Feature sensing signal fusion detection system and method for high-voltage switch cabinet

Abstract

The invention provides a feature sensing signal fusion detection system and method for a high-voltage switch cabinet, and belongs to the technical field of internal insulation defect detection of the high-voltage switch cabinet. Comprising the steps that an experiment platform is built, and the experiment platform is composed of a high-voltage switch cabinet model, an intelligent sensor and a signal processor; the high-voltage switch cabinet model is divided into a defect-free model and a model with a single insulation defect inside, and an intelligent sensor is arranged in the high-voltage switch cabinet model; the intelligent sensor is connected with N gas sensors and P temperature and humidity sensors; the high-voltage switch cabinet model is operated, the intelligent sensor carries out sensing signal fusion according to attributes, two attribute fusion results are obtained and then fused into a global state Ym, and the global state Ym is sent to the signal processor; collecting the signal processors into a total set {Ym}, and defining an interval in which the {Ym} is located to indicate an insulation defect type; in actual detection, whether the high-voltage switch cabinet has an insulation defect and the type of the insulation defect are determined according to the Ym output by each intelligent sensor and the segment interval.

Description

Technical Field

[0001] The present invention relates to the technical field of internal insulation defect detection of high-voltage switch cabinets, and in particular to a characteristic sensing signal fusion detection system and method for high-voltage switch cabinets. Background Art

[0002] High-voltage switchgear is an important power equipment in the power transmission, transformation and distribution process of the power system. It can control and protect the power system, reduce power outages and improve power supply reliability. At the same time, high-voltage switchgear is also equipped with a variety of protection devices, which can quickly cut off the power supply under abnormal conditions to protect the safety of equipment and personnel. Its operating status is directly related to the safety and stability of the power grid. Partial discharge caused by insulation defects is one of the main causes of failure of high-voltage switchgear. Early detection and treatment of partial discharge is of great significance for preventing switchgear failure.

[0003] There are three main types of insulation defects in high-voltage switchgear: (1) Metal protrusion defects: metal burrs protruding from the inner wall of the high-voltage switchgear cavity or the high-voltage (HV) conductor; (2) Insulation layer air gap defects: air gaps appear inside the solid insulation layer due to factors such as aging and production process; (3) Metal particle defects: conductive particles left in the switchgear cavity during manufacturing and installation, and metal powder particle defects formed by friction of metal devices during operation. Partial discharge caused by insulation defects is one of the main causes of failure of high-voltage switchgear. Traditional partial discharge detection methods mainly rely on acoustics, electromagnetics and other means, and have problems such as low detection accuracy and high false alarm rate. The existing technology has found that the high-energy particles generated by the breakdown of gas during partial discharge will collide and ionize with the particles in the discharge area, causing some new gases to appear in the space. Characteristic gases (such as SF6 gas or other common gas components of electrical equipment) are important indicators reflecting the operating status, failure or abnormality of the equipment. Therefore, accurate and real-time monitoring of the characteristic gas concentration in the switchgear is crucial for early warning of equipment failure. However, a single sensor may be affected by many factors, such as sensor accuracy, aging, environmental interference, etc., which may make the measurement results of gas concentration inaccurate or unstable. Using multiple gas sensors can try to eliminate the instability problem of a single sensor. However, the differences in sensitivity, accuracy, response time, environmental factors, etc. of each sensor lead to the fact that their signal curves are inconsistent with each other in the same scenario. There are reading differences (different sensors may have large differences in gas concentration values ​​at the same time), curve inconsistency (the output signal curve of the sensor may show different fluctuations, peaks or trends within the same time range), and timing inconsistency (different sensor response speeds and delays may cause them to respond inconsistently to changes in gas concentration). It is difficult to exclude abnormal data, so it is impossible to completely eliminate the problem of inaccurate or unstable measurement results caused by using a single sensor, which leads to inaccurate detection of insulation defects in high-voltage switchgear. Summary of the invention

[0004] In view of this, the present invention provides a characteristic sensor signal fusion detection system and method for high-voltage switchgear, which solves the problem that the measurement results of the traditional single sensor system are inaccurate or unstable, resulting in the inability to accurately reflect the actual insulation defect state of the switchgear. Redundant sensors and sensor signal fusion technology are used to provide higher-precision insulation defect type detection results, thereby improving the detection efficiency and accuracy of insulation defect detection in high-voltage switchgear.

[0005] The technical solution adopted by the embodiment of the present invention to solve the technical problem is:

[0006] A characteristic sensing signal fusion detection method for a high-voltage switch cabinet, comprising:

[0007] Step S1, building an experimental platform, the experimental platform consists of a high-voltage switch cabinet model, an intelligent sensor, and a signal processor; wherein the high-voltage switch cabinet model is further divided into a defect-free model and a model with a single insulation defect inside, and M intelligent sensors distributed in different positions are arranged in the high-voltage switch cabinet model, and the insulation defect type is one of a metal protrusion defect, a metal particle defect, and an insulation layer air gap defect; the intelligent sensor is connected to N gas sensors and P temperature and humidity sensors, wherein any one gas sensor is regarded as a main sensor, and the remaining N-1 gas sensors and P temperature and humidity sensors are regarded as redundant sensors;

[0008] Step S2, running the high-voltage switch cabinet model, each of the intelligent sensors collects the sensor signal of the high-voltage switch cabinet through the probe and performs signal fusion according to the attribute, and then obtains the two attribute fusion results and fuses them into the global state Y m , sending the global state to the signal processor; the two attribute fusion results are the gas sensor signal fusion result and the temperature and humidity sensor signal fusion result;

[0009] Step S3: the signal processor collects the global state Y of each smart sensor m For a total set {Y m}, {Y m The interval where} is located is defined as the data interval indicating the single insulation defect type, so as to obtain the Y corresponding to different insulation defect types. m Value segment interval; where:

[0010] The interval [0, A) indicates that no insulation fault of any type occurs in the high-voltage switchgear, the interval [A, B) indicates that there is a metal protrusion defect in the high-voltage switchgear, the interval [B, C) indicates that there is a metal particle defect in the high-voltage switchgear, and the interval [C, +∞) indicates that there is an insulation layer air gap defect in the high-voltage switchgear;

[0011] Step S4, arranging each of the intelligent sensors in the high-voltage switch cabinet, and in actual detection, according to the Y output by each of the intelligent sensors m And determine in segmented intervals whether there are insulation defects in the high-voltage switchgear and the corresponding insulation defect types.

[0012] Preferably, the fusion process of step S2 includes:

[0013] Step S21, for a group of sensor signals collected by a single sensor, first preprocess the group of sensor signals, wherein the preprocessing includes noise removal, missing value filling, data standardization and normalization processing; then the preprocessed sensor signals are reduced in dimension into low-dimensional data by PCA method;

[0014] Step S22, extracting the mean μ and variance σ of the preprocessed sensor signal data 2 as a set of signal characteristics of sensor signals;

[0015] Step S23, calculating the similarity between each pair of signals; wherein the similarity between two sensor signals of the same type of sensors A and B is calculated using the Euclidean distance:

[0016]

[0017] In the formula, d represents the dimension number, D represents the number of dimensions of the signals collected by sensors x and y of the same type; d and d is the mean of two sensor signals of the same type in dimension d; and represents the variance of two sensor signals of the same type in d dimension;

[0018] The similarity r between two sensor signals of different types of sensors x and z xz The Pearson correlation coefficient was calculated, r xz =0:

[0019] Step S24, calculating the conflict metrics between the sensors and the total conflict metrics of the sensors according to the similarities between the sensors; wherein:

[0020] The conflict measure C between two sensor signals of the same type xz The calculation formula is:

[0021]

[0022] The conflict measure C between two sensor signals of different types of sensors xz The calculation formula is:

[0023] C xz =1-r xz =1

[0024] The total conflict metric C for a single gas sensor n n The calculation formula is:

[0025] C n =∑C ni ,n∈[1,N],i∈[1,N+P],n≠i;

[0026] The total conflict metric C of a single temperature and humidity sensor p p The calculation formula is:

[0027] C p =∑C pj ,p∈[1,P],j∈[1,N+P],p≠j;

[0028] Step S25, assigning a weight w to each sensor according to the total conflict metric of each sensor i : Take the inverse of each total conflict metric and normalize it to get w i , where ∑w i =1; i∈[1,N+P]

[0029] Step S26, using a weighted average method, signal fusion is performed based on the total conflict metric of each sensor by attribute:

[0030]

[0031] In the formula, y mn is the gas sensing signal fusion result of smart sensor m, y mp It is the fusion result of temperature and humidity sensor signals;

[0032] Step S27, using the (implicit importance weighted ordered weighted average) IIWOWA method to aggregate the fusion results of the two attributes into a global state Y m , specifically: Get the important measurements of the two attributes as weight values, v1 is y mn The weight of v2 is y mp The weight of the global state Y m The definition of is:

[0033] Y m =v1y mn +v2y mp

[0034] Step S28, the global state Y m sent to the signal processor.

[0035] Preferably, the gas sensor signal is three-dimensional data, and the three dimensions indicate the gas concentrations of CO, O3, and NO2 respectively; the temperature and humidity sensor signal is two-dimensional data, and the two dimensions indicate the temperature and humidity data respectively.

[0036] Preferably, the step S27 of obtaining the weight values ​​of the two attributes includes:

[0037] According to the total conflict metric C of a single gas sensor n n Calculate the important metric v1 of the gas sensor signal fusion result, based on the total conflict metric C of a single temperature and humidity sensor p p Calculate the important metric v2 of the gas sensor signal fusion result, and the calculation formula is:

[0038]

[0039] Preferably, in the experimental platform, the installation position of the gas sensor includes the upper air window of the switch cabinet and the cabinet body gap; the installation position of the temperature and humidity sensor is located at the upper air window of the air switch cabinet.

[0040] Preferably, the step S4 comprises:

[0041] Step S41: the smart sensor performs signal fusion according to the process of step S2 to obtain Y m and send it to the information processor;

[0042] Step S42, the information processor establishes {Y m}, confirm each Y m The interval where all Y m If all of them are in the interval [0, A), it is confirmed that the high-voltage switchgear has no insulation defects; if there is a Y falling in the interval [A, B) m , it is confirmed that there is a metal protrusion defect in the high-voltage switch cabinet; if there is a Y falling into the interval [B,C) m , it is confirmed that there are metal particle defects in the high-voltage switch cabinet; if there is a Y falling into the interval [C, +∞) m , it is confirmed that there is an air gap defect in the insulation layer in the high-voltage switch cabinet.

[0043] Preferably, {Y m} m Falling into at least three intervals respectively indicates that the high-voltage switchgear has at least two insulation defects.

[0044] A characteristic sensor signal fusion detection system for a high-voltage switch cabinet comprises a simulation device, a high-voltage switch cabinet, M intelligent sensors, and a signal processor. The simulation device is used to simulate four situations of a high-voltage switch cabinet model to obtain a value range of a sensor signal fusion result in a switch cabinet with different insulation defect types; in actual application, each sensor probe of the intelligent sensor is installed in the high-voltage switch cabinet, and the signal processor is used to generate a global state Y output by each of the intelligent sensors. m , determine whether the high-voltage switch cabinet has an insulation defect and the corresponding insulation defect type, and the system is used to implement the above method.

[0045] Preferably, in the experimental platform, the installation position of the gas sensor includes the upper air window of the switch cabinet and the cabinet body gap; the installation position of the temperature and humidity sensor is located at the upper air window of the air switch cabinet.

[0046] Preferably, {Y m} m Falling into at least three intervals respectively indicates that the high-voltage switchgear has at least two insulation defects.

[0047] It can be seen from the above technical scheme that the characteristic sensing signal fusion detection method and system for high-voltage switchgear provided in the embodiment of the present invention simulates the defect-free model of the high-voltage switchgear and three models with a single insulation defect inside, adopts redundant sensors and combines the influence of temperature and humidity on the gas concentration in the high-voltage switchgear, and fuses the same type of sensor signals according to the attributes through intelligent sensors, and then fuses the fusion signals of different attributes into a global state again, and uses each global state to determine the value range indicating a single insulation defect. In practical applications, the value range is used to confirm whether there is an insulation defect inside the high-voltage switchgear and the type of insulation defect. The present invention takes into account the influence of temperature and humidity on the gas concentration in the high-voltage switchgear, effectively improves the internal defect detection accuracy and detection efficiency of the high-voltage switchgear, and even when some sensors fail, the system can still work normally, significantly improving the reliability and safety of switchgear monitoring, and providing strong support for equipment fault prediction and early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a schematic diagram of the characteristic sensing signal fusion detection system for high-voltage switchgear of the present invention.

[0049] Figure 2 The sensor diagram.

[0050] Figure 3 Schematic diagram of the sensor information data fusion process. DETAILED DESCRIPTION

[0051] The technical scheme and technical effects of the present invention are further elaborated in detail below in conjunction with the accompanying drawings of the present invention.

[0052] The present invention designs a detection system of distributed intelligent sensors and information fusion for collecting characteristic gases inside high-voltage switch cabinets. Since any single sensor or measurement cannot accurately reflect the overall characteristics of the characteristic gas, the distributed sensors and modular self-diagnosis information fusion are used to solve the problem that the characteristics cannot be reflected by a single sensor. The adopted distributed architecture can effectively improve the detection accuracy and reliability of insulation defect types in high-voltage switch cabinets.

[0053] refer to Figure 1 As shown, the present invention provides a characteristic sensing signal fusion detection method for a high-voltage switch cabinet, comprising:

[0054] Step S1, building an experimental platform, which consists of a high-voltage switch cabinet model, an intelligent sensor, and a signal processor; wherein the high-voltage switch cabinet model is further divided into a defect-free model and a model with a single insulation defect inside, and M intelligent sensors distributed in different positions are arranged in the high-voltage switch cabinet model, and the insulation defect type is one of a metal protrusion defect, a metal particle defect, and an insulation layer air gap defect; the intelligent sensor is connected to N gas sensors and P temperature and humidity sensors, wherein any one gas sensor is regarded as a main sensor, and the remaining N-1 gas sensors and P temperature and humidity sensors are regarded as redundant sensors;

[0055] Step S2, running the high-voltage switch cabinet model, each intelligent sensor collects the sensor signal of the high-voltage switch cabinet through the probe and performs signal fusion according to the attribute, and then obtains the two attribute fusion results and fuses them into the global state Y m , the global state is sent to the signal processor; the two attribute fusion results are the gas sensor signal fusion result and the temperature and humidity sensor signal fusion result;

[0056] Step S3: The signal processor collects the global state Y of each smart sensor. m For a total set {Y m}, {Y m The interval where} is located is defined as the data interval indicating a single insulation defect type, so as to obtain the Y corresponding to different insulation defect types. m Value segment interval; where:

[0057] The interval [0, A) indicates that no insulation fault of any type occurs in the high-voltage switchgear, the interval [A, B) indicates that there is a metal protrusion defect in the high-voltage switchgear, the interval [B, C) indicates that there is a metal particle defect in the high-voltage switchgear, and the interval [C, +∞) indicates that there is an insulation layer air gap defect in the high-voltage switchgear;

[0058] Step S4: Arrange various intelligent sensors in the high-voltage switch cabinet. In actual detection, according to the Y output by each intelligent sensor, m And determine in segmented intervals whether there are insulation defects in the high-voltage switchgear and the corresponding insulation defect types.

[0059] As an implementation method, refer to Figure 3 The specific implementation of the fusion process of step S2 includes:

[0060] Step S21, for a group of sensor signals collected by a single sensor, first preprocess the group of sensor signals, the preprocessing includes noise removal, missing value filling, data standardization and normalization processing (refer to Figure 2 Then, the pre-processed sensor signal is processed into low-dimensional data by PCA method;

[0061] Step S22, extracting the mean μ and variance σ of the preprocessed sensor signal data 2 as a set of signal characteristics of sensor signals;

[0062] Step S23, calculating the similarity between each pair of signals; wherein the similarity between two sensor signals of the same type of sensors A and B is calculated using the Euclidean distance:

[0063]

[0064] In the formula, d represents the dimension number, D represents the number of dimensions of the signals collected by sensors x and y of the same type; d and d is the mean of two sensor signals of the same type in dimension d; and represents the variance of two sensor signals of the same type in d dimension;

[0065] The similarity r between two sensor signals of different types of sensors x and z xz The Pearson correlation coefficient was calculated, r xz =0:

[0066] Step S24, calculating the conflict metrics between the sensors and the total conflict metrics of the sensors according to the similarities between the sensors; wherein:

[0067] The conflict measure C between two sensor signals of the same type xz The calculation formula is:

[0068]

[0069] The conflict measure C between two sensor signals of different types of sensors xz The calculation formula is:

[0070] C xz =1-r xz =1 (3)

[0071] The total conflict metric C for a single gas sensor n n The calculation formula is:

[0072] C n =∑C ni ,n∈[1,N],i∈[1,N+P],n≠i; (4)

[0073] The total conflict metric C of a single temperature and humidity sensor p p The calculation formula is:

[0074] C p =∑C pj,p∈[1,P],j∈[1,N+P],p≠j; (5)

[0075] Step S25, assigning a weight w to each sensor according to the total conflict metric of each sensor i : Take the inverse of each total conflict metric and normalize it to get w i , where ∑w i =1; i∈[1,N+P]

[0076] Step S26, using a weighted average method, signal fusion is performed based on the total conflict metric of each sensor by attribute:

[0077]

[0078] In the formula, y mn is the gas sensing signal fusion result of smart sensor m, y mp It is the fusion result of temperature and humidity sensor signals;

[0079] Step S27, using the (implicit importance weighted ordered weighted average) IIWOWA method to aggregate the fusion results of the two attributes into a global state Y m , specifically: Get the important measurements of the two attributes as weight values, v1 is y mn The weight of v2 is y mp The weight of the global state Y m The definition of is:

[0080] Y m =v1y mn +v2y mp (8)

[0081] Step S28, the global state Y m Sent to the signal processor.

[0082] Preferably, the gas sensor signal is three-dimensional data, and the three dimensions indicate the gas concentrations of CO, O3, and NO2 respectively; the temperature and humidity sensor signal is two-dimensional data, and the two dimensions indicate the temperature and humidity data respectively.

[0083] As an implementation mode, the specific implementation of step S27 of obtaining the weight values ​​of the two attributes includes:

[0084] According to the total conflict metric C of a single gas sensor n n Calculate the important metric v1 of the gas sensor signal fusion result, based on the total conflict metric C of a single temperature and humidity sensor p p Calculate the important metric v2 of the gas sensor signal fusion result, and the calculation formula is:

[0085]

[0086] Preferably, in the experimental platform, the installation position of the gas sensor includes the upper air window of the switch cabinet and the cabinet gap; the installation position of the temperature and humidity sensor is located at the upper air window of the air switch cabinet.

[0087] As an implementation mode, the specific implementation of step S4 includes:

[0088] Step S41: the intelligent sensor performs signal fusion according to the process of step S2 to obtain Y m and sent to the information processor;

[0089] Step S42, the information processor establishes {Y m}, confirm each Y m The interval where all Y m If all of them are in the interval [0,A), it is confirmed that there is no insulation defect in the high-voltage switchgear; if there is a Y falling in the interval [A,B) m , it is confirmed that there is a metal protrusion defect in the high-voltage switch cabinet; if there is a Y falling into the interval [B,C) m , it is confirmed that there are metal particle defects in the high-voltage switch cabinet; if there is a Y falling into the interval [C, +∞) m , it is confirmed that there is an insulation layer air gap defect in the high-voltage switch cabinet. In practical applications, the high-voltage switch cabinet is very large, and some locations may not detect insulation defect gas indicators. Multiple locations can detect gas indicators of different insulation defect types. Therefore, 1-3 insulation defects may exist at the same time. For example, {Y m} m Falling into at least three intervals respectively can indicate that the high-voltage switchgear has at least two insulation defects.

[0090] A characteristic sensor signal fusion detection system for a high-voltage switch cabinet comprises a simulation device, a high-voltage switch cabinet, M intelligent sensors, and a signal processor. The simulation device is used to simulate four situations of a high-voltage switch cabinet model to obtain a value range of a sensor signal fusion result in a switch cabinet with different insulation defect types. In actual application, each sensor probe of the intelligent sensor is installed in the high-voltage switch cabinet. The signal processor is used to generate a global state Y output by each intelligent sensor. m , determine whether there are insulation defects in the high-voltage switchgear and the corresponding insulation defect type.

[0091] In the experimental platform, the gas sensors are installed at the upper air vents of the switch cabinet and the cabinet gaps; the temperature and humidity sensors are installed at the upper air vents of the air switch cabinet. In actual applications, due to the large size of the high-voltage switch cabinet, sensors far away from the actual location of the insulation defect may not be able to detect the insulation defect gas indication. Therefore, {Ym All values ​​of} can fall within the interval [0, A), and the high-voltage switchgear has no insulation defects; if there is a Y falling within the interval [A, B) m , it is confirmed that there is a metal protrusion defect in the high-voltage switch cabinet; if there is a Y falling into the interval [B,C) m , it is confirmed that there are metal particle defects in the high-voltage switch cabinet; if there is a Y falling into the interval [C, +∞) m , it is confirmed that there is an air gap defect in the insulation layer in the high-voltage switch cabinet. m} m Falling into at least three intervals respectively, it indicates that the high-voltage switchgear has at least two insulation defects at the same time.

[0092] Intelligent systems require information about the environment in which they operate, and sensors are used to maintain the intelligent machine's internal description of the external world, allowing the system to interact effectively with its environment. In order to provide the system with different information (e.g., information about different variables) and information about the same variable but at different locations in space (complementary information). Each node of each intelligent sensor can be regarded as a cyber-physical system (CPS), which can reduce the amount and dimensionality of data, extract information from the acquired data, and realize the adaptive and self-diagnostic capabilities of intelligent nodes regarding conflicting sensor values.

[0093] For a fusion unit set up in an intelligent sensor for fusion function: the input includes data / information source, auxiliary information and prior external knowledge, and the output is the fusion result. The architecture of the fusion unit includes (1) a single fusion device; (2) a parallel architecture: each fusion unit processes sensor data and passes the result to the next level fusion unit to fuse the processing results from multiple sensors; (3) a serial architecture: each fusion unit receives data from the sensor and the output from the previous unit, and passes the result to the next level unit. It is possible to have a hybrid arrangement that combines parallel and serial fusion architectures into one fusion system.

[0094] The selection of data fusion algorithms is based on the abstraction level: signal level, feature level and decision level. Signal level: The fusion process provides more accurate data in terms of certain data quality metrics (such as the ratio of characteristic gases), and the fused results will be further processed for feature extraction and classification (identity declaration); Feature level fusion: Get more discriminating features to meet the needs of other tasks; Decision fusion: Take symbolic representation as the source and combine them to obtain more accurate decisions. And design an intelligent adaptive sensor system to grasp specific environmental or state variables, collect system errors or faulty sensors on site through signal fusion methods, and make corresponding compensation.

[0095] Simplify the sensor and information fusion system through intelligent communication networks. The application provides the required functions for the system, including sensor objects and fusion blocks; the middleware layer provides a standardized communication interface for the application, which is independent of the underlying real-time communication system and is responsible for decoupling the application layer and the underlying connection layer, application data exchange and node self-configuration, which are described in each application object and provided through a standardized interface; the connection layer is used to exchange application process data with deterministic time boundaries, and can provide different real-time communication channels according to the requirements of the application.

[0096] Therefore, the intelligent sensor node, self-configuring real-time communication network, and sensor and information fusion system for condition monitoring of the present invention use multiple redundant sensors to generate the same information and fuse them, while eliminating the uncertainty of the measurement results of a single sensor and providing more accurate information about the environment, effectively solving the problem that a single sensor or measurement cannot accurately reflect the characteristics of the characteristic gas in the switch cabinet.

[0097] What is disclosed above is only a preferred embodiment of the present invention, which certainly cannot be used to limit the scope of rights of the present invention. A person skilled in the art can understand that all or part of the processes of the above embodiments and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A characteristic sensing signal fusion detection method for high-voltage switchgear, characterized in that: include: Step S1, building an experimental platform, the experimental platform consists of a high-voltage switch cabinet model, an intelligent sensor, and a signal processor; wherein the high-voltage switch cabinet model is further divided into a defect-free model and a model with a single insulation defect inside, and M intelligent sensors distributed in different positions are arranged in the high-voltage switch cabinet model, and the insulation defect type is one of a metal protrusion defect, a metal particle defect, and an insulation layer air gap defect; the intelligent sensor is connected to N gas sensors and P temperature and humidity sensors, wherein any one gas sensor is regarded as a main sensor, and the remaining N-1 gas sensors and P temperature and humidity sensors are regarded as redundant sensors; Step S2, running the high-voltage switch cabinet model, each of the intelligent sensors collects the sensor signal of the high-voltage switch cabinet through the probe and performs signal fusion according to the attribute, and then obtains the two attribute fusion results and fuses them into the global state Y m , sending the global state to the signal processor; the two attribute fusion results are the gas sensor signal fusion result and the temperature and humidity sensor signal fusion result; Step S3: the signal processor collects the global state Y of each smart sensor m For a total set {Y m }, {Y m The interval where} is located is defined as the data interval indicating the single insulation defect type, so as to obtain the Y corresponding to different insulation defect types. m Value segment interval; where: The interval [0, A) indicates that no insulation fault of any type occurs in the high-voltage switchgear, the interval [A, B) indicates that there is a metal protrusion defect in the high-voltage switchgear, the interval [B, C) indicates that there is a metal particle defect in the high-voltage switchgear, and the interval [C, +∞) indicates that there is an insulation layer air gap defect in the high-voltage switchgear; Step S4: Arrange each of the intelligent sensors in the high-voltage switch cabinet. In actual detection, according to the Y output by each of the intelligent sensors, m And determine in segmented intervals whether there are insulation defects in the high-voltage switchgear and the corresponding insulation defect types.

2. The characteristic sensing signal fusion detection method for high-voltage switchgear according to claim 1, characterized in that: The fusion process of step S2 includes: Step S21, for a group of sensor signals collected by a single sensor, first preprocess the group of sensor signals, wherein the preprocessing includes noise removal, missing value filling, data standardization and normalization processing; then the preprocessed sensor signals are reduced in dimension into low-dimensional data by PCA method; Step S22, extracting the mean μ and variance σ of the preprocessed sensor signal data 2 as a set of signal characteristics of sensor signals; Step S23, calculating the similarity between each pair of signals; wherein the similarity between two sensor signals of the same type of sensors A and B is calculated using the Euclidean distance: In the formula, d represents the dimension number, D represents the number of dimensions of the signals collected by sensors x and y of the same type; d and d is the mean of two sensor signals of the same type in dimension d; and represents the variance of two sensor signals of the same type in d dimension; The similarity r between two sensor signals of different types of sensors x and z xz The Pearson correlation coefficient was calculated, r xz =0: Step S24, calculating the conflict metrics between the sensors and the total conflict metrics of the sensors according to the similarities between the sensors; wherein: The conflict measure C between two sensor signals of the same type xz The calculation formula is: The conflict measure C between two sensor signals of different types of sensors xz The calculation formula is: C xz ＝1-r xz ＝1 The total conflict metric C for a single gas sensor n n The calculation formula is: C n =∑C ni ,n∈[1,N],i∈[1,N+P],n≠i; The total conflict metric C of a single temperature and humidity sensor p p The calculation formula is: C p =∑C pj ,p∈[1,P],j∈[1,N+P],p≠j; Step S25, assigning a weight w to each sensor according to the total conflict metric of each sensor i : Take the inverse of each total conflict metric and normalize it to get w i , where ∑w i =1; i∈[1,N+P] Step S26, using a weighted average method, signal fusion is performed based on the total conflict metric of each sensor by attribute: In the formula, y mn is the gas sensing signal fusion result of smart sensor m, y mp It is the fusion result of temperature and humidity sensor signals; Step S27, using the (implicit importance weighted ordered weighted average) IIWOWA method to aggregate the fusion results of the two attributes into a global state Y m , specifically: Get the important measurements of the two attributes as weight values, v1 is y mn The weight of v2 is y mp The weight of the global state Y m The definition of is: Y m =v1y mn +v2y mp Step S28, the global state Y m sent to the signal processor.

3. The characteristic sensing signal fusion detection method for high-voltage switchgear according to claim 2, characterized in that: The gas sensor signal is three-dimensional data, and the three dimensions indicate the gas concentrations of CO, O3, and NO2 respectively; the temperature and humidity sensor signal is two-dimensional data, and the two dimensions indicate the temperature and humidity data respectively.

4. The characteristic sensing signal fusion detection method for high-voltage switch cabinet according to claim 3 is characterized in that: The step S27 of obtaining the weight values ​​of the two attributes includes: According to the total conflict metric C of a single gas sensor n n Calculate the important metric v1 of the gas sensor signal fusion result, based on the total conflict metric C of a single temperature and humidity sensor p p Calculate the important metric v2 of the gas sensor signal fusion result, and the calculation formula is:

5. The characteristic sensing signal fusion detection method for high-voltage switch cabinet according to claim 4, characterized in that: In the experimental platform, the installation positions of the gas sensor include the upper air window of the switch cabinet and the cabinet gap; the installation position of the temperature and humidity sensor is located at the upper air window of the air switch cabinet.

6. The characteristic sensing signal fusion detection method for high-voltage switch cabinet according to claim 5, characterized in that: The step S4 comprises: Step S41: the smart sensor performs signal fusion according to the process of step S2 to obtain Y m and send it to the information processor; Step S42, the information processor establishes {Y m }, confirm each Y m The interval where all Y m If all of them are in the interval [0, A), it is confirmed that the high-voltage switchgear has no insulation defects; if there is a Y falling in the interval [A, B) m , it is confirmed that there is a metal protrusion defect in the high-voltage switch cabinet; if there is a Y falling into the interval [B,C) m , it is confirmed that there are metal particle defects in the high-voltage switch cabinet; if there is a Y falling into the interval [C, +∞) m , it is confirmed that there is an air gap defect in the insulation layer in the high-voltage switch cabinet.

7. The characteristic sensing signal fusion detection method for high-voltage switch cabinet according to claim 6, characterized in that: {Y m } m Falling into at least three intervals respectively indicates that the high-voltage switchgear has at least two insulation defects.

8. A characteristic sensing signal fusion detection system for high-voltage switchgear, characterized in that: It includes a simulation device, a high-voltage switch cabinet, M intelligent sensors, and a signal processor. The simulation device is used to simulate four situations of the high-voltage switch cabinet model to obtain the value range of the sensor signal fusion result in the switch cabinet with different insulation defect types; In actual application, each sensor probe of the intelligent sensor is installed in the high-voltage switch cabinet, and the signal processor is used to generate a global state Y according to the output of each intelligent sensor. m , determine whether the high-voltage switchgear has insulation defects and the corresponding insulation defect type, and the system is used to implement the method described in any one of claims 1-7.

9. The characteristic sensing signal fusion detection system for high-voltage switchgear according to claim 8, characterized in that: In the experimental platform, the installation positions of the gas sensor include the upper air window of the switch cabinet and the cabinet gap; the installation position of the temperature and humidity sensor is located at the upper air window of the air switch cabinet.

10. The characteristic sensing signal fusion detection system for high-voltage switch cabinet according to claim 9, characterized in that: {Y m } m Falling into at least three intervals respectively indicates that the high-voltage switchgear has at least two insulation defects.

Images