Rotary Kiln Combustion State Recognition Based on Convolutional Neural Network

Abstract

The accurate recognition of combustion state is an important part of combustion control system. The rotary kiln is nonlinear, large lag, multi-disturbance and multivariable in the combustion process, so it is difficult to realize the intelligent control of the rotary kiln. Moreover, the traditional image recognition method is easily affected by the external environment, and the accuracy of recognition depends on the extraction of artificial features, so the satisfactory results can not be obtained. In order to solve these problems, a method of flame image recognition of combustion state in rotary kiln based on convolution neural network is proposed in this paper. In this method, convolution neural network vgg-16 model is used for feature migration, and the flame images of different combustion states in rotary kiln are trained and tested by network in order to achieve the purpose of automatic recognition of combustion state. The experimental results show that the combustion control system based on convolution neural network image recognition is effective, robust and has high accuracy. Compared with the traditional image recognition method, this method can ahieve more effective accurate and more reliable automatic classification effect.