CN105371957A - Transformer station equipment infrared temperature registration positioning and method - Google Patents

Abstract

The invention discloses an infrared temperature registration and positioning system for substation equipment, which includes: an infrared camera, a visible light camera, a video server, and a data processing and analysis unit, which receives the infrared thermal image and visible light of the substation equipment collected by the infrared camera and the visible light camera. image, and register the infrared thermal image and the visible light image of the same target scene, so as to match the measurement points of the infrared thermal image and the visible light image of the same target scene; the data processing and analysis unit establishes the radial basis nerve Network, and predict the temperature prediction value of each measurement point on the infrared thermal image through the radial basis neural network, and correspond the temperature prediction value to each measurement of the visible light image that matches each measurement point on the infrared thermal image Point. The invention also discloses an infrared temperature registration and positioning method for substation equipment.

Description

Infrared temperature registration and positioning system and method for substation equipment

technical field

The invention relates to a system and method for temperature prediction of substation equipment, in particular to a system and method for realizing temperature prediction of substation equipment by registering and positioning infrared thermal images and visible light images.

Background technique

Many devices in the power industry operate under high voltage and high current conditions, which are closely related to heat. In many power outage accidents, power outage maintenance due to local overheating of equipment occurs from time to time. Therefore, discovering the heating defects of the equipment in time and eliminating the heating defects in the initial state is the key to ensure the safe operation of the equipment, reduce accidents, and avoid forced power outages. The basic principle of infrared detection is to obtain the thermal state characteristics of the object by detecting the infrared radiation signal of the object, and judge the state of the object according to the thermal state characteristics and the corresponding judgment basis. Because infrared detection technology has the characteristics of long-distance, non-contact, real-time and fast, it is of great significance to realize on-line monitoring and fault diagnosis of power equipment. The power equipment of the substation mainly adopts two kinds of infrared detection technologies, infrared thermal imaging and infrared point temperature measurement, so as to realize the comprehensive monitoring of the operation performance and heat generation of the main power equipment of the whole substation.

Although infrared thermal imaging technology has many of the advantages mentioned above, compared with ordinary images, infrared thermal imaging is affected by factors such as working principle, external environment and its own devices, the visual effect is not clear enough, and the contrast between the target device and the background is poor. The analysis and handling of faults caused a lot of inconvenience. In addition, the infrared monitoring equipment market is almost occupied by large foreign companies. Due to industry monopoly, technology blockade and other reasons, power companies can only passively purchase and use the analysis software that comes with the equipment, which cannot better meet individual requirements, which seriously restricts The improvement of fault diagnosis level of power equipment in substation is not conducive to the safe and stable operation of smart grid.

Contents of the invention

The purpose of the present invention is to provide an infrared temperature registration and positioning system for substation equipment, which can predict the temperature prediction value of each measurement point on the infrared thermal image of the substation equipment, and realize the temperature prediction of each measurement point on the infrared thermal image and visible light image of the substation equipment. Matching of measurement points and sharing of information, the information includes temperature prediction values for each measurement point.

Another object of the present invention is to provide an infrared temperature registration and positioning method for substation equipment, which also has the above functions.

In order to achieve the above purpose, the present invention proposes an infrared temperature registration and positioning system for substation equipment, which includes:

An infrared camera, which collects an infrared thermal image of a target scene;

A visible light camera, which collects visible light images of the target scene;

A video server, which is connected to the infrared camera and the visible light camera respectively through video cables;

The data processing and analysis unit is usually a computer, which is connected to the video server, receives the infrared thermal image and visible light image of the substation equipment collected by the infrared camera and the visible light camera, and analyzes the infrared thermal image of the same target scene. The image image and the visible light image are registered to match each measuring point of the infrared thermal image image and the visible light image of the same target scene; the data processing and analysis unit establishes a radial basis neural network, and predicts The temperature prediction value of each measurement point on the infrared thermal image is obtained, and the temperature prediction value is corresponding to each measurement point of the visible light image matched with each measurement point on the infrared thermal image.

The infrared temperature registration and positioning system for substation equipment according to the present invention is based on the registration of the infrared thermal image and the visible light image of the same target scene, so as to realize the matching of each measuring point on the infrared thermal image of the substation equipment and the visible light image, and at the same time Fit the pixels and temperatures of the measurement points of the infrared thermal image to obtain the temperature prediction value of each measurement point on the infrared thermal image through the radial basis neural network, and correspond the temperature prediction value to the infrared thermal image Each measurement point on the map matches each measurement point of the visible light image. The registration is the process of matching and superimposing two or more images acquired at different times, different sensors (imaging devices) or under different conditions; the commonly used methods are gray-scale information-based method, transform domain method and feature-based method, etc., its technology is relatively mature, and it is a technology well known to those skilled in the art, and the present invention will not introduce it in detail. The fitting method is to first carry out temperature prediction training to the radial basis neural network, then use the radial basis neural network trained by temperature prediction as a fitting model, and use the pixels of the measurement points of the infrared thermal image and The temperature is used as the input and output of the fitting model respectively to realize the fitting. The method for temperature prediction training is to use several pixel values as input samples, and use the temperature bar temperature values corresponding to the several pixel values as the corresponding output of the input samples, and perform temperature prediction training on the radial basis neural network , to solve the parameters of the radial basis neural network.

In terms of specific applications, the infrared temperature registration and positioning system for substation equipment described in the present invention can realize dual-channel monitoring of infrared thermal images and visible light images by setting a human-computer interaction interface: by clicking on the infrared thermal image or visible light image For any measuring point, obtain information such as temperature prediction value and coordinates of the measuring point.

The infrared temperature registration and positioning system for substation equipment described in the present invention can perform all-round infrared temperature monitoring and registration and positioning for substation equipment, and obtain information such as temperature prediction values and coordinates of each measurement point by directly consulting visible light images, so it can be Reduce the difficulty of locating thermal anomalies caused by blurred infrared thermal images, reduce positioning accuracy errors, and simplify the difficulty of subsequent fault analysis and processing in substation infrared temperature monitoring, thereby improving the intelligence level of substations.

Further, in the infrared temperature registration and positioning system for substation equipment according to the present invention, the infrared camera is also directly connected to the data processing and analysis unit through a network cable, so that the data processing and analysis unit transmits control parameters to the infrared camera, thereby controlling the infrared camera. The camera performs operations such as focusing, aperture discharge, setting area, etc.

Further, the infrared temperature registration and positioning system for substation equipment described in the present invention or above also includes a rotatable pan-tilt, the infrared camera and the visible light camera are arranged on the pan-tilt, and the pan-tilt is connected to a video server , to receive the control signal from the video server, so that the rotation of the pan/tilt can be controlled by the video server.

Further, the infrared temperature registration and positioning system for substation equipment described in the present invention or above also includes a battery pack and an inverter connected to the battery pack, and the inverter is connected to an infrared camera, a visible light camera, a video server and a data processing The analysis units are respectively connected to convert the direct current provided by the battery pack into alternating current for the infrared camera, the visible light camera, the video server and the data processing and analysis unit.

Correspondingly, the present invention also provides an infrared temperature registration and positioning method for substation equipment, which includes the steps of:

(1) Obtain the infrared thermal image and visible light image of the same target scene of the substation equipment respectively;

(2) Match the pixel matrix of the visible light image with the infrared thermal image, so that each measuring point on the visible light image completely corresponds to each measuring point on the infrared thermal image;

(3) Construct radial basis neural network and carry out temperature prediction training to it;

(4) Take the pixel value corresponding to the measurement point on the infrared thermal image as the input of the radial basis neural network trained by temperature prediction, and solve the output of the radial basis neural network trained by temperature prediction. The temperature prediction value of the measurement point on the image map is also the temperature prediction value of the measurement point on the visible light image.

The concept of the infrared temperature registration and positioning method for substation equipment in the present invention is consistent with the concept of the infrared temperature registration and positioning system for substation equipment in the present invention, and will not be repeated here.

The infrared temperature registration and positioning method for substation equipment described in the present invention can predict the temperature prediction value of each measurement point on the infrared thermal image of the substation equipment, and realize the matching of each measurement point on the infrared thermal image of the substation equipment and the visible light image. Information sharing, the information includes the temperature prediction value of each measuring point.

Further, in the infrared temperature registration and positioning method for substation equipment according to the present invention, an image preprocessing step is also included between the steps (1) and (2), and the image preprocessing step at least includes Image equalization processing and filtering processing are performed on image images and visible light images. This processing process belongs to the basic content of digital image processing and is well known to those skilled in the art, so the present invention will not be explained in detail.

Further, in the infrared temperature registration and positioning method of substation equipment according to the present invention, the step (3) includes the following steps:

(3a) Construct hidden layer basis functions;

(3b) Constructing a Radial Basis Neural Network based on the hidden layer basis functions;

(3c) Use several pixel values as input samples, use the temperature bar temperature values corresponding to the several pixel values as the corresponding output of the input samples, carry out temperature prediction training on the radial basis neural network, and solve the radial basis neural network parameters of the network.

In the above scheme, the hidden layer basis function of the radial basis neural network has various forms, and the most commonly used one is the Gaussian kernel function.

Further, in the above-mentioned substation equipment infrared temperature registration positioning method,

The hidden layer basis function constructed in step (3a) is a Gaussian kernel function, and its expression is:

${\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; p</span>}$

Among them, X is the n-dimensional input vector, X=[x ₁ ,x ₂ ,…,x _n ], n is the number of neurons in the input layer; c _j is the center of the basis function of the jth hidden layer, which is the same as X has a vector of the same dimension; R _j (Xc _j ) is the output value of the jth hidden layer neuron, p is the number of hidden layer neurons; σ _j is a standardized constant, that is, the variance of the Gaussian kernel function ;

The expression of the radial basis neural network constructed in step (3b) is:

${\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; p</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}$

Among them, y _k is the output value of the kth output layer neuron, m is the number of output layer neurons; wj _,i is the distance between the jth hidden layer neuron and the ith input layer neuron connection weight;

The parameters in step (3c) include the data center c _j of the hidden layer basis function, the normalization constant σ _j and the connection weight w _j,i , which are solved by the least square method.

In the above scheme, the parameters are more than the number of equations, and it is an overdetermined equation to be solved, which needs to be solved by using a numerical solution method, and the present invention adopts the least squares method. Solving overdetermined equations using the least square method is a method well known to those skilled in the art, and will not be described in detail here.

Further, in the infrared temperature registration and positioning method for substation equipment mentioned above, the value range of σ _j is [0.01, 0.05].

Compared with the prior art, the infrared temperature registration and positioning system for substation equipment described in the present invention has the following beneficial effects:

1) It can predict the temperature prediction value of each measuring point on the infrared thermal image of the substation equipment, and realize the matching and information sharing of each measuring point on the infrared thermal image of the substation equipment and the visible light image;

2) It can carry out all-round infrared temperature monitoring and registration and positioning of substation equipment, and obtain the temperature prediction value and coordinates of each measurement point by directly consulting the visible light image, so it can reduce the heat loss caused by the blurring of the infrared thermal image. It is difficult to locate the abnormal position, reduce the error of positioning accuracy, and simplify the difficulty of subsequent fault analysis and processing in the infrared temperature monitoring of the substation, thereby improving the intelligence level of the substation.

The infrared temperature registration and positioning method for substation equipment described in the present invention also has the above effects.

Description of drawings

FIG. 1 is a schematic diagram of an overall architecture of an infrared temperature registration and positioning system for substation equipment according to an embodiment of the present invention.

Fig. 2 is a working flow chart of an infrared temperature registration and positioning system for substation equipment according to the present invention in an embodiment.

detailed description

In the following, the system and method for infrared temperature registration and positioning of substation equipment according to the present invention will be further explained and described in conjunction with the accompanying drawings and specific embodiments.

Fig. 1 schematically shows the overall architecture of the infrared temperature registration and positioning system for substation equipment according to an embodiment of the present invention.

As shown in Figure 1, this embodiment includes: an infrared camera, which collects an infrared thermal image of a target scene; a visible light camera, which collects a visible light image of a target scene; a video server, which is connected to the infrared camera and the visible light camera through video lines respectively ; As a data processing and analysis unit, the computer is connected to the video server, receives the infrared thermal image and the visible light image of the substation equipment collected by the infrared camera and the visible light camera, and configures the infrared thermal image and the visible light image of the same target scene Accurate, to match the measurement points of the infrared thermal image and the visible light image of the same target scene; the computer establishes a radial basis neural network, and predicts the temperature of each measurement point on the infrared thermal image through the radial basis neural network Prediction value, and the temperature prediction value is corresponding to each measurement point of the visible light image matching with each measurement point on the infrared thermal image. In this embodiment, the infrared camera is also directly connected to the computer through a network cable, so that the computer transmits control parameters to the infrared camera, thereby controlling the infrared camera to perform operations such as focusing, aperture discharge, and area setting. This embodiment also includes a rotatable pan-tilt. The infrared camera and visible light camera are arranged on the pan-tilt. The pan-tilt is connected to the video server to receive the control signal from the video server, so that the rotation of the pan-tilt can be controlled by the video server. This embodiment also includes a lithium battery pack and an inverter connected to the lithium battery pack, and the inverter is connected to an infrared camera, a visible light camera, a rotatable pan/tilt, a video server, and a computer respectively, so that the direct current provided by the battery pack It is converted into alternating current and supplied to infrared cameras, visible light cameras, rotatable pan-tilts, video servers and computers.

Fig. 2 schematically illustrates the workflow of the infrared temperature registration and positioning system for substation equipment in an embodiment of the present invention. As shown in Figure 2, the workflow of this embodiment is:

The computer receives the infrared thermal image and visible light image of the same target scene of the substation equipment collected by the infrared camera and the visible light camera, and performs image equalization, filtering preprocessing and registration fusion on the infrared thermal image and visible light image to obtain the The measurement points of the infrared thermal image and the visible light image are matched; the computer outputs the human-computer interaction interface through the display, and the human-computer interaction interface includes two windows that can be clicked to select the measurement point. The window displays the infrared thermal image; the computer establishes a radial basis neural network for temperature prediction, including steps:

The basis function of the hidden layer is constructed based on the Gaussian kernel function, and its expression is:

${\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; p</span>}$

Wherein, X is an n-dimensional input vector, X=[x ₁ , x ₂ ,…, x _n ], in the present invention, input refers to the pixel value (R, G, B) of each point in the image; n is The number of neurons in the input layer, because the pixel value of each point in the image has only 3, so n=3 in the present invention; cj is the center of the _jth hidden layer basis function, and has the same dimension as X The value range of the c _j component is the same as the pixel value component of the input image, that is, 0≤c _j ≤255; R _j (Xc _j ) is the output value of the jth hidden layer neuron, p is the hidden The number of neurons in the layer; σ _j is the normalization constant, that is, the variance of the Gaussian kernel function;

Construct radial basis neural network based on the above hidden layer basis function, the expression is:

${\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; p</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}$

Wherein, y _k is the output value of the kth output layer neuron, and m is the number of output layer neurons, and the output layer neuron represents the temperature prediction value among the present invention, so m=1; wj _{, i} is the jth The connection weight between a hidden layer neuron and the i-th input layer neuron;

Take a number of pixel values as input samples, take the temperature bar temperature values corresponding to the several pixel values as the corresponding output of the input samples, conduct temperature prediction training on the radial basis neural network, and solve the parameters of the radial basis neural network, which include hidden The data center c _j , the normalization constant σ _j and the connection weight w _j,i with the layer basis function are solved by the least square method:

The selection of the number p of neurons in the hidden layer takes into account the accuracy and computational complexity and is set to 45; in order to simplify the difficulty of solving, it is assumed that σ _j in the basis function of the hidden layer is equal, and the value range of σ _j is between 0.01 and 0.05 In this embodiment, 0.028 is used; in the iterative process of temperature prediction training, any integer value between 0 and 255 is randomly selected, and the c _j obtained by the final solution is:

R G B c ₁ 98 124 16 c ₂ 211 45 78 c ₃ 33 121 92 c ₄ 204 133 145 c ₅ 59 twenty three 245 c ₆ 238 231 190 c ₇ 195 226 169 c ₈ 211 112 133

c ₉ 146 199 66 c ₁₀ 202 38 245 c ₁₁ 84 158 138 c ₁₂ 57 66 8 c ₁₃ 80 114 178 c ₁₄ 149 215 133 c ₁₅ 212 50 15 c ₁₆ 74 77 227 c ₁₇ 103 123 84 c ₁₈ 220 86 59 c ₁₉ 157 204 29 c ₂₀ 253 252 79 c ₂₁ 52 41 58 c ₂₂ 211 60 166 c ₂₃ 172 179 17 c ₂₄ 63 96 70 c ₂₅ 121 248 72 c ₂₆ 102 248 224 c ₂₇ 153 164 113 c ₂₈ 204 219 193 c ₂₉ 27 102 154 c ₃₀ 209 161 200 c ₃₁ 214 251 29 c ₃₂ 90 143 250 c ₃₃ 110 238 216 c ₃₄ 146 184 13 c ₃₅ 179 123 119 c ₃₆ 189 163 83 c ₃₇ 193 226 161 c ₃₈ 99 51 59

c ₃₉ 109 101 148 c ₄₀ 244 253 154 c ₄₁ 146 103 153 c ₄₂ 217 168 114 c ₄₃ 70 230 9 c ₄₄ 159 254 131 c ₄₅ 150 167 104

The obtained connection weight w _j,i table is (j corresponds to the row in the table, and i corresponds to the column in the table):

221.5928652 179.6582716 61.6122449 233.9444619 230.3126588 183.8514565 202.7950514 104.755969 41.71688667 226.9271137 215.6851312 108.7026239 174.1148756 46.45351002 94.09030251 255 217 235 127.0308205 19.00102848 127.9914406 213.0395243 156.088203 49.01078632 233.9841563 227.0374844 147.3673979 181.7492711 67.59766764 67.22157434 219.7627689 183.7478517 63.73127693 62.78984947 10.91224745 108.8839429 241.6939286 234.3474573 211.9652781 164.911797 37.93877551 115.0612245 100.426183 17.32960756 121.4033098 223.0571531 198.1744171 71.52620932 200.9854227 118.1428571 37.98542274 189.9642921 99.11218965 42.83484773 233.5897373 233.969171 198.0237146 212.9773564 163.7101633 48.302986 31.23217367 7.147030574 87.30131153 149.0072674 22.04602674 127.1647273 201.9852867 110.9412065 39.98448776 233.0763542 227.9741434 162.8921623 138.607485 20.98342527 125.8541424 118.8619963 16.88569389 126.9539988 229.029112 221.8932843 130.6620881 209.8709721 147.8709721 46.85781435 50 0 36

190.8705556 90.56609916 48.575959 170.0405698 39.03399094 104.8717116 218.5510204 171.4460641 55.63848397 209.9913301 145.4614489 47.99133014 190.2831133 82.42055606 52.82347491 226.2851023 211.2114 93.2726245 206.637702 135.6938011 45.70145093 177.7475202 60.67346939 77.4364848 110.9545597 16 123.6137579 221.9953761 202.3076354 81.41629763 222.9987675 193.9473349 68.00855936 153.9825073 24.03498542 124.2099125 158.7777457 30.58002193 116.5233534 202.453561 125.2524798 42.60166257 39.07679623 8.863679249 101.6191808 56.73435388 13.65833964 107.2892672

After the computer establishes the radial basis neural network for temperature prediction, the computer can use the pixel value corresponding to the measurement point on the infrared thermal image as the input of the radial basis neural network trained by temperature prediction to solve the radial basis neural network trained by temperature prediction. The output to the basic neural network is the temperature prediction value of the measurement point on the infrared thermal image, that is, the temperature prediction value of the measurement point on the visible light image. The operator clicks the "select visible light image" or "select infrared thermal image" button to select the visible light image or infrared thermal image through the human-computer interaction interface:

In this embodiment, first select the visible light image, click on a measuring point, and the computer will mark the red circle 1, and the computer will judge whether the measuring point exceeds the measuring point range of the matching infrared thermal image. Then the computer obtains the matching infrared thermal image measuring point through the matching relationship between the measuring points, and obtains the temperature prediction value of the infrared thermal image measuring point through the radial basis neural network prediction, and uses the temperature prediction value as the red The temperature prediction value of the measuring point of the visible light image at circle 1 is output and displayed, and the figure shows that the output value of this embodiment is 21.6311;

In this embodiment, select the infrared thermal image again, click a measurement point, the computer will mark the green circle 2, and the computer will mark the measurement point of the visible light image corresponding to the green circle 2 as a yellow circle 3, and output its coordinates X axis 123, Y Axis 100: The predicted temperature values of the measuring points of the infrared thermal image at the two green circles are obtained through radial basis neural network prediction and output and displayed. The figure shows that the output value of this embodiment is 8.293.

The infrared temperature registration and positioning method of the substation equipment described in the present invention can take the workflow of the above system as an implementation mode, which will not be repeated here.

It should be noted that the above examples are only specific embodiments of the present invention, and obviously the present invention is not limited to the above embodiments, and there are many similar changes accordingly. All modifications directly derived or associated by those skilled in the art from the content disclosed in the present invention shall belong to the protection scope of the present invention.

Claims (9)

1. a substation equipment infrared temperature registered placement system, is characterized in that, comprising:

Thermal camera, it gathers the Infrared Thermogram of target scene;

Visible light camera, it gathers the visible images of target scene;

Video server, it is connected by video line respectively with thermal camera and visible light camera;

Data Management Analysis unit, it is connected with described video server, it receives Infrared Thermogram and the visible images of the substation equipment of thermal camera and visible light camera collection, and registration is carried out, to mate the Infrared Thermogram of same target scene and each measuring point of visible images to the Infrared Thermogram of same target scene and visible images; Described Data Management Analysis unit sets up radial base neural net, and obtained the temperature prediction value of each measuring point on Infrared Thermogram by radial base neural net prediction, and this temperature prediction value is corresponded on each measuring point of the visible images mated with each measuring point on Infrared Thermogram.

2. substation equipment infrared temperature registered placement system as claimed in claim 1, it is characterized in that, described thermal camera is also directly connected by netting twine with Data Management Analysis unit, to make Data Management Analysis unit to thermal camera transmission control parameters.

3. substation equipment infrared temperature registered placement system as claimed in claim 1 or 2, it is characterized in that, also comprise rotating The Cloud Terrace, described thermal camera and visible light camera are arranged on described The Cloud Terrace, described The Cloud Terrace is connected with video server, with the control signal of receiver, video server.

4. substation equipment infrared temperature registered placement system as claimed in claim 1, it is characterized in that, the inverter also comprising electric battery and be connected with electric battery, described inverter is connected respectively with thermal camera, visible light camera, video server and Data Management Analysis unit, changes alternating current into be supplied to thermal camera, visible light camera, video server and Data Management Analysis unit with direct current electric battery provided.

5. a substation equipment infrared temperature registered placement method, is characterized in that, comprise step:

(1) Infrared Thermogram and the visible images of the same target scene of substation equipment is obtained respectively;

(2) visible images is mated with the picture element matrix of Infrared Thermogram, to make each measuring point on each measuring point complete Corresponding matching Infrared Thermogram on visible images;

(3) build radial base neural net and temperature prediction training is carried out to it;

(4) using pixel value corresponding to the measuring point on Infrared Thermogram as the input of the radial base neural net of training through temperature prediction, solve the output of the radial base neural net through temperature prediction training, this output is the temperature prediction value of the measuring point on Infrared Thermogram, is also the temperature prediction value of the measuring point on visible images.

6. substation equipment infrared temperature registered placement method as claimed in claim 5, it is characterized in that, between described step (1) and (2), also comprise Image semantic classification step, described Image semantic classification step at least comprises carries out image equalization process and filtering process to Infrared Thermogram and visible images.

7. the substation equipment infrared temperature registered placement method as described in claim 5 or 6, it is characterized in that, described step (3) comprises the steps:

(3a) hidden layer basis function is built;

(3b) radial base neural net is built based on described hidden layer basis function;

(3c) with some pixel values for input amendment, export accordingly as described input amendment using the temperature strip temperature value that described some pixel values are corresponding, temperature prediction training carried out to described radial base neural net, solves the parameter of radial base neural net.

8. substation equipment infrared temperature registered placement method as claimed in claim 7, is characterized in that:

The hidden layer basis function built in step (3a) is gaussian kernel function, and its expression formula is:

$R_j(X-c_j)=\exp (-||X-c_j|{|}^2/2{\mathrm{\&sigma;}}_j^2),j=1,2,...,p$

Wherein, X is that n ties up input vector, X=[x ₁, x ₂..., x _n], n is the number of input layer; c _jfor the center of a jth hidden layer basis function, it is the vector with X with same dimension; R _j(X-c _j) be the output valve of a jth hidden layer neuron, p is the number of hidden layer neuron; σ _jfor generalized constant, i.e. the variance of gaussian kernel function;

The expression formula of the radial base neural net built in step (3b) is:

$y_k=\underset{j=1}{\overset{p}{\&Sigma;}}{w}_{j,i}\exp (-||x-c_j|{|}^2/2{\mathrm{\&sigma;}}_j^2);i=1,2,...,n;k=1,2,...,m$

Wherein, y _kfor the neuronic output valve of a kth output layer, m is the neuronic number of output layer; w _j,ifor the connection weights between a jth hidden layer neuron and i-th input layer;

Described in step (3c), parameter comprises the data center c of hidden layer basis function _j, generalized constant σ _jand connection weight w _j,i, by least square method, it is solved.

9. substation equipment infrared temperature registered placement method as claimed in claim 8, is characterized in that: σ _jspan be [0.01,0.05].