基于全矢支持向量回归的设备频谱成分预测研究

Equipment Spectral Composition Prediction Based on Full Vector Support Vector Regression

支持向量机(SVM)在解决小样本、非线性及高维问题方面表现突出,支持向量回归(SVR)目前被广泛应用于设备状态趋势预测中用于故障定量分析。故障预测用于定性分析的相对较少,为进一步提高其预测精度,结合基于同源信息融合的全矢谱技术进行倍频成分预测。该方法采用全矢谱技术融合双通道信息,相比传统单通道信号提取方法,保障了SVR预测数据特征提取的完整性,提高预测精度。对特征频率进行分别预测,然后重新生成预测的频谱图。该方法应用于某电厂1号汽轮机振动数据的预测,实验结果表明,全矢支持向量回归(FVSVR)频谱成分预测方法具有较高的预测精度,可以对一些故障定性分析。

Support vector machine （SVM） is used to solve the problem of small samples, nonlinear and high dimensional problems, and support vector regression （SVR）has been widely used in the field of equipment state trend prediction. For the qualitative analysis, the fault prediction is seldom. In order to improve the accuracy of the prediction, combined with the full vector spectrum based on the fusion of homologous information to carry on the frequency multiplication component prediction. Compared with the traditional single channel signal extraction method, this method uses the full vector spectrum technology to fuse the dual channel information, and can ensure the integrity of the SVR prediction data feature extraction and improve the prediction accuracy. The characteristic frequency is predicted respectively, and then the spectrogram is regenerated in this paper. The method is applied to the prediction of the vibration data of NO.1 steam turbine in a power plant. The experimental results show that the full vector support vector regression （ FVSVR ）spectral composition prediction method has high prediction accuracy and can be used for qualitative analysis of some faults.