论文部分内容阅读
喷氨量大小不仅影响超临界锅炉选择性催化还原(selective catalytic reduction,SCR)烟气脱硝装置的效率,过量喷氨也会导致下游空预器受热面的积灰、腐蚀和造成资源浪费、二次污染,且在变负荷时,传统PID控制方式很难实现最佳控制。通过引入混结构隐含层,改善传统RBF神经网络变工况控制时的非线性和扰动适应能力,设计了基于混结构RBF神经网络(MS-RBFNN)的喷氨流量最优控制系统,用MS-RBFNN综合学习当前主要相关状态参数,以SCR脱硝装置出口NOx排放量最小作为学习训练信号,实时并行计算出最优喷氨控制流量。实验结果表明,此优化方案相对传统PID控制,具有更好的NOx排放控制效果和变工况适应能力,同时节约了喷氨量。
The amount of ammonia injection not only affects the efficiency of the selective catalytic reduction (SCR) flue gas denitrification device in the supercritical boiler, but also results in fouling, corrosion and waste of resources in the heated surface of the downstream air preheater. Pollution, and in variable load, the traditional PID control method is difficult to achieve optimal control. By introducing the hidden layer of mixed structure and improving the nonlinear and disturbance adaptability of traditional RBF neural network under variable working conditions, an optimal ammonia flow control system based on mixed structure RBF neural network (MS-RBFNN) -RBFNN comprehensive study of the current main relevant state parameters to SCR denitrification device exit NOx emissions as a minimum learning and training signals, real-time parallel calculation of optimal ammonia injection control flow. The experimental results show that this optimization scheme has better NOx emission control effect and adaptability to conditions than the traditional PID control, and saves the amount of ammonia injection.