Application Of Generalized Predictive Control Based On Improved Particle Swarm Optimization In The Water-level System Of Boilers

With the rapid development of industry, the boiler, as an important power device, has been widely applied in metallurgy, electric power generation, chemical industry and other fields, which results in energy consumption and security being the focus of research. As an important part of the safe operation of boilers, drum water-level control system is the most important point in the study. In practical applications, the boiler water-level control system is mainly through the traditional PID control, but the control effect often doesn’t completely reach the actual requirements. Faced with the interference of the nonlinear coupling, hysteresis, and other unknown factors, we are often at the end of our resources. At present, the intelligent control of the mostly boilers are not good, which wastes resources and pollutes the environment. In addition, a large number of manual operations pose a threat to personal safety.In order to ensure that the boiler operates safely and has control well, this thesis makes a detailed analysis of the procedure of boiler operation and dynamic characteristics of water-level control, and propose to improve application of generalized predictive control Particle Swarm Optimization (PSO) and put this application into the boiler drum water-level control system. The controlled objects, influenced by time, nonlinear and other complicated external factors in generalized predictive control, result in the low precise parameter value and a large amount of online calculations.With readjustment of the standard PSO algorithm inertia weight, fitness function, population initialization and introduction of neighborhood operator, the above improvements enhance the convergence rate and the accuracy of the parameters.Generalized predictive control based on improved particle swarm optimization is applied to the boiler water-level control system and boiler drum water-level control system model is gained by the use of the principle of the mechanism modeling. So we get the mathematical model and could analyze its dynamic characteristics. By put improved particle swarm optimization into parameter identification, we can have access to input and output data accurately and rapidly. Then identified dynamic model is brought into a generalized predictive controller. We gain controlled increment of the minimum objective function, and realize a boiler drum water-level control by applying the rolling optimization strategies. In the end, through the analysis of MATLAB simulated experiments, the drum level control system is robust and convergence is fast. When the water-level mutates, it can make adjustments quickly and stably with high anti-interference ability and great practical value.