Quality Inspection Data-Driven Sampling Strategy and Path Optimization of Coordinate Measuring Machines for Body-in-White

The detection of shape accuracy of body-in-white (BIW) via coordinate measuring machines (CMM) is a critical process in automobile manufacturing quality management. Traditional CMM inspection for a whole BIW requires an exhaustive examination of over 2000 measuring points and takes up to eight hours. To improve the inspection efficiency, in this paper, a novel sampling strategy and detection path optimization method that leverages historical quality inspection data instead of computer-aided design (CAD) models is proposed. In the proposed method, three key attributes of measuring points are first defined based on quality data, and a sampling point optimization model is established. Next, genetic algorithm (GA) is used to determine the most efficient and effective sampling point subsets for reducing the number of measuring points. Based on the selected points, the detection path planning problem is then formulated as a traveling salesman problem (TSP) and solved via particle swarm optimization (PSO). Based on a real dataset, the proposed method is proved effective in reducing the number of sampling points while preserving the integrity of defect detections.

I. Introduction

CMMs have been widely used in the automobile industry to ensure the production quality of cars due to their high accuracy in quality inspection [1]. The CMM uses a probe to detect specific points on the BIW and acquires their position coordinates to determine whether they meet design requirements [2]. Typically, a complete inspection of a BIW involves more than 2,000 measuring points and takes up to eight hours. Thus, efficient sampling point selection and detection path planning have always been core research aspects of quality management in automobile manufacturing.