Research On Modeling And Control Optimization Of Ultra-supercritical Power Unit Coordinated System

With the growing demand of electricity and increasing contradiction between energy and environment, ultra-supercritical units which have higher power generation efficiency have become the mainstream of thermal power construction in China. The thermal power unit is a kind of typical nonlinear controlled object with large time delay, large inertia, strong coupling, time-varying parameters characteristics. With the implementation of the sustainable development strategy in electric power industry, the power gid integrated automation system has put forward higher requirements to coordinated control system (CCS) of power unit. At the same time, with that the new energy power generation may cause peak fluctuation to power grid, designing a safe, efficient and adaptive control scheme has great significance for maintaining grid stability and power quality, reducing the difficulty of the staff regulation and improving the economic efficiency of power plants.This article makes some studies of ultra-supercritical unit from the perspective of modeling and coordinated control. The main contents include the following two aspects.First, unit modeling.A second-order plus dead link is proposed to simplify the thermal object with large inertia characteristics, using genetic algorithm to identify the transfer function parameters. For basic genetic algorithm "premature" problem, an improved genetic algorithm is proposed, from the perspective of adaptive crossover and mutation probability. For the parameters of unit model varying with the load, a method based on recursive Bayesian probability-weighted of multi-model dynamic modeling is introduced. For global nonlinear characteristics of the system, a neural network algorithm based on nonlinear compensation is proposed, and further improve the modeling accuracy and adaptability of unit.Second, coordinated control.Analysis of the basic principles and its shortcomings of the traditional linear GPC is made, and a nonlinear GPC algorithm based on RBF neural network compensation is proposed for the global nonlinear problems of unit. The simulation results prove that the nonlinear GPC has a stronger ability to track setting signal of the nonlinear system. For traditional GPC actual reference trajectory is unpredictable, a kind of reference trajectory optimization algorithm based on the expected response is put forward to make the actual reference trajectory glide along the expected as much as possible, ensuring both the respose speed and overshoot suppression.The basic principle of fuzzy PID control is introduced, and analysis of quantitative restrictions of the classical fuzzy rules is made.In the full summary based on the results of previous studies, a simple and efficient region-vary fuzzy PID control strategy based on the error classification is porposed. By reasonably feedback "bidirectional" coordinated signal, with improved GPC controlling the power and improved fuzzy PID controlling the main steam pressure, the coordinated control of unit has been realized.