基于小波滤波的神经网络变形预测

运用小波滤波的的优越性，消除数据噪声，使数据更加的接近真实的数据和更具规律性，有利于我们对数据发展趋势的预测。对消噪后的数据，利用BP神经网络强大的学习能力建立预测网络。在建立网络时，输入样本为监测k时段序列k和第k-1时段变形量与再k-2时段变形量之差组成的二维向量，目标样本为小波滤波后的变形量。并与GM（1，1）。模型和回归模型进行了对比。

Prediction of Neural Network Deformation Based on Wavelet Filter

The article studies a new model for deformation prediction, which is more accurate and precise.The superiority of wavelet filter is to eliminate the data noise and make it closer to the real data and more regular, in that way it can be beneficial to the prediction of the data development trend.After wavelet filter, prediction network can be created based on the powerful learning ability of BP neural network.When creating the network, the input sample is the two dimensional vector of the deformation of k mon...