人工智能技术在电站锅炉燃烧优化中的应用研究

电站锅炉的运行面临降低运行成本与降低污染物排放的双重要求，高效低污染的优化决策问题日益引起关注。由于锅炉设备结构庞大，运行条件复杂，燃料性质多变等因素，建立电站锅炉排放特性的函数模型难度极大，为满足锅炉高效低污染燃烧优化研究的需要，该文借助优化燃烧特性试验数据，建立了电站锅炉热效率与NOx排放的响应特性的神经网络与解析函数的混合模型。文中使用了非函数形式的响应模型，燃烧优化采用了十进制遗传算法。优化数值解表明，该方法可针对锅炉热效率和NOx排放的不同优化目标，给出可行的调整各风门开度等操作量的优化控制方案。

RESEARCH ON APPLICATIONS OF ARTIFICIAL INTELLIGENCE TO COMBUSTION OPTIMIZATION IN A COAL-FIRED BOILER

Coal-fired boiler operation is confronted with two requirements to reduce its operation cost and to lower its emission. In order to improve the efficiency and to reduce the emission in combustion, a model of a coal-fired boiler for NOx emission and efficiency response characteristics is needed. Such a modeling is quite difficult, due to the huge boiler architecture, complicated operating conditions, coal sort variation and etc. Based on the data of optimal combustion experiment, a new approach to combine neural network with function-type mode is developed, which results in a mixed model of NOx emission and boiler efficiency response characteristics model. Based on the model, we apply decimal genetic algorithm to solve the control problem of high efficiency and low emission boiler combustion. The solution shows that the proposed optimal algorithm may provide feasible optimal control strategy to regulate wind door aperture according to different optimal object of NOx emission and boiler efficiency.