摘要
SO2的脱硫率是电站湿式石灰石/石膏脱硫系统脱硫效率的关键。为保证较高的脱硫效率,吸收塔内浆液pH的控制是其中最重要的控制参数之一。采用递推遗忘因子最小二乘法(REFLS),用于神经网络在线学习,并结合广义预测控制(GPC)方法应用到吸收塔浆液pH值控制中。仿真结果表明,采用基于神经网络的非线性系统广义预测控制(NNGPC),系统的输出能很好地跟踪设定值,可以满足系统实时控制要求,在工程实践上具有较强的实用性。
High desulfurization efficiency of SO2 is very important in limestone/plaster desulfurization system,which can be insured by controlling the slurry pH of absorbing tower within certain limits. In this paper, recursive forgetting factor least-square algorithm is applied in online learning of neural network,which is combined with general predictive control （GPC） to control the slurry pH. Simulation result shows that the GPC of nonlinear systems （NNGPC）based on neural network is better in respect of tracking the set value,and it can satisfy the real-time control demand,which is more practical in engineering practice.
出处
《锅炉技术》
北大核心
2005年第6期74-78,共5页
Boiler Technology
