摘要
对于像电站这样的复杂工业过程,基于数学模型的故障诊断方法难以应用。为此,提出了一种基于动态时间归整(DTW)技术的故障诊断方法。该方法不依赖过程数学模型,而是基于过程数据分析来进行故障诊断,因而适合在电站生产过程中应用。首先,通过历史数据分析和过程知识,建立故障模式库。然后在故障诊断过程中利用DTW技术将检测样本与故障模式库进行模式匹配。最后根据相似性尺度找出故障模式库中与之最匹配的故障样本,从而得出诊断结果。以电站主汽温控制过程为例对该方法进行了仿真研究。实验结果表明,该方法具有较高的诊断精度,并对电站生产过程的时变特性具有良好的鲁棒性。
Considering the difficulties of applying mathematical models for fault diagnosis of so complicated process, as existing in power plants, a diagnosing method, based on the dynamic time warping （DTW） technique, is being proposed, which doesn＇t depend on mathematical models but is based on data analysis of relevant processes, and therefore is suitable for power plants. First of all, a fault pattern bank is formed through historical data analysis and process knowledge. Then later on in the course of fault diagnosing, detected samples are subjected to pattern matching with the fault pattern bank by making use of the DTN technique. Finally, a best matching fault sample, stored in the bank, is found according to its measure of comparability and with it the result of diagnosis. Taking fresh steam temperature control process as an example, a simulation test of the mentioned method has been performed, which shows that, considering the versatility of behavior of operational processes in power plants, the said method is featured by relatively high diagnosing precision and robustness. Figs 2, tables 3 and refs 8.
出处
《动力工程》
EI
CSCD
北大核心
2006年第3期396-399,共4页
Power Engineering
基金
国家自然科学基金资助项目(50576022)
关键词
自动控制技术
故障诊断
动态时间归整
电站过程
相似性尺度
鲁棒性
automatic control technique
fault diagnosis
dynamic time warping
power plant process
comparability measure
robustness
