Performance analysis of sequential tests in process control
- Northwestern Univ., Evanston, IL (United States). Dept. of Industrial Engineering/Management Science
In recent years, a great deal of emphasis has been placed on quality control of industrial processes. In particular, numerous statistical techniques exist which are designed to continually check an industrial process for machine or component failure, thereby determining if the process is under control, or if it is becoming out of control. In this study, the authors consider a very powerful class of quality control techniques known as sequential tests. Sequential tests classify a set of observations in a manner similar to statistical hypothesis tests, but are characterized by a random sample size. Perhaps the best known sequential test is Wald`s sequential probability ratio test. The sequential probability ratio test (SPRT) is a log likelihood ratio based test for simple or composite hypotheses. After taking each observation, the SPRT decides whether to accept the null hypothesis, reject the null hypothesis, or continue sampling. For the purpose of quality control, an SPRT can be conducted repeatedly over time as incoming observations are received. In the following section, the authors reviews the sequential probability ratio test and state some of its properties. In Section 3, they formulate a Markov additive model which allows them to study the sequential probability ratio test under various types of process behavior. Section 4 develops the theoretical results and the methodological approach that allows them to bound the first passage time distributions of their model. Section 5 illustrates their techniques through numerical examples.
- Research Organization:
- Argonne National Lab., IL (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- W-31109-ENG-38
- OSTI ID:
- 10141727
- Report Number(s):
- ANL/RA/CP--82382; CONF-9404136--1; ON: DE94009801
- Country of Publication:
- United States
- Language:
- English
Similar Records
Markov sequential pattern recognition : dependency and the unknown class.
Computer simulation of the Sequential Probability Ratio Test for nuclear safeguards