First Time Quality Diagnostics and Improvement through Data Analysis: A Study of a Crankshaft Line

Abstract

Current enhancement of data availability and cloud technology provide a tremendous opportunity and platform for data-driven modeling and analyses in smart manufacturing. This paper proposes a data analysis framework to diagnose the root causes of first-time quality (FTQ), where FTQ is the quality of a part when it is measured the first time after all processes/operations in a production line. Due to the fact that too many inline factors in a production line with multiple operations may impact FTQ directly or indirectly, finding the key factors is essential yet difficult. An automotive crankshaft line is studied as an example. Production real-time inline data are collected and fetched from the data warehouse. After data preprocessing and exploratory analysis, a prediction model for FTQ is built through machine learning algorithms in Python. The data analysis identifies the significant process factors that influence the part quality, consequently improvement solutions and adjustment strategies could be generated based on the analysis. This study indicates the effectiveness and sustainability of data analysis and machine learning, as applied for quality diagnostics and improvement in manufacturing systems.