水轮发电机组改进型非线性广义预测控制器设计

Design of Improved Nonlinear Generalized Predictive Controller for Hydro-generator Unit

为提高水轮发电机组调速器的鲁棒性和稳定性,设计了一种基于瞬时线性化模型的水轮发电机组改进型非线性广义预测控制器(JNGPC)。为了使初始状态的建模误差最小,依据机组历史运行数据,采用遗忘因子递推增广最小二乘法辨识控制系统的初始参数。在此基础上,引入一种改进广义预测控制算法,并对控制系统在每个采样时刻进行瞬时线性化,设计了一种基于控制系统CARIMA模型瞬时线性化的改进的非线性广义预测控制器。以我国某水电站为试验对象,对水轮发电机组调节系统不同水头下开机启动进行仿真试验,并与PID、FOPID、GPC控制器进行比较。结果表明,与PID、FOPID控制器相比,所设计的JNGPC能有效抑制开机启动工况下的转速振荡,具有更强的鲁棒性和稳定性;与GPC控制器相比,JNGPC具有更高的计算效率,更符合工业现场实时性要求。

In order to improve the robustness and stability of the governor of hydro-generator unit,an improved nonlinear generalized predictive controller(JNGPC)based on instantaneous linearization model is designed.Firstly,in order to minimize the modeling error of the initial state,the initial parameters of the control system are identified by the forgetting factor recursive augmented least squares method based on the historical operation data of the unit.And then,an improved generalized predictive control algorithm is introduced,and the instantaneous linearization of the control system at each sampling time is performed to design an improved nonlinear generalized predictive controller based on the instantaneous linearization of the CARIMA model of the control system.Taking a hydropower station as the test object,the startup simulation experiments of hydro-generator unit governing system under different water heads are carried out.Compared with the PID and FOPID controllers,the results show that the designed JNGPC can effectively suppress the speed oscillation under start-up operating conditions,and has stronger robustness and stability;Compared with the GPC controller,the JNGPC has higher computational efficiency,and is more in line with the real-time requirements of industrial field.