基于改进型Elman神经网络和遗传算法的锅炉在线燃烧优化

On-line Optimizing Combustion for the Boiler Based on Modified Elman Neural Network and Genetic Algorithm

目前,电站锅炉燃烧优化要求在保证燃烧效率的基础上降低NOx的排放,即同时满足电站经济性和环保的要求。利用改进型Elman网络结构简单、计算量小、容易收敛和动态学习的特点,建立锅炉燃烧预测模型,并结合遗传算法的寻优特性,在线地搜寻出一定工况下可操作量的最优控制方案,可实时地指导运行人员。仿真试验结果表明,该预测模型实现了锅炉高效低氮燃烧优化寻优,且满足实时性的要求。

In the present, combustion optimization for the boilers in power station is required by reduce NOx emissions which based on ensure combustion efficiency. That is to say, economics and environmental protection should be satisfied at the same time. In this paper, modified Elman neural network is used to constitute prediction model for boiler＇s combustion, which has such characteristic as simple structure, low computational complexity, easy to convergence, dynamic learning. Ability of searching optimum point of Genetic Algorithm （GA） is used too, which is combined with modified Elman neural network. In this way, optimal control scheme of governable parameters in operating comditions can be searched, which can realize on-line guiding operator to operate. Simulation result shows that this model is effective, which can achieve optimum searching of high efficiency, low NOx combustion in the boiler, and can satisfy the demand for real-time.