论文部分内容阅读
本文介绍了最小二乘支持向量机(LS-SVM)回归的基本原理,提出了一种基于LS-SVM回归的时间序列预测器,并将其用于传感器的故障检测和数据恢复。论述了LS-SVM预测器的实现方法和步骤,并且将其应用于压力传感器的故障检测和数据恢复,同线性神经网络预测器、RBF神经网络预测器和BP神经网络预测器的比较结果表明,LS-SVM预测器具有更高的预测精度,更好的外推能力,计算效率最高,因此,LS-SVM预测器是传感器故障检测和短期数据恢复的一种有效方法。
This paper introduces the basic principle of least square support vector machine (LS-SVM) regression, and proposes a time series predictor based on LS-SVM regression, which is used for sensor fault detection and data recovery. The implementation methods and steps of the LS-SVM predictor are discussed and applied to the fault detection and data recovery of the pressure sensor. The results of comparison with the linear neural network predictor, the RBF neural network predictor and the BP neural network predictor show that, LS-SVM predictor has higher prediction accuracy, better extrapolation capability and highest computational efficiency. Therefore, LS-SVM predictor is an effective method for sensor fault detection and short-term data recovery.