Neural Network Based Modeling And Predictive Optimal Control For The Coordinated System Of Supercritical Power Unit

In order to meet the requirements of regional power grid, a supercritical powergenerating unit often operates under wide-range load-changing conditions. Since asupercritical boiler unit is a multi-variable non-linear system with large time-delayand strong coupling characteristics, which often makes the coordinated controlquality deteriorate under wide-range load-changing conditions, and can’t well meetthe unit load and main steam pressure control demand. Considering safety andstability of the power grid and economy of the power unit, it is of great significanceto improve the coordinated control quality of a power unit with advanced intelligentcontrol strategies.Based on deep analysis of the unit’s operating characteristics， a ModelPredictive Optimal Control (MPOC) method based on neural network modeling andParticle Swarm Optimization (PSO) algorithm is proposed for the coordinatedsystem control of the supercritical power unit. A BP-network based nonlinearautoregressive moving average (NARMA) model is adopted to set up thehigh-precision predictive model for the unit’s load and the main steam pressurecharacteristics. Abundant operation data over wide-range load-changing conditionsare used for model training. A high-efficiency simplified PSO (sPSO) algorithm isapplied for optimal search, which can greatly improve the convergence speed andaccuracy. An elastic search space is updated dynamically based on real-time loadand the steam pressure error feedbacks and load demand feedforward to avoidoscillation in PSO optimizing process and to improve the control precision. Theproposed MPOC scheme is programmed with MATLAB software and tested byextensive control simulation experiments in the full-scope simulator of a600MWsupercritical power generating unit. The simulation results show that the proposedMPOC method can greatly improve the load dynamic response speed of thesupercritical power unit, and at the same time keep other key parameters, such asmain steam pressure within safety limits.