Particle Swarm Optimization Applied to Generalized Predictive Control of a Solar Power Plant

In this paper, generalized predictive control based on particle swarm optimization (PSO-GPC) was applied to the control of linear Fresnel distributed collector system, of which the purpose was to reduce the system error in the dynamic process. The main goal of solar thermal power generation system control was to control the collector outlet temperature in a certain range. Using the the rate of heat transfer oil in heat collector as input, outlet temperature as output and solar radiation intensity as a disturbance signal, model of Controlled Auto-Regressive Integrated Moving Average(CARIMA ) was established. The control task of generalized predictive control (GPC) was to make the actual outlet temperature of systems as close as possible to the desired output trajectory. Gradient optimization without constraints was used to obtain the optimal control input, and gradient optimization with constraints and PSO optimization were matched with each other. A multi-mode hybrid optimization method was formed, which can obtain the optimal control increment of system quickly and accurately. GPC and PSO-GPC technology was applied to Lanzhou 200kW linear Fresnel solar thermal power generation demonstration system. Through the simulation results obtained from actual data we can see that the PSO-GPC control technology can reduce the error of the system dynamic process compared to GPC control technology.