Mathematical Modeling And Optimization Control Research Of Hot Water Boiler Combustion System

In northern China, the winter is very cold, heating is the important needs of the residents. Therefore, the heating boiler is widely used. However, the heating boiler combustion control system is by manual operation, with a low degree of automation, resulting in a low energy use and serious air pollution. The heating boiler combustion system is difficult to control with strong coupling, large delay, time-varying characteristics. According to the change of load, how to reasonably and effectivly control the heating boiler combustion system to ensure the boiler’s efficiency, environmental protection and safe operation is of great significance.This paper researches against the chain coal-fired hot water boiler, analysis its main work process and principles as well as the boiler combustion system’s main tasks. On this basis, through the study of the hot water boiler combustion’s characteristics and control methods, establish the combustion control system and clarify the relationship between variables, put forword a whole combustion control system comprising by load control, air-input control and air-output control three relatively independent subsystem.This paper has the Dalian XinYu Company’s20t/h chain coal-fired hot water boiler heating as experimental subject, through the analysis of the main parameters of the combustion system, the boiler combustion system three inputs and three outputs model is given. By analyzing the principle of multi-input multi-output system parameter identification, the hot water boiler three inputs and three outputs system is transformed into three-input single-output system. In the identification test, take M sequence as incentive input signal, model the boiler combustion system using recursive least squares parameter estimation method according to the actual operation data of the boiler, and achieve good modeling results.Against the combustion control system multi-input multi-output characteristics and the coupling between these parameters, PID neural network control algorithm is put forward,but because that their initial weights are randomly obtained, control result is easy to fall into local optimum, this paper proposes a new adaptive mutation particle swarm optimization algorithm. Taking the system identification model as a control object to conduct matlab simulation, Compared to PID neural network, Particle Swarm Optimization PID neural network,adaptive particle swarm optimization PID neural network not only responds quickly, approaches the target time shorter, but also the steady-state relative error is smaller, the problem that PID neural network algorithm’s initial weights easy to fall into local optimum and particle swarm optimization premature convergence is solved with better control effect.