An empirical wavelet transform and important points based time-series trend analysis method in tobacco producing process

In tobacco producing process, there are many quality indicators including moisture, weight, draw resistance and ventilation, which directly affect the final cigarette product quality. This paper proposes a time series trend analysis method based on empirical wavelet transform and important points. Empirical Wavelet Transform (EWT) decomposes the time series of tobacco quality indicators, which is to extract global trend points. The important points segmentation method based on piecewise linear regression (PLR) is adopted to extract local trend points. The proposed method overcomes the limitations of traditional methods which cannot consider both local and global trends. It significantly improves the accuracy of trend analysis, which identifies and understands the patterns of these quality indicators, thereby optimizing the tobacco producing process and enhancing the quality of cigarette products.