Insulator temperature monitoring method and system
By constructing a random tree and calculating the numerical and density similarity, selecting suitable random trees for abnormal detection, the problem of the impact of historical data quality in insulator state monitoring is solved, and the accurate monitoring and detection accuracy of insulator state is achieved.
Patent Information
- Application Number
- CN202411262414.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-10
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-09-10
AI Technical Summary
When monitoring the insulator status of ultra-high voltage DC transmission lines, the prior art relies on the quality and accuracy of historical data and is susceptible to missing values or abnormal data, resulting in unreliable identification results and complex model maintenance, affecting monitoring accuracy.
By constructing multiple random trees, numerical similarity and density similarity between the current temperature data and the nodes, filter random trees that meet preset conditions for abnormal detection, monitor the insulator status in real time, and improve detection accuracy.
Accurate monitoring of insulator states is achieved, the accuracy of abnormal detection is improved, the timely maintenance of insulator states is ensured, and the complexity of model maintenance is reduced.
Smart Images

Figure CN118779814B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data processing technology. More specifically, the present invention relates to an insulator temperature monitoring method and system. Background Art
[0002] Based on the advantages of UHV DC transmission, such as large transmission capacity, long transmission distance, and small transmission loss, the special voltage level and harsh working environment will greatly increase the probability of failure of insulators in UHV DC transmission lines. Preventing insulator contamination flashover accidents has also become an important research topic for maintaining the safe and stable operation of power systems. Insulator leakage current is affected by various environmental factors, especially temperature and humidity. Therefore, the status of the insulator can be monitored by monitoring the temperature during the operation of the insulator.
[0003] In the related art, for example, a patent application document with publication number CN114896831A discloses an insulator state identification method and system. The method includes: obtaining the effective value of leakage current, corona current probability, leakage current peak value and leakage current pulse frequency according to the historical operation data of the insulator, inputting them into the identification model to output the coding value, and then obtaining the corresponding insulator state identification result according to different permutations and combinations of the coding values.
[0004] However, the recognition result of the insulator status by the above scheme depends on the quality and accuracy of historical data. If there are missing values or abnormal data in the historical data, the recognition result of the insulator status may be unreliable. The set recognition model needs to design and maintain complex output libraries and encoding rules, which will also affect the recognition result of the insulator status to a certain extent, and may not accurately monitor the status of the insulator. Summary of the invention
[0005] In order to solve the problem that the state of an insulator cannot be accurately monitored, the present invention provides an insulator temperature monitoring method and system.
[0006] According to a first aspect of an embodiment of the present invention, there is provided a method for monitoring temperature of an insulator, comprising:
[0007] Obtain the temperature data sequence of the insulator before the current moment, and construct multiple random trees of the temperature data sequence, with the number of nodes in each random tree being the same;
[0008] Determine the numerical difference and change rate difference between the current temperature data and any node in any random tree, calculate the sum of the squares of the numerical difference and the change rate difference, and take the reciprocal of the arithmetic square root of the sum of squares as the numerical similarity between the current temperature data and any node;
[0009] Calculate the density similarity between the current temperature data and any node. The density similarity reflects the difference between the data distribution density in a preset local area of the current temperature data and the node distribution density of the random tree where any node is located.
[0010] The final similarity between the current temperature data and any node is calculated. The final similarity is positively correlated with the numerical similarity and the density similarity. The final similarity between the current temperature data and each node in each random tree is used to screen the random trees that meet the preset conditions, so as to perform anomaly detection on the current temperature data according to the screened random trees, and monitor the temperature of the insulator according to the detection results.
[0011] The present invention constructs a random tree and inserts the current temperature data into the selected random tree for abnormality detection, so that the state of the insulator can be monitored in real time, so that the insulator can be maintained in time; and when the abnormality detection is performed on the current temperature data, the constructed random tree is screened by using the numerical similarity and density similarity between the current temperature data and any node in any random tree, so that each node in the selected random tree has a similar numerical value and spatial distribution characteristics to the current temperature data, so that the selected random tree performs abnormality detection on the current temperature data, which can improve the accuracy of the abnormality detection on the current temperature data, and further accurately monitor the state of the insulator.
[0012] In an exemplary embodiment of the present invention, a method for obtaining a change rate difference includes:
[0013] Taking the difference between the current temperature data and the previous temperature data as the first difference, calculating the ratio of the first difference to the current temperature data, and obtaining a first change rate;
[0014] Taking the difference between any node and the parent node of any node as the second difference, calculating the ratio of the second difference to any node, and obtaining the second change rate;
[0015] The difference between the first change rate and the second change rate is calculated to obtain a change rate difference.
[0016] The present invention can determine the difference between the current temperature data and the dynamic change characteristics of any node, so as to accurately capture the trend change between the data.
[0017] In an exemplary embodiment of the present invention, a method for obtaining data distribution density includes:
[0018] The sequence composed of the current temperature data and the previous temperature data, whose length is the same as the number of nodes in any random tree, is taken as the local area of the current temperature data;
[0019] The temperature data in the local area are clustered to obtain a cluster center, and the mean of the distance between the temperature data in the local area and the cluster center is calculated to obtain the data distribution density.
[0020] The present invention determines the data distribution density by clustering, can identify the distribution pattern and structure of the data itself, and ensures the accuracy of the determined data distribution density.
[0021] In an exemplary embodiment of the present invention, a method for obtaining node distribution density includes:
[0022] Calculate the average value of the difference between all nodes except any node in the random tree where any node is located and any node, and use the average value as the node distribution density of the random tree where any node is located.
[0023] In an exemplary embodiment of the present invention, the density similarity between the current temperature data and any node satisfies the following relationship:
[0024] ;
[0025] In the formula, The current temperature data and The random tree Density similarity of nodes; is the data distribution density in the local area of the current temperature data; The node is located at The node distribution density of a random tree; The natural constant An exponential function with base .
[0026] The calculation formula of density similarity provided by the present invention can limit the range of density similarity to 0-1, thereby reducing the calculation difficulty when determining the final similarity.
[0027] In an exemplary embodiment of the present invention, the final similarity between the current temperature data and any node satisfies the following relationship:
[0028] ;
[0029] In the formula, The current temperature data and The random tree The final similarity of nodes; The current temperature data and The random tree The numerical similarity of nodes; The current temperature data and The random tree Density similarity of nodes; is the normalization function.
[0030] In an exemplary embodiment of the present invention, the final similarity between the current temperature data and each node in each random tree is used to screen the random trees that meet the preset conditions, including:
[0031] Calculate the average value of the final similarity between the current temperature data and each node in any random tree, and use the node in any random tree whose final similarity with the current temperature data is greater than the average value as the reference node;
[0032] Count the number of reference nodes in each random tree, sort the random trees from large to small according to the number of corresponding reference nodes, and select a preset number of random trees from front to back as the target random tree.
[0033] The target random tree selected by the present invention can accurately detect anomalies in current temperature data.
[0034] In an exemplary embodiment of the present invention, performing anomaly detection on current temperature data according to the selected random tree includes:
[0035] Calculate the complexity change of any target random tree before and after inserting the current temperature data, and obtain the abnormal score of the current temperature data in any target random tree;
[0036] The average of the anomaly scores of the current temperature data in all target random trees is calculated to obtain the collusion anomaly score of the current temperature data. When the collusion anomaly score is greater than the preset threshold, the current temperature data is determined to be abnormal data.
[0037] According to a second aspect of an embodiment of the present invention, an insulator temperature monitoring system is provided. The insulator temperature monitoring system includes a memory and a processor. A computer program is stored in the memory. The processor executes the computer program to implement the steps of the first aspect of the embodiment of the present invention.
[0038] The present invention has the following effects:
[0039] When calculating the numerical similarity, the present invention comprehensively considers the similarity of static values of data and the similarity of dynamic change characteristics of data, thereby ensuring the accuracy of the determined numerical similarity; and when performing abnormality detection on the current temperature data, the constructed random tree is screened in combination with the final similarity determined by the numerical similarity and the density similarity, so that each node in the screened random tree has a similar numerical value, similar change characteristics and similar spatial distribution characteristics with the current temperature data, so that the current temperature data is detected for abnormality based on the screened random tree, the accuracy of abnormality detection on the current temperature data can be improved, and the state of the insulator can be accurately monitored. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:
[0041] Figure 1 The present invention is a schematic diagram of the steps of an insulator temperature monitoring method according to an embodiment of the present invention. DETAILED DESCRIPTION
[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0043] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.
[0044] Reference Figure 1 , an insulator temperature monitoring method, comprising steps S1 to S4, specifically as follows:
[0045] S1: Obtain the temperature data sequence of the insulator before the current moment, and construct multiple random trees of the temperature data sequence, with the same number of nodes in each random tree.
[0046] Specifically, a temperature sensor can be used to collect temperature data of the insulator during use to obtain a temperature data sequence consisting of temperature data before the current moment, and the frequency of collecting temperature data can be set to 1HZ. This embodiment does not specifically limit the collection frequency.
[0047] A random tree refers to a binary tree constructed based on a subset randomly selected from a temperature data sequence. For example, multiple subsets, such as 100, containing a certain number, such as 30 temperature data, can be randomly selected from the temperature data sequence, and then the Robust Random Cut Forest (RRCF) algorithm is used to construct a binary tree corresponding to each subset to obtain multiple random trees of the temperature data sequence. This embodiment does not specifically limit the number of random trees constructed and the number of nodes contained in any random tree. It should be noted that the process of constructing multiple random trees based on the algorithm is a prior art and is not described in detail in this embodiment.
[0048] S2: Determine the numerical difference and change rate difference between the current temperature data and any node in any random tree, calculate the sum of the squares of the numerical difference and the change rate difference, and use the reciprocal of the arithmetic square root of the sum of squares as the numerical similarity between the current temperature data and any node.
[0049] It should be noted that since the temperature change of insulators is often dynamic and nonlinear, that is, the collected temperature data may fluctuate significantly with time, environmental conditions such as wind speed, humidity, solar radiation, etc., and changes in equipment status, therefore, the comprehensive numerical difference and change rate difference can more accurately capture the trend changes between the data.
[0050] In an exemplary embodiment of the present invention, the determination of the difference between the current temperature data and the rate of change of any node can be achieved by the following steps:
[0051] The difference between the current temperature data and the previous temperature data is taken as the first difference, and the ratio of the first difference to the current temperature data is calculated to obtain the first change rate; the difference between any node and the parent node of any node is taken as the second difference, and the ratio of the second difference to any node is calculated to obtain the second change rate; the difference between the first change rate and the second change rate is calculated to obtain the change rate difference.
[0052] The current temperature data refers to the temperature data collected at the current moment, the previous temperature data refers to the temperature data collected at the previous moment, and the time interval between the current temperature data and the previous temperature data is the time interval corresponding to the collection frequency.
[0053] The first rate of change , is the current temperature data, is the previous temperature data, is the absolute value sign; the second rate of change , For the The random tree The value of a node, For the The random tree The value of the parent node of the node; the current temperature data and the The random tree The difference in the rate of change between nodes .
[0054] Furthermore, the current temperature data is The random tree The numerical similarity of the nodes , is the current temperature data, For the The random tree The value of a node, For the The random tree The value of the parent node of a node, is the absolute value symbol.
[0055] It should be noted that two temperature data with corresponding large numerical similarity are very likely to be processed by similar cutting rules. If the numerical similarity between the current temperature data and any node is large, the current temperature data will be inserted into the random tree where the node is located. Abnormal patterns similar to the current temperature data can be captured, thereby improving the accuracy of anomaly detection for the current temperature data.
[0056] S3: Calculate the density similarity between the current temperature data and any node, where the density similarity reflects the difference between the data distribution density in a preset local area of the current temperature data and the node distribution density of the random tree where any node is located.
[0057] Among them, density similarity refers to the data that measures the similarity between the data distribution of the area where the current temperature data is located and the data distribution of the area where any node is located. If the difference between the data distribution density in the local range of the current temperature data and the node distribution density of the random tree where any node is located is smaller, it means that the current temperature data is more similar to the node in space, and the corresponding density similarity is greater.
[0058] In an exemplary embodiment of the present invention, the determination of the data distribution density within a local range of the current temperature data can be achieved by the following steps:
[0059] The sequence consisting of the current temperature data and the previous temperature data, whose length is the same as the number of nodes in any random tree, is taken as the local area of the current temperature data; the temperature data in the local area are clustered to obtain a cluster center, and the mean of the distance between each temperature data in the local area and the cluster center is calculated to obtain the data distribution density.
[0060] Exemplarily, when the number of nodes in any random tree is 30, the local area of the current temperature data is a data segment with a length of 30 consisting of the current temperature data and the previous 29 temperature data.
[0061] Optionally, the k-means clustering algorithm can be used to cluster the temperature data in the local range, and the k value is set to 1, so that a cluster center can be obtained. Of course, other clustering algorithms, such as the DBSCAN clustering algorithm, can also be used. This embodiment does not specifically limit the selected clustering algorithm.
[0062] Optionally, the mean of the Euclidean distances between each temperature data in the local area and the cluster center can be calculated, or the mean of the Manhattan distances between each temperature data in the local area and the cluster center can be calculated to determine the data distribution density. This embodiment does not specifically limit the type of distance selected when determining the data distribution density.
[0063] In an exemplary embodiment of the present invention, the node distribution density of the random tree where any node is located can be determined by the following steps:
[0064] Calculate the average value of the differences between all nodes except any node in the random tree where any node is located and any node, and use the average value as the node distribution density of the random tree where any node is located.
[0065] Specifically, the node distribution density of the random tree where any node is located , For the In the random trees, except The node outside The value of a node, For the The random tree The value of a node, is the absolute value symbol, For the The number of nodes in a random tree, is the summation symbol.
[0066] Furthermore, after determining the data distribution density in the local area of the current temperature data and the node distribution density of the random tree where any node is located, the density similarity between the current temperature data and the node can be calculated. Specifically, the density similarity between the current temperature data and any node satisfies the following relationship:
[0067] ;
[0068] In the formula, The current temperature data and The random tree Density similarity of nodes; is the data distribution density in the local area of the current temperature data; The node is located at The node distribution density of a random tree; The natural constant An exponential function with base .
[0069] It should be noted that the density similarity reflects the similarity of the spatial distribution of the two temperature data. If the density similarity between the current temperature data and any node is large, the current temperature data will be inserted into the random tree where the node is located, which can reflect the distribution characteristics of the current temperature data, thereby improving the accuracy of anomaly detection for the current temperature data to a certain extent.
[0070] S4: Calculate the final similarity between the current temperature data and any node. The final similarity is positively correlated with the numerical similarity and the density similarity. Use the final similarity between the current temperature data and each node in each random tree to screen the random trees that meet the preset conditions, so as to perform anomaly detection on the current temperature data according to the screened random trees, and monitor the temperature of the insulator according to the detection results.
[0071] It should be noted that in the RRCF algorithm, each binary tree is constructed by randomly cutting data. The structures of different binary trees often have different performances for the same temperature data. Inserting the current temperature data into the random tree where the node with the highest final similarity to the current temperature data is located can better reflect the characteristics of the current temperature data, thereby more accurately detecting anomalies in the current temperature data.
[0072] Specifically, the final similarity between the current temperature data and any node satisfies the following relationship:
[0073] ;
[0074] In the formula, The current temperature data and The random tree The final similarity of nodes; The current temperature data and The random tree The numerical similarity of nodes; The current temperature data and The random tree Density similarity of nodes; is the normalization function.
[0075] In another embodiment, the final similarity between the current temperature data and any node may also be calculated by other formulas, such as by a product formula instead of a summation formula.
[0076] In an exemplary embodiment of the present invention, the selected random tree can be determined by the following steps:
[0077] Calculate the average value of the final similarity between the current temperature data and each node in any random tree, and use the nodes in any random tree whose final similarity with the current temperature data is greater than the average value as reference nodes; count the number of reference nodes in each random tree, sort the random trees from large to small according to the number of corresponding reference nodes, and select a preset number of random trees from front to back as the target random tree.
[0078] Optionally, when the number of constructed random trees is 100, the random trees can be sorted from large to small according to the number of corresponding reference nodes, and then a certain number, such as the top 5%, that is, the top 5 random trees, are selected as target binary trees, so as to use the screened target binary trees to perform abnormality detection on the current temperature data. Of course, an appropriate number of random trees can also be selected as target random trees according to specific circumstances. This embodiment does not specifically limit the number of target random trees.
[0079] In an exemplary embodiment of the present invention, abnormality detection of current temperature data can be achieved by following the steps below:
[0080] (1) Calculate the complexity change of any target random tree before and after inserting the current temperature data, and obtain the anomaly score of the current temperature data in any target random tree;
[0081] Specifically, the calculation process of the complexity change of any target random tree before and after the current temperature data is inserted is as follows: first, the difference in path length before and after the current temperature data is inserted into any node in any target random tree is calculated; then, the difference in path length before and after the current temperature data is inserted into all nodes in the target random tree is calculated, and the difference is summed to obtain the complexity change of the target random tree before and after the current temperature data is inserted; and then the complexity change of all target random trees before and after the current temperature data is determined.
[0082] Furthermore, the complexity change of each target random tree before and after the current temperature data is inserted may be used as the abnormality score of the current temperature data in the corresponding target random tree.
[0083] (2) Calculate the average of the anomaly scores of the current temperature data in all target random trees to obtain the collusion anomaly score of the current temperature data. When the collusion anomaly score is greater than the preset threshold, the current temperature data is judged to be anomaly data.
[0084] Optionally, the 95% quantile of the anomaly scores of the current temperature data in all target random trees can be calculated, and then the quantile is used as a threshold. If the average of the anomaly scores of the current temperature data in all target random trees is greater than the threshold, the current temperature data is determined to be abnormal data.
[0085] Furthermore, when the current temperature data is determined to be abnormal data, it means that the temperature of the insulator is abnormal at the current moment. The information that the temperature of the insulator exceeds the safety value at the current moment can be promptly fed back to relevant personnel via text messages, emails or alarms, so that corresponding measures can be taken in time to deal with it, thereby realizing the monitoring of the insulator temperature.
[0086] The present invention also provides an insulator temperature monitoring system, the system includes a memory and a processor, and a computer program is stored in the memory, the computer program integrates the function of an insulator temperature monitoring method, when the computer program is executed, the accuracy of abnormal detection of insulator temperature data can be improved through an insulator temperature monitoring method, so that the state of the insulator can be accurately monitored.
[0087] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.
[0088] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.
Claims
1. A method for monitoring insulator temperature, It is characterized in that include: Acquire a temperature data sequence of the insulator before the current moment, and construct multiple random trees of the temperature data sequence, wherein the number of nodes in each of the random trees is the same; Determine the numerical difference and the change rate difference between the current temperature data and any node in any random tree, calculate the sum of the squares of the numerical difference and the change rate difference, and use the reciprocal of the arithmetic square root of the sum of squares as the numerical similarity between the current temperature data and any node; Calculating density similarity between the current temperature data and any of the nodes, wherein the density similarity reflects the difference between the data distribution density in a preset local area of the current temperature data and the node distribution density of the random tree where any of the nodes are located; The final similarity between the current temperature data and any of the nodes is calculated, and the final similarity is positively correlated with the numerical similarity and the density similarity. The final similarity between the current temperature data and each node in each of the random trees is used to screen the random trees that meet the preset conditions, so as to perform abnormality detection on the current temperature data according to the screened random trees, and monitor the temperature of the insulator according to the detection results.
2. The insulator temperature monitoring method according to claim 1, It is characterized in that The method for obtaining the change rate difference includes: Taking the difference between the current temperature data and the previous temperature data as a first difference, calculating the ratio of the first difference to the current temperature data to obtain a first change rate; Taking the difference between any node and the parent node of any node as the second difference, calculating the ratio of the second difference to the any node, and obtaining a second change rate; The difference between the first change rate and the second change rate is calculated to obtain the change rate difference.
3. The insulator temperature monitoring method according to claim 1, It is characterized in that The method for obtaining the data distribution density comprises: The sequence composed of the current temperature data and the previous temperature data, whose length is the same as the number of nodes in any random tree, is taken as the local area of the current temperature data; Clustering is performed on each temperature data in the local area to obtain a cluster center, and the mean value of the distance between each temperature data in the local area and the cluster center is calculated to obtain the data distribution density.
4. The insulator temperature monitoring method according to claim 1, It is characterized in that The method for obtaining the node distribution density includes: The average value of the differences between all nodes except the any node in the random tree where any node is located and the any node is calculated, and the average value is used as the node distribution density of the random tree where the any node is located.
5. A method for monitoring insulator temperature according to claim 3 or 4, It is characterized in that The density similarity between the current temperature data and any node satisfies the following relationship: ; In the formula, The current temperature data and The random tree Density similarity of nodes; is the data distribution density in the local area of the current temperature data; The node is located at The node distribution density of a random tree; The natural constant An exponential function with base .
6. The insulator temperature monitoring method according to claim 1, It is characterized in that The final similarity between the current temperature data and any node satisfies the following relationship: ; In the formula, The current temperature data and The random tree The final similarity of nodes; The current temperature data and The random tree The numerical similarity of nodes; The current temperature data and The random tree Density similarity of nodes; is the normalization function.
7. The insulator temperature monitoring method according to claim 1, It is characterized in that The method of using the final similarity between the current temperature data and each node in each of the random trees to screen a random tree that meets a preset condition includes: Calculate the average value of the final similarity between the current temperature data and each node in any random tree, and use the node in any random tree whose final similarity with the current temperature data is greater than the average value as a reference node; The number of reference nodes in each of the random trees is counted, the random trees are sorted from large to small according to the number of corresponding reference nodes, and a preset number of random trees are selected from front to back as target random trees.
8. The insulator temperature monitoring method according to claim 7, It is characterized in that The abnormality detection of the current temperature data according to the selected random tree includes: Calculate the complexity change of any target random tree before and after the current temperature data is inserted, and obtain the abnormal score of the current temperature data in any target random tree; The average value of the anomaly scores of the current temperature data in all target random trees is calculated to obtain the collusion anomaly score of the current temperature data. When the collusion anomaly score is greater than a preset threshold, the current temperature data is determined to be anomaly data.
9. An insulator temperature monitoring system, It is characterized in that The insulator temperature monitoring system comprises a memory and a processor, wherein a computer program is stored in the memory, and the processor executes the computer program to implement the steps of the method according to any one of claims 1 to 8.
Citation Information
Cited By
Temperature monitoring system and temperature monitoring method of water jacket heater
CN119538165A
Temperature monitoring system and temperature monitoring method for water jacket heating furnace
CN119538165B
