摘要
利用优化递归的BP神经网络进行锅炉飞灰含碳量建模,并对锅炉二次风配风方式的影响进行敏感性分析,同时采用群体复合形法对运行工况寻优,获得各种工况下二次风开度的优化调整方式。优化递归神经网络是以遗传算法来确定神经网络的权值,克服了BP算法易陷入局部极小等缺陷,提高了网络学习速度和精度。通过对某台300 MW机组现场试验与计算表明,该方法可以指导运行人员进行二次风开度的优化调整,降低飞灰含碳量,同时也解决了锅炉变工况下运行参数基准值的问题。
Sensitivity analysis of effect of the secondary air distribution mode on the unburned carbon is studied in this paper, meanwhile, a muhi-complex algorithm is employed to perform a search to determine the optimum solution of optimal recursive neural network （ORNN） ,which is used to set up a boiler property model,so as to obtain currently optimum distribution mode of secondary air. The weighs of ORNN are trained by the GA,and this algorithm overcomes the disadvantage that BP algorithm is easy to being trapped in local optimums. By comparing the results of in site experiments with calculation, the rationality of the method to adjust the secondary air distribution mode is proved, and the problem of the operating standard value of boiler parameters also can be solved.
出处
《锅炉技术》
北大核心
2006年第3期9-14,共6页
Boiler Technology
关键词
飞灰含碳量
二次风配风
遗传神经网络
群体复合形法
敏感性分析
the unburned carbon
secondary air distribution mode
GA- BP
multicomplex algorithm
sensitivity analysis
