Limestone-gypsum Wet Flue Gas Desulphurization Efficiency Of Soft Measurement Technology

World air quality continues to deteriorate, the main polluting gases emitted as SO2and CO2, present, China is still the thermal power generation companies, power plants asthe SO2emitters, controlling SO2emissions have become important issues facing thepower generation business, which predict the desulfurization efficiency has also beenmonitoring the plant attention. Limestone-gypsum wet FGD technology in China powerplant desulfurization dominant, accounting for85%to90%, affecting more wetdesulfurization efficiency factors, and has relevance, and therefore more difficult topredict wet desulfurization efficiency. Soft-sensing technology is applied the automaticcontrol theory to the production process, it is the application process for the situations ofsome important variables are difficult to measure in practice or is not easy to measure, itshould through the high-end science and technology such as the computer, and so on.Firstly，to measurement the results of the measurement variables which is easily available,and then use some mathematical relationships to infer and estimate important variables.This measurement technology has the functions of substitute or auxiliary hardware.In this paper the application of partial least squares method, respectively, fuzzyneural networks and support vector regression machine desulfurization efficiencyestablished by the forecast model. Analysis of limestone-gypsum wet desulphurizationefficiency factors, the use of a power plant from Jiangxi DCS control system runsdesulfurization equipment collected historical data, the model in the MATLAB platformfor training and testing, to get a more accurate prediction model desulfurizationefficiency，and three kinds of prediction model were compared. The results showed that:based on support vector regression prediction model established by the desulfurizationefficiency and test sets can basically consistent sample data, the maximum relative erroris less than0.6%, with a high degree of accuracy, significantly better than the other twomethod, and the method introduction, has applied in the project is feasible.