小波和稀疏分解在非连续性薄膜去噪中的应用

为了在传感器测量锂电池非连续性膜厚前不需测量C型机构的固有频率和扫描振动频率，采用3层小波-阈值判断-稀疏分解信号处理去噪方法，进行了理论分析和实验验证。该方法不需固有频率和扫描振动频率的先验知识，在不同C型机构扫描速率模式下，通过迭代选取最佳匹配的原子序列保留锂电池薄膜厚度分布，滤除局部噪声波动，实现稀疏迭代去噪。结果表明，相对于小波算法，在缺乏先验知识的条件下，稀疏分解算法具有较好的去噪性能，其均方差值达5μm~7μm，是一种操作简单、可行有效的方法。

Applications of wavelets and sparse decomposition in non-continuous film de-noising

In order to avoid measuring the inherent frequency and the scanning vibration frequency of C-dynamic scanning system before measuring discontinuous film thickness of lithium battery with laser sensors, the 3-layer wavelet-threshold judgment-sparse decomposition signal processing de-noising method was used. Theoretical analysis and experimental verification were made. Without prior knowledge of the inherent frequency and the scanning vibration frequency and under different C-dynamic scanning mode, the best-matching atomic sequence was selected by iteration and the film thickness distribution of lithium battery was reserved, fluctuations of the local noise were filtered and sparse iterative de-noising was realized. The results show that comparing with the wavelet algorithm and in the absence of the prior knowledge, sparse decomposition algorithm has better de-noising performance and is a simple, practical and effective method. Mean square error of sparse decomposition algorithm is 5μm~7μm.