改进DRNN在单元机组协调控制系统参数整定中应用

Application of modified DRNN algorithm in parameter adjusting of coordinated control system

为协调单元机组机炉特性差异、提高机组运行水平,针对单元机组非线性、强耦合性和参数时变的特点,设计并仿真研究了一种基于改进对角递归神经网络(DRNN)的单元机组协调控制器。该控制器采用DRNN网络整定的PID控制器代替协调解耦控制系统中的常规PID控制器,并在常规DRNN中引入带动量项的PID梯度优化算法,使DRNN网络具有更快的跟踪性和收敛性;在此基础上,针对某300MW直流燃煤机组在100%、70%负荷工况下的线性模型和某330MW机组非线性动态模型,分别进行了负荷扰动和主蒸汽压力2种扰动下的仿真研究。仿真结果表明,与常规PID协调控制器相比,所提出的协调控制器具有响应速度快、自适应能力强、抗干扰能力强等特点,表现出更好的静态性能和动态性能。

As the coordination control system of fossil-fired unit is nonlinear, strong coupled and parameter time-varying,a coordination controller based on the modified DRNN（Diagonal Recurrent Neural Network） is designed to coordinate the difference between boiler and turbine. Its PID parameters are tuned by DRNN ,and the PID gradient descent algorithm with momentum term is introduced to improve the convergence property and response performance of DRNN. Simulation research of load disturbance and main steam pressure disturbance are carried out with the linear models of a 300 MW once-through boiler at 100 % and 70 % load level and the nonlinear dynamic model of a 330 MW boiler. Results show that, compared with the conventional PID controller,the proposed controller presents better static and dynamic performance,with faster response,better self-adaptability and stronger anti-disturbance capability.