一种基于数据挖掘的入炉燃料发热量在线智能软诊断方法研究

A Study of the Method for an On-line Intelligent Soft Diagnosis of In-furnace Fuel Low Heating Values Based on Data-mining

国内许多火电站燃烧煤质波动对锅炉的稳定燃烧和安全运行构成了严重威胁,主要矛盾反映在燃料低位发热量太低,经常在炉内无法稳定燃烧。针对锅炉燃料发热量大范围波动工况下的燃烧数据特性,提出了关联信息算法和非线性映射网络的混合模型。利用此模型对国内某300 MW电站锅炉现场燃烧数据进行了计算和分析,得到了判断燃料发热量变化的诊断规则知识,可以较好地预测燃料发热量的变化,优化运行人员的操作。该方法实施性强,投入成本小,而且还可以无缝地集成至现有的SIS平台中,完善锅炉系统的实时性能诊断模块,提升SIS系统二次开发的空间。

Coal quality fluctuations in many Chinese coal-fired power plants have posed a serious threat to boiler stable combustion and safe operation.The main contradiction lies in an excessively low heating value of fuel.Taking account of the characteristics of combustion data under the conditions of wide-range fluctuations in boiler fuel heating value,the authors have presented a hybrid model composed of a correlation information algorithm and a non-linear mapping network.By making use of the above model,a calculation and analysis has been conducted of the on-site combustion data of a 300 MW utility boiler in China.As a result,obtained was certain diagnostic knowledge governing the change in fuel heating values,which can lead to a better prediction of the change in fuel heating values and an optimized operation by operating personnel.The method can be used conveniently with a low input of outlays.Moreover,it lends itself to be seamlessly integrated into an existing SIS platform to improve the real-time performance diagnostic module of a boiler system and expand the space for the secondary development of a SIS system.