Study On Diagnosis Of Combustion State In Refuse Incinerator Based On Digital Image Processing And Artificial Intelligence

With the nation’s economic and social development and urbanization, the increasing quantity of municipal solid waste is becoming a difficult problem plaguing urban development. Waste incineration is a treatment of reduction, utilization and decontamination. There is no doubt that burning will be the most popular processing method. Combustion in the refuse incinerator is usually unsteady in practical operation because of the property of solid waste, which has a high moisture, changeable composition and fluctuant heat value over a wide range. Unsteady combustion brings a series of problem of economy and safety. What is more, it may cause pollution exceeding the standard. So it is necessary to monitoring the combustion state of garbage in real time, in order to adjust the combustion timely. At present, monitoring methods of garbage burning flame is mainly by artificial observation, which cannot meet the need in practical operation. In this paper, digital image processing technology and artificial intelligence are introduced to diagnose the combustion state of municipal solid waste in the refuse incinerator. The main content and results of this paper are listed as follows:(1) Seven typical combustion states of refuse incinerator and their corresponding combustion adjustment strategies are determined based on expert experience and flame image sample database.(2) The definition and algorithm model of 12 characteristic quantities of flame image are proposed.(3) Based on rough set theory, a attribute importance based algorithm that combines forward search and backing pruning is proposed. The 12 characteristic quantities are reduced to 7 dominated characteristic quantities, which compose the optimal combination of characteristic quantities. Typical combustion state characteristic sample set is established.(4) A diagnostic model based on artificial neural network is established. The sample set consisting of the optimal combination of characteristic quantities is used as the input of the network, 10-fold cross validation experiment is conducted. Accuracy of classification in this experiment is 99.05%, and variance is 2.07. The experimental result show that the diagnosis accuracy rate of the method proposed in this paper is high, and diagnosis result is rather steady. So this model is feasible in combustion diagnosis of municipal solid waste.(5). A new diagnostic method of the combustion state of refuse incinerator based on digital image processing and artificial intelligence is proposed.