CN-122023239-A - High-precision machine vision-based energy storage power station hot spot dynamic monitoring method and device

CN122023239ACN 122023239 ACN122023239 ACN 122023239ACN-122023239-A

Abstract

The invention discloses a high-precision machine vision-based method and a high-precision machine vision-based device for dynamically monitoring hot spots of an energy storage power station, wherein the method comprises the steps of arranging an image acquisition unit based on the layout of the energy storage power station equipment, and synchronously acquiring visible light and multispectral images of the equipment in the energy storage power station; the method comprises the steps of carrying out median filtering denoising, graying and histogram equalization preprocessing on an original image, utilizing a YOLO series algorithm to realize accurate identification and positioning of equipment, adopting a U-Net algorithm to divide the image of the equipment, screening out hot spot candidate areas through gray mean and variance thresholds, carrying out accurate judgment by combining a multispectral image fusion technology and an SVM classification algorithm, and utilizing a Kalman filtering algorithm to realize dynamic tracking on confirmed hot spots, and predicting future temperature change trend based on a ARIM model. The whole flow forms a closed-loop feedback mechanism, the system performance is continuously improved through parameter tuning and feedback optimization, and finally, the high-precision, real-time and intelligent monitoring of the hot spot of the energy storage power station is realized.

Inventors

JIA YUJIE
YANG YI
LI MENG
HUO XUANMIN
WANG JIAHAN
JIA BINCHENG
ZHANG DONGYUE
LIU YING

Assignees

国网河南省电力公司新乡供电公司

Dates

Publication Date: 20260512
Application Date: 20251216

Claims (10)

1. The utility model provides a energy storage power station hot spot dynamic monitoring method based on high accuracy machine vision which characterized in that, the method includes: The method comprises the steps of arranging an image acquisition unit comprising high-precision imaging equipment, auxiliary lighting equipment and multispectral imaging equipment based on the layout of the energy storage power station equipment, wherein the image acquisition unit synchronously acquires an original visual image and a multispectral image of target equipment in the energy storage power station through the high-precision imaging equipment and the multispectral imaging equipment according to a set time interval or a trigger condition; The image acquisition unit is connected with the data processing server through the gigabit Ethernet, performs preprocessing operation on the acquired original visual image, and improves the definition of equipment details in the image; Performing equipment characteristic extraction and positioning on the preprocessed image by adopting a deep learning target detection algorithm, and outputting category information of equipment and position coordinates in the image; dividing a target equipment region in the preprocessed image by adopting a deep learning image segmentation algorithm based on the position coordinates, calculating characteristic parameters of each sub-region, and marking hot spot candidate regions by threshold judgment based on a characteristic parameter threshold range of a normal equipment region; Carrying out fusion processing on the original visual image and the multispectral image, extracting the multidimensional feature vector of the candidate region of the hot spot in the fusion image, and outputting a judging result of whether the candidate region is the hot spot or not based on the classification model; Dynamically tracking the confirmed hot spots by adopting a Kalman filtering target tracking algorithm, and predicting trend of temperature change data of the hot spots by utilizing a time sequence analysis method; and setting hot spot temperature early warning thresholds of different levels, and carrying out early warning treatment when the detected hot spot temperature reaches or exceeds the early warning threshold.
2. The method for dynamically monitoring the hot spots of the energy storage power station based on the high-precision machine vision according to claim 1 is characterized in that the high-precision imaging equipment is a high-precision industrial camera with resolution not lower than 4K and frame rate not lower than 30fps and is provided with an optical lens with adjustable focal length, the auxiliary lighting equipment is an LED lighting equipment with adjustable light intensity and angle and has a light intensity and angle self-adaptive adjusting function, and the acquisition range of the spectral imaging equipment comprises visible light and near infrared bands.
3. The method for dynamically monitoring hot spots in an energy storage power station based on high-precision machine vision according to claim 1, wherein the preprocessing operation is performed on the collected original visual image to improve the definition of the details of the equipment in the image, and the method comprises the following steps: And denoising the original visual image by adopting a median filtering algorithm, converting the color image into a gray image, and enhancing the contrast of the image by a histogram equalization method, thereby improving the definition of equipment details in the image.
4. The high-precision machine vision-based method for dynamically monitoring the hot spots of the energy storage power station according to claim 1, wherein the deep learning target detection algorithm is a YOLO series algorithm, and the model is trained through a training set marked with equipment type and position information, so that the model can identify equipment for image acquisition.
5. The high-precision machine vision-based method for dynamically monitoring hot spots of an energy storage power station according to claim 1, wherein the deep learning image segmentation algorithm is a U-Net algorithm, and the threshold is determined by statistical analysis of gray-scale mean values and variances of a large number of normal areas and hot spot areas.
6. The high-precision machine vision-based method for dynamically monitoring the hot spots of the energy storage power station according to claim 1, wherein an original visual image and a multispectral image are fused by adopting an image fusion algorithm based on weighted average, the multidimensional feature vector comprises gray features, texture features and spectral features, and the classification model is a support vector machine model.
7. The high-precision machine vision-based energy storage power station hot spot dynamic monitoring method according to claim 1 is characterized in that the Kalman filtering target tracking algorithm can be combined with a time interval between image frames, the position of the hot spot in a next frame image is predicted according to the position and the motion state of the hot spot in the current frame image, the accurate tracking of the hot spot is realized by continuously updating and correcting a predicted value, the time sequence analysis method is an autoregressive integral sliding average model, and the autoregressive integral sliding average model can be established according to historical temperature data to predict the future temperature change trend of the hot spot.
8. The method for dynamically monitoring hot spots in an energy storage power station based on high-precision machine vision according to claim 1, wherein the step of setting hot spot temperature early warning thresholds of different levels, and performing early warning processing when the detected hot spot temperature reaches or exceeds the early warning threshold comprises the steps of: setting a two-stage early warning threshold, namely starting a first-stage early warning when the temperature exceeds 80 ℃, sending out an alarm through an audible and visual alarm device, and starting a second-stage early warning when the temperature exceeds 90 ℃, and informing workers through audible and visual alarm, short messages and mails.
9. The method for dynamically monitoring the hot spot of the energy storage power station based on the high-precision machine vision of claim 1, further comprising the step of generating corresponding treatment measures according to the position, the type and the temperature change trend information of the hot spot and combining the operation state and the emergency plan of the energy storage power station.
10. High-precision machine vision-based energy storage power station hot spot dynamic monitoring device is characterized in that the device comprises: the image acquisition unit comprises a high-precision industrial camera, self-adaptive auxiliary lighting equipment and multispectral imaging equipment based on the distributed arrangement of the energy storage power station equipment; the data communication module adopts a gigabit Ethernet protocol to realize high-speed data transmission between the image acquisition module and the data processing server and supports real-time image stream transmission; The image preprocessing module is used for preprocessing the acquired original visual image and improving the definition of equipment details in the image; The device identification positioning module integrates a deep learning target detection algorithm to extract and position device characteristics of the preprocessed image and output category information of the device and position coordinates in the image; the hot spot candidate identification module is used for dividing the target equipment area in the preprocessed image into subareas based on the position coordinates by adopting a deep learning image segmentation algorithm, calculating characteristic parameters of each subarea, and judging and marking the hot spot candidate area through a threshold value; The multispectral fusion judging module is used for carrying out fusion processing on the original visual image and the multispectral image, extracting gray scale, texture and spectral feature vectors of the candidate region, and outputting a hot spot judging result based on a support vector machine classification model; The dynamic tracking prediction module integrates a Kalman filtering tracking algorithm and a time sequence analysis model, so that motion trail tracking and temperature change trend prediction of the confirmed hot spots are realized; And the intelligent early warning decision module is used for setting early warning thresholds of hot spot temperatures of different levels, and carrying out early warning processing when the detected hot spot temperature reaches or exceeds the early warning threshold.

Description

High-precision machine vision-based energy storage power station hot spot dynamic monitoring method and device Technical Field The invention belongs to the technical field of energy storage power station safety monitoring, and particularly relates to a high-precision machine vision-based method and a high-precision machine vision-based device for dynamically monitoring hot spots of an energy storage power station. Background Along with the increasing global demand for clean energy, energy storage power stations are widely used as key facilities for realizing energy storage and flexible allocation. However, during operation of the energy storage power station, the devices such as the battery pack, the electrical connection components and the like may generate hot spots due to aging, poor contact, abnormal heat dissipation and the like. If the hot spots cannot be found and treated in time, serious accidents such as fire and explosion can be caused, and huge economic loss and casualties are caused. At present, the hot spot monitoring method of the energy storage power station mainly comprises a contact type temperature measuring method and a non-contact type temperature measuring method. The non-contact temperature measuring method such as infrared thermal imaging temperature measurement can realize non-contact measurement and can rapidly acquire a large-area temperature distribution image, but has the problems that the measuring precision is greatly influenced by environmental factors, tiny hot spots cannot be accurately identified, dynamic tracking is difficult to realize and the like. In addition, most of the existing monitoring methods can only perform static temperature detection, and cannot effectively analyze and predict the development trend of hot spots. The high-precision machine vision technology has strong capability in the field of target detection and image analysis, and can realize the precise identification and positioning of a target object through the feature extraction and analysis of images. The high-precision machine vision technology is applied to monitoring the hot spots of the energy storage power station, so that the defects of the prior art are hopeful to be overcome, and the high-precision and dynamic monitoring of the hot spots is realized. Disclosure of Invention The invention aims to provide a high-precision machine vision-based method and a high-precision machine vision-based device for dynamically monitoring the hot spot of an energy storage power station, which solve the problems that the existing method for monitoring the hot spot of the energy storage power station is low in measurement precision and cannot realize dynamic tracking and trend prediction, realize high-precision, real-time and dynamic monitoring of the hot spot of the energy storage power station, and improve the operation safety and reliability of the energy storage power station. The invention adopts the following technical scheme for solving the technical problems: The invention provides a high-precision machine vision-based method for dynamically monitoring hot spots of an energy storage power station, The method comprises the following steps: The method comprises the steps of arranging an image acquisition unit comprising high-precision imaging equipment, auxiliary lighting equipment and multispectral imaging equipment based on the layout of the energy storage power station equipment, synchronously acquiring an original visual image and a multispectral image of target equipment in the energy storage power station through the high-precision imaging equipment and the multispectral imaging equipment according to a set time interval or a trigger condition, wherein the image acquisition unit is connected with a data processing server through a gigabit Ethernet; The method comprises the steps of preprocessing an acquired original visual image, improving the definition of equipment details in the image, extracting and positioning equipment characteristics of the preprocessed image by adopting a deep learning target detection algorithm, and outputting category information of equipment and position coordinates in the image; Carrying out fusion processing on the original visual image and the multispectral image, extracting the multidimensional feature vector of the candidate region of the hot spot in the fusion image, and outputting a judging result of whether the candidate region is the hot spot or not based on the classification model; Dynamically tracking the confirmed hot spots by adopting a Kalman filtering target tracking algorithm, and predicting trend of temperature change data of the hot spots by utilizing a time sequence analysis method; and setting hot spot temperature early warning thresholds of different levels, and carrying out early warning treatment when the detected hot spot temperature reaches or exceeds the early warning threshold. In one embodiment, the high-precision imaging device is a high-prec