A Feature Extraction Method for Shearer Cutting Pattern Recognition Based on Improved Local Mean Decomposition and Multi-Scale Fuzzy Entropy

Aiming at accurately identifying shearer cutting patterns, this article proposes a new feature extraction method based on improved local mean decomposition (LMD) and multi-scale fuzzy entropy (MFE). The cubic trigonometric Hermite interpolation was used to calculate local mean and envelope estimate functions to improve LMD decomposition results and a sum of product functions was acquired. Furthermore, MFE, referring to the calculation of fuzzy entropy over a range of scales, was designed to measure the complexity and self-similarity of vibration signals and extract the features from the decomposition results. Subsequently, the obtained feature vectors were fed into two classifiers of support vector machine and back propagation neutral network to realize the cutting pattern recognition. The experimental results indicate the applicability and effectiveness of the methodology and demonstrate that the proposed algorithm could perform better in identifying different cutting categories of shearer.