Applied Study On Combustion Optimizing Of Supercritical Thermal Power Plant Boiler Based On Artificial Intelligence And Numerical Simulation

The development of supercritical units is very quickly in China. Manufacturing technology of supercriticle boiler were introduced to China but its operation technology is still in research, so how to achieve combustion optimizing operation technology of (ultra) supercriticle boiler is still domestic electric power industry issues of concern. The efficiency, safety and environmental protection are three important part during combustion in boiler. This paper aimed at combustion optimization application on domestic (ultra) supercriticle boiler , by using the method of the study of least squares support vector machines (LS-SVM), the genetic algorithm (GA) , the Multi-Objective Differential Evolution Algorithm (MODE) and the numerical simulation method.In the economic aspect, boiler efficiency is significant influenced by burning heat loss of q2 and q4, so this research respectively analyzes the factors of exhaust temperature which was close relationship with q2 and the carbon-in-ash content which was close relationship with q4. Through actual combustion data of an 1000MW ultra supercritical boiler, respectively established the Least Squares Support Vector Machine (LS-SVM) model of exhaust temperature and carbon-in-ash content. Then the RBF kernel parameters and regularization parameters of LS-SVM model were optimized by programing. The influence rule for feed water temperature, oxygen content, air leakage rate of air preheater and the total air to exhaust temperature were analyzed. The influence rule for primary wind temperature, second wind temperature, oxygen content and secondary air baffle opening to carbon-in-ash content were analyzed. Morever, by using genetic algorithm (GA) for the two LS-SVM model, the exhaust temperature and carbon-in-ash content after adjusting the parameters were optimized in particular conditions.Then, the influencing factors of NOx emission concentrations and boiler efficiency were analyzed. With a 600MW supercritical boiler as the research object, through adjusting every layer of burn-out air baffle opening, its effect on NOx emission concentration has been studied by experiments. And according to the real-time running data of the boiler, the multi-objective model of least squares support vector machine for boiler efficiency and NOx emission concentrations was established. Using Multi-Objective Differential Evolution Algorithm (MODE) to optimize the model, Pareto solutions was obitained, and the multi-objective optimization adjustment strategy was put forward.Based on combustion safety, with a 600MW supercritical boiler as the research object, the flue gas deviation at the horizontalflue was studied. The research established the model of the boiler, meshed the model to prevent false diffusion and adopt related calculation formula. The the flue gas deviation at the horizontalflue was simulated under four different working conditions. The simulation results show that: in the right side of horizontal flue, flue gas temperature and velocity are both significantly higher than the left side of the supercritical boiler of corner tangential firing; the reversed tangential SOFA (Separated OFA) can effectively reduce the flue gas deviation. The more reversed tangential angle increase and the SOFA were further to furnace's outlet, the more significantly of reducing flue gas deviation.The study of supercritical boiler combustion provides a reference for the optimization method and has certain directive significance for combustion optimization and adjustment.