A fuzzy c-means based hybrid evolutionary approach to the clustering of supply chain
Computers and Industrial Engineering
| Hi-index | 0.00 |
Feature extraction and selection of signals is crucial to finding useful information from large volume of raw data and eventually achieving dimension reduction and effective decision making, which directly affects final results of process analysis. Effective feature extraction methodologies can make the number of selected features as small as possible and facilitate further data fusion. In this paper, we adopt a distinctive visual angle, applied group technology to reduce superfluous information and extract feature of non-stationary signals. As an overall similarity measure, the cosine of generalized angle is employed to process the array composed by multi-band energies which are obtained by wavelet packet method, further more, information grouping and dimension reduction process are performed. On the other hand, we improved fundamental grouping method and obtained the optimum core element under same similarity condition. As a result, each group has the more elements which satisfy the preset similarity threshold thereby the number of groups reaches the fewest. Comparing with fuzzy clustering method, this algorithm is with better convergence speed and a simple structure. Vibration signals acquired from a cutting tool degradation testbed was employed to evaluate the functionalities of this method, the results have shown the effectiveness.