摘要
提出一种新的火电机组运行过程的稳态检测方法。通过滑动窗滤波将数据动态趋势标准化,建立了BP神经网络提取趋势,得到稳定性模糊隶属度矢量,通过模糊推理计算出系统的稳定因子(SF),从而判断系统工况是否稳定。在某负荷变化过程的稳态检测实例中,与多种方法进行了对比,结果表明稳态趋势检测方法具有较高的准确性。将该方法用于某电厂125MW机组抽汽系统的多传感器故障检测系统中,有效地降低了过渡过程带来的误诊率,表明该方法具有一定的工程实用价值。
A new way of ascertaining operational steadiness of fossil fired power plants is being proposed. The dynamic tendency of data is normalized by screening with a moving window; tendency distillation is effected by a BP neural network and a steadiness fuzzy affiliation vector is obtained. The steadiness factor （SF） of the system is then calculated by fuzzy inference, by which the steadiness of the system＇ s mode of operation can be judged. By a steadiness ascertaining example of load changing process, a number of different methods are compared. Results indicate that the steadiness tendency distillation method exhibits a higher accuracy. Thanks to the application of this method to the fault detection system of a 125 MW set＇ s steam extraction system, equipped with quite a number of sensors, the erroneous diagnosing rate of transient processes has effectively been reduced, which indicates that this method has a certain engineering practicability. Figs 4,table 1 and refs 16.
出处
《动力工程》
CSCD
北大核心
2006年第4期503-506,共4页
Power Engineering
关键词
自动控制技术
火电机组
稳态检测
故障诊断
趋势提取
autocontrol technique
fossil fired power set
state steadiness detection
failure diagnosis
tendency distillation
