新阈值及阈值函数的评价和变形预测实验

The Evaluation of New Thresholding and Threshold Function and Experiment of Deformation Prediction

变形监测数据是判断对象是否会发生异常变化的重要依据,但实测数据会被带入噪声,将隐藏其实际变化规律。本文利用含有高斯白噪声的信号来评价两种新的阈值及阈值函数的去噪效果,并评价两种新的阈值及阈值函数在提高预测精度方面的效果。实验表明,较传统的硬/软阈值函数,两种新阈值及阈值函数的去噪效果更优,而且该组合方法提高了预测的可靠性和精度。

Deformation monitoring data is the significant basis to judge whether the target appear certain abnormal changes, but the observed data contain noise that could hide the actual change. In this paper, the signal pretreated by white Gaussian noise is applied to evaluate de - noising effect and assess whether two thresholds and corresponding threshold function are helpful to improve prediction accuracy. Experiments show that the their performance is superior to the hard and soft threshold function and this combined method promote the the reliability and prediction accuracy.