A New Systematic Method for Selecting Continuous Sampling Plan Based on the Boundary Feasible Plan for In-Control Process

摘要

Continuous sampling plans (CSPs) are extensively adopted in manufacturing systems to improve outgoing quality as well as reduce inspection costs. Existing CSPs formulating inspection scheme with the maximum value of average outgoing quality limit (AOQL) contour neglect some feasible sampling plans for in-control process. A new systematic method, designated as dynamic continuous sampling plan (DCSP), is proposed for establishing boundary feasible inspection schemes and supplying all feasible inspection schemes for the practitioners. Selecting inspection scheme according to produce lot size and AOQL in CSPs leads to the complexity and perplexity for the practitioners. DCSP solves the problem by demonstrating the-bigger-the-better rule on sampling frequency for inspection scheme selection. In all CSPs, which interval CSPs can work effectively and when CSPs should be stopped are two pendent issues. DCSP successfully solves the two problems by the foundation of effective working interval and stopping rule. Unlike partial optimization in the literature, DCSP can incorporate all process control tools into a whole integer by these characteristics, such as effective working interval, stopping rule, et al to realize process quantitative and qualitative control. Simultaneously DCSP realizes closed-loop and in-process control by the natural estimator of the probability of non-conformance. The original definition of the probability of acceptance in CSPs is unsuitable for DCSP due to the different value of the probability of acceptance for the processes with same outgoing quality. The probability of acceptance in DCSP is redefined as accepting or rejecting the outgoing product flow according to average outgoing quality (AOQ). The effects of the parameters in DCSP are discussed. The results comparing DCSP with CSP-1 show the stability and controllability in outgoing quality for DCSP. A numerical example is given at last to verify the proposed method.