摘要
提出了基于小波分析和支持向量机的风机故障早期预测方法。通过小波分解,将风机原始振动时间序列依尺度分解到不同层次,对每层分别采用支持向量机(SVM)预测,最后合成得到原始序列的预测值。对某铝厂排送风机的运行状态进行预测,并与其它预测方法进行了对比,结果表明该方法预测精度更高。应用该预测方法可合理安排维修时间,减少维修费用。
A forecasting method for fan faults based on wavelet analysis and support vector machine was proposed. It decomposes the original time series of fan into different layers according to the scale by wavelet analysis method and forecasts each layer separately by means of support vector machine to finally obtain the forecasting result of the original time se- ries by composion. It was used to forecast the running conditions of the blowers in an aluminum plant and obtained results having higher accuracy than those got by other forecasting methods. Its application can help arrange the maintenance time in a.reasonable way and reduce the maintenance cost.
出处
《金属矿山》
CAS
北大核心
2006年第4期55-58,共4页
Metal Mine
基金
广西自然科学基金(0447003)
广西教育厅科研项目基金(2004)资助项目。
关键词
小波分析
支持向量机
风机
故障
预测
Wavelet analysis, Support vector machine, Fan, Fault, Forecasting
