利用改进型DRNN神经网络控制锅炉的负压和风量

An Improved DRNN Neural-Network for Controlling Suction Pressure and Air Flow of Boiler

火电厂锅炉负压和送风系统是具有多变量、非线性及时变参数的受控对象,利用改进型DRNN神经网络来辨识系统模型,进而对PID控制器参数进行整定,实现多变量解耦控制。对负压和送风控制系统进行了设计和仿真研究,通过计算机仿真给出了仿真曲线。结果表明:系统达到了解耦目的,系统能做到稳定性高,鲁棒性强,调节及时,反应速度快,动态偏差小,无静态偏差。该控制方法适合于负压和送风控制系统,符合工程实际,控制品质好,有实用价值。

Suction and forced draft systems of utility boilers are multi-variable control objects with time varying, nonlinear parameters. System models can be identified by improved DRNN, followed by tuning the parameters of a PID controller, thereby realizing muhivariable decoupling control. A suction and forced draft system was designed and relevant simulating studies have been conducted. Simulation curves were provided by the computer. Results show that decoupling is possible and that the system can achieve high stability, strong robustness, with opportune regulation, rapid response, slight dynamic deviations and without steady state divergence. This way of control suits to suction and forced draft control systems. It can meet engineering requirements with high control and is therefore of practical value. Figs 11 and refs 4.