Bayesian Quality Control Models for Decision-Making Under Uncertainty

Abstrak

This review examines Bayesian quality control as a framework for decision-making under these conditions. It synthesizes its inferential and decision-theoretic foundations; reviews Bayesian control charts, hierarchical models, process-capability analysis, sequential updating, Bayesian networks, and computational methods; and examines applications in real-time monitoring, quality prediction, small-sample and high-mix production, predictive maintenance and reliability, and Lean Six Sigma. Across these applications, Bayesian models represent uncertainty through posterior and predictive distributions, allowing quality teams to update estimates as sensor, inspection, or failure data accumulate. This is particularly relevant when production volumes are low, evidence comes from several sources, or decisions depend on the probability of a shift, nonconformance, or reliability shortfall. The review also finds that Bayesian methods complement established quality practices: they can extend control charts, capability assessment, and continuous-improvement work by making uncertainty explicit. Their practical value, however, depends on more than inference. Computational demand can delay online decisions; weakly grounded priors can influence posterior conclusions; probabilistic outputs can be difficult to interpret; and legacy data systems, application-specific methods, and absent benchmarking standards limit transfer across manufacturing settings.