基于BP神经网络和GA-PID的超超临界锅炉系统解耦控制研究

Study on Decoupling and Control of Ultra Supercritical Boiler System based on BP Neural Network and GA-PID

直流锅炉运行中,给水调节和燃料调节十分重要,但其各变量之间存在强耦合关系。本文针对1 000 MW超超临界机组直流锅炉中燃料和给水协调控制对象参数多变、强耦合的特点,提出了一种改进权值调整的BP神经网络分散解耦智能方法,实现系统解耦,然后采用遗传算法PID（GA-PID）控制方法对解耦后近似独立的两组对象进行控制。仿真结果表明：BP神经网络分散解耦算法具有很强的非线性映射能力和自适应解耦能力,GA-PID具有良好的控制效果,所设计的系统具有较强的鲁棒性,解耦控制方案能够达到理想的效果。

The feed water regulation and the fuel regulation are very important in once-through boiler,but variable parameters are significantly coupled.For the strong parameter variability and coupling characteristics of the fuel and water supply coordinated control objects in a 1 000 MW ultra supercritical boiler,an algorithm is proposed based on the BP neural network with improved weight adjustment.The PID controller optimized by genetic algorithm（GA-PID） is used to control two single input-single output systems which are approximately independent.Experiment and simulation results show that the BP neural network decoupling algorithm has strong nonlinear mapping ability and adaptive decoupling capacity,and GA-PID has good control effect.The proposed system has strong robustness and can achieve the desired results.