Composite insulator temperature abnormal data detection method and system based on evidence distance

Abstract

The present application relates to the technical field of power equipment state monitoring, in particular to a composite insulator temperature abnormal data detection method and system based on evidence distance. The method comprises the following steps: collecting infrared temperature data, field humidity data and wind speed data of the composite insulator and preprocessing; constructing a recognition framework including normal state, slight abnormal state and serious abnormal state; constructing temperature evidence based on the infrared temperature data, constructing humidity evidence by correcting the temperature difference threshold value according to the humidity data, and constructing wind speed evidence by correcting the temperature difference threshold value according to the wind speed data; calculating the improved Jousselme distance based on the Jaccard coefficient matrix of block processing, calculating the credibility weight and weighting the evidence; fusing the corrected evidence by using the Dempster combination rule, and determining the temperature state by the minimum distance criterion. The present application effectively improves the accuracy of temperature abnormality detection.

Description

Technical Field

[0001] This invention relates to the field of power equipment condition monitoring technology, specifically to a method and system for detecting abnormal temperature data of composite insulators based on evidence distance. Background Technology

[0002] Composite insulators are core insulation components of power transmission lines, and abnormal temperatures in them directly reflect potential faults such as aging, moisture, and partial discharge. Infrared drone inspection technology provides an efficient means of monitoring the temperature of composite insulators, but accurately identifying temperature anomalies remains a pressing technical challenge.

[0003] Existing methods for detecting temperature anomalies have the following main shortcomings: First, they rely on single infrared temperature data to determine anomalies, which is easily affected by environmental factors. For example, condensation on the surface of insulators in high humidity environments can lead to higher infrared temperature readings and false alarms, while convective heat dissipation in windy environments can mask minor anomalies and cause missed alarms. Second, multi-source data fusion methods use simple weighted averages or traditional evidence theories, which are difficult to effectively handle conflicts between infrared temperature and environmental data. Traditional Jousselme distance cannot distinguish between conflicts between single and multiple subsets, resulting in inaccurate assessment of the credibility of the evidence. Third, no coupling correction mechanism between temperature thresholds and environmental parameters has been established, and fixed thresholds cannot adapt to changes in the field environment. Summary of the Invention

[0004] This invention provides a method and system for detecting temperature anomalies in composite insulators based on evidence distance, which solves the problem of low detection accuracy caused by environmental interference in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The present invention provides a method for detecting temperature anomalies in composite insulators based on evidence distance, comprising:

[0007] S100: Collects infrared temperature data, on-site humidity data, and wind speed data of composite insulators, and preprocesses the collected data;

[0008] S200: Construct a framework for identifying the temperature state of composite insulators, including normal state, slightly abnormal state and severely abnormal state;

[0009] S300: Construct temperature evidence based on preprocessed infrared temperature data, adjust the temperature difference threshold downward based on on-site humidity data and construct humidity evidence, and adjust the temperature difference threshold upward based on wind speed data and construct wind speed evidence.

[0010] S400: Based on the Jaccard coefficient matrix of block processing, calculate the improved Jousselme distance between temperature evidence, humidity evidence and wind speed evidence, calculate the credibility weights of temperature evidence, humidity evidence and wind speed evidence according to the improved Jousselme distance, and use the credibility weights to perform weighted correction on temperature evidence, humidity evidence and wind speed evidence.

[0011] S500: The modified temperature evidence, humidity evidence, and wind speed evidence are fused using the Dempster synthesis rule to obtain fused evidence;

[0012] S600: Construct target evidence corresponding to normal state, slightly abnormal state and severely abnormal state respectively, calculate the improved Jousselme distance between the fused evidence and the target evidence of normal state, slightly abnormal state and severely abnormal state, and take the state corresponding to the minimum distance as the temperature state determination result.

[0013] As a preferred embodiment of the present invention, the preprocessing of the collected data includes:

[0014] use The criteria remove outliers from infrared temperature data, on-site humidity data, and wind speed data that exceed the mean plus or minus three standard deviations.

[0015] The infrared temperature data, on-site humidity data, and wind speed data after removing outliers were smoothed using the moving average method.

[0016] As a preferred embodiment of the present invention, the construction of temperature evidence based on preprocessed infrared temperature data includes:

[0017] The temperature difference is obtained by calculating the difference between the average temperature of the key parts of the composite insulator and the ambient temperature.

[0018] Set a first temperature difference threshold and a second temperature difference threshold, wherein the first temperature difference threshold is less than the second temperature difference threshold;

[0019] When the temperature difference is less than the first temperature difference threshold, the basic probability value of the normal state is greater than the basic probability value of the slightly abnormal state, the basic probability value of the slightly abnormal state is greater than the basic probability value of the severe abnormal state, and a certain basic probability value is assigned to the entire set.

[0020] When the temperature difference is greater than or equal to the first temperature difference threshold and less than the second temperature difference threshold, the basic probability value of the slight abnormal state is greater than the basic probability value of the normal state and the severe abnormal state, and a certain basic probability value is assigned to the entire set.

[0021] When the temperature difference is greater than or equal to the second temperature difference threshold, the basic probability value of the severe abnormal state is greater than the basic probability value of the slight abnormal state, the basic probability value of the slight abnormal state is greater than the basic probability value of the normal state, and a certain basic probability value is assigned to the entire set.

[0022] The basic probability values ​​satisfy the normalization condition and constitute temperature evidence.

[0023] As a preferred embodiment of the present invention, the step of adjusting the temperature difference threshold based on on-site humidity data and constructing humidity evidence includes:

[0024] When the on-site humidity data is higher than the humidity threshold, the temperature difference threshold will be lowered by a preset ratio, and the basic probability values ​​for minor and severe abnormal states will be increased.

[0025] When the on-site humidity data is not higher than the humidity threshold, the temperature difference threshold remains unchanged;

[0026] Based on the revised temperature difference threshold, basic probability values ​​are assigned to normal, slightly abnormal, and severely abnormal states to form humidity evidence.

[0027] As a preferred embodiment of the present invention, the step of adjusting the temperature difference threshold upward based on wind speed data and constructing wind speed evidence includes:

[0028] When the wind speed data is higher than the wind speed threshold, the temperature difference threshold will be increased by a preset ratio, and the basic probability value of normal state will be reduced.

[0029] When the wind speed data is not higher than the wind speed critical value, the temperature difference threshold remains unchanged;

[0030] Based on the revised temperature difference threshold, basic probability values ​​are assigned to normal, slightly abnormal, and severely abnormal states to form wind speed evidence.

[0031] As a preferred embodiment of the present invention, the block processing includes:

[0032] Based on the single-subset and multi-subset relationships of states in the identification framework, the Jaccard coefficient matrix is ​​divided into single-subset internal similarity blocks, single-subset and multi-subset similarity blocks, multi-subset internal similarity blocks, and whole-set similarity blocks.

[0033] As a preferred embodiment of the present invention, the calculation of the improved Jousselme distance between temperature evidence, humidity evidence, and wind speed evidence based on the Jaccard coefficient matrix of block processing includes:

[0034] For any two pieces of evidence, obtain the basic probability values ​​of each state and construct the corresponding basic probability assignment vector.

[0035] Calculate the difference vector between two basic probability assignment vectors;

[0036] The difference vector is subjected to a quadratic operation with the block-processed Jaccard coefficient matrix to obtain the improved Jousselme distance between the two pieces of evidence.

[0037] The matrix elements of the similarity blocks within a single subset, the similarity blocks between a single subset and multiple subsets, the similarity blocks within multiple subsets, and the similarity blocks of the entire set are determined according to the intersection and union relationships between focal elements.

[0038] As a preferred embodiment of the present invention, the calculation of the credibility weights of temperature evidence, humidity evidence, and wind speed evidence based on the improved Jousselme distance includes:

[0039] Similarity between evidence is calculated based on the improved Jousselme distance between temperature evidence, humidity evidence, and wind speed evidence.

[0040] Global support for temperature, humidity, and wind speed evidence is calculated based on the similarity between evidence.

[0041] The credibility weights of temperature evidence, humidity evidence, and wind speed evidence are obtained by normalizing the global support.

[0042] As a preferred embodiment of the present invention, the fusion of the corrected temperature evidence, humidity evidence, and wind speed evidence using the Dempster synthesis rule includes:

[0043] Calculate the conflict coefficient between the corrected temperature evidence, humidity evidence, and wind speed evidence;

[0044] Based on the conflict coefficient, the intersection of the focal elements of the corrected temperature evidence, humidity evidence, and wind speed evidence is normalized to obtain fused evidence.

[0045] This invention also proposes a composite insulator temperature anomaly data detection system based on evidence distance, comprising:

[0046] The data preprocessing module is used to collect infrared temperature data, on-site humidity data, and wind speed data of composite insulators, and to preprocess the collected data.

[0047] The framework construction module is used to build an identification framework for the temperature state of composite insulators, including normal state, slightly abnormal state and severely abnormal state;

[0048] The evidence construction module is used to construct temperature evidence based on preprocessed infrared temperature data, adjust the temperature difference threshold downward based on the on-site humidity data and construct humidity evidence, and adjust the temperature difference threshold upward based on the wind speed data and construct wind speed evidence.

[0049] The evidence correction module is used to calculate the improved Jousselme distance between temperature evidence, humidity evidence and wind speed evidence based on the Jaccard coefficient matrix processed in blocks, calculate the credibility weights of temperature evidence, humidity evidence and wind speed evidence based on the improved Jousselme distance, and use the credibility weights to perform weighted correction on temperature evidence, humidity evidence and wind speed evidence.

[0050] The evidence fusion module is used to fuse the corrected temperature evidence, humidity evidence, and wind speed evidence using the Dempster synthesis rule to obtain fused evidence;

[0051] The state determination module is used to construct target evidence corresponding to normal state, slightly abnormal state and severely abnormal state respectively, calculate the improved Jousselme distance between the fused evidence and the target evidence of normal state, slightly abnormal state and severely abnormal state, and take the state corresponding to the minimum distance as the temperature state determination result.

[0052] The beneficial effects of this invention are:

[0053] 1. This invention establishes a coupling relationship between temperature thresholds and environmental parameters through a dynamic threshold mechanism that corrects for humidity reduction and wind speed increase, solving the problem that traditional fixed thresholds cannot adapt to environmental interference. This mechanism significantly reduces the false alarm rate caused by environmental factors, particularly in real-world scenarios involving false alarms due to condensation in high-humidity environments and missed alarms due to heat dissipation in high-wind environments.

[0054] 2. This invention improves the Jousselme distance by employing a block-based Jaccard coefficient matrix. By distinguishing the similarity differences within a single subset, between a single subset and multiple subsets, within multiple subsets, and between the entire set, it accurately quantifies the degree of conflict between pieces of evidence. Compared to the traditional Jousselme distance, this method effectively solves the problem of failing to distinguish between different levels of conflict, making the calculation of evidence credibility weights more accurate.

[0055] 3. This invention combines environmentally corrected dynamic threshold evidence construction with weighted fusion of improved distance metrics, achieving end-to-end optimization from evidence source to fusion decision. This innovative combination fully leverages the complementary advantages of multi-source data, significantly improving the accuracy and robustness of temperature anomaly detection in complex transmission line inspection scenarios with challenging environmental interference. Attached Figure Description

[0056] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0057] Figure 1 This is a flowchart illustrating the method for detecting temperature anomalies in composite insulators based on evidence distance according to the present invention.

[0058] Figure 2 This is a schematic diagram of the composite insulator temperature anomaly data detection system based on evidence distance according to the present invention;

[0059] Figure 3 This is a schematic diagram of an unmanned aerial vehicle (UAV) inspection scenario according to the present invention. Detailed Implementation

[0060] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0061] Example 1: As Figure 1 As shown, the method for detecting temperature anomalies in composite insulators based on evidence distance includes:

[0062] S100: Collects infrared temperature data, on-site humidity data, and wind speed data of composite insulators, and preprocesses the collected data;

[0063] Furthermore, the preprocessing of the collected data includes:

[0064] use The criteria remove outliers from infrared temperature data, on-site humidity data, and wind speed data that exceed the mean plus or minus three standard deviations.

[0065] The infrared temperature data, on-site humidity data, and wind speed data after removing outliers were smoothed using the moving average method.

[0066] Specifically, non-contact temperature measurement of the composite insulator is performed using an infrared thermal imager to collect infrared temperature data. The measurement targets include key components of the composite insulator and the ambient temperature, with key components including areas prone to temperature anomalies such as the skirts, core rods, and hardware connections. Simultaneously, a humidity sensor collects real-time relative humidity data, and an anemometer measures the wind speed.

[0067] After obtaining the raw monitoring data, preprocessing is required to improve data quality. Preprocessing begins with outlier removal. This involves using... The criteria perform outlier detection on infrared temperature data, on-site humidity data, and wind speed data separately. For a given data category, the mean of that dataset is calculated. and standard deviation Determine whether each data point falls within the range of 0. Within the specified interval. If a data point exceeds the range of this interval, it is identified as an outlier and removed. The criteria are based on the statistical properties of the normal distribution. Under the normal distribution, the probability of a value falling within the range of the mean plus or minus three standard deviations is 99.73%. Data points outside this range are considered to be low-probability events, which are likely caused by factors such as measurement error, equipment failure, or sudden environmental changes.

[0068] After outlier removal, the data is smoothed using a moving average method to eliminate random fluctuations and noise. A fixed-length sliding window is set and moved sequentially across the data sequence. The arithmetic mean of the data points within the window is calculated at each step and used as the smoothed value at the window's center. Let the sliding window length be... For time The smoothed values ​​of the data points are:

[0069] ;

[0070] in, This is the original data. For the smoothed data, Odd numbers are usually chosen to ensure window center symmetry. Sliding window length The choice requires a trade-off between smoothing effect and data response speed. A larger value results in better smoothing but reduces the temporal resolution of the data. Smaller values ​​result in faster data response but limited smoothing effects. A suitable window length should be selected based on the data acquisition frequency and on-site conditions, with an optimal range of 5 to 11. Preprocessed data exhibits higher reliability and stability, providing a high-quality data foundation for subsequent evidence construction and state identification.

[0071] S200: Construct a framework for identifying the temperature state of composite insulators, including normal state, slightly abnormal state and severely abnormal state;

[0072] Specifically, an identification framework for the temperature states of composite insulators is constructed based on Dempster-Shafer evidence theory. In Dempster-Shafer evidence theory, the identification framework is the set of all possible states, using... Based on the actual conditions of abnormal temperature in composite insulators, the temperature state is divided into three mutually exclusive states: normal state, slightly abnormal state, and severely abnormal state.

[0073] Define the identification framework as ,in Indicates a normal state. Indicates a slightly abnormal condition. The following conditions indicate a severe abnormal state: A normal state indicates the composite insulator temperature is within a reasonable range with no abnormal heating. A slight abnormal state indicates the composite insulator is slightly overheating, possibly caused by poor contact, minor contamination, etc., requiring attention but not yet reaching a dangerous level. A severe abnormal state indicates the composite insulator is significantly overheating, possibly caused by internal defects, severe contamination, or electrical faults, posing a safety hazard and requiring immediate attention.

[0074] Identification Frame power set Including all possible combinations of states, there are a total of Elements. In DS evidence theory, elements in a power set whose basic probability assignment is non-zero are called focal elements. A subset such as... Represents complete determination of a specific state, multiple subsets such as Indicates the state is and There is uncertainty between them, the complete series This indicates complete uncertainty.

[0075] This identification framework provides a unified theoretical basis for subsequent evidence construction, evidence fusion, and state determination, enabling multi-source information from different sensors to be expressed and processed within a unified framework.

[0076] S300: Construct temperature evidence based on preprocessed infrared temperature data, adjust the temperature difference threshold downward based on on-site humidity data and construct humidity evidence, and adjust the temperature difference threshold upward based on wind speed data and construct wind speed evidence.

[0077] Furthermore, the construction of temperature evidence based on preprocessed infrared temperature data includes:

[0078] The temperature difference is obtained by calculating the difference between the average temperature of the key parts of the composite insulator and the ambient temperature.

[0079] Set a first temperature difference threshold and a second temperature difference threshold, wherein the first temperature difference threshold is less than the second temperature difference threshold;

[0080] When the temperature difference is less than the first temperature difference threshold, the basic probability value of the normal state is greater than the basic probability value of the slightly abnormal state, the basic probability value of the slightly abnormal state is greater than the basic probability value of the severe abnormal state, and a certain basic probability value is assigned to the entire set.

[0081] When the temperature difference is greater than or equal to the first temperature difference threshold and less than the second temperature difference threshold, the basic probability value of the slight abnormal state is greater than the basic probability value of the normal state and the severe abnormal state, and a certain basic probability value is assigned to the entire set.

[0082] When the temperature difference is greater than or equal to the second temperature difference threshold, the basic probability value of the severe abnormal state is greater than the basic probability value of the slight abnormal state, the basic probability value of the slight abnormal state is greater than the basic probability value of the normal state, and a certain basic probability value is assigned to the entire set.

[0083] The basic probability values ​​satisfy the normalization condition and constitute temperature evidence.

[0084] Specifically, the temperature difference between the average temperature of key parts of the composite insulator and the ambient temperature is calculated to obtain the temperature difference. Temperature difference reflects the temperature rise of the composite insulator relative to the environment and is a direct indicator of whether abnormal heating exists. Based on temperature difference... The relationship between the magnitude of the temperature difference and the preset temperature threshold is normal. Minor abnormal conditions and severe abnormal conditions Assign basic probability values.

[0085] Set two temperature difference thresholds and ,in .when At that time, the temperature rise is small, and the composite insulator is likely to be in normal condition. Assign a higher base probability value ,for and Assign a lower base probability value and At the same time, it is the complete series. Assign a certain basic probability value To indicate uncertainty in the judgment. When At that time, the temperature rise was moderate, but there might be slight anomalies. Assign a higher base probability value ,for and Assign a lower basic probability value and for the entire set Assign a certain basic probability value. When If the temperature rises excessively, it may indicate a serious anomaly. Assign a higher base probability value ,for and Assign a lower basic probability value and for the entire set Assign a certain basic probability value. The sum of the basic probability values ​​of all focal elements satisfies the normalization condition, i.e. This constitutes evidence of temperature. .

[0086] Furthermore, the step of adjusting the temperature difference threshold based on on-site humidity data and constructing humidity evidence includes:

[0087] When the on-site humidity data is higher than the humidity threshold, the temperature difference threshold will be lowered by a preset ratio, and the basic probability values ​​for minor and severe abnormal states will be increased.

[0088] When the on-site humidity data is not higher than the humidity threshold, the temperature difference threshold remains unchanged;

[0089] Based on the revised temperature difference threshold, basic probability values ​​are assigned to normal, slightly abnormal, and severely abnormal states to form humidity evidence.

[0090] Specifically, humidity has a significant impact on the temperature characteristics of composite insulators. In high humidity environments, a water film easily forms on the insulator surface, generating leakage current and causing heat generation under the influence of an electric field. Therefore, the temperature difference judgment standard needs to be lowered under high humidity conditions. When the on-site humidity data exceeds the humidity threshold... At that time, the temperature difference threshold will be set. and Adjust the preset ratio respectively The corrected temperature difference threshold is obtained. and , The preferred range is 5% to 15%. When allocating basic probability values, this equates to an increase in minor anomalous states compared to temperature evidence. and severe abnormal conditions The basic probability value reduces the normal state. The basic probability value is used to reflect the increased risk of anomalies in high humidity environments. When the on-site humidity data is not higher than the humidity threshold, the temperature difference threshold remains unchanged, and the basic probability value allocation method is the same as that for temperature evidence. Based on the corrected temperature difference threshold and temperature difference... The size relationship is and the complete series Assign basic probability values ​​that satisfy the normalization condition. This constitutes evidence of humidity. .in, These represent the basic probability assignments of humidity evidence for normal, slightly abnormal, and severely abnormal states, respectively. This represents the assignment of basic probabilities to the entire set.

[0091] Furthermore, the step of adjusting the temperature difference threshold upwards based on wind speed data and constructing wind speed evidence includes:

[0092] When the wind speed data is higher than the wind speed threshold, the temperature difference threshold will be increased by a preset ratio, and the basic probability value of normal state will be reduced.

[0093] When the wind speed data is not higher than the wind speed critical value, the temperature difference threshold remains unchanged;

[0094] Based on the revised temperature difference threshold, basic probability values ​​are assigned to normal, slightly abnormal, and severely abnormal states to form wind speed evidence.

[0095] Specifically, wind speed affects the heat dissipation of composite insulators. In high winds, convective heat transfer is enhanced, and even if internal heating exists, the surface temperature may not rise significantly due to good heat dissipation. Therefore, the temperature difference judgment standard needs to be raised under high wind speed conditions. When the wind speed data exceeds the critical wind speed value... At that time, the temperature difference threshold will be set. and Increase the preset ratio respectively The corrected temperature difference threshold is obtained. and . The preferred range is 10% to 20%. When allocating basic probability values, this represents a decrease in the normal state compared to temperature evidence. The basic probability value improves the severity of minor abnormal states. and severe abnormal conditions The basic probability value is used to reflect situations where anomalies may be masked under high wind speed conditions. When the wind speed data is not higher than the wind speed threshold, the temperature difference threshold is kept unchanged, and the basic probability value allocation method is the same as that for temperature evidence. Based on the corrected temperature difference threshold and temperature difference... The size relationship is and the complete series Assign basic probability values ​​that satisfy the normalization condition. This constitutes evidence of wind speed. .

[0096] Through the above process, adaptive consideration of environmental factors is achieved, making the evidence construction more consistent with the actual operating conditions of composite insulators and improving the accuracy and reliability of anomaly detection.

[0097] S400: Based on the Jaccard coefficient matrix of block processing, calculate the improved Jousselme distance between temperature evidence, humidity evidence and wind speed evidence, calculate the credibility weights of temperature evidence, humidity evidence and wind speed evidence according to the improved Jousselme distance, and use the credibility weights to perform weighted correction on temperature evidence, humidity evidence and wind speed evidence.

[0098] Furthermore, the block processing includes:

[0099] Based on the single-subset and multi-subset relationships of states in the identification framework, the Jaccard coefficient matrix is ​​divided into single-subset internal similarity blocks, single-subset and multi-subset similarity blocks, multi-subset internal similarity blocks, and whole-set similarity blocks.

[0100] Furthermore, the improved Jousselme distance calculation based on the Jaccard coefficient matrix of block processing, which calculates the temperature evidence, humidity evidence, and wind speed evidence, includes:

[0101] For any two pieces of evidence, obtain the basic probability values ​​of each state and construct the corresponding basic probability assignment vector.

[0102] Calculate the difference vector between two basic probability assignment vectors;

[0103] The difference vector is subjected to a quadratic operation with the block-processed Jaccard coefficient matrix to obtain the improved Jousselme distance between the two pieces of evidence.

[0104] The matrix elements of the similarity blocks within a single subset, the similarity blocks between a single subset and multiple subsets, the similarity blocks within multiple subsets, and the similarity blocks of the entire set are determined according to the intersection and union relationships between focal elements.

[0105] Specifically, the improved Jousselme distance between temperature, humidity, and wind speed evidence is first calculated based on the block-processed Jaccard coefficient matrix. Traditional Jousselme distance uses the Jaccard coefficient matrix to measure the similarity between focal elements, but it does not fully consider the different properties of single subsets, multiple subsets, and the entire set within the identification framework. This invention performs block processing on the Jaccard coefficient matrix and calculates similarity based on the type differences of the focal elements.

[0106] Based on the subset and subset relationships of states in the identification framework, the Jaccard coefficient matrix is ​​divided into four blocks: subset intra-subset similarity block, subset-matrix similarity block, subset intra-subset similarity block, and global similarity block. The subset intra-subset similarity block corresponds to... The similarity between subsets is determined by the fact that subsets are independent and mutually exclusive. Elements in the similarity matrix take values ​​of 0 or 1; 1 indicates two identical focal elements, and 0 indicates two different focal elements. The subset and multi-subset similarity blocks correspond to the similarity between a single subset and a subset containing multiple states. The similarity is determined by the cardinality ratio of the intersection to the union of focal elements. ,in and These represent two focal elements. The cardinality of the set is represented. The similarity block within multiple subsets corresponds to the similarity between multiple subsets, calculated using the cardinality ratio of the intersection to the union. The similarity block for the entire set corresponds to the cardinality of the entire set. The similarity to other focal elements is calculated as follows: Since the union of the universal set and any focal element is the universal set itself, the similarity is... .

[0107] Constructing the Jaccard coefficient matrix through block processing Then, calculate the improved Jousselme distance between any two pieces of evidence. For temperature evidence... Evidence of humidity Obtain the basic probability values ​​for each state and construct the corresponding basic probability assignment vector. and The dimension of the basic probability assignment vector is equal to the number of subsets with non-zero basic probability assignments in the power set of the identification frame, with each element corresponding to the basic probability value of a subset. The difference vector between two basic probability assignment vectors is calculated. By performing a quadratic operation on the difference vector and the block-processed Jaccard coefficient matrix, the improved Jousselme distance between the temperature and humidity evidence is obtained:

[0108] ;

[0109] Similarly, the improved Jousselme distance between temperature evidence and wind speed evidence was calculated. And the improved Jousselme distance between humidity evidence and wind speed evidence. The improved Jousselme distance reflects the degree of difference between pieces of evidence; the smaller the distance, the more similar the two pieces of evidence are, and the larger the distance, the greater the difference between the two pieces of evidence.

[0110] Furthermore, the confidence weights for calculating temperature, humidity, and wind speed evidence based on the improved Jousselme distance include:

[0111] Similarity between evidence is calculated based on the improved Jousselme distance between temperature evidence, humidity evidence, and wind speed evidence.

[0112] Global support for temperature, humidity, and wind speed evidence is calculated based on the similarity between evidence.

[0113] The credibility weights of temperature evidence, humidity evidence, and wind speed evidence are obtained by normalizing the global support.

[0114] Specifically, the credibility weights of temperature, humidity, and wind speed evidence are calculated based on the improved Jousselme distance. The basic idea is that evidence with high similarity to other evidence has higher credibility, while evidence with significant differences from other evidence has lower credibility. First, the similarity between evidence is calculated using the improved Jousselme distance. For the evidence... and evidence Improved Jousselme distance between The similarity is defined as:

[0115] ;

[0116] in, This is a similarity adjustment parameter that controls the mapping relationship between distance and similarity. Using an exponential function allows the similarity to decay rapidly as distance increases. The value of affects the similarity decay rate, and the preferred range is 1 to 10.

[0117] The global support of each piece of evidence is calculated based on the similarity between the pieces of evidence. (Evidence) The global support of an evidence is defined as the sum of its similarity to all other pieces of evidence:

[0118] ;

[0119] in, The total number of pieces of evidence in this invention Global support reflects the degree to which a piece of evidence is supported by other evidence. The higher the support, the better the consistency between the evidence and other evidence, and the higher its credibility.

[0120] Normalize the global support to obtain the credibility weight of each piece of evidence:

[0121] ;

[0122] Credibility weight satisfies For temperature evidence, humidity evidence, and wind speed evidence, their credibility weights were calculated respectively. , and .

[0123] Finally, credibility weights were used to weight and adjust the temperature, humidity, and wind speed evidence. For the evidence... Basic probability assignment Its corrected basic probability assignment for:

[0124] ;

[0125] in, Let represent any focal element in the power set of the identification frame. The corrected basic probability assignments need to be renormalized to satisfy the requirements of the probability axioms:

[0126] ;

[0127] By using weighted adjustments, highly credible evidence plays a greater role in the fusion process, while the influence of less credible evidence is appropriately suppressed, thereby improving the reliability and accuracy of evidence fusion.

[0128] S500: The modified temperature evidence, humidity evidence, and wind speed evidence are fused using the Dempster synthesis rule to obtain fused evidence;

[0129] Furthermore, the fusion of the corrected temperature evidence, humidity evidence, and wind speed evidence using the Dempster synthesis rule includes:

[0130] Calculate the conflict coefficient between the corrected temperature evidence, humidity evidence, and wind speed evidence;

[0131] Based on the conflict coefficient, the intersection of the focal elements of the corrected temperature evidence, humidity evidence, and wind speed evidence is normalized to obtain fused evidence.

[0132] Specifically, the conflict coefficient among the three pieces of evidence after modification is first calculated. The conflict coefficient reflects the degree of contradiction between different pieces of evidence and is used to measure the feasibility of evidence fusion. For the modified temperature evidence... Humidity evidence Evidence of wind speed Its conflict coefficient Defined as:

[0133] ;

[0134] The summation iterates through all focal triples whose intersection is an empty set. , , , These are the focal elements of the three revised pieces of evidence. The conflict coefficient represents the sum of the products of the probabilities assigned to mutually exclusive states by the three pieces of evidence, reflecting the inconsistency between the evidence. The range of values ​​is , The closer to 0, the smaller the conflict between the pieces of evidence; the closer to 1, the greater the conflict.

[0135] The three pieces of evidence are fused based on the conflict coefficient. For any focal element in the power set of the identification framework... The basic probability assignment after fusion is performed on all sets satisfying the intersection. The summation of the probability products of the focal triples and normalization yields:

[0136] ;

[0137] For the empty set, the basic probability after merging is assigned a value of 0, that is... Normalization factor Ensure that the basic probability assignments after fusion satisfy the probability axioms, so that the sum of the basic probability values ​​of all focal elements is 1.

[0138] Fusion of evidence The assessment integrates information from three sources: temperature, humidity, and wind speed, preserving the valid information from each piece of evidence. At the same time, it suppresses conflicts and uncertainties through credibility weighting and Dempster's synthesis rules, providing a more reliable basis for subsequent state determination.

[0139] It should be noted that Dempster's rules of composition require that the coefficient of conflict between pieces of evidence be less than 1, i.e. When the conflict coefficient is close to 1, it indicates a serious conflict between pieces of evidence, and Dempster's composition rule may fail. This invention, through the aforementioned credibility weighting correction, can effectively reduce the degree of conflict between pieces of evidence, ensuring the applicability of Dempster's composition rule.

[0140] S600: Construct target evidence corresponding to normal state, slightly abnormal state and severely abnormal state respectively, calculate the improved Jousselme distance between the fused evidence and the target evidence of normal state, slightly abnormal state and severely abnormal state, and take the state corresponding to the minimum distance as the temperature state determination result.

[0141] Specifically, target evidence is first constructed corresponding to normal, slightly abnormal, and severely abnormal states, respectively. Target evidence is an ideal representation of evidence, used as a reference benchmark for judgment. For target evidence in the normal state... Normal state The basic probability value for is set to 1, and the basic probability values ​​for other focal elements are set to 0, i.e. This means that under ideal normal conditions, the system... It possesses complete certainty. Similarly, constructing evidence for targets with slight anomalies. ,make The remaining focal elements have a basic probability value of 0. Construct evidence of a severely anomalous target state. ,make The basic probability value of the remaining focal elements is 0.

[0142] Next, the improved Jousselme distance between the fused evidence and each target piece of evidence is calculated. The improved Jousselme distance calculation method based on block processing of the Jaccard coefficient matrix, established in S400, is used. For the fused evidence... Evidence of targets in normal state Construct their respective basic probability assignment vectors and Calculate the difference vector The improved Jousselme distance is obtained through quadratic form operations:

[0143] ;

[0144] in, This represents the Jaccard coefficient matrix after block processing constructed in S400. Similarly, the improved Jousselme distance between the fused evidence and the evidence for the slightly anomalous target is calculated. :

[0145] ;

[0146] in, Calculate the improved Jousselme distance between fused evidence and evidence of severe anomalies. :

[0147] ;

[0148] in, .

[0149] Finally, the state corresponding to the minimum distance is taken as the temperature state determination result. (Comparison) , and Find the minimum value among the given values.

[0150] ;

[0151] like If so, the composite insulator is determined to be in normal condition. If so, the composite insulator is determined to be in a slightly abnormal state, and it is recommended to strengthen monitoring or arrange maintenance. If the composite insulator is in a severely abnormal state, it is determined that immediate measures should be taken to address the issue.

[0152] By comparing the fused evidence with the ideal state and selecting the closest state as the judgment result, the above-mentioned evidence distance-based judgment method is used to accurately identify and classify the temperature anomalies of composite insulators, providing a reliable basis for the operation and maintenance decisions of power equipment.

[0153] Example 2: This example uses the composite insulator temperature anomaly data detection system based on evidence distance of the present invention to detect temperature anomalies in a composite insulator of a 110kV transmission line, in order to verify the effectiveness of the method of the present invention.

[0154] The detection system of the present invention is as follows Figure 2As shown, the system includes a data preprocessing module, a framework construction module, an evidence construction module, an evidence correction module, an evidence fusion module, and a state determination module. These modules work together to complete the entire detection process from data acquisition to state determination. It should be noted that the composite insulator temperature anomaly data detection system based on evidence distance provided in this embodiment two is used to execute all the process steps of the composite insulator temperature anomaly data detection method based on evidence distance in embodiment one described above.

[0155] Specifically, the data preprocessing module controls a drone equipped with an infrared thermal imager to fly along a preset cruise path along a 110kV transmission line, performing fixed-point infrared imaging of the target composite insulator, such as... Figure 3 As shown. Figure 3 This is a real-time image taken during a drone inspection. The yellow areas represent high-temperature zones, the cyan areas represent normal-temperature zones, and the green triangles and squares mark key monitoring points on the composite insulator automatically identified by the system. The red lines indicate the boundaries of the inspection area. The system extracted temperature data from five sampling points on the key insulator components (such as the edge of the skirt and the junction of the skirt and the core rod). At the same time, the ambient temperature at the site was collected. relative humidity Wind speed .

[0156] through After outlier removal and smoothing using the moving average method, the average temperature of the critical parts of the insulator was calculated. Temperature difference .

[0157] The framework building module establishes the identification framework. The evidence construction module sets a temperature difference threshold for 110kV insulators. .

[0158] Temperature evidence: because ,distribute .

[0159] Humidity evidence: because If the humidity threshold is exceeded, the system will trigger the first temperature difference threshold. Second temperature difference threshold The temperature difference thresholds were adjusted by 10% respectively to obtain the corrected threshold values. ;distribute .

[0160] Evidence of wind speed: due to ,distribute .

[0161] The evidence construction results show that the temperature and wind speed evidence tend to indicate a normal state, while the humidity evidence, after environmental correction, provides higher support for an abnormal state. The three pieces of evidence conflict and their credibility needs to be assessed through evidence distance evaluation.

[0162] The evidence correction module uses a block-based Jaccard coefficient matrix calculation to improve the Jousselme distance: the distance between temperature and humidity evidence. The distance between temperature evidence and wind speed evidence The distance between humidity evidence and wind speed evidence It accurately measures the degree of difference between pieces of evidence.

[0163] Calculate global support: Normalization yields the credibility weights , , Temperature evidence received the highest weight, consistent with its status as the most direct and reliable basis for judgment.

[0164] Using credibility weights The three pieces of evidence are weighted and adjusted to obtain the basic probability value of the temperature evidence for the normal state. The basic probability value of the corrected humidity evidence for the normal state. The revised wind speed evidence represents the basic probability value for the normal state. By using credibility-weighted correction, the impact of high-quality evidence is enhanced, while the impact of conflicting evidence is suppressed, thus improving detection accuracy.

[0165] The evidence fusion module uses Dempster's rules of composition to fuse three pieces of evidence and calculate the conflict coefficient. This indicates that contradictions among the evidence have been effectively controlled. After fusion, we obtain... .

[0166] The state determination module calculates the improved Jousselme distance between the fused evidence and each target piece of evidence: the distance between the fused evidence and the target evidence in the normal state. Distance between fused evidence and evidence of minor anomalies in the target state The distance between fused evidence and evidence of severely anomalous targets Take the minimum distance. The temperature of the insulator was determined to be normal.

[0167] This embodiment demonstrates the technical advantages of the present invention: the environmental correction mechanism reflects high humidity ( The impact of anomaly risks is mitigated to avoid underreporting; credibility assessment based on evidence distance identifies evidence discrepancies, reasonably balances the contribution of each piece of evidence, and avoids false alarms; the improved Jousselme distance accurately quantifies the degree of conflict, and the Dempster synthesis rule effectively integrates multi-source information.

[0168] Compared with traditional methods that rely solely on weighted averages or infrared temperature, this invention achieves more accurate and reliable detection in complex environments, effectively reducing false alarms and missed alarms, and improving the intelligence level of condition monitoring for composite insulators in power transmission lines.

[0169] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting temperature anomalies in composite insulators based on evidence distance, characterized in that, include: S100: Collects infrared temperature data, on-site humidity data, and wind speed data of composite insulators, and preprocesses the collected data; S200: Construct a framework for identifying the temperature state of composite insulators, including normal state, slightly abnormal state, and severely abnormal state; S300: Construct temperature evidence based on preprocessed infrared temperature data, adjust the temperature difference threshold downward based on on-site humidity data and construct humidity evidence, and adjust the temperature difference threshold upward based on wind speed data and construct wind speed evidence. S400: Based on the Jaccard coefficient matrix of block processing, calculate the improved Jousselme distance between temperature evidence, humidity evidence and wind speed evidence, calculate the credibility weights of temperature evidence, humidity evidence and wind speed evidence according to the improved Jousselme distance, and use the credibility weights to perform weighted correction on temperature evidence, humidity evidence and wind speed evidence. S500: The modified temperature evidence, humidity evidence, and wind speed evidence are fused using the Dempster synthesis rule to obtain fused evidence; S600: Construct target evidence corresponding to normal state, slightly abnormal state and severe abnormal state respectively, calculate the improved Jousselme distance between the fused evidence and the target evidence of normal state, slightly abnormal state and severe abnormal state, and take the state corresponding to the minimum distance as the temperature state determination result. The step of adjusting the temperature difference threshold based on on-site humidity data and constructing humidity evidence includes: When the on-site humidity data is higher than the humidity threshold, the temperature difference threshold will be lowered by a preset ratio, and the basic probability values ​​for minor and severe abnormal states will be increased. When the on-site humidity data is not higher than the humidity threshold, the temperature difference threshold remains unchanged; Based on the corrected temperature difference threshold, basic probability values ​​are assigned to normal, slightly abnormal, and severely abnormal states to form humidity evidence. The step of adjusting the temperature difference threshold upwards based on wind speed data and constructing wind speed evidence includes: When the wind speed data is higher than the wind speed threshold, the temperature difference threshold will be increased by a preset ratio, and the basic probability value of normal state will be reduced. When the wind speed data is not higher than the wind speed critical value, the temperature difference threshold remains unchanged; Based on the revised temperature difference threshold, basic probability values ​​are assigned to normal, slightly abnormal, and severely abnormal states to form wind speed evidence.

2. The method for detecting temperature anomalies in composite insulators based on evidence distance according to claim 1, characterized in that, The preprocessing of the collected data includes: use The criteria remove outliers from infrared temperature data, on-site humidity data, and wind speed data that exceed the mean plus or minus three standard deviations. The infrared temperature data, on-site humidity data, and wind speed data after removing outliers were smoothed using the moving average method.

3. The method for detecting temperature anomalies in composite insulators based on evidence distance according to claim 1, characterized in that, The construction of temperature evidence based on preprocessed infrared temperature data includes: The temperature difference is obtained by calculating the difference between the average temperature of the key parts of the composite insulator and the ambient temperature. Set a first temperature difference threshold and a second temperature difference threshold, wherein the first temperature difference threshold is less than the second temperature difference threshold; When the temperature difference is less than the first temperature difference threshold, the basic probability value of the normal state is greater than the basic probability value of the slightly abnormal state, the basic probability value of the slightly abnormal state is greater than the basic probability value of the severe abnormal state, and a certain basic probability value is assigned to the entire set. When the temperature difference is greater than or equal to the first temperature difference threshold and less than the second temperature difference threshold, the basic probability value of the slight abnormal state is greater than the basic probability value of the normal state and the severe abnormal state, and a certain basic probability value is assigned to the entire set. When the temperature difference is greater than or equal to the second temperature difference threshold, the basic probability value of the severe abnormal state is greater than the basic probability value of the slight abnormal state, the basic probability value of the slight abnormal state is greater than the basic probability value of the normal state, and a certain basic probability value is assigned to the entire set. The basic probability values ​​satisfy the normalization condition and constitute temperature evidence.

4. The method for detecting temperature anomalies in composite insulators based on evidence distance according to claim 1, characterized in that, The block processing includes: Based on the single-subset and multi-subset relationships of states in the identification framework, the Jaccard coefficient matrix is ​​divided into single-subset internal similarity blocks, single-subset and multi-subset similarity blocks, multi-subset internal similarity blocks, and whole-set similarity blocks.

5. The method for detecting temperature anomalies in composite insulators based on evidence distance according to claim 4, characterized in that, The improved Jousselme distance calculation based on the block-processed Jaccard coefficient matrix for temperature, humidity, and wind speed evidence includes: For any two pieces of evidence, obtain the basic probability values ​​of each state and construct the corresponding basic probability assignment vector. Calculate the difference vector between two basic probability assignment vectors; The difference vector is subjected to a quadratic operation with the block-processed Jaccard coefficient matrix to obtain the improved Jousselme distance between the two pieces of evidence. The matrix elements of the similarity blocks within a single subset, the similarity blocks between a single subset and multiple subsets, the similarity blocks within multiple subsets, and the similarity blocks of the entire set are determined according to the intersection and union relationships between focal elements.

6. The method for detecting temperature anomalies in composite insulators based on evidence distance according to claim 1, characterized in that, The confidence weights for calculating temperature, humidity, and wind speed evidence based on the improved Jousselme distance include: Similarity between evidence is calculated based on the improved Jousselme distance between temperature evidence, humidity evidence, and wind speed evidence. Global support for temperature, humidity, and wind speed evidence is calculated based on the similarity between evidence. The credibility weights of temperature evidence, humidity evidence, and wind speed evidence are obtained by normalizing the global support.

7. The method for detecting temperature anomalies in composite insulators based on evidence distance according to claim 1, characterized in that, The process of fusing the corrected temperature, humidity, and wind speed evidence using the Dempster synthesis rule includes: Calculate the conflict coefficient between the corrected temperature evidence, humidity evidence, and wind speed evidence; Based on the conflict coefficient, the intersection of the focal elements of the corrected temperature evidence, humidity evidence, and wind speed evidence is normalized to obtain fused evidence.

8. A composite insulator temperature anomaly data detection system based on evidence distance, characterized in that, The system is used to execute the composite insulator temperature anomaly data detection method based on evidence distance according to any one of claims 1-7, the system comprising: The data preprocessing module is used to collect infrared temperature data, on-site humidity data, and wind speed data of composite insulators, and to preprocess the collected data. The framework construction module is used to build an identification framework for the temperature state of composite insulators, including normal state, slightly abnormal state and severely abnormal state; The evidence construction module is used to construct temperature evidence based on preprocessed infrared temperature data, adjust the temperature difference threshold downward based on the on-site humidity data and construct humidity evidence, and adjust the temperature difference threshold upward based on the wind speed data and construct wind speed evidence. The evidence correction module is used to calculate the improved Jousselme distance between temperature evidence, humidity evidence and wind speed evidence based on the Jaccard coefficient matrix processed in blocks, calculate the credibility weights of temperature evidence, humidity evidence and wind speed evidence based on the improved Jousselme distance, and use the credibility weights to perform weighted correction on temperature evidence, humidity evidence and wind speed evidence. The evidence fusion module is used to fuse the corrected temperature evidence, humidity evidence, and wind speed evidence using the Dempster synthesis rule to obtain fused evidence; The state determination module is used to construct target evidence corresponding to normal state, slightly abnormal state and severely abnormal state respectively, calculate the improved Jousselme distance between the fused evidence and the target evidence of normal state, slightly abnormal state and severely abnormal state, and take the state corresponding to the minimum distance as the temperature state determination result.

Images