基于煤耗小指标和NO_x排放的电站锅炉混合建模与优化  

Hybrid Modeling and Optimization of Power Station Boiler on the Basis of Small Coal Consumption Index and NO_x Emission

作　　者：练纯青 郭俊 洪喜生 LIAN Chunqing;GUO Jun;HONG Xisheng(Inner Mongolia Jingtai Electric Power Generation Co.,Ltd.,Fengzhen 012100,Inner Mongolia,China)

机构地区：[1]内蒙古京泰发电有限责任公司,内蒙古丰镇012100

出　　处：《能源与节能》2018年第7期2-5,50,共5页Energy and Energy Conservation

基　　金：国家自然科学基金项目(61364009)

摘　　要：提出一种基于BP网络和自适应模糊神经网络(AFNN)的混合建模方法,并采用遗传模拟退火算法(GA-SA)进行在线寻优,实现锅炉的实时优化燃烧指导。BP与AFNN的结合克服了过程数据与实验室化验数据不同步,数据量偏差过大的问题,建模中提出基于煤耗小指标和NO_x排放建立锅炉热效率,相比传统方法精度更高。最后,通过在内蒙古京隆电厂锅炉机组上应用该算法,结果表明在保证NO_x排放达标的基础上,煤耗与优化前相比降低了大约1g/(kW·h),从而验证了所提出算法的可行性和有效性。This paper proposed a kind of hybrid modeling method based on BP network and adaptive fuzzy neutral network （AFNN） and it sought for online optimization by adopting genetic algorithm-simulated annealing algorithm（GA-SA） so as to real- ize the real-time optimized combustion guidance of boilers. The combination of BP and AFNN had overcome the problems that process data did not synchronize with laboratory test data and excessive deviation of data volume. In modeling, it put forward that establishing thermal efficiency of the boiler on basis of small coal consumption index and NOx emission enjoyed higher precision compared with traditional methods. At last, by applying the above algorithm on boiler unit in Jinglong Power Station of Inner Mongolia, the result showed that coal consumption was reduced by about 1 g/（kW .h） after optimization on the basis of ensuring the NOx emission standards, which proved the feasibility and effectiveness of the proposed algorithm.

关 键 词：电站锅炉 自适应模糊神经网络 遗传模拟退火算法 煤耗 NOx 

分 类 号：TK227[动力工程及工程热物理—动力机械及工程]