Fault detection device for power transformation equipment and use method

Abstract

The invention relates to a fault detection device for power transformation equipment, which comprises a terminal and a mobile terminal, and is characterized in that the terminal adopts a power supply module as a power supply and comprises a multi-target fault detection node, a data processing unit, a thermal infrared imager and a data mining module; the multi-target fault detection node comprises at least three infrared detectors, and the at least three infrared detectors are arranged on at least three positions of the periphery of detected power transformation equipment in a linear array; each infrared detector is connected with the thermal infrared imager; each infrared detector is also connected with the data mining module; the thermal infrared imager and the data mining module send data and information to the data processing unit; the method is small in noise influence, employs multiple targets to expand the detection range, improves the fault detection accuracy, and introduces the index analysis model to quantitatively estimate the fault rate.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system fault detection, and in particular relates to a fault detection device for substation equipment and a use method thereof. Background Art

[0002] Substations are the basic units for the production and operation of power systems. If the potential safety hazards of substation equipment turn into faults, the relay protection device will automatically trip to protect the faulty line. In this case, it may cause small or even large-scale power outages, causing inconvenience to people's lives and property losses. Inside the substation, various substation device components work continuously for a long time and are in contact with the external environment. They are easily disturbed by environmental factors, resulting in poor contact, cracks, aging, oxidation and other faults. Traditional substation equipment operation fault detection forms are mostly single structures. Although they can achieve the expected fault detection goals and tasks, they are easily affected by the external environment and specific factors, resulting in large errors in the final detection results. It is difficult to accurately control each detection link and adjust the detection plan. In addition, with the deepening of understanding of equipment and detection With the advancement of means, equipment maintenance is no longer limited to judging the equipment failure rate based on the operating years. Instead, it is necessary to infer the possibility of equipment failure based on the actual status of the equipment, and then formulate a corresponding maintenance plan. This work is very difficult. The main reasons are: first, there are few statistical samples, and it is almost difficult to provide sufficient correlation data between equipment status and equipment failure on site, making statistical analysis difficult; second, equipment safety is placed in an extremely important position in the power system. Once a defect occurs in the equipment, measures are often taken immediately to eliminate the defect, which reduces the credibility of the correlation data; therefore, it is very necessary to provide a fault detection device and a method for use of substation equipment that is less affected by noise, uses multiple targets to expand the detection range, improves fault detection accuracy, and introduces an exponential analysis model to quantitatively estimate the failure rate. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art and to provide a fault detection device and a method for use of substation equipment that is less affected by noise, uses multiple targets to expand the detection range, improves fault detection accuracy, and introduces an exponential analysis model to quantitatively estimate the fault rate.

[0004] The objective of the present invention is achieved as follows: a fault detection device for substation equipment, comprising a terminal and a mobile terminal, wherein the terminal adopts a power supply module as a power supply, and the terminal comprises a multi-target fault detection node, a data processing unit, an infrared thermal imager and a data mining module; the multi-target fault detection node comprises at least three infrared detectors, and the at least three infrared detectors are arranged in a linear array at at least three positions on the periphery of the substation equipment to be detected; each of the infrared detectors is connected to the infrared thermal imager, and is used to collect infrared radiation data of the substation equipment in real time, and convert it into an electrical signal to generate a thermal infrared image of the substation equipment; each of the infrared detectors is also connected to the data mining module, and is used to classify and screen the data and information collected in real time from the substation equipment, complete quantitative processing, and divide the data into variable and non-variable quantities, and calibrate the location where the abnormality occurs according to the fault data; the infrared thermal imager and the data mining module send the data and information to the data processing unit.

[0005] The data processing unit includes a data preprocessing module, a data mining fault detection module, an adaptive fault detection module and a signal transmission module; the data preprocessing module is connected to the infrared thermal imager and the data mining module respectively, the data preprocessing module is connected to the data mining fault detection module, the data mining fault detection module is connected to the adaptive fault detection module, and the adaptive fault detection module is connected to the signal transmission module.

[0006] The data processing unit may also be connected to an external display terminal, and the display terminal is used to display the current fault status of the substation equipment.

[0007] A fault detection device for substation equipment also includes a target detection algorithm module A and a target detection algorithm module B, wherein the target detection algorithm module A is connected to a terminal; and the target detection algorithm module B is connected to a mobile terminal.

[0008] A fault detection device for substation equipment also includes a thermal fault intelligent diagnosis algorithm module and an equipment fault estimation algorithm module. The thermal fault intelligent diagnosis algorithm module is connected to a target detection algorithm module A and a target detection algorithm module B respectively, and the equipment fault estimation algorithm module is connected to the thermal fault intelligent diagnosis algorithm module.

[0009] A method for using a fault detection device for a power substation, using the fault detection device for a power substation as described above, comprises the following steps:

[0010] Step 1: Expand the fault detection coverage by adopting a multi-target approach, thereby deploying multi-target fault detection nodes;

[0011] Step 2: Use a linear array of infrared detectors to determine the scanning area of ​​the substation equipment;

[0012] Step 3: The data mining module collects and preprocesses fault data based on the requirements and standards of fault detection; the infrared thermal imager converts the infrared radiation energy of the substation equipment into electrical signals and collects images of the substation equipment;

[0013] Step 4: The data processing module of the data processing unit calculates the pixel points of the infrared image through the probability density function, and pre-processes the infrared image of the substation equipment using the median filtering principle;

[0014] Step 5: Implementing fault detection of the substation equipment according to the relative temperature difference of the fault point of the substation equipment;

[0015] Step 6: The data mining fault detection module establishes a data mining fault detection model, and completes fault detection by using an adaptive fault detection module using an adaptive recognition processing method;

[0016] Step 7: The data processing unit fuses the above two fault detection data to obtain fault detection and diagnosis information, and sends the fault detection and diagnosis information to the display terminal through the signal transmission module to display the fault status of the current substation equipment;

[0017] Step 8: The target detection algorithm module A adopts the target detection method of infrared images of power equipment based on the improved YOLOv4 model to achieve accurate positioning and classification of power equipment;

[0018] Step 9: The target detection algorithm module B adopts the target detection method of infrared images of power equipment based on the improved PP-YOLO--Tiny lightweight model to realize the target detection and diagnosis of power equipment on the mobile terminal;

[0019] Step 10: The thermal fault intelligent diagnosis algorithm module combines the target detection algorithm module A and the target detection algorithm module B, and adopts a temperature extraction method based on colorimetric strip analysis to realize intelligent diagnosis of thermal faults of substation equipment;

[0020] Step 11: The equipment failure estimation algorithm module introduces an index analysis model based on the substation equipment status information of the thermal fault intelligent diagnosis algorithm module, associates the equipment failure rate with the equipment status, and realizes quantitative estimation of the equipment failure rate;

[0021] Step 12: This enables the possibility of equipment failure within a certain evaluation period to be determined based on the current status of the equipment.

[0022] The step 2 is specifically as follows: using a linear array of infrared detectors to identify the scanning area of ​​the substation equipment in different ways. Assuming that the linear array of infrared detectors can divide the scanning space of the substation equipment into n units, the frame images of each unit are scanned one by one in a certain order. The calculation formula of the unit scanning time is: Where, T p Indicates the scanning period; f p Indicates the frame rate of the scan.

[0023] The step 4 is specifically as follows: filtering the infrared image of the substation equipment using a filtering algorithm. Assuming that there are 6 grayscale images in the image, the pixel points of the infrared image are calculated using a probability density function. The calculation formula is: In the formula, a represents the black noise in the infrared image; b represents the white noise in the infrared image; u a Indicates the probability corresponding to the black noise point; u b Indicates the probability corresponding to the white noise point;

[0024] The median filter is introduced to sort the window range points of the infrared image of the substation equipment, and the gray value of the infrared image is used as the middle value of the sequence center. The extraction formula is: R(i,j)=median[g(k)], where g represents the gray value sequence of the infrared image; k represents the number of pixels in the infrared image;

[0025] The median filter is composed of the R trigger level signal and the W trigger level signal, which can be expressed as: In the formula, H med Represents the grayscale median of the infrared image; H xy Represents the coordinate gray value of the infrared image; H min Indicates the minimum gray value of the infrared image; H max Indicates the maximum gray value of the infrared image.

[0026] The step 5 is specifically: calculating the relative temperature difference at the fault point of the substation equipment, the formula is: In the formula, T1 represents the temperature of the heating point of the substation equipment; T0 represents the ambient temperature of the substation equipment; T2 represents the standard temperature in the environment; the fault detection of the substation equipment is realized according to the calculation result of the relative temperature difference at the fault point.

[0027] The step 11 is specifically: the correlation between the equipment status and the equipment failure rate, and its approximate model can be expressed as: p = ke cH , where p is the current failure rate of the equipment; k and c are unknown constants; and H is the equipment state parameter.

[0028] Beneficial effects of the present invention: The present invention is a fault detection device for substation equipment and a method for using the device. In use, the present invention adopts a multi-target form to expand the coverage of fault detection, so as to deploy multi-target fault detection nodes, and use a linear array of infrared detectors to collect infrared images of substation equipment and pre-process them to achieve fault detection of substation equipment; the data mining module is combined with the needs and standards of fault detection to collect and pre-process fault data, and then a data mining fault detection model is established, and the fault detection is completed by adaptive recognition processing; the fault detection information is integrated to effectively improve the accuracy of fault detection of substation equipment, and the fault of substation equipment can be accurately detected so that the staff can go to the site for maintenance in time to ensure the safe and stable operation of the substation equipment; the present invention also expands the fault identification range of the equipment, can accurately lock the location where the abnormality or fault occurs, strengthen the control of detection errors, and reduce the false recognition rate; in addition, the present invention introduces an index analysis model to associate the equipment failure rate with the equipment state, thereby providing a quick method for quantitatively estimating the equipment failure rate, and realizing the possibility of equipment failure within a certain evaluation period according to the current state of the equipment; the present invention has the advantages of being less affected by noise, using multiple targets to expand the detection range, improving the accuracy of fault detection, and introducing an index analysis model to quantitatively estimate the failure rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a structural diagram of the present invention.

[0030] Figure 2 It is a flow chart of the present invention.

[0031] Figure 3 This is a flow chart of the multi-objective fault detection node deployment of the present invention.

[0032] Figure 4 It is a schematic diagram of the data mining fault detection model of the present invention.

[0033] Figure 5 Schematic diagram of the network structure of the improved YOLOv4 model of the present invention.

[0034] Figure 6 It is a schematic diagram of the SPP-CSP structure and characteristic diagram size change of the present invention.

[0035] Figure 7 Schematic diagram of the network structure of the improved PP-YOLO-Tiny model of the present invention.

[0036] Figure 8 It is a schematic diagram of the LCNet backbone network structure of the present invention. DETAILED DESCRIPTION

[0037] The present invention will be further described below in conjunction with the accompanying drawings.

[0038] Example 1

[0039] like Figure 1-8 As shown, a fault detection device for substation equipment includes a terminal and a mobile terminal. The terminal uses a power supply module as a power supply. The terminal includes a multi-target fault detection node, a data processing unit, an infrared thermal imager and a data mining module. The multi-target fault detection node includes at least three infrared detectors, and the at least three infrared detectors are arranged in a linear array at at least three positions on the periphery of the substation to be detected. Each of the infrared detectors is connected to the infrared thermal imager to collect infrared radiation data of the substation equipment in real time, convert it into an electrical signal, and generate a thermal infrared image of the substation equipment. Each of the infrared detectors is also connected to the data mining module to classify and screen the data and information collected from the substation equipment in real time, complete quantitative processing, and divide the data into variable and non-variable quantities, and calibrate the location where the abnormality occurs according to the fault data. The infrared thermal imager and the data mining module send the data and information to the data processing unit.

[0040] The data processing unit includes a data preprocessing module, a data mining fault detection module, an adaptive fault detection module and a signal transmission module; the data preprocessing module is connected to the infrared thermal imager and the data mining module respectively, the data preprocessing module is connected to the data mining fault detection module, the data mining fault detection module is connected to the adaptive fault detection module, and the adaptive fault detection module is connected to the signal transmission module.

[0041] The data processing unit may also be connected to an external display terminal, and the display terminal is used to display the current fault status of the substation equipment.

[0042] A fault detection device for substation equipment also includes a target detection algorithm module A and a target detection algorithm module B, wherein the target detection algorithm module A is connected to a terminal; and the target detection algorithm module B is connected to a mobile terminal.

[0043] A fault detection device for substation equipment also includes a thermal fault intelligent diagnosis algorithm module and an equipment fault estimation algorithm module. The thermal fault intelligent diagnosis algorithm module is connected to a target detection algorithm module A and a target detection algorithm module B respectively, and the equipment fault estimation algorithm module is connected to the thermal fault intelligent diagnosis algorithm module.

[0044] The present invention is a fault detection device for substation equipment and a method for using the device. In use, the present invention adopts a multi-target form to expand the coverage of fault detection, so as to deploy multi-target fault detection nodes, and uses a linear array of infrared detectors to collect infrared images of substation equipment and pre-process them to realize fault detection of substation equipment; a data mining module is also used to combine the needs and standards of fault detection, to collect and pre-process fault data, and then a data mining fault detection model is established, and fault detection is completed in an adaptive recognition processing manner; the fault detection information is integrated to effectively improve the fault detection accuracy of substation equipment, and the fault of substation equipment can be accurately detected, so that the staff can go to the site for repair in time to ensure the safe and stable operation of the substation equipment; the present invention also expands the fault recognition range of the equipment, can accurately lock the location where the abnormality or fault occurs, strengthen the control of detection errors, and reduce the false recognition rate; in addition, the present invention introduces an index analysis model to associate the equipment failure rate with the equipment state, thereby providing a quick equipment failure rate quantitative estimation method, and realizing the possibility of equipment failure within a certain evaluation period according to the current state of the equipment; the present invention has the advantages of being less affected by noise, adopting multiple targets to expand the detection range, improving the accuracy of fault detection, and introducing an index analysis model to quantitatively estimate the fault rate.

[0045] Example 2

[0046] like Figure 1-8 As shown, a method for using a fault detection device for a substation equipment, using the above-mentioned fault detection device for a substation equipment, comprises the following steps:

[0047] Step 1: Expand the fault detection coverage by adopting multi-target method to deploy multi-target fault detection nodes. The deployment process of multi-target fault detection nodes is as follows: Figure 3 As shown, combined with mining technology, the detection range of each location node is clarified, and the multi-dimensional selection of control parameters is completed;

[0048] Step 2: Use a linear array of infrared detectors to determine the scanning area of ​​the substation equipment;

[0049] In this embodiment, specifically: a linear array infrared detector is used to identify the scanning area of ​​the substation equipment in different ways. It is assumed that the linear array infrared detector can divide the scanning space of the substation equipment into n units. In a certain order, the frame image of each unit is scanned one by one. The calculation formula of the unit scanning time is: Where, T p Indicates the scanning period; f pIndicates the frame rate of the scan; use an infrared detector to scan the substation equipment, obtain the infrared radiation of the substation equipment, and convert the radiation into electrical signals; if the substation equipment is scanned under the same scanning cycle, the signal-to-noise ratio of the infrared image of the substation equipment will be increased, and a high-quality substation equipment image will be obtained.

[0050] Step 3: The data mining module collects and preprocesses fault data based on the requirements and standards of fault detection; the infrared thermal imager converts the infrared radiation energy of the substation equipment into electrical signals and collects images of the substation equipment;

[0051] Step 4: The data processing module of the data processing unit calculates the pixel points of the infrared image through the probability density function, and pre-processes the infrared image of the substation equipment using the median filtering principle;

[0052] In this embodiment, specifically: when collecting infrared images of substation equipment, the ambient temperature in the air will affect the signal-to-noise ratio of the image, resulting in the inability to effectively extract features from the infrared images of substation equipment, so the infrared images of substation equipment must be preprocessed;

[0053] The filtering algorithm is used to filter the infrared image of the substation equipment. Assuming that there are 6 grayscale images in the image, the pixel points of the infrared image are calculated by the probability density function. The calculation formula is: In the formula, a represents the black noise in the infrared image; b represents the white noise in the infrared image; u a Indicates the probability corresponding to the black noise point; u b Indicates the probability corresponding to the white noise point;

[0054] The median filter is introduced to sort the window range points of the infrared image of the substation equipment, and the gray value of the infrared image is used as the middle value of the sequence center. The extraction formula is: R(i,j)=median[g(k)], where g represents the gray value sequence of the infrared image; k represents the number of pixels in the infrared image;

[0055] The median filter is composed of the R trigger level signal and the W trigger level signal, which can be expressed as: In the formula, H med Represents the grayscale median of the infrared image; H xy Represents the coordinate gray value of the infrared image; H min Indicates the minimum gray value of the infrared image; H max Indicates the maximum gray value of the infrared image.

[0056] When R1>0 and R2<0, the R trigger level signal cannot be formed. Level Convert to W Level Or expand the filter window C xy When W1>0 and W2<0, Hxy As the filter window C xy Output value, or H med As the filter window C xy The output value of the substation equipment is obtained; the median filtering algorithm is used to pre-process the infrared image of the substation equipment to provide a basis for fault detection of the substation equipment.

[0057] Step 5: Implementing fault detection of the substation equipment according to the relative temperature difference of the fault point of the substation equipment;

[0058] In this embodiment, specifically: infrared thermal imaging is based on the overall temperature difference of the infrared image of the substation equipment, by comparing it with the normally operating substation equipment, to determine the location of the fault point, excluding the influence of factors such as the environment and the operation of the substation equipment, and calculating the relative temperature difference of the fault point of the substation equipment. The formula is: In the formula, T1 represents the temperature of the heating point of the substation equipment; T0 represents the ambient temperature of the substation equipment; T2 represents the standard temperature in the environment; the fault detection of the substation equipment is realized according to the calculation result of the relative temperature difference at the fault point.

[0059] Step 6: The data mining fault detection module establishes a data mining fault detection model, and completes fault detection by using an adaptive fault detection module using an adaptive recognition processing method;

[0060] In this embodiment, the deployed multi-objective fault detection nodes are used to collect abnormal data and information in real time, and the data mining technology is integrated to construct the fault detection model of the substation equipment; the fault detection range of the equipment can generally be divided into two parts: the time domain and the frequency domain. The time domain is generally used to identify, detect and calibrate faults in different time backgrounds, while the frequency domain is directly related to the operating frequency of the primary equipment. It is necessary to use the acquired basic data to calculate the fault identification limit difference and set it as the fault detection benchmark value of the model. The fault detection model based on data mining is as follows Figure 4 As shown;

[0061] After completing the establishment of the data mining fault detection model, the adaptive recognition method is used to complete the fault detection processing of the substation equipment; the adaptive recognition mainly corrects the fuzzy detection results output by the model, eliminates the existing detection errors, and ensures that the final fault detection results are true and reliable.

[0062] Step 7: The data processing unit fuses the above two fault detection data to obtain fault detection and diagnosis information, and sends the fault detection and diagnosis information to the display terminal through the signal transmission module to display the fault status of the current substation equipment;

[0063] Step 8: The target detection algorithm module A adopts the target detection method of infrared images of power equipment based on the improved YOLOv4 model to achieve accurate positioning and classification of power equipment;

[0064] In this embodiment, the improved YOLOv4 model is as follows Figure 5 As shown in Figure 2, the CSP structure is introduced into the SPP module at the neck of the network to improve the feature extraction ability of the model; the first CBM layer in the SPP-CSP structure is the last layer of the CSPDarkNet-53 backbone network (the structure and feature map size changes of the SPP-CSP module are shown in Figure 2). Figure 6 As shown in the figure); the NMS algorithm can filter out a large number of overlapping prediction boxes and retain the optimal prediction box; in the post-processing part of the model, the original NMS algorithm is replaced by Matrix NMS; Soft NMS is implemented by parallel computing, which can not only improve the detection accuracy, but also does not cause a decrease in detection speed;

[0065] When there are almost overlapping objects of the same category in the image, the classic NMS algorithm may remove the prediction box with a low confidence score. To solve this problem, the Sof NMS algorithm introduces a penalty factor, taking into account the degree of overlap between the confidence score and the border. This soft filtering method reduces the probability of overlapping objects of the same category being removed. The formula is: s k =s j ×decay j , where IoU i,j represents the intersection-over-union ratio of prediction boxes i and j; s represents the confidence score of the prediction box; f(x) is the Gaussian function; decay is the decay coefficient; s k is the updated confidence score; the confidence of the prediction box greater than the threshold is not set to zero, but a penalty factor is added so that the confidence score is continuously updated during each screening, and only the prediction box less than the confidence threshold is finally eliminated; Matrix NMS reduces the time consumed by the Sof NMS algorithm by parallel computing IoU.

[0066] Step 9: The target detection algorithm module B adopts the target detection method of infrared images of power equipment based on the improved PP-YOLO--Tiny lightweight model to realize the target detection and diagnosis of power equipment on the mobile terminal;

[0067] In this embodiment, based on PPYOLO-Tiny, the LCNet-1.0x backbone network is used as a new backbone network to improve the feature extraction ability of the model; the neck adopts PP-YOLO-Tiny-FPN, and the LRelu activation function is replaced by the H-Swish activation function to improve the expression ability of the final feature map without affecting the detection speed. The overall structure is as follows: Figure 7 As shown;

[0068] The LCNet backbone network consists of one ordinary convolution and five depth-wise separable convolutions, and SE modules are added to the last two depth-wise separable convolutions, as shown in Figure 8 As shown in the figure; the LCNet backbone network also uses depthwise separable convolution and H-Swish activation function; when using depthwise separable convolution, the LCNet backbone network does not add residual shortcut operations, so there are no operations such as "add"; due to the existence of the fully connected layer, the SE module will increase the inference time of the model, so the LCNet backbone network only places the SE module in the last two layers of the network, which can bring a good balance between accuracy and speed.

[0069] Step 10: The thermal fault intelligent diagnosis algorithm module combines the target detection algorithm module A and the target detection algorithm module B, and adopts a temperature extraction method based on colorimetric strip analysis to realize intelligent diagnosis of thermal faults of substation equipment;

[0070] In this embodiment, the matrix data obtained by the infrared thermal imager contains the temperature information of the object, which is converted from the temperature matrix to the RGB infrared image through pseudo-color transformation; the present invention takes the colorimetric bar in the infrared image as the research object, and obtains the function curve that can represent the relationship between the gray value and the temperature value of the pixel point in the infrared image by fitting the function;

[0071] The temperature values ​​within the upper and lower limits of the colorimetric bar are evenly distributed to each pixel in the middle column, and the functional relationship between the two is fitted to obtain the functional relationship between the grayscale value and the temperature value of the pixel; in addition, since the upper and lower limits of temperature in infrared images are often different, the temperature value array and the grayscale value array are normalized respectively, so that the fitting function can be applied to different upper and lower limits of temperature in infrared images; in the process of fitting the function, the grayscale value is used as the independent variable and the temperature value is used as the dependent variable; since the data distribution shows a linear trend, a linear function is selected for fitting, and the fitting function and the temperature calculation formula are as follows: y = f(x )=1.00872×x-0.00208, x=G / 255, G=0.299Red+0.587Green+0.114Blue, T=(ab)×y+b, where x is the normalized gray value of the pixel; y is the normalized temperature value; G is the gray value of the pixel, and its value range is 0 to 255. The larger the gray value, the greater the brightness of the pixel; Red, Green, and Blue are the red component, green score, and blue component of the pixel, respectively; T is the temperature value obtained using the fitting function; a and b are the upper and lower limits of the temperature of the infrared image color bar, respectively;

[0072] R 2 The coefficient of determination is used to measure the goodness of fit. The maximum value is 1. The closer it is to 1, the better the function fits the data. 2 The calculation formula is as follows: In the formula, t i is the normalized temperature value; is the mean of the normalized temperature array; f i is the fitted value.

[0073] In infrared images, if the grayscale value of a pixel and the upper and lower temperature limits of the colorimetric bar are known, the temperature value of the pixel can be obtained. In order to realize the intelligent temperature extraction process, the Tesseract-OCR engine is used to automatically identify and extract the upper and lower temperature limits. The positions of the upper and lower temperature limits in infrared images are fixed. In order to prevent interference from other texts, only the smallest area where the numbers are located is identified. By setting the grayscale value on the rectangular frame for color conversion, white numbers are converted to black and the background is converted to white, thereby eliminating background interference and improving recognition accuracy.

[0074] Step 11: The equipment failure estimation algorithm module introduces an index analysis model based on the substation equipment status information of the thermal fault intelligent diagnosis algorithm module, associates the equipment failure rate with the equipment status, and realizes quantitative estimation of the equipment failure rate;

[0075] In this embodiment, specifically, the correlation between the device status and the device failure rate can be expressed in an approximate model as follows: p = ke cH, where p is the current failure rate of the equipment; k and c are unknown constants; H is the equipment state parameter;

[0076] The formula essentially states that when the equipment condition deteriorates, the equipment failure rate should increase exponentially, a trend that is consistent with field experience.

[0077] k and c are constants related to many factors such as equipment type and operating environment, and are difficult to obtain directly. Considering that statistical samples are often insufficient to support the evaluation of equipment failure rate under a single state, the present invention provides the following inversion calculation method:

[0078] Assume that there is a batch of status and fault sample data, where the status classification H has 1 to N types, and the corresponding number of devices is M1 to M N , the number of faulty units in a historical evaluation period is Q, then according to the above formula, it should satisfy: This will result in an equation for k and c. If there are more than two years of data, k and c can be solved to provide an equipment failure rate estimation model based on equipment status. If the equation group has no numerical solution, consider using least squares fitting, such as the Gauss-Newton method and Levenberg-Marquardt method in nonlinear least squares.

[0079] Step 12: This enables the possibility of equipment failure within a certain evaluation period to be determined based on the current status of the equipment.

[0080] The present invention is a fault detection device for substation equipment and a method for using the device. In use, the present invention adopts a multi-target form to expand the coverage of fault detection, so as to deploy multi-target fault detection nodes, and uses a linear array of infrared detectors to collect infrared images of substation equipment and pre-process them to realize fault detection of substation equipment; a data mining module is also used to combine the needs and standards of fault detection, to collect and pre-process fault data, and then a data mining fault detection model is established, and fault detection is completed in an adaptive recognition processing manner; the fault detection information is integrated to effectively improve the fault detection accuracy of substation equipment, and the fault of substation equipment can be accurately detected, so that the staff can go to the site for repair in time to ensure the safe and stable operation of the substation equipment; the present invention also expands the fault recognition range of the equipment, can accurately lock the location where the abnormality or fault occurs, strengthen the control of detection errors, and reduce the false recognition rate; in addition, the present invention introduces an index analysis model to associate the equipment failure rate with the equipment state, thereby providing a quick equipment failure rate quantitative estimation method, and realizing the possibility of equipment failure within a certain evaluation period according to the current state of the equipment; the present invention has the advantages of being less affected by noise, adopting multiple targets to expand the detection range, improving the accuracy of fault detection, and introducing an index analysis model to quantitatively estimate the fault rate.

Claims

1. A fault detection device for substation equipment, comprising a terminal and a mobile terminal, wherein the terminal uses a power supply module as a power supply, and is characterized in that: The terminal includes a deployment multi-target fault detection node, a data processing unit, an infrared thermal imager and a data mining module; The multi-target fault detection node includes at least three infrared detectors, which are arranged in a linear array at at least three positions on the periphery of the substation equipment to be detected; each of the infrared detectors is connected to the infrared thermal imager, and is used to collect infrared radiation data of the substation equipment in real time, and convert it into electrical signals to generate thermal infrared images of the substation equipment; each of the infrared detectors is also connected to the data mining module, and is used to classify and screen the data and information collected in real time from the substation equipment, complete quantitative processing, and divide the data into variable and non-variable quantities, and calibrate the location where the abnormality occurs according to the fault data; the infrared thermal imager and the data mining module send the data and information to the data processing unit.

2. A fault detection device for substation equipment according to claim 1, characterized in that: The data processing unit includes a data preprocessing module, a data mining fault detection module, an adaptive fault detection module and a signal transmission module; the data preprocessing module is connected to the infrared thermal imager and the data mining module respectively, the data preprocessing module is connected to the data mining fault detection module, the data mining fault detection module is connected to the adaptive fault detection module, and the adaptive fault detection module is connected to the signal transmission module.

3. A fault detection device for substation equipment according to claim 2, characterized in that: The data processing unit may also be connected to an external display terminal, and the display terminal is used to display the current fault status of the substation equipment.

4. A fault detection device for substation equipment as claimed in claim 3, characterized in that: It also includes a target detection algorithm module A and a target detection algorithm module B, wherein the target detection algorithm module A is connected to the terminal; and the target detection algorithm module B is connected to the mobile terminal.

5. A fault detection device for substation equipment according to claim 4, characterized in that: It also includes a thermal fault intelligent diagnosis algorithm module and an equipment fault estimation algorithm module. The thermal fault intelligent diagnosis algorithm module is connected to the target detection algorithm module A and the target detection algorithm module B respectively, and the equipment fault estimation algorithm module is connected to the thermal fault intelligent diagnosis algorithm module.

6. A method for using a fault detection device for a substation, using the fault detection device for a substation as claimed in any one of claims 1 to 5, characterized in that: The following steps are involved: Step 1: Expand the fault detection coverage by adopting a multi-target approach, thereby deploying multi-target fault detection nodes; Step 2: Use a linear array of infrared detectors to determine the scanning area of ​​the substation equipment; Step 3: The data mining module collects and preprocesses fault data based on the requirements and standards of fault detection; the infrared thermal imager converts the infrared radiation energy of the substation equipment into electrical signals and collects images of the substation equipment; Step 4: The data processing module of the data processing unit calculates the pixel points of the infrared image through the probability density function, and pre-processes the infrared image of the substation equipment using the median filtering principle; Step 5: Implementing fault detection of the substation equipment according to the relative temperature difference of the fault point of the substation equipment; Step 6: The data mining fault detection module establishes a data mining fault detection model, and completes fault detection by using an adaptive fault detection module using an adaptive recognition processing method; Step 7: The data processing unit fuses the above two fault detection data to obtain fault detection and diagnosis information, and sends the fault detection and diagnosis information to the display terminal through the signal transmission module to display the fault status of the current substation equipment; Step 8: The target detection algorithm module A adopts the target detection method of infrared images of power equipment based on the improved YOLOv4 model to achieve accurate positioning and classification of power equipment; Step 9: The target detection algorithm module B adopts the target detection method of infrared images of power equipment based on the improved PP-YOLO--Tiny lightweight model to realize the target detection and diagnosis of power equipment on the mobile terminal; Step 10: The thermal fault intelligent diagnosis algorithm module combines the target detection algorithm module A and the target detection algorithm module B, and adopts a temperature extraction method based on colorimetric strip analysis to realize intelligent diagnosis of thermal faults of substation equipment; Step 11: The equipment failure estimation algorithm module introduces an index analysis model based on the substation equipment status information of the thermal fault intelligent diagnosis algorithm module, associates the equipment failure rate with the equipment status, and realizes quantitative estimation of the equipment failure rate; Step 12: This enables the possibility of equipment failure within a certain evaluation period to be determined based on the current status of the equipment.

7. The method for using the fault detection device for substation equipment according to claim 6, characterized in that: The step 2 is specifically as follows: using a linear array of infrared detectors to identify the scanning area of ​​the substation equipment in different ways. Assuming that the linear array of infrared detectors can divide the scanning space of the substation equipment into n units, the frame images of each unit are scanned one by one in a certain order. The calculation formula of the unit scanning time is: Where, T p Indicates the scanning period; f p Indicates the frame rate of the scan.

8. The method for using the fault detection device for substation equipment according to claim 7, characterized in that: The step 4 is specifically as follows: filtering the infrared image of the substation equipment using a filtering algorithm. Assuming that there are 6 grayscale images in the image, the pixel points of the infrared image are calculated using a probability density function. The calculation formula is: In the formula, a represents the black noise in the infrared image; b represents the white noise in the infrared image; u a Indicates the probability corresponding to the black noise point; u b Indicates the probability corresponding to the white noise point; The median filter is introduced to sort the window range points of the infrared image of the substation equipment, and the gray value of the infrared image is used as the middle value of the sequence center. The extraction formula is: R(i,j)=median[g(k)], where g represents the gray value sequence of the infrared image; k represents the number of pixels in the infrared image; The median filter is composed of the R trigger level signal and the W trigger level signal, which can be expressed as: In the formula, H med Represents the grayscale median of the infrared image; H xy Represents the coordinate gray value of the infrared image; H min Indicates the minimum gray value of the infrared image; H max Indicates the maximum gray value of the infrared image.

9. The method for using the fault detection device for substation equipment according to claim 8, characterized in that: The step 5 is specifically: calculating the relative temperature difference at the fault point of the substation equipment, the formula is: In the formula, T1 represents the temperature of the heating point of the substation equipment; T0 represents the ambient temperature of the substation equipment; T2 represents the standard temperature in the environment; the fault detection of the substation equipment is realized according to the calculation result of the relative temperature difference at the fault point.

10. The method for using the fault detection device for substation equipment according to claim 9, characterized in that: The step 11 is specifically: the correlation between the equipment status and the equipment failure rate, and its approximate model can be expressed as: p = ke cH , where p is the current failure rate of the equipment; k and c are unknown constants; and H is the equipment state parameter.

Images