小波多尺度分析方法在磨辊磨损检测中的应用  

Application of Wavelet Multi-scale Analysis for Wear Characteristics

作　　者：曾德良 刘继伟 刘吉臻 柳玉 谢谢 

机构地区：[1]工业过程测控新技术与系统北京市重点实验室(华北电力大学),北京市昌平区102206

出　　处：《中国电机工程学报》2012年第23期126-131,160,共6页Proceedings of the CSEE

基　　金：国家科技支撑计划(2011BAA04B03)~~

摘　　要：实现预知维修对电厂经济、安全运行具有重要意义,设备状态的准确获取是预知维修的前提。针对电厂设备状态难以准确获取的问题,提出利用基于模型的小波多尺度分析方法。该方法以电厂海量历史数据为基础,从不同尺度研究模型预测值与实际值构成的残差序列,提取有效成分反映设备运行状态。由于现场情况复杂多变,对模型结构以及小波基的选取问题进行深入讨论以及定量分析。对某机组磨煤机进行实例分析表明,该方法可以正确检测出设备状态,为实现设备预知维修,提高运行经济性、安全性提供技术依据。Predictive maintenance is of great significance for power plants＇ economic and safe operation. Accurately obtaining equipments＇ status is the premise. In order to overcome the difficulty of obtaining equipment＇s status, the method of wavelet multi-scale analysis based on model was proposed. Based on masses of historical data, residuals composed by prediction model and actual value were analyzed in different scales to reflect equipment＇s status. Considering the complex situation, the chosen of model structure and best wavelet base were discussed. The result shows that, the equipment＇s status can be obtained, which provides technical basis for condition maintenance economic, Safe operation

关 键 词：状态检修 神经网络 最佳小波基 多尺度分析 磨损指数 

分 类 号：TK38[动力工程及工程热物理—热能工程]