Research Of Soft Sensor Measuring System On Flue Gas Flow Of Fossil Power Station

The technology of soft Sensor Measuring on Flue Gas Flow of Fossil Power Station is the development direction of the research area of flue gas monitoring in the future. BP network is a kind of Multilayer feed forward neural network, which has the ability of self-study and generalize. Genetic algorithm has great properties and shows unlimited development potential on the aspect on the optimization of neural network. This thesis is mainly choosing BP network and genetic algorithm to study on the prediction of SO2.First of all, this thesis introduces monitor technology of the prediction of SO2at present. Then it gives the basic introduction of sensor measuring and neural network, putting forward the feasibility on the measure of the prediction of SO2in Flue Gas Flow of Fossil Power Station. After that, this thesis builds up the model using BP neural network, proving the availability of BP neural network, and pointing out the disadvantage of BP neural network which is easy to fail on training when it traps-in local extreme.Second, this thesis improves the disadvantage on BP network sensor measuring-model. Specifically, it using genetic algorithm to configure and optimize the link weights and threshold value of BP network. And the situation of trapping the model into local extreme is avoided.Final, the improved sensor measuring model is trained and learned. After the comparative analysis of predictions between the before optimization and after optimization, it comes into the conclusion. The advanced model is more stable and more accurate. As a result, it inspects and verifies the feasibility and superiority of the sensor measuring model by using genetic algorithm to optimize the BP network.