基于GA和模糊关联规则的锅炉脱硝经济性优化

Economical Optimization of a Boiler Denitration System Based on GA and Fuzzy Association Rules

在某660 MW火电机组的厂级监控信息系统(SIS)中选取历史运行数据,利用最小二乘支持向量机(LSSVM)方法建立脱硝经济性预测模型,并基于该模型采用遗传算法进行常运行负荷点的离线寻优以建立离线最优专家数据库(OOED).采用模糊关联规则挖掘(FARM)算法从OOED中提取各调整变量的最优设定值与机组负荷的关联关系,实现电网负荷调度指令下各参数的在线优化调整.结果表明:所提出的脱硝经济性优化方法的优化效果与遗传算法寻优结果接近,且优化时间短,适合火电机组的在线优化控制.

A prediction model of denitrification cost was established using least squares support vector ma- chines （LSSVM） according to historical data taken from the supervisory information system of a 660 MW coal-fired boiler. Based on the model, an optimal expert database was built up for off-line optimization of the frequently-operating load points by genetic algorithm, from which the associations between unit load and adjustment variables were extracted using fuzzy association rule mining （FARM） algorithm, so as to achieve online adjustment and optimization of various parameters under power grid dispatching conditions. Results show that the algorithm proposed has a close optimization effect and a shorter simulation time on denitrification cost when compared with the genetic algorithm, which therefore is suitable for online opti-mization and control of thermal power units.