GIS sulfur hexafluoride gas pressure data on-line monitoring system

Abstract

The invention relates to the technical field of sulfur hexafluoride gas pressure monitoring, in particular to a GIS sulfur hexafluoride gas pressure data online monitoring system which comprises a plurality of wireless visual recognition sensors, an edge computing gateway, a displayer and a station end intelligent auxiliary platform. The plurality of wireless visual identification sensors are respectively in communication connection with the edge computing gateway, the station end intelligent auxiliary platform is in communication connection with the edge computing gateway through a web, and the display is connected with the station end intelligent auxiliary platform. The plurality of wireless visual identification sensors are respectively and fixedly arranged on a dial plate of the sulfur hexafluoride density relay so as to acquire an image of the dial plate, and the edge computing gateway analyzes the acquired dial plate image, identifies current gas pressure data of sulfur hexafluoride and sends the current gas pressure data to the wireless visual identification sensors. And the station end intelligent auxiliary platform reads the gas pressure data of sulfur hexafluoride, performs recording and trend analysis, and solves the leakage trend and the annual leakage rate of the sulfur hexafluoride gas.

Description

Technical Field

[0001] The present invention relates to the technical field of sulfur hexafluoride gas pressure monitoring. Specifically, it relates to an on-line monitoring system for GIS sulfur hexafluoride gas pressure data. Background Art

[0002] The leakage of sulfur hexafluoride gas is mainly due to equipment defects caused by damage or aging in parts such as the body of the combined electrical equipment, flange seals, gas meters, and pipelines due to manufacturing or installation process problems. When the sulfur hexafluoride gas pressure is lower than the set value, it will endanger the normal insulation strength of the equipment, and thus pose a danger to the safe and reliable operation of the power system. At present, the defects of sulfur hexafluoride gas leakage in combined electrical equipment are mainly discovered by recording gas pressure data through daily manual inspections and comparing them with the previously recorded data to find a decrease in air pressure, or when the equipment emits an alarm signal. In addition, due to the layout of the equipment structure, the installation positions of most sulfur hexafluoride density relays in gas compartments are relatively high, and there are situations such as the dial being blocked, which easily leads to problems such as inability to read the data or inaccurate readings during manual inspections. Therefore, leakage faults cannot be discovered in a timely manner, posing a hidden danger to the stable operation of the equipment. Based on this, in response to the above problems, we have designed an on-line monitoring system for GIS sulfur hexafluoride gas pressure data. Summary of the Invention

[0003] The purpose of the present invention is to provide an on-line monitoring system for GIS sulfur hexafluoride gas pressure data, which relies on the real-time monitoring of wireless vision recognition sensors to replace the original method of manual inspection and data transcription. By combining an edge computing gateway with a pre-set sulfur hexafluoride gas pressure recognition model to analyze the monitoring data, it can obtain the gas pressure data of sulfur hexafluoride, which not only reduces the manual labor cost, but also ensures the accuracy of the data, is more conducive to trend analysis, and greatly reduces the workload of on-site maintenance personnel.

[0004] The embodiments of the present invention are implemented through the following technical solutions:

[0005] An on-line monitoring system for GIS sulfur hexafluoride gas pressure data includes: a plurality of wireless vision recognition sensors, an edge computing gateway, a display, and a station-side intelligent auxiliary platform; the plurality of wireless vision recognition sensors are respectively communicatively connected to the edge computing gateway, the station-side intelligent auxiliary platform is communicatively connected to the edge computing gateway through the web, and the display is connected to the station-side intelligent auxiliary platform. Among them, the plurality of wireless vision recognition sensors are respectively fixedly installed on the dials of the sulfur hexafluoride density relays to obtain images of the dials. The edge computing gateway analyzes the collected dial images and identifies the current gas pressure data of sulfur hexafluoride. The station-side intelligent auxiliary platform reads the gas pressure data of sulfur hexafluoride and performs recording and trend analysis to solve the leakage trend and annual leakage rate of sulfur hexafluoride gas.

[0006] Optionally, before the edge computing gateway analyzes the collected dial image, it further includes image preprocessing of the dial image, where the image preprocessing includes: denoising, grayscale conversion, and binarization.

[0007] Optionally, a sulfur hexafluoride gas pressure recognition model is preset in the edge computing gateway. The sulfur hexafluoride gas pressure recognition model is specifically based on a convolutional neural network model and consists of an input layer, a convolutional layer, a pooling layer, and a fully connected layer. Among them, the input layer is used to receive the preprocessed dial image, the convolutional layer is used to extract the features of the dial image, the pooling layer is used to reduce the spatial dimension of the feature map, and the fully connected layer is used to output the final sulfur hexafluoride gas pressure data.

[0008] Optionally, the calculation formula of the convolutional layer is:

[0009]

[0010] where f(x, y) is the pixel value of the output feature map, g(x, y) is the pixel value of the input dial image, k(i, j) is the weight of the convolutional kernel, and a and b are the ranges of the convolutional kernel respectively;

[0011] The calculation formula of the pooling layer is:

[0012] B(u, v) = max(A(s * (u - 1) + a, t * (v - 1) + b))

[0013] where u and v are the row index and column index of the output feature map B respectively, s and t are the step sizes in the vertical and horizontal directions respectively, and 1 ≤ a ≤ p, 1 ≤ b ≤ q traverse the p×q pooling window;

[0014] The calculation formula of the fully connected layer is:

[0015]

[0016] where x is the input vector, W′ j , W k are the weight matrices of different layers respectively, b′ j , b k are the bias vectors of different layers respectively, σ is the activation function, α k , β j are the scaling factors used to adjust the influence of the outputs of different layers, K and J are the numbers of neurons in different layers respectively, and y is the sulfur hexafluoride gas pressure data finally output by the sulfur hexafluoride gas pressure recognition model.

[0017] Optionally, the training process of the sulfur hexafluoride gas pressure recognition model is:

[0018] Input the dial image dataset of the historical sulfur hexafluoride density relay;

[0019] After preprocessing, divide the historical dial image dataset into a training set and a test set;

[0020] Input the training set into the sulfur hexafluoride gas pressure recognition model for training, evaluate the performance of the sulfur hexafluoride gas pressure recognition model through the test set, and iteratively update the parameters of the sulfur hexafluoride gas pressure recognition model through the SGD optimizer until the iteration reaches the set number of times, and then output the trained sulfur hexafluoride gas pressure recognition model.

[0021] Optionally, when evaluating the performance of the sulfur hexafluoride gas pressure recognition model through the test set, the calculation formula is:

[0022]

[0023] Among them, N is the number of categories, y c is the true label, is the predicted output of the sulfur hexafluoride gas pressure recognition model, and L is the loss function.

[0024] Optionally, when iteratively updating the weights of the sulfur hexafluoride gas pressure recognition model through the SGD optimizer, the calculation formula is:

[0025]

[0026] Among them, δ is the parameter of the sulfur hexafluoride gas pressure recognition model before update, δ' is the parameter of the sulfur hexafluoride gas pressure recognition model after update, η is the learning rate, is the gradient of the loss function L with respect to the parameter of the sulfur hexafluoride gas pressure recognition model.

[0027] Optionally, solve the leakage trend and annual leakage rate of the sulfur hexafluoride gas. Among them, the calculation formula for the annual leakage rate is:

[0028]

[0029] Among them, F y is the annual leakage rate, m 1 , m 2 are the relative gauge pressures before and after the pressure drop respectively, and ΔT is the time interval for the gas to leak to m 2 .

[0030] The technical solution of the embodiment of the present invention has at least the following advantages and beneficial effects:

[0031] In the embodiments of the present invention, relying on the real-time monitoring of wireless vision recognition sensors to replace the original manual inspection and data transcription method, and combining with an edge computing gateway preset with a sulfur hexafluoride gas pressure recognition model to analyze the monitoring data, the gas pressure data of sulfur hexafluoride can be obtained, which not only reduces the labor cost but also ensures the accuracy of the data, is more conducive to trend analysis, and greatly reduces the workload of on-site maintenance personnel. Brief Description of the Drawings

[0032] Figure 1 It is a system schematic diagram of an on-line monitoring system for GIS sulfur hexafluoride gas pressure data provided by an embodiment of the present invention. Detailed Embodiments

[0033] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations.

[0034] As Figure 1 shown, the present invention provides one of the embodiments: an on-line monitoring system for GIS sulfur hexafluoride gas pressure data, including: a plurality of wireless vision recognition sensors, an edge computing gateway, a display, and a station-side intelligent auxiliary platform; the plurality of wireless vision recognition sensors are respectively communicatively connected to the edge computing gateway, the station-side intelligent auxiliary platform is communicatively connected to the edge computing gateway through the web, and the display is connected to the station-side intelligent auxiliary platform. Among them, the plurality of wireless vision recognition sensors are respectively fixedly installed on the dial of the sulfur hexafluoride density relay to obtain the image of the dial. The edge computing gateway analyzes the collected dial image and identifies the current gas pressure data of sulfur hexafluoride. The station-side intelligent auxiliary platform reads the gas pressure data of sulfur hexafluoride and performs recording and trend analysis to solve the leakage trend and annual leakage rate of sulfur hexafluoride gas.

[0035] In this embodiment, this embodiment specifically includes a wireless vision recognition sensor, an edge computing gateway, a display, and a station-side intelligent assistance platform. The wireless vision recognition sensor installed on the dial of the equipment SF6 density relay takes pictures of the dial according to instructions and transmits them to the edge computing gateway. The edge computing gateway analyzes the collected photos to identify the current SF6 gas pressure data. The station-side intelligent assistance platform accesses the edge computing gateway through the web, reads the SF6 gas pressure data, records and trends the data, and calculates the SF6 leakage trend and annual leakage rate. The system sets an early warning trigger value. When the recognized data reaches the trigger value, the system generates an early warning signal and pushes it to the server, thereby realizing alarms for low pressure and excessive leakage rate.

[0036] Specifically, through statistical analysis, the pressure of the SF6 equipment gas chamber basically does not decrease under normal operating conditions. Therefore, the system sets the low alarm set value of the gas chamber SF6 pressure to the rated pressure value. When the system issues a low air pressure alarm, in the case of checking that there is no leakage in the gas chamber, the gas chamber can be refilled in combination with the unit maintenance later. The system calculates the annual gas leakage rate using the pressure drop method and sets the alarm threshold to 0.5% according to the standard.

[0037] More specifically, before the edge computing gateway analyzes the collected dial image, it also includes image preprocessing of the dial image. Among them, the image preprocessing includes: denoising, grayscale conversion, and binarization.

[0038] In this embodiment, a SF6 gas pressure recognition model is preset in the edge computing gateway. The SF6 gas pressure recognition model is specifically based on a convolutional neural network model and consists of an input layer, a convolutional layer, a pooling layer, and a fully connected layer. Among them, the input layer is used to receive the preprocessed dial image, the convolutional layer is used to extract the features of the dial image, the pooling layer is used to reduce the spatial dimension of the feature map, and the fully connected layer is used to output the final SF6 gas pressure data.

[0039] Among them, the calculation formula of the convolutional layer is:

[0040]

[0041] Among them, f(x, y) is the pixel value of the output feature map, g(x, y) is the pixel value of the input dial image, k(i, j) is the weight of the convolutional kernel, and a and b are the ranges of the convolutional kernel respectively;

[0042] The calculation formula of the pooling layer is:

[0043] B(u, v) = max(A(s*(u - 1) + a, t*(v - 1) + b))

[0044] Among them, u and v are respectively the row index and column index of the output feature map B, s and t are respectively the strides in the vertical and horizontal directions, and 1 ≤ a ≤ p, 1 ≤ b ≤ q traverse the pooling window of p×q;

[0045] The calculation formula of the fully connected layer is:

[0046]

[0047] Among them, x is the input vector, W′ j , W k are the weight matrices of different layers respectively, b′ j , b k are the bias vectors of different layers respectively, σ is the activation function, α k , β j are the scaling factors respectively, used to adjust the influence of the outputs of different layers, K and J are the numbers of neurons in different layers respectively, and y is the sulfur hexafluoride gas pressure data finally output by the sulfur hexafluoride gas pressure recognition model.

[0048] In the specific application of this embodiment, the training process of the sulfur hexafluoride gas pressure recognition model is as follows:

[0049] Input the dial image dataset of the historical sulfur hexafluoride density relay;

[0050] After preprocessing, divide the historical dial image dataset into a training set and a test set;

[0051] Input the training set into the sulfur hexafluoride gas pressure recognition model for training, evaluate the performance of the sulfur hexafluoride gas pressure recognition model through the test set, and iteratively update the parameters of the sulfur hexafluoride gas pressure recognition model through the SGD optimizer until the iteration reaches the set number of times, and then output the trained sulfur hexafluoride gas pressure recognition model.

[0052] The formula for evaluating the performance of the sulfur hexafluoride gas pressure recognition model through the test set is:

[0053]

[0054] Among them, N is the number of categories, y c is the true label, is the predicted output of the sulfur hexafluoride gas pressure recognition model, and L is the loss function.

[0055] Specifically, the formula for iteratively updating the weights of the sulfur hexafluoride gas pressure recognition model through the SGD optimizer is:

[0056]

[0057] Among them, δ is the parameter of the sulfur hexafluoride gas pressure recognition model before update, δ' is the parameter of the sulfur hexafluoride gas pressure recognition model after update, η is the learning rate, is the gradient of the loss function L with respect to the parameters of the sulfur hexafluoride gas pressure recognition model.

[0058] In this embodiment, the definition of the pressure drop method and its existing approximate calculation errors may cause certain errors. Therefore, the system stores the recorded dial pressure data, draws a curve graph for trend analysis. For some gas chambers with minor leaks where obvious leak points cannot be found and the pressure has been slowly decreasing, and for equipment where the leakage rate calculated by the system is at the edge of exceeding and not exceeding the annual leakage rate standard, it can assist in judgment, thereby avoiding the influence brought by calculation errors, that is, the following calculation formula:

[0059]

[0060] Among them, F y is the annual leakage rate, m 1 , m 2 are the relative gauge pressures before and after the pressure drop, in MPa, ΔT is the time interval from gas leakage to m 2 , in months.

[0061] Furthermore, before the system runs, it is necessary to calibrate m 1 for each dial. It should be noted that under different climatic conditions and different loads, the actual gas pressure value of the equipment will change accordingly. Therefore, when calibrating the relative pressure m 1 before the pressure drop, the lowest air pressure value obtained from inspections at different times within 1 day should be used for calibration. In addition, if gas replenishment operation is performed on a certain gas chamber, it is necessary to re - calibrate m 1 . In addition, to avoid the influence brought by long - term gas pressure changes in the gas chamber, it is necessary to re - calibrate m 1 every two years.

[0062] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A GIS sulfur hexafluoride gas pressure data online monitoring system, characterized in that: include: Multiple wireless visual recognition sensors, edge computing gateways, displays, and station-side intelligent assistance platforms; The multiple wireless visual recognition sensors are respectively communicated with the edge computing gateway, the station-side intelligent auxiliary platform is communicated with the edge computing gateway via the web, and the display is connected to the station-side intelligent auxiliary platform, wherein the multiple wireless visual recognition sensors are respectively fixedly installed on the dial of the sulfur hexafluoride density relay to obtain the image of the dial, the edge computing gateway analyzes the collected dial image and identifies the current sulfur hexafluoride gas pressure data, the station-side intelligent auxiliary platform reads the sulfur hexafluoride gas pressure data, and performs recording and trend analysis to solve the leakage trend and annual leakage rate of sulfur hexafluoride gas.

2. The GIS sulfur hexafluoride gas pressure data online monitoring system according to claim 1 is characterized in that: Before the edge computing gateway analyzes the collected dial image, it also includes image preprocessing on the dial image, wherein the image preprocessing includes: denoising, grayscale and binarization.

3. The GIS sulfur hexafluoride gas pressure data online monitoring system according to claim 2 is characterized in that: A sulfur hexafluoride gas pressure recognition model is preset in the edge computing gateway. The sulfur hexafluoride gas pressure recognition model is specifically based on a convolutional neural network model, and is composed of an input layer, a convolutional layer, a pooling layer, and a fully connected layer, wherein the input layer is used to receive a preprocessed dial image, the convolutional layer is used to extract features of the dial image, the pooling layer is used to reduce the spatial dimension of the feature map, and the fully connected layer is used to output the final sulfur hexafluoride gas pressure data.

4. The GIS sulfur hexafluoride gas pressure data online monitoring system according to claim 3 is characterized in that: The calculation formula of the convolutional layer is: Among them, f(x,y) is the pixel value of the output feature map, g(x,y) is the pixel value of the input dial image, k(i,j) is the weight of the convolution kernel, and a and b are the ranges of the convolution kernel respectively; The calculation formula of the pooling layer is: B(u,v)=max(A(s*(u-1)+a,t*(v-1)+b)) Among them, u and v are the row index and column index of the output feature map B, s and t are the steps in the vertical and horizontal directions, 1≤a≤p, 1≤b≤q traverse the pooling window of p×q; The calculation formula of the fully connected layer is: Where x is the input vector, W j ′、W k are the weight matrices of different layers, b′ j 、b k are the bias vectors of different layers, σ is the activation function, α k , β j They are scaling factors used to adjust the influence of outputs of different layers, K and J are the number of neurons in different layers, and y is the sulfur hexafluoride gas pressure data finally output by the sulfur hexafluoride gas pressure recognition model.

5. The GIS sulfur hexafluoride gas pressure data online monitoring system according to claim 4 is characterized in that: The training process of the sulfur hexafluoride gas pressure recognition model is as follows: Input the dial image dataset of the historical sulfur hexafluoride density relay; After preprocessing, the historical dial image dataset is divided into a training set and a test set; The training set is input into the sulfur hexafluoride gas pressure recognition model for training, the performance of the sulfur hexafluoride gas pressure recognition model is evaluated through the test set, and the parameters of the sulfur hexafluoride gas pressure recognition model are iteratively updated through the SGD optimizer until the set number of iterations is reached, and the trained sulfur hexafluoride gas pressure recognition model is output.

6. The GIS sulfur hexafluoride gas pressure data online monitoring system according to claim 5 is characterized in that: The performance of the sulfur hexafluoride gas pressure recognition model is evaluated by the test set, and the calculation formula is: Where N is the number of categories, y c is the true label, is the predicted output of the sulfur hexafluoride gas pressure identification model, and L is the loss function.

7. The GIS sulfur hexafluoride gas pressure data online monitoring system according to claim 6, characterized in that: The weight of the sulfur hexafluoride gas pressure identification model is iteratively updated by the SGD optimizer, and the calculation formula is: Among them, δ is the sulfur hexafluoride gas pressure identification model parameter before updating, δ' is the sulfur hexafluoride gas pressure identification model parameter after updating, η is the learning rate, is the gradient of the loss function L with respect to the parameters of the sulfur hexafluoride gas pressure identification model.

8. The GIS sulfur hexafluoride gas pressure data online monitoring system according to claim 7 is characterized in that: The leakage trend and annual leakage rate of sulfur hexafluoride gas are solved, wherein the calculation formula of the annual leakage rate is: Among them, F y is the annual leakage rate, m1 and m2 are the relative gauge pressures before and after the pressure drop, and ΔT is the time interval from leakage to m2.

Images