基于最小二乘支持向量机的声波测距系统消噪处理

The Acoustic Ranging Signal De-Noising Based on Least Squares Support Vector Machines

针对声波测距系统噪声复杂,淹没在噪声中的回波难以检测的问题,以机器统计学习理论为基础,采用最小二乘支持向量机(LS-SVM)建立系统模型,实现了声波测距系统异常值的预测和噪声的消除,并与传统的时间序列分析方法建立的自回归滑动平均求和模型(ARIMA)的消噪效果进行了仿真对比。仿真结果表明,利用最小二乘支持向量机建立的模型预测精度高,能有效地抑制声波测距系统中的噪声。

Aiming at the problem of complex noise in acoustic ranging system and weak echo difficuhly detected from signals with high noise,least squares support vector machines based on the statistical learning theory are used in model building to realizes the unusual values detection and noise elimination. Comparing the deoising result with the traditional autoregressive integrated moving average （ARIMA）, simulated results show that the proposed method can improve the prediction accuracy and restrain the noise of acoustic ranging system.