Feature-Preserving Data Compression of Stamping Tonnage Information Using Wavelets

Abstract

Tonnage information is referred to as stamping force measurement in a complete forming cycle. Tonnage data contains rich information and features of stamping process failures. Due to its nonstationary nature and lack of physical engineering models, tonnage information cannot be effectively compressed using conventional data-compression techniques. This article presents a statistical method for “feature-preserving” data compression of tonnage information using wavelets. The technique provides more effcient data-compression results while maintaining key information and features for process monitoring and diagnosis. Detailed criteria, algorithms, and procedures are presented. A real case study is provided to illustrate the developed concepts and algorithms.

KEY WORDS:

• Data shrinkage
• Feature-lossless
• Multiresolution analysis
• Process monitoring and diagnosis
• Signal segmentation
• Stamping tonnage information processing