A data-driven method for performance analysis and improvement in production systems with quality inspection

The advance of new generation of IT and sensor technologies results in data enriched production environment. However, there is a lack of an effective utilization of the data to improve productivity while reducing quality management cost. Therefore, this paper proposes a systematic method to analyze the production dynamics, and presents an event-based method to quantitatively evaluate the impact of various disruptions on system throughput, including machine breakdown and quality failure. It is proved that the impact of the events can be measured with system loss which is the summation of the production loss of the slowest machine and the overall number of defective parts produced in the subsystem where the slowest machine locates in. The data-driven method is integrated into an optimization method to exploit the optimal quality inspection allocations. In the method, a non-linear optimization problem is formulated and solved with an adaptive genetic algorithm to trade off the penalty cost of production loss and the investment cost of quality inspection. The research results in a comprehensive understanding of production dynamics subject to quality inspection and rework. It is of critical importance to boost productivity with better quality inspection allocations. Simulation studies are performed to validate the proposed methods.