Objective-oriented optimal sensor allocation strategy for process monitoring and diagnosis by multivariate analysis in a Bayesian network

Measurement strategy and sensor allocation have a direct impact on the product quality, productivity, and cost. This article studies the couplings or interactions between the optimal design of a sensor system and quality management in a manufacturing system, which can improve cost-effectiveness and production yield by considering sensor cost, process change detection speed, and fault diagnosis accuracy. Based on an established definition of sensor allocation in a Bayesian network, an algorithm named Best Allocation Subsets by Intelligent Search (BASIS) is developed in this article to obtain the optimal sensor allocation design at minimum cost under different specified Average Run Length (ARL) requirements. Unlike previous approaches reported in the literature, the BASIS algorithm is developed based on investigating a multivariate T 2 control chart when only partial observations are available. After implementing the derived optimal sensor solution, a diagnosis ranking method is proposed to find the root cause variables by ranking all of the identified potential faults. Two case studies are conducted on a hot forming process and a cap alignment process to illustrate and evaluate the developed methods.