Abstract
A Statistical Modal Analysis (SMA) methodology is developed for geometric variation characterization, modeling, and applications in manufacturing quality monitoring and control. The SMA decomposes a variation (spatial) signal into modes, revealing the fingerprints engraved on the feature in manufacturing with a few truncated modes. A discrete cosine transformation approach is adopted for mode decomposition. Statistical methods are used for model estimation, mode truncation, and determining sample strategy. The emphasis is on implementation and application aspects, including quality monitoring, diagnosis, and process capability study in manufacturing. Case studies are conducted to demonstrate application examples in modeling, diagnosis, and process capability analysis.