Research On Combustion Stability Modeling Of Boiler Based On Multi-Information Fusion

Boiler combustion stability plays a major role in the factors that have an impact on the safety and economy of boiler operations in coal-fired power plants.It is very necessary and important to monitor the combustion status in the furnace in real time.At present,based on the experience of operating personnel to judge the combustion state of the boiler,the combustion state can be modeled according to the real-time parameters of the boiler or the extracted features of the flame image to achieve the purpose of real-time monitoring of the combustion state of the boiler.This paper presents a neural network modeling method based on multiple data fusion,which built a boiler combustion stability model to achieve the function of boiler combustion stability judgment.First,a set of detection parameters and flame image feature quantities that have obvious influence on the combustion stability of the boiler are obtained through mechanism analysis and correlation analysis.The image feature quantities include the black dragon length and the mixing strength,which are divided into the influence on the combustion state of the boiler of the factors that can reflect the combustion status of the boiler and reflect the factors.Adopting the self-organizing neural network to build the boiler combustion preliminary stability judgment model,the features of the reflect factors as the input.When the preliminary judgment of stability result was obtained,the result was compared with the experience of the operator to judge its validity.Then the time series neural network is used to build a further combustion stability judgment model.The influencing factors are taken as input and the Euclidean distance in the initial stability judgment model is taken as output to train the neural network.When the model was built,the validity of the model was verified.After testing,it is found that the number of iterations and the error of the model are within the specified range,with excellent judgments.Finally,the conclusions and prospects of the research are summarized.