Insulator accumulated dirt on-line monitoring identification and discharge risk prediction device and method

Abstract

The invention relates to the technical field of insulator on-line monitoring, and provides an insulator dirt retention on-line monitoring identification and discharge risk prediction device and method.The device is installed at the position, close to an insulator, of an electric power tower, a no-load control insulator is arranged, and the insulator dirt degree is monitored and identified in real time; insulator discharge risk prediction is carried out in combination with environmental rainfall and humidity data, and insulator flashover discharge accidents of a power system can be prevented; meanwhile, in order to improve the monitoring efficiency and ensure the result reliability, a light shield capable of being opened and closed is used to meet the requirements of no-load control insulator accumulated dirt and photographing of the hyperspectral probe module; according to the invention, solar energy is matched with the storage battery to realize uninterrupted power supply, and the device can communicate with background software to realize remote checking of insulator pollution degree and discharge risk judgment results.

Description

Technical Field

[0001] The present invention relates to the technical field of insulator online monitoring, and in particular to an insulator contamination online monitoring identification and discharge risk prediction device and method. Background Art

[0002] With the continuous development of power systems, the issue of safe and stable operation has received more and more attention. Insulators are important components that play a role in mechanical connection support and electrical insulation for electrical equipment and are an indispensable part of the power system. The insulation performance of insulators directly affects whether the power system can maintain stable operation.

[0003] Since insulators work in an outdoor environment, dirty dust will accumulate on the surface of insulators that have been in operation for a long time. When the dirt accumulation on the surface of the insulator is serious and the surrounding humidity is high or there is rainfall, the surface resistance of the insulator will decrease and the insulation performance will decline, resulting in an increased risk of discharge and short circuit tripping of the power system. In order to avoid discharge, it is necessary to conduct online identification and monitoring of the dirt accumulation status of the insulator, and determine the discharge risk in combination with rainfall and humidity data.

[0004] At present, there are many devices in the industry for online monitoring of insulator contamination status and discharge risk prediction, but they often have defects such as complex equipment and poor anti-interference performance, and cannot fully meet the needs of insulator contamination measurement and discharge risk prediction.

[0005] In summary, in order to solve the problems of contamination degree identification, data processing, discharge risk prediction and data transmission in insulator contamination degree measurement and discharge risk prediction, the present invention proposes an insulator contamination online monitoring identification and discharge risk prediction device and method. Summary of the invention

[0006] The purpose of the present invention is to solve the deficiencies of the prior art and to provide an insulator contamination online monitoring identification and discharge risk prediction device and method to improve the measurement accuracy, range and use safety.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] An insulator contamination online monitoring identification and discharge risk prediction device, comprising: a mounting frame, an information collection module, a data processing and control module, and a power supply module;

[0009] The bottom of the mounting frame is fixed on the power pole tower near the position where the insulator is mounted on the power pole tower, an unloaded control insulator is placed on the top of the mounting frame, and a closable light shield is provided on one side of the top of the mounting frame;

[0010] The information collection module includes a digital rain gauge, a humidity sensor and a hyperspectral probe, which input the collected precipitation, relative humidity and hyperspectral image information into the data processing and control module; the hyperspectral probe is installed inside an openable sunshade;

[0011] The data processing and control module can control the opening and closing of the closable light shield according to the needs of data collection, and after the closable light shield is closed, control the hyperspectral probe to take pictures, and input the hyperspectral image information into the data processing and control module; the data processing and control module first converts the hyperspectral image information into the surface contamination degree of the insulator, and then combines the surface contamination degree of the insulator, precipitation and relative humidity to comprehensively judge the risk level of discharge of the insulator under the current environmental conditions and contamination degree, and sends the risk level signal of discharge of the insulator to the background software;

[0012] The power module is electrically connected to the openable light shield, the information collection module and the data processing and control module.

[0013] Preferably, the openable and closable sunshade includes a shielding cover rotatably connected to one side of the top of the mounting frame through a rotating shaft, a large driving wheel fixedly connected to the center position of one side of the shielding cover, a small driving wheel meshing with the large driving wheel, and a motor with an output shaft fixedly connected to the small driving wheel; the shielding cover and the mounting frame form a closed cavity; the axis of the large driving wheel coincides with the outer edge of the shielding cover, the rotating shaft is provided with two groups, and the two groups are symmetrically installed on both sides of the large driving wheel, and the large driving wheel rotates and penetrates into the accommodating space provided at the bottom of the mounting frame.

[0014] Preferably, a cap socket is provided on the top of the mounting frame, and the cap socket is detachably connected to the no-load control insulator.

[0015] Preferably, a buckle for mounting a hyperspectral probe is provided on the top of the shielding cover cavity.

[0016] Preferably, the digital rain gauge adopts the ABS tipping bucket rain gauge of Preson.

[0017] Preferably, the power module selects the Zhier YXP-60W photovoltaic panel-20A battery combination.

[0018] Another object of the present invention is to provide a method for current measurement using the above-mentioned insulator contamination online monitoring identification and discharge risk prediction device, comprising the following steps:

[0019] S1. The device is started, and the staff remotely sets the working cycle T through the background software;

[0020] S2, the shielding cover is opened to keep the unloaded control insulator in a polluted working state, and the rainfall and relative humidity data are continuously collected through the digital rain gauge and humidity sensor, and the rainfall and relative humidity data are input into the data processing and control module;

[0021] S3, after a single working cycle T ends, the shield is closed, so that the unloaded control insulator is changed to a photographing working state, and the hyperspectral probe takes pictures. After the photographing is completed, the hyperspectral image information is input into the data processing and control module, and the shield is opened to restore the unloaded control insulator to a contamination working state;

[0022] S4, the data processing and control module first converts the hyperspectral image information into the surface contamination degree d of the insulator, calculates the average rainfall p and the average relative humidity RH within the period T, and determines the insulator discharge risk R value based on the average rainfall p and the average relative humidity RH, combined with the surface contamination degree d; the determination is made according to the following formula:

[0023] R=f(d,p,RH)

[0024] Where f uses a machine learning classification algorithm function, such as a decision tree, a random forest algorithm, SVM, or a naive Bayes Bayes model;

[0025] S5. The data processing and control module sends the d value and R value data to the background software via wireless network transmission, and the software displays the current insulator contamination level and discharge risk information.

[0026] Preferably, the prediction function f of the insulator discharge risk level R in step 4 is obtained by constructing a 3D naive Bayesian network, comprising the following steps:

[0027] S4.1. Obtain (d, p, RH, R) data to construct training and test sets;

[0028] S4.2. In the insulator contamination flashover discharge model, it is considered that the three parameters of insulator surface contamination degree, rainfall and relative humidity have independent effects on the discharge risk level; a 3D naive Bayes classification model is used as a machine learning model;

[0029] S4.3. For each set of data (d, p, RH), let data X = (d, p, RH), where X1 = d, X2 = p, X3 = RH; then according to Bayes' theorem, we can get

[0030]

[0031] where k = 0, 1, 2, 3; c k = k represents all possible values ​​of R; considering the independence of each parameter, it can be further simplified to the following prediction function

[0032]

[0033] Where P(R=c k ) is the frequency of each value of R in all training set data, P(X j =x j |R=c k ) is the number of all training set data when the jth component in X is x j The conditional probability when ;

[0034] S4.4. Use the test set data to test, input the (d, p, RH) of the test set data as the value of X into the prediction function, and calculate the probability of the discharge risk state being no risk, medium risk, and high risk under the conditions of the insulator contamination level, rainfall, and relative humidity, that is, the values ​​of P(R=0|X=x), P(R=1|X=x), P(R=2|X=x), and P(R=3|X=x); c corresponding to the maximum value of the four probabilities k The value is the R value obtained by the final judgment; the accuracy of the model is tested by the above method. If the accuracy of the test set exceeds 90%, it is considered that the naive Bayes classification model is reliable.

[0035] Preferably, obtaining (d, p, RH, R) data to construct a training set and a test set in step S4.1 includes the following steps:

[0036] a. Select a string of insulators suitable for a new power tower and apply artificial pollution coating on the insulators. The specific pollution coating method refers to "Insulators for Power Equipment and Their Pollution Levels" (GB / T 15166-2018), and apply five levels of pollution from 0 to IV respectively; correspondingly, the d value representing the pollution level is determined according to the following standards:

[0037]

[0038] b. Adjust the humidity and precipitation in the artificial climate chamber, raise the voltage to the standard working voltage of the insulator string and keep it for 1 hour to check whether flashover discharge will occur; repeat the experiment 10 times at each humidity, precipitation and pollution level, record the number of discharges as n, then the discharge probability P is

[0039]

[0040] c. Determine the discharge risk level R under the current humidity, precipitation and pollution level conditions according to the value of P. The specific principles are as follows:

[0041]

[0042] Among them, when R is 0, 1, 2, and 3, it means that the discharge risk is no risk, low risk, medium risk, and high risk respectively;

[0043] d. Repeat steps S1 to S3, change the pollution level, relative humidity and precipitation conditions to repeat the discharge experiment, and obtain at least 1000 (d, p, RH, R) data points; half of these data points are used as training sets for the naive Bayes model, and the other half are used as test sets for the recognition effect of the naive Bayes model; after the training is completed, a function f can be obtained for judging the insulator discharge risk level R according to the insulator surface pollution level d, rainfall p and ambient relative humidity RH.

[0044] The present invention discloses an insulator contamination online monitoring identification and discharge risk prediction device having the following beneficial effects.

[0045] The present invention installs the device at a position near the insulator of the power tower and sets an unloaded control insulator to monitor and identify the degree of insulator contamination in real time, and predicts the risk of insulator discharge in combination with environmental rainfall and humidity data, which can be used to prevent the occurrence of insulator flashover discharge accidents in the power system; at the same time, in order to improve the monitoring efficiency and ensure the reliability of the results, the present invention uses an openable and closable sunshade to simultaneously meet the needs of unloaded control insulator contamination accumulation and high-spectral probe module photography; the present invention uses solar energy in combination with a battery to achieve uninterrupted power supply, and can also communicate with background software to realize remote viewing of the insulator contamination degree and discharge risk judgment results. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a structural schematic diagram of the insulator contamination online monitoring identification and discharge risk prediction device of the present invention.

[0047] Figure 2 It is a schematic diagram of the pollution accumulation working state of the no-load control insulator of the present invention.

[0048] Figure 3 This is a schematic diagram of the working state of the no-load control insulator of the present invention.

[0049] Figure 4 Another schematic diagram of the working state of the no-load control insulator of the present invention.

[0050] Figure 5 for Figure 4 A local enlarged schematic diagram of point B in the middle.

[0051] Figure 6 Flow chart of the method for online monitoring, identification and discharge risk prediction of insulator contamination working in the present invention

[0052] In the attached figure: 1. mounting frame; 11. shielding cover; 12. rotating shaft; 13. large driving wheel; 14. small driving wheel; 15. motor; 2. information collection module; 21. digital rain gauge; 22. humidity sensor; 23. hyperspectral probe; 3. data processing and control module; 4. power module; 41. solar panel; 42. battery; 5. pole tower; 6. insulator string; 7. no-load control insulator. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0054] In the description of the present invention, it is necessary to understand that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0055] Example 1

[0056] Please refer to Figure 1 , an insulator contamination online monitoring identification and discharge risk prediction device, comprising: a mounting frame 1, an information collection module 2, a data processing and control module 3 and a power supply module 4;

[0057] like Figure 2 and Figure 4 As shown, the mounting frame 1 is fixed on the power pole 5 near the position where the insulator string 6 is mounted on the power pole 5 by a clamp, an unloaded control insulator 7 is placed on the top of the mounting frame 1, and a closable light shield is provided on one side of the top of the mounting frame 1; the closable light shield can rotate along one side of the top of the mounting frame 1, and the rotation angle is 0-130°. When the rotation angle of the closable light shield is 0°, as shown in FIG. Figure 3 and Figure 4 As shown, the closable light shield is buckled on the top of the mounting frame 1 so that the no-load reference insulator 7 is in a dark state. When the closable light shield is rotated to an angle of 130°, as shown in FIG. Figure 2As shown, the closable light shield is transferred to one side of the top of the mounting frame 1, and the unloaded control insulator 7 is not blocked. Preferably, in this embodiment, the closable light shield includes a shielding cover 11 rotatably connected to one side of the top of the mounting frame 1 through a rotating shaft 12, a large driving wheel 13 fixedly connected to the center position of one side of the shielding cover 11, a small driving wheel 14 meshing with the large driving wheel 13, and a motor 15 whose output shaft is fixedly connected to the small driving wheel 14; the shielding cover 11 and the top of the mounting frame 1 form a closed cavity; the axis of the large driving wheel 13 coincides with the outer edge of the shielding cover 11, and the rotating shaft 12 is provided with two groups, which are symmetrically installed on both sides of the large driving wheel 13. After the large driving wheel 13 rotates, it penetrates into the accommodation space provided at the bottom of the mounting frame 1, and the installation position of the motor 15 should avoid interference with the shielding cover 11 after it is opened. A cap socket is provided at the top of the mounting frame 1, and the cap socket is detachably connected to the unloaded control insulator. A buckle for installing the hyperspectral probe 23 is provided at the top of the cavity of the shielding cover 11.

[0058] The information collection module 2 includes a digital rain gauge 21, a humidity sensor 22 and a hyperspectral probe 23. The digital rain gauge 21, the humidity sensor 22 and the hyperspectral probe 23 input the collected precipitation, relative humidity and hyperspectral image information into the data processing and control module 3; the hyperspectral probe 23 is installed inside the openable sunshade; the digital rain gauge 21 adopts the ABS tipping bucket rain gauge of Preson, with a resolution of 0.2mm, which meets the accuracy requirements and can send the rainfall data to the data processing and control module 3 through the signal line; the hygrometer selects the Jianda Renke 485 hygrometer, which is connected to the data processing and control module 3 through the signal line. The hyperspectral probe 23 selects the Jiu Enbo FS-200 transmission grating visible light hyperspectral camera, with an image resolution of 1920*1920 and a single-frame shooting time of less than 15 seconds, which can meet the requirements of insulator contamination identification. This type of camera can be connected to the data processing and control module 3 through the USB3.0 port.

[0059] The data processing and control module 3 can control the opening and closing of the openable hood according to the needs of data collection. After the openable hood is closed, the hyperspectral probe 23 is controlled to take pictures and the hyperspectral image information is input into the data processing and control module 3; the data processing and control module 3 first converts the hyperspectral image information into the surface contamination degree of the insulator, and then combines the surface contamination degree, precipitation and relative humidity of the insulator to comprehensively judge the risk level of discharge of the insulator under the current environmental conditions and contamination degree and send the risk level signal of the insulator discharge to the background software; the data processing and control module 3 selects the Spartan UltraScale+ sixth-generation FPGA chip, and is matched with the Ankerui AWT100 wireless transmission component with two communication modes of 4G and NB-IoT, so as to realize the real-time processing and transmission of data transmitted from the hyperspectral probe 23 module and the environmental information collection subsystem.

[0060] The power module 4 is electrically connected to the openable sunshade, the information collection module 2, and the data processing and control module 3. The power module 4 selects the Zhier YXP-60W photovoltaic panel-20A battery combination. This type of solar panel 41 and battery 42 can meet the power supply needs of the whole set of equipment, and its battery 42 can continuously power the equipment for more than 48 hours in the absence of light, so that the lighting conditions in most parts of my country can enable the whole set of equipment to work continuously 24 hours a day.

[0061] Example 2

[0062] Please refer to Figure 5 , an insulator contamination online monitoring identification and discharge risk prediction method, comprising the following steps:

[0063] S1. The device is started, and the staff remotely sets the working cycle T through the background software;

[0064] S2, the shielding cover 11 is opened, so that the unloaded control insulator 7 is kept in the pollution working state, and the rainfall and relative humidity data are continuously collected through the digital rain gauge 21 and the humidity sensor 22, and the rainfall and relative humidity data are input into the data processing and control module 3;

[0065] S3, after the single working cycle T ends, the shielding cover 11 is closed, so that the unloaded control insulator 7 is changed to the photographing working state, and the hyperspectral probe 23 takes pictures. After the photographing is completed, the hyperspectral image information is input into the data processing and control module 3, and the shielding cover 11 is opened, so that the unloaded control insulator 7 is restored to the contamination accumulation working state;

[0066] S4. The image captured by the hyperspectral probe 23 includes the entire visible light spectrum range from 400-700nm. The data processing and control module divides the spectral information within this range into 150 bands for collection and processing. By analyzing the intensity level of the visible light signal in a special band, the hyperspectral image information can be converted into the surface contamination degree d of the insulator, and d is divided into five levels from 0 to IV; the average rainfall p and the average relative humidity RH within the period T are calculated, and the insulator discharge risk R value is judged based on the average rainfall p and the average relative humidity RH, combined with the surface contamination degree d; the judgment is made according to the following formula:

[0067] R=f(d,p,RH)

[0068] Where f uses a machine learning classification algorithm function, such as a decision tree, a random forest algorithm, SVM, or a naive Bayes Bayes model;

[0069] In step 4, the prediction function f of the insulator discharge risk level R is obtained by constructing a 3D naive Bayesian network, including the following steps:

[0070] S4.1. Obtain (d, p, RH, R) data to construct training and test sets;

[0071] Obtaining (d, p, RH, R) data to construct training sets and test sets in step S4.1 includes the following steps:

[0072] a. Select a string 6 of insulators suitable for a new power tower, and apply artificial pollution coating on the insulators of the string. The specific pollution coating method refers to "Insulators for Power Equipment and Their Pollution Levels" (GB / T 15166-2018), and apply five levels of pollution from 0 to IV respectively; correspondingly, the d value representing the pollution level is determined according to the following standards:

[0073]

[0074] b. Adjust the humidity and precipitation in the artificial climate chamber, raise the voltage to the standard working voltage of the insulator string and keep it for 1 hour to check whether flashover discharge will occur; repeat the experiment 10 times at each humidity, precipitation and pollution level, record the number of discharges as n, then the discharge probability P is

[0075]

[0076] c. Determine the discharge risk level R under the current humidity, precipitation and pollution level conditions according to the value of P. The specific principles are as follows:

[0077]

[0078] Among them, when R is 0, 1, 2, and 3, it means that the discharge risk is no risk, low risk, medium risk, and high risk, respectively.

[0079] d. Repeat steps S1 to S3, change the pollution level, relative humidity and precipitation conditions to repeat the discharge experiment, and obtain at least 1000 (d, p, RH, R) data points; half of these data points are used as training sets for the naive Bayes model, and the other half are used as test sets for the recognition effect of the naive Bayes model; after the training is completed, a function f can be obtained for judging the insulator discharge risk level R according to the insulator surface pollution level d, rainfall p and ambient relative humidity RH.

[0080] S4.2. In the insulator contamination flashover discharge model, it is considered that the three parameters of insulator surface contamination degree, rainfall and relative humidity have independent effects on the discharge risk level; a 3D naive Bayes classification model is used as a machine learning model;

[0081] S4.3. For each set of data (d, p, RH), let data X = (d, p, RH), where X1 = d, X2 = p, X3 = RH; then according to Bayes' theorem, we can get

[0082]

[0083] where k = 0, 1, 2, 3; c k = k represents all possible values ​​of R. Considering the independence of each parameter, it can be further simplified to the following prediction function:

[0084]

[0085] Where P(R=c k ) is the frequency of each value of R in all training set data, P(X j =x j |R=c k ) is the number of all training set data when the jth component in X is x j The conditional probability when ;

[0086] S4.4. Use the test set data to test, input the (d, p, RH) of the test set data as the value of X into the prediction function, and calculate the probability of the discharge risk state being no risk, medium risk, and high risk under the conditions of the insulator contamination level, rainfall, and relative humidity, that is, the values ​​of P(R=0|X=x), P(R=1|X=x), P(R=2|X=x), and P(R=3|X=x); c corresponding to the maximum value of the four probabilities k The value is the final R value. The accuracy of the model is tested by the above method. If the accuracy of the test set exceeds 90%, the naive Bayes classification model is considered to be reliable.

[0087] S5. The data processing and control module sends the d value and R value data to the background software via wireless network transmission, and the software displays the current insulator contamination level and discharge risk information.

[0088] The above are only preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The substitution may be a substitution of a part of the structure, device, method step, or a complete technical solution. The equivalent replacement or change according to the technical solution and the inventive concept of the present invention shall be covered within the protection scope of the present invention.

Claims

1. An insulator contamination online monitoring and identification and discharge risk prediction device, characterized in that: include: Mounting frame, information collection module, data processing and control module and power module; The bottom of the mounting frame is fixed on the power pole tower near the position where the insulator is mounted on the power pole tower, an unloaded control insulator is placed on the top of the mounting frame, and a closable light shield is provided on one side of the top of the mounting frame; The information collection module includes a digital rain gauge, a humidity sensor and a hyperspectral probe, which input the collected precipitation, relative humidity and hyperspectral image information into the data processing and control module; the hyperspectral probe is installed inside an openable sunshade; The data processing and control module can control the opening and closing of the closable light shield according to the needs of data collection, and after the closable light shield is closed, control the hyperspectral probe to take pictures, and input the hyperspectral image information into the data processing and control module; the data processing and control module first converts the hyperspectral image information into the surface contamination degree of the insulator, and then combines the surface contamination degree of the insulator, precipitation and relative humidity to comprehensively judge the risk level of discharge of the insulator under the current environmental conditions and contamination degree, and sends the risk level signal of discharge of the insulator to the background software; The power module is electrically connected to the openable light shield, the information collection module and the data processing and control module.

2. The insulator contamination online monitoring identification and discharge risk prediction device according to claim 1, characterized in that: The openable and closable sunshade includes a shielding cover rotatably connected to one side of the top of the mounting frame through a rotating shaft, a large driving wheel fixedly connected to the center position of one side of the shielding cover, a small driving wheel meshing with the large driving wheel, and a motor with an output shaft fixedly connected to the small driving wheel; the shielding cover and the mounting frame form a closed cavity; the axis of the large driving wheel coincides with the outer edge of the shielding cover, the rotating shaft is provided with two groups, and the two groups are symmetrically installed on both sides of the large driving wheel. After the large driving wheel rotates, it penetrates into the accommodating space provided at the bottom of the mounting frame.

3. The insulator contamination online monitoring, identification and discharge risk prediction device according to claim 1 or 2, characterized in that: A cap socket is provided on the top of the mounting frame, and the cap socket is detachably connected to the no-load control insulator.

4. The insulator contamination online monitoring identification and discharge risk prediction device according to claim 2, characterized in that: A buckle for installing a hyperspectral probe is provided on the top of the shielding cover cavity.

5. The insulator contamination online monitoring identification and discharge risk prediction device according to claim 1, characterized in that: The digital rain gauge adopts the ABS tipping bucket rain gauge of Preson.

6. The insulator contamination online monitoring identification and discharge risk prediction device according to claim 1, characterized in that: The power module is a combination of Zhier YXP-60W photovoltaic panel and 20A battery.

7. An insulator contamination online monitoring identification and discharge risk prediction method, using the insulator contamination online monitoring identification and discharge risk prediction device according to any one of claims 1 to 6, characterized in that: The steps include: S1. The device is started, and the staff remotely sets the working cycle T through the background software; S2, the shielding cover is opened to keep the unloaded control insulator in a polluted working state, and the rainfall and relative humidity data are continuously collected through the digital rain gauge and humidity sensor, and the rainfall and relative humidity data are input into the data processing and control module; S3, after a single working cycle T ends, the shield is closed, so that the unloaded control insulator is changed to a photographing working state, and the hyperspectral probe takes pictures. After the photographing is completed, the hyperspectral image information is input into the data processing and control module, and the shield is opened to restore the unloaded control insulator to a contamination working state; S4, the data processing and control module first converts the hyperspectral image information into the surface contamination degree d of the insulator, calculates the average rainfall p and the average relative humidity RH within the period T, and determines the insulator discharge risk R value based on the average rainfall p and the average relative humidity RH, combined with the surface contamination degree d; the determination is made according to the following formula: R=f(d,p,RH) Where f uses a machine learning classification algorithm function, such as a decision tree, a random forest algorithm, SVM, or a naive Bayesian Bayesian model; S5. The data processing and control module sends the d value and R value data to the background software via wireless network transmission, and the software displays the current insulator contamination level and discharge risk information.

8. The method for online monitoring, identification and discharge risk prediction of insulator contamination according to claim 7, characterized in that: The prediction function f of the insulator discharge risk level R in step 4 is obtained by constructing a 3D naive Bayesian network, which includes the following steps: S4.

1. Obtain (d, p, RH, R) data to construct training and test sets; S4.

2. In the insulator contamination flashover discharge model, it is considered that the three parameters of insulator surface contamination degree, rainfall and relative humidity have independent effects on the discharge risk level; a 3D naive Bayes classification model is used as a machine learning model; S4.

3. For each set of data (d, p, RH), let data X = (d, p, RH), where X1 = d, X2 = p, X3 = RH; then according to Bayes' theorem, we can get where k = 0, 1, 2, 3; c k = k represents all possible values ​​of R; considering the independence of each parameter, it can be further simplified to the following prediction function Where P(R=c k ) is the frequency of each value of R in all training set data, P(X j =x j |R=c k ) is the number of all training set data when the jth component in X is x j The conditional probability when ; S4.

4. Use the test set data to test, input the (d, p, RH) of the test set data as the value of X into the prediction function, and calculate the probability of the discharge risk state being no risk, medium risk, and high risk under the conditions of the insulator contamination level, rainfall, and relative humidity, that is, the values ​​of P(R=0|X=x), P(R=1|X=x), P(R=2|X=x), and P(R=3|X=x); c corresponding to the maximum value of the four probabilities k The value is the R value obtained by the final judgment; the accuracy of the model is tested by the above method. If the accuracy of the test set exceeds 90%, it is considered that the naive Bayes classification model is reliable.

9. The insulator contamination online monitoring identification and discharge risk prediction device according to claim 8, characterized in that: Acquiring (d, p, RH, R) data to construct a training set and a test set in step S4.1 includes the following steps: a. Select a string of insulators suitable for a new power tower and apply artificial pollution coating on the insulators. The specific pollution coating method refers to "Insulators for Power Equipment and Their Pollution Levels" (GB / T 15166-2018), and apply five levels of pollution from 0 to IV respectively; correspondingly, the d value representing the pollution level is determined according to the following standards: b. Adjust the humidity and precipitation in the artificial climate chamber, raise the voltage to the standard working voltage of the insulator string and keep it for 1 hour to check whether flashover discharge will occur; repeat the experiment 10 times at each humidity, precipitation and pollution level, record the number of discharges as n, then the discharge probability P is c. Determine the discharge risk level R under the current humidity, precipitation and pollution level conditions according to the value of P. The specific principles are as follows: Among them, when R is 0, 1, 2, and 3, it means that the discharge risk is no risk, low risk, medium risk, and high risk respectively; d. Repeat steps S1 to S3, change the pollution level, relative humidity and precipitation conditions to repeat the discharge experiment, and obtain at least 1000 (d, p, RH, R) data points; half of these data points are used as training sets for the naive Bayes model, and the other half are used as test sets for the recognition effect of the naive Bayes model; after the training is completed, a function f can be obtained for judging the insulator discharge risk level R according to the insulator surface pollution level d, rainfall p and ambient relative humidity RH.

Images