Intelligent Analysis Of Infrared Images Of Key Substation Equipment Based On Improved AlexNet Model

Power equipment is the fundamental component of the power grid system.Power equipment condition detection is an indispensable part of power reform.With the continuous expansion of the scale of the power grid and the increase in the demand for intelligence,the stability,reliability and safety of power equipment operations have become top priorities.Infrared detection technology is the only way to reveal the operating status of power equipment through the visualization of equipment temperature information.It can detect power equipment without power outages,regularly or in real time.It has the characteristics of non-contact,fast and accurate,effectively find hidden faults of operating equipment,so it has received widespread attention and applications in power companies.However,the current infrared image analysis of power equipment still has the shortcomings of large human factors and low image analysis efficiency.The infrared detection technology alone cannot meet the needs of power equipment detection.The combination of infrared technology with image processing technology and artificial neural network method has become the breakthrough.This article takes the 500 k V key substation equipment as the research object.It firstly introduces the principle of infrared radiation,providing a theoretical basis for detecting the temperature distribution of the equipment using infrared radiation,and combs the principle of infrared imaging,the characteristics of infrared images and related parameters of the infrared camera,etc.Secondly,clarify the key substation equipment studied in this paper,including the working principle,basic structure,fault type and infrared image characteristics of substation equipment,and divide the equipment further according to the fault type of the substation equipment.Third,unify the image pixel size and extract the image target device to construct an image sample database;By understanding the advantages and characteristics of the Alex Net network,and combining the relevant characteristics of infrared images in this article,an improved Alex Net algorithm is constructed to study the infrared image classification of substation equipment,and realize the automatic selection of the processing mode of different equipment operating states.By comparing with BP neural network,the superiority of the algorithm in this paper is determined.Fourth,based on the infrared diagnostic application specifications for the charged equipment,the corresponding operating state intelligent analysis methods are determined for different types of heating of the substation equipment under study.If it is a current-heated device,the background is separated based on the RGB and HSV color space conversion,the infrared image temperature is imported,the separated component area temperature data is extracted,and the surface temperature judgment method is used to detect the state of the power equipment;if it is a voltage-heated device,the temperature rise is not obvious,and its temperature distribution and location have a non-linear relationship.On the basis of the background separation of color space conversion,this type of devices need to be separated the background again based on the improved Ostu algorithm,input the temperature data,use adaptive meshing method to extract temperature characteristics,compare the temperature difference with the normal state equipment,and determine the equipment operating state.The paper establishes a complete intelligent analysis method and process of infrared images of key substation equipment.Finally,case studies on the substation equipment show that the research method proposed in this paper has a good effect on improving the accuracy of the infrared image state detection of the substation equipment.