Diagnostic method and diagnostic device based on temperature vision

Abstract

The present invention relates to the field of power detection technology, and discloses a diagnostic method and a diagnostic device based on temperature vision, which include the following four parts: collection and annotation of temperature vision data sets; building a power equipment target detection model based on the temperature vision data set; building a power equipment component detection model based on the temperature vision data set; and performing fault diagnosis on the power equipment based on intelligent recognition results. The present invention locates the position of the power equipment through a target detection network based on temperature vision information, and locates the position of the power equipment components through a feature point network, so as to accurately locate the position of each component of the power equipment and obtain the corresponding temperature information.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power detection, and in particular to a temperature vision-based diagnosis method and a diagnosis device. Background Art

[0002] At present, the detection method of power equipment is mainly based on infrared images to study the problem of abnormal heating of power equipment. Since infrared images contain multiple pseudo-colors, the complex shooting environment causes serious interference from the photographed equipment, and there are few types of training data, the existing methods have defects such as low detection accuracy and poor model generalization ability. Summary of the invention

[0003] 1. Technical issues to be resolved

[0004] In view of the shortcomings of the prior art, the present invention provides a diagnostic method and a diagnostic device based on temperature vision, which have the advantages of accurately locating power equipment based on temperature vision information and obtaining corresponding temperature information, and solves the problems of low detection accuracy and poor model generalization ability of the existing methods.

[0005] (II) Technical solution

[0006] In order to achieve the above-mentioned purpose of accurately locating power equipment based on temperature visual information and obtaining corresponding temperature information, the present invention provides the following technical solutions:

[0007] The diagnostic method based on temperature vision includes the following steps:

[0008] S1. Collection and annotation of temperature vision dataset: including temperature vision, dataset collection and dataset annotation.

[0009] S2. Power equipment target detection: Based on the temperature vision dataset, build a power equipment target detection model, and use the target detection architecture to build a power equipment recognition model.

[0010] S3. Power equipment component detection: Based on the temperature vision data set, a power equipment component detection model is built to detect the location of components in each type of power equipment for fault diagnosis of power equipment.

[0011] S4. Fault diagnosis: Based on the intelligent identification results, the power equipment is diagnosed for faults. The intelligent identification of the power equipment is realized through the above steps. After the locations of the components of the power equipment are determined, the power equipment fault diagnosis is performed through the following methods: including maximum value detection method, average value detection method, variance detection method, regional variance detection method and equipment temperature difference detection method.

[0012] The step S1 is specifically as follows:

[0013] S101. Temperature vision: The infrared temperature sensor reflects the infrared radiation energy distribution diagram of the target received by the detector to the photosensitive element of the infrared detector, and the processor processes it to obtain a .jpg file in the infrared general data file storage format, that is, an infrared thermal image.

[0014] The step S1 is specifically as follows:

[0015] S102, data set acquisition: using an infrared thermal imager, collect information on power equipment, obtain infrared general data file storage format images, and then extract temperature visual data from them, and obtain temperature visual data of 106 types of power equipment in total.

[0016] The step S1 is specifically as follows:

[0017] S103, data set annotation: including device frame annotation and feature point annotation.

[0018] The step S103 is specifically as follows:

[0019] Equipment frame labeling: Based on the collected temperature visual data, the target frame is labeled for the power equipment located at the center of the temperature vision.

[0020] The step S103 is specifically as follows:

[0021] Feature point marking: Based on the collected temperature visual data, the feature points of the components in the power equipment located at the center of the temperature visual data are marked.

[0022] The step S2 is specifically as follows:

[0023] First, the original input temperature visual data is standardized. Then, the temperature visual dataset with equipment box annotations is trained using target detection networks such as R-CNN, Fast R-CNN, Faster R-CNN, YOLO, SSD, RetinaNet, and RefineDet to obtain the power equipment target detection model.

[0024] The step S4 is specifically as follows:

[0025] Maximum value detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the maximum value of all elements of the matrix is ​​obtained; the maximum temperature threshold is set for each component of the power equipment, and the maximum value obtained from the above process is compared with the maximum temperature threshold. If the threshold is exceeded, it is judged as a fault.

[0026] The step S4 is specifically as follows:

[0027] Average value detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the average value of all elements of the matrix is ​​obtained; the average temperature threshold is set for each component of the power equipment, and the average value obtained from the above process is compared with the average temperature threshold. If the threshold is exceeded, it is judged as a fault.

[0028] The step S4 is specifically as follows:

[0029] Variance detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the variance of all elements of the matrix is ​​obtained; the variance temperature threshold is set for each component of the power equipment, and the variance obtained from the above process is compared with the variance temperature threshold. If the threshold is exceeded, it is judged as a fault.

[0030] The step S4 is specifically as follows:

[0031] Regional variance detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and then it is divided into several small matrices with the same length and width according to the length and width of the matrix; the maximum value of all elements of each small matrix is ​​obtained, and the variance of these maximum values ​​is calculated, which is called the regional variance; the regional variance temperature threshold is set for each component of the power equipment, and the regional variance obtained from the above process is compared with the regional variance temperature threshold. If the threshold is exceeded, it is judged as a fault.

[0032] The step S4 is specifically as follows:

[0033] Equipment temperature difference detection method: For the component frames of the same type of equipment output by intelligent identification during the same shooting process, the maximum value detection method, average value detection method, variance detection method, and regional variance detection method are used to obtain their maximum value, average value, variance, and regional variance; for the same component frames of the same type of equipment, the difference between the maximum value, average value, variance, and regional variance of each component and the average value of all equipment is obtained; an equipment temperature difference threshold is set for each component of the power equipment, and the difference obtained from the above process is compared with the equipment temperature difference threshold. If the threshold is exceeded, it is judged as a fault.

[0034] The diagnostic device based on temperature vision comprises a device body and a detector matched with the device body. A display screen is rotatably mounted on the top of the device body, a probe is arranged on one side of the detector, and a handle is installed on the bottom of the detector.

[0035] (III) Beneficial effects

[0036] Compared with the prior art, the present invention provides a diagnostic method and a diagnostic device based on temperature vision, which have the following beneficial effects:

[0037] The present invention as a whole includes four parts: the collection and annotation of temperature visual data sets; the establishment of a power equipment target detection model based on the temperature visual data sets; the establishment of a power equipment component detection model based on the temperature visual data sets; the power equipment fault diagnosis based on the intelligent recognition results, the positioning of the power equipment through the target detection network based on the temperature visual information, the positioning of the power equipment components through the feature point network, so as to accurately locate the positions of the components of the power equipment and obtain the corresponding temperature information. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is the overall flow chart of temperature visual intelligent recognition of the present invention;

[0039] Figure 2 This is a table diagram of the storage format of infrared universal data files of the present invention;

[0040] Figure 3 It is the infrared thermal image of the present invention;

[0041] Figure 4 It is the temperature visual diagram of the present invention;

[0042] Figure 5 This is an example diagram of the target frame annotation of the present invention;

[0043] Figure 6 This is an example diagram of the characteristic point marking of the present invention;

[0044] Figure 7 The overall flow chart of building an intelligent identification model for power equipment for the present invention;

[0045] Figure 8 It is a flow chart of the small classification model of the present invention;

[0046] Fig. 9 It is a feature point detection model and a component frame detection flow chart of the present invention;

[0047] Fig.10 It is a schematic diagram of the structure of the diagnostic device of the present invention.

[0048] In the figure: 1. Device body; 2. Detector; 3. Display screen; 4. Probe; 5. Handle. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] See also Figure 1-9The embodiment of the present invention provides a temperature vision-based diagnosis method, comprising the following steps:

[0051] S1. Collection and annotation of temperature vision dataset: including temperature vision, dataset collection and dataset annotation.

[0052] Step S1 is specifically as follows:

[0053] S101. Temperature vision: The infrared temperature sensor reflects the infrared radiation energy distribution diagram of the target received by the detector to the photosensitive element of the infrared detector, and the processor processes it to obtain a .jpg file in the infrared general data file storage format, that is, the infrared thermal image. The data format is as follows Figure 2 As shown, it contains: file header, calibration data, temperature measurement parameters, temperature data, imaging parameters, analysis data, sound and user-defined data and file tail. Figure 3 As shown in the figure, the temperature data part is a float type two-dimensional matrix with the same length and width as the infrared thermal image resolution, which can image the ambient temperature in the form of numerical values, that is, temperature vision, such as Figure 3 As shown, the value of an element in the matrix corresponds to the actual temperature value (in degrees Celsius) of an area in the environment. For example, if the temperature value in the upper left corner of the matrix is ​​37.0, then the actual temperature of the upper left corner area in the corresponding environment is 37.0℃.

[0054] S102, data set acquisition: Use infrared thermal imager to collect information of power equipment, obtain infrared general data file storage format image, and then extract temperature visual data from it, and obtain temperature visual data of 106 types of power equipment, including GIS gas chamber, neutral point voltage transformer, JP cabinet, bushing, low voltage pile head, damping resistor box, combined filter, high voltage side bushing, lighting box, metal wire, incoming line warehouse, convergence box, pump, insulator, convergence box, GIS bushing, combined mutual inductor, transfer box, lightning arrester, coupling mutual inductor, junction box, transformer, battery pack, mechanism box, switch cabinet, filter, power box, wall bushing, lightning rod, terminal cabinet, magnetic reactor, busbar, capacitor box, station transformer, busbar, resistor box, low voltage side pile head, supporting insulator, metering box, reactor, pole vacuum switch, intelligent control cabinet, cable terminal, coupling capacitor, Intelligent control cabinet, power cable, air switch, intelligent component cabinet, current transformer, compensation device, intelligent terminal cabinet, capacitor, fuse, panel cabinet, oil tank, vertical insulator, maintenance power box, arc extinguishing device, wire, maintenance power line box, voltage transformer, radiator, capacitor box, connecting pipe, wire clamp, control box, resistor, arc extinguishing coil, control cabinet, drop-out fuse, suspension insulator, distribution box, connecting bridge, oil pillow, terminal box, soft connection, low resistance, damping box, circuit breaker, inflatable bushing, regulating cabinet, discharge gap, power box, discharge coil, pole-mounted circuit breaker, electric control box, voltage divider, pole-mounted disconnector, component cabinet, cable terminal tail pipe, pile head, switch box, high-voltage fuse, wave arrester, wind control box, high-voltage bushing, pipe mother, low-voltage switch box, disconnector, AC filter, low-voltage side switch box, converter valve, grounding transformer, maintenance box and cold control box.

[0055] S103, data set annotation: including device frame annotation and feature point annotation.

[0056] Step S103 is specifically as follows:

[0057] Equipment box annotation: For the collected temperature visual data, the target box is annotated for the power equipment located in the center of the temperature visual data, such as Figure 4 Shown is an example of target box annotation for power equipment.

[0058] Feature point annotation: For the collected temperature visual data, the feature points of the components in the power equipment located at the center of the temperature visual data are annotated, such as Figure 5 The figure shows an example of marking characteristic points of electric power equipment.

[0059] S2. Power equipment target detection: Based on the temperature vision dataset, build a power equipment target detection model, and use the target detection architecture to build a power equipment recognition model.

[0060] Step S2 is specifically as follows:

[0061] like Figure 6As shown, in order to effectively identify and locate 106 types of power equipment, the present invention uses a target detection architecture to build a power equipment recognition model. First, the original input temperature visual data is standardized to reduce the impact of the large numerical differences between the temperature visual data on the results. Then, using target detection networks such as R-CNN, Fast R-CNN, Faster R-CNN, YOLO, SSD, RetinaNet, RefineDet, etc., the temperature visual data set annotated with the equipment box is trained to obtain a power equipment target detection model. The model is used to detect the input temperature visual data, obtain the location of the power equipment therein, and which type of power equipment it belongs to among the 106 types of power equipment, and output the location information and type information.

[0062] S3. Power equipment component detection: Based on the temperature vision data set, a power equipment component detection model is built to detect the location of components in each type of power equipment for fault diagnosis of power equipment.

[0063] Step S3 is specifically as follows:

[0064] In order to more accurately obtain the position of each component in the power equipment, the present invention divides the power equipment component detection method into two steps:

[0065] The first step is to classify each power equipment into multiple small categories for the above 106 types of power equipment, since each type of power equipment has differences in component composition structure, appearance shape, etc. Then use convolutional neural networks such as LeNet, AlexNet, CaffeNet, ZFNet, VggNet, ResNet, NiN, GoogLeNet, DenseNet, Senet, BAM, MobileNet, etc. to crop the annotation box area of ​​the temperature visual target box annotation data set of 106 types of power equipment, and train the obtained power equipment temperature visual data set (remove the background part) to obtain 106 classification models for dividing different appearance devices under the same type of equipment. The temperature visual data obtains the device location information and type information through the visual inspection detection model, and the temperature visual data of the device area is cropped according to the device location information, and then the corresponding classification model is called according to the type information for classification. This step will make a more detailed division of each of the 106 types of power equipment according to the appearance. The present invention refers to this step as the division of power equipment into small categories, and the model used for small classification is called a small classification model. Taking the casing as an example, Figure 7 As shown, the temperature visual data is first cropped and then sent to the network for training to obtain a small classification model for the casing.

[0066] In the second step, for each small category of power equipment, based on convolutional neural networks such as LeNet, AlexNet, CaffeNet, ZFNet, VggNet, ResNet, NiN, GoogLeNet, DenseNet, Senet, BAM, MobileNet, etc., it is modified into a feature point extraction network. The network outputs the feature point coordinates of each component of the power equipment, and trains the temperature visual data set marked with feature points to obtain a feature point extraction model. The feature point detection model detects the temperature visual data of the power equipment, outputs the location of the feature points of the power equipment, and finally converts the feature points into component boxes to frame the locations of the components of the power equipment for subsequent fault diagnosis, such as Figure 8 shown.

[0067] S4. Fault diagnosis: Based on the intelligent identification results, the power equipment is diagnosed for faults. The intelligent identification of the power equipment is realized through the above steps. After the locations of the components of the power equipment are determined, the power equipment fault diagnosis is performed through the following methods: including maximum value detection method, average value detection method, variance detection method, regional variance detection method and equipment temperature difference detection method.

[0068] Step S4 is specifically as follows:

[0069] Maximum value detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the maximum value of all elements of the matrix is ​​obtained; the maximum temperature threshold is set for each component of the power equipment, and the maximum value obtained from the above process is compared with the maximum temperature threshold. If the threshold is exceeded, it is judged as a fault.

[0070] Step S4 is specifically as follows:

[0071] Average value detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the average value of all elements of the matrix is ​​obtained; the average temperature threshold is set for each component of the power equipment, and the average value obtained from the above process is compared with the average temperature threshold. If the threshold is exceeded, it is judged as a fault.

[0072] Step S4 is specifically as follows:

[0073] Variance detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the variance of all elements of the matrix is ​​obtained; the variance temperature threshold is set for each component of the power equipment, and the variance obtained from the above process is compared with the variance temperature threshold. If the threshold is exceeded, it is judged as a fault.

[0074] Step S4 is specifically as follows:

[0075] Regional variance detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and then it is divided into several small matrices with the same length and width according to the length and width of the matrix; the maximum value of all elements of each small matrix is ​​obtained, and the variance of these maximum values ​​is calculated, which is called the regional variance; the regional variance temperature threshold is set for each component of the power equipment, and the regional variance obtained from the above process is compared with the regional variance temperature threshold. If the threshold is exceeded, it is judged as a fault.

[0076] Step S4 is specifically as follows:

[0077] Equipment temperature difference detection method: For the component frames of the same type of equipment output by intelligent identification during the same shooting process, the maximum value detection method, average value detection method, variance detection method, and regional variance detection method are used to obtain their maximum value, average value, variance, and regional variance; for the same component frames of the same type of equipment, the difference between the maximum value, average value, variance, and regional variance of each component and the average value of all equipment is obtained; an equipment temperature difference threshold is set for each component of the power equipment, and the difference obtained from the above process is compared with the equipment temperature difference threshold. If the threshold is exceeded, it is judged as a fault.

[0078] See also Fig.10 The embodiment of the present invention provides a temperature vision-based diagnostic device, including a device body 1 and a detector 2 matched with the device body 1. A display screen 3 is rotatably installed on the top of the device body 1, and a probe 4 is provided on one side of the detector 2. When in use, the detector 2 is used to detect the equipment, and the device body 1 analyzes and processes the data and displays the results on the display screen 3. A handle 5 is installed at the bottom of the detector 2, and the handle 5 is used to facilitate the use of the handheld detector 2.

[0079] Working principle: Based on temperature visual data, the present invention can accurately identify and locate 106 types of power equipment. The mAP for identifying 106 types of power equipment is 98%, the mAP for identifying small categories of 106 types of power equipment is 99%, and the IoU for locating the feature points of 106 types of power equipment components is 95%.

[0080] The best use state of the present invention is that there is a temperature difference between the temperature of the target detection device and the temperature of the surrounding environment.

[0081] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise one" do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0082] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A diagnostic method based on temperature vision, characterized in that: The following steps are involved: S1. Collection and annotation of temperature vision dataset: including temperature vision, dataset collection and dataset annotation; The step S1 is specifically as follows: S101, temperature vision: the infrared temperature sensor reflects the infrared radiation energy distribution diagram of the target received by the detector to the photosensitive element of the infrared detector, and the processor processes it to obtain a .jpg file in the infrared general data file storage format, that is, an infrared thermal image; S102, data set collection: using an infrared thermal imager to collect information on power equipment, obtain infrared general data file storage format images, and then extract temperature visual data from them, obtaining temperature visual data of 106 types of power equipment in total; S103, data set annotation: including device frame annotation and feature point annotation, device frame annotation: for the collected temperature visual data, annotate the target frame for the power equipment located at the center of the temperature visual data; Feature point marking: Based on the collected temperature visual data, feature points are marked on the components of the power equipment located at the center of the temperature visual data; S2. Power equipment target detection: Based on the temperature vision dataset, build a power equipment target detection model, and use the target detection architecture to build a power equipment recognition model; The step S2 is specifically as follows: First, the original input temperature visual data is standardized. Then, the R-CNN, Fast R-CNN, Faster R-CNN, YOLO, SSD, RetinaNet, and RefineDet target detection networks are used to train the temperature visual data set annotated with the equipment box to obtain the power equipment target detection model. S3. Power equipment component detection: Based on the temperature vision data set, a power equipment component detection model is built to detect the location of components in each power equipment for fault diagnosis of power equipment; Each power equipment is classified into multiple small categories. Then, LeNet, AlexNet, CaffeNet, ZFNet, VggNet, ResNet, NiN, GoogLeNet, DenseNet, Senet, BAM, and MobileNet convolutional neural networks are used to crop the labeled box area of ​​the temperature visual target box annotation dataset of 106 types of power equipment. The obtained power equipment temperature visual dataset is trained to obtain 106 classification models for dividing different appearance devices of the same type. The temperature visual data obtains the device location information and type information through the visual inspection model. The temperature visual data of the device area is cropped according to the device location information, and then the corresponding classification model is called for classification according to the type information. For each small category of power equipment, based on LeNet, AlexNet, CaffeNet, ZFNet, VggNet, ResNet, NiN, GoogLeNet, DenseNet, Senet, BAM, and MobileNet convolutional neural networks, they are modified into feature point extraction networks. The network outputs the feature point coordinates of each component of the power equipment, and trains the temperature visual data set annotated with the feature points to obtain a feature point extraction model. The feature point detection model detects the temperature visual data of the power equipment, outputs the location of the feature points of the power equipment, and finally converts the feature points into component frames to frame the locations of the components of the power equipment for subsequent fault diagnosis. S4. Fault diagnosis: Based on the intelligent identification results, the power equipment is diagnosed for faults. The intelligent identification of the power equipment is realized through the above steps. After the locations of the components of the power equipment are determined, the power equipment fault diagnosis is performed through the following methods: including maximum value detection method, average value detection method, variance detection method, regional variance detection method and equipment temperature difference detection method.

2. The temperature vision-based diagnostic method according to claim 1, characterized in that: The step S4 is specifically as follows: Maximum value detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the maximum value of all elements of the matrix is ​​obtained; the maximum temperature threshold is set for each component of the power equipment, and the maximum value obtained from the above process is compared with the maximum temperature threshold. If the threshold is exceeded, it is judged as a fault.

3. The temperature vision-based diagnostic method according to claim 1, characterized in that: The step S4 is specifically as follows: Average value detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the average value of all elements of the matrix is ​​obtained; the average temperature threshold is set for each component of the power equipment, and the average value obtained from the above process is compared with the average temperature threshold. If the threshold is exceeded, it is judged as a fault.

4. The temperature vision-based diagnostic method according to claim 1, characterized in that; The step S4 is specifically as follows: Variance detection method: According to the component box output by intelligent recognition, the two-dimensional matrix corresponding to the component box is intercepted from the temperature visual data, and the variance of all elements of the matrix is ​​obtained; the variance temperature threshold is set for each component of the power equipment, and the variance obtained from the above process is compared with the variance temperature threshold. If it exceeds the threshold, it is judged as a fault; The step S4 is specifically as follows: Regional variance detection method: According to the component frame output by intelligent recognition, the two-dimensional matrix corresponding to the component frame is intercepted from the temperature visual data, and then the matrix is ​​divided into several small matrices with the same length and width according to the length and width of the matrix; Get the maximum value of all elements of each small matrix, and calculate the variance of these maximum values, which is called regional variance; set regional variance temperature thresholds for each component of the power equipment, compare the regional variance obtained from the above process with the regional variance temperature threshold, and determine that a fault occurs if the threshold is exceeded; The step S4 is specifically as follows: Equipment temperature difference detection method: For the component frames of the same type of equipment output by intelligent identification during the same shooting process, the maximum value detection method, average value detection method, variance detection method, and regional variance detection method are used to obtain their maximum value, average value, variance, and regional variance; for the same component frames of the same type of equipment, the difference between the maximum value, average value, variance, and regional variance of each component and the average value of all equipment is obtained; an equipment temperature difference threshold is set for each component of the power equipment, and the difference obtained from the above process is compared with the equipment temperature difference threshold. If the threshold is exceeded, it is judged as a fault.

Images