Research On Intelligent Detection Method Of Flue Gas Desulfurization PH Value Based On Multivariables

In the flue gas desulphurization system of thermal power plants, the PH value of seriflux in absorption tower directly affects the desulphurization efficiency which is the only indicator of whether the emission of SO2 in the thermal power plants reaches the standard. Therefore, it is important to make a timely and accurate measure of the PH value of seriflux in the absorption tower. In engineering practice, the PH value of seriflux in the absorption tower, usually measured by PH value tester, represents the potential of hydrogen of the solution in the absorption tower. However, the PH value tester goes out of operation from time to time, which will easily causes the unbalance of the potential of hydrogen of the seriflux in the absorption tower. The seriflux in the absorption tower which loses the balance of the potential of hydrogen will speed up the damage of equipments of the desulphurization system, and even causes the paralysis of the whole desulphurization system. In order to minimize the damage to the whole desulphurization system when the tester goes out of operation, the desulphurization system will adopt temporary methods to measure the PH value of seriflux in the absorption tower. Generally, the frequently-used temporary method is to measure the PH value of seriflux in the absorption tower by using redundant equipments or field sampling. Given that the traditional temporary method costs much and is time-consuming, it is important to develop a new method to measure the PH value of seriflux in the absorption tower.This paper studies the flue gas desulfurization process of wet limestone-gypsum method and comes up with the idea that artificial intelligence prediction model should be applied in the monitoring of the PH value of seriflux in the absorption tower. First, regard the influence factors of the PH value of seriflux in the absorption tower of the wet desulfurization system, such as the flue gas flow rate, the SO2 concentration, the O2 content, the powder content, the flue gas temperature, the density of the seriflux in the absorption tower and the density of the limestone slurry, as input variables, and regard the PH value of seriflux in the absorption tower as an output variable, and then establish models such as PLS model, PSO-BP neural net model, SA-SVM model and the optimized GALS-SVM model respectively. Second, predict the PH value of seriflux in the absorption tower of the desulphurization system through the existed four models with the timely monitoring data from a thermal power plant with a 600MW installed capacity in the southwest region and confirm the feasibility of the artificial intelligence model. Third, test the above-mentioned four artificial intelligence models with sample data from the same group and compare the accuracy of these models. The study and analysis shows that comparing with the other three artificial intelligence models, PSO-BP neural net model has the minimum relative error and better prediction ability. Fourth, in order to get a more accurate outcome, this paper introduces clipping average method to handle the prediction result of the four artificial intelligence models. Finally, apply the optimized artificial intelligence model to the above-mentioned plant and prove that this artificial intelligence model makes a great influence in controlling the operating cost of the desulphurization system.The research results show that artificial intelligence models can be used in the study of the prediction of the PH value of seriflux in the absorption tower through flue gas desulfurization system of wet limestone-gypsum method and can provide a more reliable guarantee for the safety production, energy saving and emission reduction and cost control of the desulfurization system.