The Research Of Boiler Combustion State Intelligent Monitoring Based On Image Processing

The installed capacity of the current and the control of industrial boiler complexity continues to increase, this will give the safe and economic operation of boilers and put forward higher requirements, and real-time monitoring of boiler combustion based on flame image, to overcome the traditional flame monitoring low real-time, single function, the reliability is not high, recognition rate is low, for the promotion of industrial boiler control quality and ensure the production safety of boiler combustion, has important significance for economy and environmental protection. Industrial boiler control system for real-time monitoring of the flame image, without considering the combustion characteristic information of the flame itself, therefore this topic based on the flame image monitoring of boiler combustion status, through the introduction of artificial intelligence algorithm to achieve the real-time, accurate and intelligent monitoring boiler combustion state.The industrial boiler is the most important and complex pulverized coal boiler as the research object, research content also has significance for the oil and gas industry boiler flame image processing, in order to improve the accuracy and speed of the algorithm for the purpose, mainly from the following several aspects of the research and analysis:(1) this paper has made the improvement to the traditional flame image acquisition system, puts forward the flame image acquisition system based on the flame monitoring and full furnace flame monitoring single burner in one, and the hardware flame collection and the reliability of the system is improved. The flame image processing system based on HSI color space, analyzed the denoising, image enhancement, segmentation and pseudo color processing algorithm for fire flame image. In the aspect of flame image noise processing presents a fast adaptive median filter algorithm based on neighborhood related, at the same time, the algorithm is effective in the noise reduction can keep the flame image edge detail information well, in the coarse segmentation of flame image variance algorithm for maximum between class based on flame image segmentation and fine segmentation of flame image, based on C-V contour model algorithm, through the experiment to achieve a fast and accurate segmentation of flame image.(2) this topic first analyzed the characteristics of the flame combustion state, amount of flame image feature extraction algorithm is proposed, and the correlation of the characteristic of reliability analysis and diagnosis on the basis of choosing extract six characteristic data. Application of BP neural network model to build the boiler combustion state diagnosis model of this topic, and USES the extracted more than six group characteristics of flame image data validate the experiment, the result of simulation shows that this algorithm has higher identification accuracy.(3) this paper presents the boiler combustion system architecture state intelligent monitoring based on LavVIEW software platform, and the design of the boiler combustion state intelligent monitoring software, realizes the boiler intelligent monitoring, monitoring of boiler combustion diagnosis, parameter setting and combustion parameters real-time display and other functions, the application of LavVIEW software to capture the features of the flame image data the simulation experiment.