Feature-Preserving Data Compression of Stamping Tonnage Information Using Wavelets

Simple linear profiles monitoring in the presence of within profile autocorrelation

“…Kazemzadeh, Noorossana, and Amiri (2008b) transformed polynomial regression to an orthogonal polynomial regression model and proposed a method based on using exponentially weighted moving average (EWMA) control charts to monitor the parameters of orthogonal polynomial regression model in phase II. Nonlinear profiles monitoring have also been discussed by several authors including Ding, Zeng, and Zhou (2006), Jin and Shi (1999), Moguerza, Muñoz, and Psarakis (2007), Walker and Wright (2002), Williams, Woodall, and Birch (2007), and Vaghefi, Tajbakhsh, and Noorossana (2009).…”

Monitoring roundness profiles based on an unsupervised neural network algorithm

“…With reference to the type of application faced, the examples of profile data include the tonnage signature signals measured in stamping processes at the equal crank angle sampling intervals (Jin & Shi, 1999, the vertical density profile of a particleboard measured at fixed vertical depths (Walker and Wrigth, 2002), the roundness profile of a machined part in turning process, measured at different sampling angles (Colosimo & Pacella, 2007Colosimo et al, 2008).…”

Latent variable based key process variable identification and process monitoring for forging

“…1 [4]. Acknowledging that the tonnage signal is one of the key process variables indicating process condition changes, patterns in a single tonnage signal were extracted for forming process monitoring [2,3,5] using principal component analysis. Wavelet analysis was performed on a single-channel or an averaged tonnage signal from multiple channels to extract faultsensitive coefficients [8].…”

“…Wavelet analysis was performed on a single-channel or an averaged tonnage signal from multiple channels to extract faultsensitive coefficients [8]. Segmentation of tonnage signals was combined with the wavelet analysis to identify and isolate patterns for specific segments [7,9] and for data reduction as well [2]. More recently, multiple channels of tonnage signals were contemplated concurrently to consider interactions among signals.…”