400: obtaining the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameters of each monitoring node; S500: calculating the fusion feature vector of each monitoring node according to the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameters; S600: applying an anomaly judgment algorithm to analyze the fusion feature vector and the power transmission line monitoring data obtained by each monitoring node to generate a preliminary anomaly judgment result; S700: converting the power transmission line monitoring data and the preliminary anomaly judgment result into basic probability assignments according to a mapping function; S800: recursively fusing multiple basic probability assignments using D-S evidence theory and outputting the final fusion confidence; S900: comparing the final fusion confidence with a preset safety threshold, and triggering an alarm information when the final fusion confidence is greater than the preset safety threshold.

[0077] S100: configuring multiple monitoring nodes along the power transmission line.

[0078] In the embodiment of the present application, the power line multi-dimensional state monitoring method based on the wideband low-power ad hoc network technology comprises configuring multiple monitoring nodes along the power line in step S100.

[0079] In the embodiment of the present application, each monitoring node comprises at least one of a visible light image sensor, an infrared thermal imaging sensor, an ultrasonic weather sensor, a conductor tension sensor, a tower inclination sensor, and a non-contact fault location sensor.

[0080] It can be understood that in the embodiment of the present application, the design of configuring multiple monitoring nodes along the power line fully considers the complexity of the line and the demand for multi-dimensional state monitoring. The distribution of the monitoring nodes should be optimized according to the geographical features, environmental conditions, and potential risk points of the power line. For example, sensors are preferentially deployed at key positions across valleys, rivers, or densely populated areas to ensure key coverage of high-risk sections. In addition, the monitoring nodes support the integration of multiple sensor types (such as visible light image sensors, infrared thermal imaging sensors, ultrasonic weather sensors, etc.), which can comprehensively collect the operating state data of the power line. This multi-sensor cooperative working mode not only improves the richness of the data, but also provides a solid foundation for subsequent multi-source data fusion.

[0081] S200: applying the wideband low-power ad hoc network technology to build an ad hoc network among the multiple monitoring nodes.

[0082] In the embodiment of the present application, the power line multi-dimensional state monitoring method based on the wideband low-power ad hoc network technology comprises, in step S200, applying the wideband low-power ad hoc network technology to build an ad hoc network among the multiple monitoring nodes.

[0083] It can be understood that in the embodiment of the present application, the application of the wideband low-power ad hoc network technology solves the limitations of the traditional power line monitoring system in network transmission. Through dynamic routing algorithms and intelligent networking mechanisms, the technology can achieve stable data transmission in complex communication environments. Specifically, the ad hoc network can automatically adapt to the changes in the terrain along the power line and signal interference conditions, and select the optimal path for data forwarding. At the same time, its low-power feature enables the monitoring nodes to operate for a long time under limited power supply conditions (such as solar or battery power supply), thereby significantly reducing maintenance costs. In addition, the dynamic topology design of the ad hoc network can effectively deal with the situation of partial node failure or signal loss, ensuring the reliability of the entire system.

[0084] S300: collecting power line monitoring data from the multiple monitoring nodes through the ad hoc network.

[0085] In the embodiment of the present application, the power line multi-dimensional state monitoring method based on the wideband low-power ad hoc network technology comprises, in step S300, collecting power line monitoring data from a plurality of monitoring nodes through an ad hoc network.

[0086] It can be understood that, in the embodiment of the present application, the power line monitoring data collected through the ad hoc network needs to be strictly preprocessed and quality controlled to ensure the effectiveness and accuracy of the data. For example, abnormal values or noise interference can be removed through data cleaning technology, or preliminary analysis and compression of the original data can be performed through edge computing, thereby reducing unnecessary data transmission volume. In addition, in order to meet the demand of real-time monitoring, the system will allocate bandwidth resources according to the importance of the monitoring nodes and the data priority, to ensure that critical data can be timely uploaded to the cloud or edge computing platform. This hierarchical processing and resource optimization strategy significantly improves the overall performance of the system.

[0087] S400: Obtain the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameter of each monitoring node.

[0088] In the embodiment of the present application, the power line multi-dimensional state monitoring method based on the wideband low-power ad hoc network technology comprises, in step S400, obtaining the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameter of each monitoring node.

[0089] It can be understood that, in the embodiment of the present application, the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameter are core indicators for evaluating the reliability of the monitoring nodes. Among them, the historical false alarm rate reflects the alarm accuracy of the sensor in long-term operation, which can be calculated by continuously recording the comparison log of the sensor alarm event and the actual fault event; the real-time signal-to-noise ratio is used to measure the quality of the current signal, which is obtained by real-time collection of the power ratio of the original signal to the background noise; and the environmental interference parameter is quantified by a standard deviation evaluation method to evaluate the influence of the external environment on the performance of the sensor. The comprehensive analysis of these parameters provides an important basis for subsequent reliability score calculation and data fusion.

[0090] In the embodiment of the present application, obtaining the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameter of each monitoring node comprises: continuously recording the comparison log of the sensor alarm event and the actual fault event at each monitoring node to calculate the historical false alarm rate of each monitoring node; real-time collecting the power ratio of the original signal to the background noise at each monitoring node to obtain the real-time signal-to-noise ratio of each monitoring node; and collecting the environmental interference parameter of each monitoring node in each preset period according to the standard deviation evaluation method.

[0091] It can be understood that the process of obtaining the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameters is an important basis for evaluating the reliability of the monitoring node. By continuously recording the comparison log of sensor alarm events and actual fault events at each monitoring node, the historical false alarm rate of each node can be accurately calculated. This process not only reflects the alarm accuracy of the sensor in the long-term operation, but also provides data support for subsequent optimization of alarm strategy. For example, for nodes with high historical false alarm rate, the false alarm probability can be reduced by adjusting the alarm threshold or performing periodic calibration, thereby improving the overall system reliability.

[0092] It can be understood that the collection of real-time signal-to-noise ratio and environmental interference parameters further complements the comprehensive evaluation of the working state of the monitoring node. By collecting the real-time signal-to-noise ratio of the original signal and the background noise power, the signal quality change of each node can be dynamically mastered, and the performance decline problem caused by external interference or equipment aging can be found in time. At the same time, the environmental interference parameters obtained by using the standard deviation evaluation method in each preset period can quantify the influence degree of external environment (such as weather, electromagnetic interference, etc.) on the performance of the sensor. This multi-dimensional parameter collection method ensures the comprehensive evaluation of the reliability of the monitoring node, and provides accurate data basis for subsequent fusion feature vector calculation and abnormality judgment. In addition, the dynamic updating mechanism of these parameters also makes the system adapt to complex operating environment, thereby improving the overall monitoring effect.

[0093] S500: Calculate the fusion feature vector of each monitoring node according to the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameters.

[0094] In the embodiments of the present application, the power line multi-dimensional state monitoring method based on wideband low-power self-organizing network technology includes, in step S500, calculating the fusion feature vector of each monitoring node according to the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameters.

[0095] In the embodiments of the present application, calculating the fusion feature vector of each monitoring node according to the historical false alarm rate, real-time signal-to-noise ratio and environmental interference parameters includes:

[0096] Calculating the reliability score, and the calculation formula of the reliability score is as follows:

[0097] r i i = α · SNR i i + β · (1 - FR i i ) + γ · (1 - NI i i )

[0098] Where r i is the reliability score of the i-th monitoring node, α, β, γ are empirical coefficients and α + β + γ = 1, SNR i is the real-time signal-to-noise ratio of the i-th monitoring node, FRi is the historical false alarm rate of the i-th monitoring node, NI i is the environmental interference parameter.

[0099] It can be understood that by calculating the reliability score, the performance status of each monitoring node can be comprehensively reflected. The formula comprehensively considers the real-time signal-to-noise ratio, historical false alarm rate and environmental interference parameter, and balances the influence of different factors through the weighted distribution of empirical coefficients. Among them, the higher the real-time signal-to-noise ratio, the lower the historical false alarm rate and the smaller the environmental interference parameter, the higher the reliability score of the node, so as to occupy a greater weight in the subsequent data fusion process. This design ensures accurate evaluation of the performance of each monitoring node, laying a foundation for subsequent fusion of feature vectors.

[0100] In the embodiments of the present application, all monitoring nodes are scored and normalized according to the reliability score to obtain the weight, and the formula for normalization is as follows:

[0101]

[0102] where w i is the normalized fusion weight, j is an index variable, indicating that all monitoring nodes are traversed when summing, r j represents the reliability score of each monitoring node.

[0103] It can be understood that by normalizing the reliability scores of all monitoring nodes, the reliability scores of different nodes can be converted into weight values in a unified range. The normalized weight w i indicates the relative importance of the i-th monitoring node in the overall system, and the greater the weight value, the higher the reliability and contribution of the node. This process not only simplifies the complexity of subsequent calculations, but also effectively avoids unreasonable results caused by excessively high or low scores of a single node, thereby improving the fairness and accuracy of data fusion.

[0104] In the embodiments of the present application, the original feature vectors extracted by the i-th node are fused according to the weight to obtain the fused feature vector, and the calculation formula of the fusion process is as follows:

[0105]

[0106] where F i is the original feature vector.

[0107] It can be understood that by fusing the original feature vectors extracted by each monitoring node according to the weight, the final fused feature vector can be generated. This process uses the normalized weight w i to weight the original feature vector F iThe weighted sum is performed to ensure that the contribution of each node is proportional to its reliability. This method not only improves the representativeness of the fused feature vector, but also effectively reduces the risk of misjudgment caused by the performance fluctuation of a single node. In addition, through the fusion of multi-source feature values, information from different sensors can be integrated to form a global and three-dimensional understanding of the state of the power transmission line, thereby providing more comprehensive and accurate data support for subsequent anomaly detection and intelligent analysis.

[0108] S600: An anomaly judgment algorithm is applied to analyze the fused feature vector and the power transmission line monitoring data obtained by each monitoring node to generate a preliminary anomaly judgment result.

[0109] In the embodiments of the present application, the power transmission line multi-dimensional state monitoring method based on the wideband low-power ad hoc network technology includes, in step S600, applying an anomaly judgment algorithm to analyze the fused feature vector and the power transmission line monitoring data obtained by each monitoring node to generate a preliminary anomaly judgment result.

[0110] It can be understood that, in the embodiments of the present application, the design of the anomaly judgment algorithm adopts a multi-stage processing mode combining target detection, tracking and classification. First, the target detection submodule quickly locates the possible abnormal area by analyzing the input original feature vector; then, the tracking submodule combines the historical trajectory information of the candidate abnormal area to predict the motion state of the target and verify its rationality; finally, the CNN cascaded submodel outputs a preliminary confidence score based on the regional features and the motion state, including an abnormal confidence score and a normal confidence score. This hierarchical processing mode not only improves the accuracy of anomaly detection, but also enhances the robustness of the system, which can effectively cope with uncertain factors in complex environments.

[0111] In the embodiments of the present application, applying an anomaly judgment algorithm to analyze the fused feature vector and the power transmission line monitoring data obtained by each monitoring node to generate a preliminary anomaly judgment result includes: inputting the original feature vector into a target detection submodule to generate a candidate abnormal area; inputting the regional features and the historical trajectory of the candidate abnormal area into a tracking submodule to predict the motion state of the target; and inputting the motion state of the target and the regional features into a CNN cascaded submodel to output a preliminary confidence score, the preliminary confidence score including an abnormal confidence score and a normal confidence score.

[0112] Specifically, in the embodiments of the present application, the anomaly judgment algorithm intelligently analyzes the state of the power transmission line through a multi-stage processing flow. First, the original feature vectors extracted by each monitoring node are input into a target detection submodule, which quickly locates the possible abnormal area based on deep learning or traditional image processing technology and generates a candidate abnormal area. This process can effectively filter out potential risk points and reduce the complexity and data volume of subsequent calculations.

[0113] It can be understood that after the candidate abnormal area is generated, the feature information (such as shape, texture, position, etc.) of the area and its historical trajectory are further input to the tracking sub-module. The tracking sub-module verifies the rationality of the candidate abnormal area by predicting the motion state of the target. For example, if a certain abnormal area does not show significant dynamic changes in consecutive time steps, it may be determined as a false alarm; on the contrary, if its motion trend is consistent with the known fault mode, it further increases the possibility of being an abnormality. This analysis method combining region features and historical trajectories significantly improves the accuracy of anomaly detection.

[0114] It can be understood that finally, the target motion state and the region feature are input to the CNN cascade sub-model to output preliminary confidence scores, including abnormal confidence scores and normal confidence scores. The CNN cascade sub-model models the complex nonlinear relationship through a multi-layer neural network structure, and can extract more discriminative information from a high-dimensional feature space. The abnormal confidence score reflects the probability that the current state belongs to an abnormal event, and the normal confidence score represents the probability that it is a normal state. Both of them constitute the preliminary abnormality judgment result, which provides a basis for subsequent fusion based on D-S evidence theory. Through the cooperative work of target detection, tracking and classification, the precise evaluation and efficient monitoring of the power line state are realized.

[0115] S700: Convert the power line monitoring data and the preliminary abnormality judgment result into basic probability distribution according to the mapping function.

[0116] In the embodiment of the present application, the power line multi-dimensional state monitoring method based on wideband low-power self-organizing network technology includes converting the power line monitoring data and the preliminary abnormality judgment result into basic probability distribution according to the mapping function in step S700.

[0117] It can be understood that in the embodiment of the present application, the role of the mapping function is to convert the preliminary abnormality judgment result into a basic probability distribution suitable for D-S evidence theory processing. By introducing a temperature coefficient to adjust the smoothness of the mapping function, the sensitivity of the output can be flexibly adjusted in different scenarios. For example, in a high-noise environment, appropriately increasing the temperature coefficient can reduce the sensitivity to abnormal confidence scores, avoiding frequent alarms caused by false alarms; while in a low-noise environment, the temperature coefficient can be reduced to improve the detection sensitivity. This method ensures the accuracy of detection while also considering the stability of the system.

[0118] In the embodiments of the present application, the transmission line monitoring data and the preliminary abnormality judgment result are converted into basic probability distribution according to a mapping function, including: obtaining abnormal confidence scores and normal confidence scores according to the transmission line monitoring data and the preliminary abnormality judgment result; setting a temperature coefficient to adjust the smoothness of the mapping function; and converting the confidence scores into the basic probability distribution by using the mapping function, and the specific calculation formula is as follows:

[0119]

[0120] wherein A represents an abnormality hypothesis, F k represents the abnormal confidence score of the kth information source, represents the normal confidence score of the kth monitoring node, and T k is a temperature coefficient.

[0121] Specifically, in the embodiments of the present application, the role of the mapping function is to convert the abnormal confidence scores and the normal confidence scores in the transmission line monitoring data and the preliminary abnormality judgment result into the basic probability distribution suitable for the D-S evidence theory processing. This process adjusts the smoothness of the mapping function by introducing a temperature coefficient T k , so as to flexibly adjust the sensitivity of the output in different scenarios.

[0122] It can be understood that, first, the abnormal confidence scores F k and the normal confidence scores are obtained according to the transmission line monitoring data and the preliminary abnormality judgment result. These two scores respectively represent the probability estimation value of the current state belonging to an abnormal event or a normal event, and are the basis for subsequent basic probability distribution calculation. In order to ensure the flexibility and adaptability of the mapping process, a temperature coefficient T k is set, which controls the steepness of the mapping function. A higher temperature coefficient will make the mapping function smoother, reduce the sensitivity to the change of the confidence score, and be suitable for a high-noise or high-uncertainty environment; while a lower temperature coefficient will make the mapping function steeper, improve the response speed to the change of the confidence score, and be suitable for a low-noise or high-deterministic scenario.

[0123] It can be understood that the specific formula for converting the confidence scores into the basic probability distribution by using the mapping function is as follows:

[0124]

[0125] The formula normalizes the abnormal confidence scores and the normal confidence scores into a probability value by using the softmax function, so that the final basic probability distribution m k(A) falls within the range of [0, 1]. This design not only guarantees the rationality of the output, but also effectively balances the sensitivity and stability of anomaly detection, avoiding misjudgment caused by extreme values. At the same time, by dynamically adjusting the temperature coefficient T k , the system can achieve better performance in different environments. This design not only retains the information of the original confidence score, but also effectively controls the degree of uncertainty through the adjustment of the temperature coefficient.

[0126] It can be understood that the smoothness adjustment mechanism of the mapping function is of great significance in practical applications. The setting of the temperature coefficient T k needs to be adjusted in combination with specific scenarios and system requirements. For example, in critical areas of power transmission lines (such as crossing valleys or densely populated areas), the sensitivity to potential anomalies can be improved by reducing the temperature coefficient; while in areas with greater environmental interference (such as near strong electromagnetic fields), the robustness of the system can be enhanced by increasing the temperature coefficient. This dynamic adjustment mechanism enables the system to maintain good performance in complex and variable environments, providing a reliable foundation for subsequent uncertainty information fusion.

[0127] S800: Apply D-S evidence theory to recursively fuse multiple basic probability assignments and output the final fused confidence.

[0128] In the embodiments of the present application, the power line multi-dimensional state monitoring method based on wideband low-power ad hoc network technology includes applying D-S evidence theory to recursively fuse multiple basic probability assignments and output the final fused confidence in step S800.

[0129] In the embodiments of the present application, applying D-S evidence theory to recursively fuse multiple basic probability assignments and output the final fused confidence includes:

[0130] Calculate the conflict factor for any two information sources, and the specific calculation formula is as follows:

[0131]

[0132] where K ij represents the conflict factor, i, j represent two different information sources, K ij represents the conflict factor, B, C are hypothesis sets, m i (B) represents the basic probability assignment of the i-th information source to the hypothesis set B, m j (C) represents the basic probability assignment of the j-th information source to the hypothesis set C.

[0133] It can be understood that in the embodiments of the present application, the conflict factor K ijThe process is an important step of evaluating the conflict degree between multiple information sources. The formula can accurately reflect the conflict level between information sources i and j by quantifying the product of the probability distribution in the case that the intersection between different hypothesis sets (B and C) is empty. A higher conflict factor value indicates that there is a greater difference between the evidence provided by the two information sources, which may be caused by sensor errors, environmental interference or data anomalies. Therefore, the calculation of the conflict factor provides a basis for conflict processing in the subsequent fusion process.

[0134] In the embodiment of the present application, a modified rule with a conflict mitigation coefficient is used for fusion, and the specific fusion formula is as follows:

[0135]

[0136] wherein A is a hypothesis, m i (A) represents the basic probability assignment of the ith information source to the hypothesis A, m j (A) represents the basic probability assignment of the jth information source to the hypothesis A, and λ is a conflict mitigation coefficient for adjusting the influence of conflict, m ij (A) represents the basic probability assignment of the fused information sources i and j to the hypothesis A, min{m i (A), m j (A)} is a conflict mitigation term.

[0137] It can be understood that in the embodiment of the present application, a modified rule with a conflict mitigation coefficient λ is used for fusion. This formula introduces a conflict mitigation mechanism on the basis of the traditional D-S evidence theory, and dynamically balances the influence of conflict by adjusting the value of λ. When the conflict factor K ij is high, appropriately increasing λ can increase the weight of the conflict mitigation term λmin{m i (A), m j (A)} so as to avoid unreasonable fusion results caused by excessive conflict. On the contrary, when the conflict is small, a smaller λ value can retain the information of the original evidence more fully. This design significantly improves the rationality and stability of the fusion result. In the embodiment of the present application, the basic probability assignment of the fused information sources i and j to the hypothesis A is recursively fused with the third information source until all information sources are merged, and the final fusion confidence is output.

[0138] It can be understood that in the embodiment of the present application, the process of recursive fusion gradually combines the fused information sources with other information sources that have not participated in the fusion until all information sources are merged. For example, the basic probability assignment m ij (A) is obtained by first fusing information sources i and j, and then m ij(A) continue to fuse with the basic probability assignment of the third information source, and so on until all information sources are included in the fusion range. This recursive process not only ensures the comprehensive integration of multi-source data, but also effectively deals with the case where the number of information sources is large in a complex scenario. The final output fusion confidence comprehensively reflects the support degree of each information source for the abnormal hypothesis, and provides a reliable basis for subsequent alarm decision.

[0139] S900: Compare the final fusion confidence with a preset safety threshold, and trigger an alarm information when the final fusion confidence is greater than the preset safety threshold.

[0140] In the embodiment of the present application, the power line multi-dimensional state monitoring method based on wideband low-power ad hoc network technology includes, in step S900, comparing the final fusion confidence with a preset safety threshold, and triggering an alarm information when the final fusion confidence is greater than the preset safety threshold.

[0141] It can be understood that, in the embodiment of the present application, the setting of the preset safety threshold needs to be scientifically planned in combination with historical data statistics and expert experience to ensure the timeliness and accuracy of the alarm information. When the final fusion confidence exceeds the threshold, the system will immediately trigger an alarm and send relevant information to the operation and maintenance personnel for processing. The content of the alarm information not only includes the time and location of the abnormal occurrence, but also covers detailed descriptions of the specific abnormal type and severity, so as to facilitate the operation and maintenance personnel to quickly locate the problem and take appropriate measures. In addition, the system also supports a closed-loop feedback mechanism to automatically correct related parameters after each alarm, thereby continuously improving the monitoring performance of the system.

[0142] In the embodiment of the present application, the method further includes dynamically adjusting the reliability score between information sources, and the specific condition formula is as follows:

[0143]

[0144] Wherein, η is an adjustment parameter, and τ is a reliability threshold.

[0145] Specifically, in the embodiment of the present application, the dynamic adjustment mechanism of the conflict relaxation coefficient λ is optimized based on the reliability score between information sources. By combining the adjustment parameter η and the reliability threshold τ, the comprehensive reliability score r i + r j of the information sources i and j can be flexibly reflected. When the reliability score of the information source is high, the value of λ will decrease, thereby reducing the weight of the conflict relaxation term and more fully retaining the information of the original evidence; on the contrary, when the reliability score is low, the value of λ will increase, thereby increasing the weight of the conflict relaxation term to reduce the influence of the conflict on the fusion result. This design ensures the rationality and stability of the fusion result under different reliability levels.

[0146] It can be understood that in the embodiments of the present application, when the conflict mitigation coefficient λ>0, the system further adjusts the value of λ to improve the rationality of the fusion result in the case of high conflict. For example, in some extreme scenarios (such as significant contradictions between multiple information sources or strong environmental interference), simply relying on the original evidence may cause the fusion result to deviate from the actual situation. At this time, appropriately increasing the value of λ can enhance the effect of the conflict mitigation mechanism and effectively alleviate the risk of misjudgment caused by high conflict. This process adjusts the value range of λ dynamically, ensuring that the system can still maintain good performance in complex and variable environments.

[0147] In the embodiments of the present application, the method further comprises: when the conflict mitigation coefficient is greater than 0, adjusting the conflict mitigation coefficient to improve the rationality of the fusion result in the case of high conflict.

[0148] Specifically, the adjustment process of the conflict mitigation coefficient λ is based on whether the basic probability assignment of the fusion result meets the preset rationality condition. For example, if it is found that the basic probability assignment of a hypothesis (such as the support degree of an abnormal hypothesis) deviates significantly from the expected range or contradicts the results of other information sources, the system will reevaluate the current conflict level and adjust the value of λ as needed. This adjustment mechanism can ensure that the fusion result still has high credibility and stability in the case of high conflict.

[0149] It can be understood that this dynamic adjustment mechanism not only improves the adaptability of the system in the case of high conflict, but also enhances its robustness to complex environments. By flexibly adjusting the conflict mitigation coefficient, the system can ensure data fusion accuracy while effectively dealing with contradictions between multiple sources of information, providing a more intelligent and reliable solution for state monitoring of power transmission lines. This method is particularly suitable for complex and variable environments along the power transmission line and can significantly improve the overall performance and long-term stability of the system.

[0150] In the embodiments of the present application, the method further comprises: after each fusion, verifying the basic probability assignment of the current fusion result; if the basic probability assignment does not meet the preset rationality condition, readjusting the conflict mitigation coefficient and reevaluating the weight of the information source participating in the fusion.

[0151] Specifically, in the embodiments of the present application, the basic probability assignment of the current fusion result is verified after each fusion to ensure that it meets the preset rationality condition. These rationality conditions usually include but are not limited to the following points: non-negativity of the basic probability assignment (i.e., the basic probability value of all hypotheses should be greater than or equal to zero), normalization requirement (the sum of the basic probabilities of all hypotheses is 1), and whether the support degree of an abnormal hypothesis is within a reasonable range. If it is found that the fusion result violates any of the above conditions, the system will start the correction mechanism.

[0152] Understandably, when the verification results show that the basic probability allocation does not meet the preset reasonableness conditions, the system will take the following measures to correct it:

[0153] 1. Readjust the conflict mitigation coefficient λ: based on the current conflict factor K. ij Size and reliability score of information source r i +r j The conflict mitigation coefficient λ is dynamically adjusted. For example, if the conflict factor is high, λ is increased appropriately to enhance the effect of the conflict mitigation term; conversely, if the conflict factor is low, λ is decreased to more fully preserve the information of the original evidence.

[0154] 2. Reassess the weights of information sources participating in the fusion: by recalculating the reliability score r of each information source. i and normalized weight w i The contribution ratio of each information source in the fusion process is optimized. For information sources with low reliability, their weight is reduced to decrease their impact on the fusion result; while for information sources with high reliability, their weight is increased to improve their contribution.

[0155] The multi-dimensional condition monitoring method for transmission lines based on broadband low-power self-organizing network technology provided in this application significantly improves the comprehensiveness and accuracy of transmission line condition monitoring through multi-sensor collaborative work and intelligent data analysis. First, this method utilizes broadband low-power self-organizing network technology to achieve dynamic networking and low-power data transmission between monitoring nodes, ensuring real-time data acquisition and transmission stability in complex terrain and harsh environments. Simultaneously, by fusing features extracted from different sensors through multi-source eigenvalue analysis and DS evidence theory, it effectively integrates outputs from target detection, tracking, and classification modules, solving the problem that a single sensor or technology cannot comprehensively reflect the operating status of the transmission line. This method not only improves the accuracy of abnormal event early warning but also provides strong support for the safe and stable operation of the power system.

[0156] Furthermore, this method achieves system self-correction and continuous optimization through a closed-loop feedback mechanism, further enhancing its long-term reliability and adaptability. In practical applications, the system can dynamically adjust the conflict mitigation coefficient based on the reliability scores between information sources and verify the rationality of the basic probability allocation after each fusion, thereby promptly identifying and correcting potential problems. This design not only reduces the false alarm rate and false negative rate but also significantly improves the system's robustness to complex environments. The multi-dimensional condition monitoring method based on broadband low-power self-organizing network technology provides an intelligent solution for transmission line condition assessment and fault early warning, significantly reducing the workload and cost of manual inspections and laying a solid foundation for the construction of smart grids and digital power systems.

[0157] The application scenario of the method is described below with an exemplary embodiment.

[0158] There is a construction site near a power transmission line, about 10 kilometers long, with complex environments along the line, including valleys, forests, and strong electromagnetic interference areas. To ensure the safe operation of the line, a multi-dimensional state monitoring method based on wideband low-power ad hoc network technology is used for real-time monitoring.

[0159] Ten monitoring nodes are deployed along the power transmission line, each equipped with the following sensors:

[0160] Visible light image sensor

[0161] Infrared thermal imaging sensor

[0162] Ultrasonic weather sensor

[0163] Conductor tension sensor

[0164] Tower inclination sensor

[0165] Step 1: Build an ad hoc network

[0166] Through the wideband low-power ad hoc network technology, a communication network is dynamically built between the 10 monitoring nodes. Assuming that the communication range of each node is 3 kilometers, the entire network is divided into 3 subnets, each containing 3-4 nodes, and cross-subnet data transmission is achieved through relay nodes.

[0167] Step 2: Collect monitoring data

[0168] Each monitoring node collects state data of the power transmission line at a preset period (every 5 minutes). The following is part of the collected data at a certain time:

[0169] Node number Temperature (°C) Tension (kN) Inclination angle (°) Image abnormal area (%) 1 35 12.5 0.8 0 2 40 13.0 1.2 0 3 38 11.8 1.5 5 ... ... ... ... ...

[0170] Step 3: Obtain reliability parameters

[0171] Calculate the historical false alarm rate, real-time signal-to-noise ratio, and environmental interference parameters of each node:

[0172] Historical false alarm rate: derived by comparing the past month's alarm events with the actual fault log, for example, the historical false alarm rate of node 3 is 10%.

[0173] Real-time signal-to-noise ratio: real-time collection of the ratio of original signal power to background noise power, for example, the signal-to-noise ratio of node 3 is 20dB.

[0174] Environmental interference parameters: calculated according to the standard deviation evaluation method, for example, the environmental interference parameter of node 3 is 0.7.

[0175] Step 4: Calculate the fusion feature vector

[0176] According to the formula

[0177] r i = a SNR i + b (1 FR i ) + g (1 NI i )

[0178] Set the experience coefficients a = 0.5, b = 0.3, g = 0.2, and calculate the reliability scores of each node:

[0179] Node 3:

[0180] r3 = 0.5 * 20 + 0.3 * (1 - 0.1) + 0.2 * (1 - 0.7) = 10.6

[0181] According to the normalization weight formula, assuming that the reliability scores of all nodes are [10.6, 9.8, 11.2, …, 10.0], the weight of node 3 is:

[0182]

[0183] Step 5: Generate preliminary anomaly judgment results

[0184] Input the fusion feature vector into the target detection sub-module to generate candidate abnormal regions (such as the construction crane being close to the power transmission line). Please refer to Figure 3 , Figure 3 is the image recognition schematic diagram provided by an embodiment of the present application. Then, the tracking sub-module combines the historical trajectory to predict the target motion state, and inputs the result into the CNN cascade sub-model to output the preliminary confidence score:

[0185] Abnormal confidence score F k = 0.82

[0186] Normal confidence score

[0187] Step 6: Convert to basic probability assignment

[0188] According to the mapping function formula, set the temperature coefficient T k = 1.0

[0189]

[0190] Step 7: Recursive fusion and output final confidence

[0191] Recursive fusion is performed on the basic probability assignments of multiple information sources, assuming that the conflict factor K ij = 0.2, and l = 0.5, then the fusion formula is:

[0192]

[0193] Continue recursive fusion until all information sources are merged, and the final output fusion confidence is 0.72.

[0194] Step 8: Trigger alarm information

[0195] The preset safety threshold is 0.7. Since 0.72>0.7, the system triggers the alarm information and sends the abnormal position, type and severity to the operation and maintenance personnel.

[0196] As can be seen from the above exemplary embodiments, the method can effectively integrate various sensor data, and use the wideband low-power ad hoc network technology and D-S evidence theory to realize high-precision transmission line state monitoring. In actual application, the method not only improves the accuracy of anomaly detection, but also significantly reduces the false positive rate and the false negative rate, thereby providing a reliable guarantee for the safe and stable operation of the power system.

[0197] Figure 2 An electronic device 20 provided by an embodiment of the present application. As shown in Figure 2 The electronic device 20 at least includes the following parts: a processor 21 and a memory 22.

[0198] In the embodiment of the present application, the memory 22 is used to store the processor 21 executable instructions, and the processor 21 is configured to execute the instructions to implement the wideband low-power ad hoc network technology-based transmission line multi-dimensional state monitoring method as Figure 1 shown in the embodiment.

[0199] In the embodiment of the present application, a computer readable storage medium includes instructions, and the instructions instruct the device to execute the method of the first aspect. For example, the instructions instruct the device to execute the wideband low-power ad hoc network technology-based transmission line multi-dimensional state monitoring method as shown in steps S100 to S900 in Figure 1 the embodiment.

[0200] The program working in the electronic device 20 related to an embodiment of the present application can be a program (a program for making a computer function) for controlling a central processing unit (CPU) and the like to realize the functions of the above-described embodiments related to one scheme of the present application. Then, the information processed by these devices is temporarily stored in a random access memory (RAM) when it is processed, and then stored in various ROMs such as a read only memory (Flash ROM), a hard disk drive (HDD), and read out, corrected, and written by a CPU as needed.

[0201] Note that a part of the electronic device 20 of the above-described embodiment can be realized by a computer. In this case, a program for realizing the control function can be recorded in a computer-readable recording medium, and the realization can be achieved by reading the program recorded in the recording medium into a computer and executing it.

[0202] Note that the "computer" referred to here means a computer built into the electronic device 20, and a computer including hardware such as an OS and a peripheral device. Further, the "computer-readable recording medium" means a removable medium such as a floppy disk, a magneto-optical disk, a ROM, a CD-ROM, and a storage device such as a hard disk built into a computer.

[0203] Further, the "computer-readable recording medium" can include a medium that dynamically stores a program for a short time, such as a communication line in the case of transmitting a program via a network such as the Internet or a communication line such as a telephone line, and a medium that stores a program for a fixed time, such as a volatile memory inside a computer that is a server or a client in this case. Further, the above-described program can be a program for realizing a part of the above-described function, and can also be a program that can realize the above-described function by being combined with a program already recorded in a computer.

[0204] Further, the electronic device 20 in the above-described embodiment can also be realized as an assembly (device group) composed of a plurality of devices. Each device constituting the device group can have a part or all of each function or each functional block of the electronic device 20 of the above-described embodiment. As the device group, all of each function or each functional block of the electronic device 20 can be possessed.

[0205] It can be understood that the power transmission line multi-dimensional state monitoring method based on the wideband low-power self-organizing network technology provided by the embodiments of the present application has significant beneficial effects. First, the method effectively solves the problems of insufficient network coverage and unstable transmission in traditional monitoring systems by configuring monitoring nodes of multiple sensor types along the power transmission line and combining wideband low-power self-organizing network technology to realize dynamic networking and data transmission. The self-organizing network technology can automatically adapt to complex terrain and harsh environments, ensuring the real-time data collection and reliable transmission. At the same time, its low-power characteristic enables the monitoring nodes to operate for a long time under limited power supply conditions, greatly reducing the maintenance cost and energy consumption of the system. In addition, through the multi-source data fusion technology, the data from different sensors can be integrated to comprehensively reflect the operation state of the power transmission line, thereby providing strong guarantee for the safe and stable operation of the power system.

[0206] It is understandable that the anomaly detection and uncertainty information fusion mechanism in this application embodiment further enhances the intelligence level of the monitoring system. By recursively fusing the basic probability allocations of multiple information sources using DS evidence theory, this method can effectively handle conflicts and uncertainties between multi-source data, improving the accuracy and robustness of anomaly judgment. In particular, the dynamic adjustment mechanism of the conflict mitigation coefficient automatically optimizes the fusion strategy based on the reliability score of the information source, ensuring the rationality of the fusion result. Furthermore, the application of the closed-loop feedback mechanism enables the system to continuously self-correct and optimize according to actual operating conditions, thereby possessing stronger adaptability and long-term stability. These designs not only improve the monitoring accuracy of the system but also enhance its reliability and anti-interference capability in complex environments, providing a more intelligent and efficient solution for the condition monitoring of transmission lines.

[0207] Those skilled in the art should recognize that the above embodiments are only used to illustrate this application and are not intended to limit this application. Any appropriate changes and variations made to the above embodiments within the essential spirit and scope of this application fall within the scope of protection claimed in this application.

Claims

1. A method for monitoring multi-dimensional state of power transmission line based on wideband low-power ad hoc network technology, applied to monitoring power transmission line, characterized in that, The method comprises: configuring a plurality of monitoring nodes along the power transmission line; applying a wideband low-power ad hoc network technology to build an ad hoc network among the plurality of monitoring nodes; collecting power transmission line monitoring data from the plurality of monitoring nodes through the ad hoc network; obtaining historical false alarm rates, real-time signal-to-noise ratios, and environmental interference parameters of each monitoring node; calculating a fusion feature vector of each monitoring node according to the historical false alarm rates, the real-time signal-to-noise ratios, and the environmental interference parameters; applying an anomaly judgment algorithm to analyze the fusion feature vector and the power transmission line monitoring data obtained by each monitoring node to generate a preliminary anomaly judgment result; converting the power transmission line monitoring data and the preliminary anomaly judgment result into basic probability assignments according to a mapping function; recursively fusing a plurality of basic probability assignments using D-S evidence theory and outputting a final fusion confidence; comparing the final fusion confidence with a preset safety threshold, and triggering an alarm information when the final fusion confidence is greater than the preset safety threshold.

2. The method according to claim 1, wherein, The method for obtaining historical false alarm rates, real-time signal-to-noise ratios, and environmental interference parameters of each monitoring node comprises: continuously recording a comparison log of sensor alarm events and actual fault events at each monitoring node to calculate the historical false alarm rate of each monitoring node; real-time acquisition of the power ratio of original signals and background noise at each monitoring node to obtain the real-time signal-to-noise ratio of each monitoring node; acquiring environmental interference parameters of each monitoring node according to a standard deviation evaluation method within each preset period.

3. The method according to claim 2, wherein, The method for calculating a fusion feature vector of each monitoring node according to the historical false alarm rates, the real-time signal-to-noise ratios, and the environmental interference parameters comprises: calculating a reliability score, and the calculation formula of the reliability score is as follows: r i = a · SNR i + b · (1 - FR i ) + g · (1 - NI i ) wherein r i is the reliability score of the i-th monitoring node, a, b, g are empirical coefficients and a+b+g = 1, SNR i is the real-time signal-to-noise ratio of the i-th monitoring node, FR i is the historical false positive rate of the i-th monitoring node, NI i is the environmental interference parameter; normalizing the scores of all monitoring nodes according to the reliability score to obtain weights, and the normalization formula is as follows: wherein w i is a normalized fusion weight, j is an index variable, indicating that all monitoring nodes are traversed when summing, r j represents the reliability score of each monitoring node; fusing the original feature vector extracted by the i-th node according to the weights to obtain a fusion feature vector, and the calculation formula of the fusion process is as follows: where F i is the original feature vector.

4. The method according to claim 3, wherein, The method for applying an anomaly judgment algorithm to analyze the fusion feature vector and the power transmission line monitoring data obtained by each monitoring node to generate a preliminary anomaly judgment result comprises: inputting the original feature vector into a target detection submodule to generate a candidate abnormal area; inputting the area features and historical trajectories of the candidate abnormal area into a tracking submodule to predict a target motion state; inputting the target motion state and the area features into a CNN cascade submodule to output a preliminary confidence score, and the preliminary confidence score includes an abnormal confidence score and a normal confidence score.

5. The method according to claim 4, wherein, The method for converting the power transmission line monitoring data and the preliminary anomaly judgment result into basic probability assignments according to a mapping function comprises: obtaining the abnormal confidence score and the normal confidence score according to the power transmission line monitoring data and the preliminary anomaly judgment result; setting a temperature coefficient to adjust the smoothness of the mapping function; applying a mapping function to convert the confidence score into a basic probability assignment, and the specific calculation formula is as follows: wherein A represents an anomaly hypothesis, F k represents an anomaly confidence score of the kth information source, represents a normal confidence score of the kth monitoring node, T k is a temperature coefficient.

6. The method according to claim 5, wherein, The application of D-S evidence theory includes: The conflict factor is calculated for any two information sources, and the specific calculation formula is as follows: where K ij denotes the conflict factor, i, j denote two different information sources, K ij denotes the conflict factor, B, C are sets of hypotheses, m i (B) denotes the basic probability assignment of the ith information source to the set of hypotheses B, m j (C) denotes the basic probability assignment of the jth information source to the set of hypotheses C; The modified rule with the conflict buffer coefficient is used for fusion, and the specific fusion formula is as follows: where A is a hypothesis, m i (A) denotes the basic probability assignment of the ith information source to hypothesis A, m j (A) denotes the basic probability assignment of the jth information source to hypothesis A, λ is a conflict relaxation coefficient used to adjust the impact of conflict, m ij (A) denotes the basic probability assignment of the fusion of information sources i and j to hypothesis A, min{m i (A), m j (A)} is the conflict relaxation term; The basic probability distribution of the information sources i and j after fusion is recursively fused with the third information source until all information sources are combined, and the final fusion confidence is output.

7. The method according to claim 6, wherein, The method further includes: The reliability score between information sources is dynamically adjusted, and the specific condition formula is as follows: Wherein, η is an adjustment parameter, and τ is a reliability threshold.

8. The method according to claim 6, wherein, The method further includes: When the conflict buffer coefficient is greater than 0, the conflict buffer coefficient is adjusted to improve the rationality of the fusion result in the case of high conflict. 9.The power transmission line multi-dimensional state monitoring method based on wideband low-power ad hoc network technology of claim 6, wherein, The method further includes: After each fusion, the basic probability distribution of the current fusion result is verified; If the basic probability distribution does not meet the preset rationality condition, the conflict buffer coefficient is adjusted and the information source weight participating in the fusion is re-evaluated.

10. The method according to claim 1, wherein, Each monitoring node includes at least one of a visible light image sensor, an infrared thermal imaging sensor, an ultrasonic weather sensor, a conductor tension sensor, a tower inclination sensor, and a non-contact fault location sensor.

Images