Design of change detection algorithms based on the generalized likelihood ratio test

Capizzi, Giovanna

doi:10.1002/env.497

Cited by 14 publications

(5 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…5 These methods are described in textbooks on statistical quality control like Montgomery, 6 Ryan, 7 Kenett et al, 8 and Vardeman and Jobe, 9 to name a few. In addition to some more or less straightforward modifications, such as the combination charts of Lucas, 10,11 schemes with adaptive parameters like Capizzi and Masarotto, 12 and generalized likelihood ratio approaches (unspecified out-of-control mean), for example, Capizzi, 13 we have been facing for the last 20 years some unfortunate developments. Quite a substantial set of charts have been proposed, which share one common feature: they often exhibit excellent out-of-control zero-state average run-length (ARL) performance, which implies quick detection of changes that happen right at the beginning of the monitoring, but their detection performance deteriorates substantially when the process change happens later.…”

Section: Introductionmentioning

confidence: 99%

The impracticality of homogeneously weighted moving average and progressive mean control chart approaches

Knoth

Tercero-Gómez

Khakifirooz

et al. 2021

Quality & Reliability Eng

View full text Add to dashboard Cite

There is growing literature on new versions of “memory‐type” control charts, where deceptively good zero‐state average run‐length (ARL) performance is misleading. Using steady‐state run‐length analysis in combination with the conditional expected delay (CED) metric, we show that the increasingly discussed progressive mean (PM) and homogeneously weighted moving average (HWMA) control charts should not be used in practice. Previously reported performance of methods based on these two approaches is misleading, as we found that performance is good only when a process change occurs at the very start of monitoring. Traditional alternatives, such as exponentially weighted moving average (EWMA) and cumulative sum (CUSUM) charts, not only have more consistent detection behavior over a range of different change points, they can also lead to better out‐of‐control zero‐state ARL performance when properly designed.

show abstract

Section: Introductionmentioning

confidence: 99%

The impracticality of homogeneously weighted moving average and progressive mean control chart approaches

Knoth

Tercero-Gómez

Khakifirooz

et al. 2021

Quality & Reliability Eng

View full text Add to dashboard Cite

show abstract

“…See Lai 6 and Runger and Testik 11 . Some applications of the GLR charts were discussed by Bordignon and Scagliarini 12 , Vega et al 13 , Apley and Shi 14 , and Capizzi 15 . A GLR approach does not make the assumption of a known change time, but uses a maximum likelihood estimate (MLE) of the change time.…”

Section: Glr Chartsmentioning

confidence: 98%

Control Charts for Monitoring Fault Signatures: Cuscore versus GLR

Runger

Testik

2003

Quality & Reliability Eng

View full text Add to dashboard Cite

It is commonly assumed that an assignable cause may shift a process mean persistently to an unknown but constant value. However, there are situations such that a mean change is not persistently constant but time varying, which is known as the signature of a fault. Different forms of fault signatures may arise, especially when a residuals control chart is used to monitor an autocorrelated or closed-loop controlled process. Incorporating a priori knowledge of a fault signature, fault-sensitized control charts may be developed. In this paper, we investigate two control charts of this kind: generalized likelihood ratio (GLR) and cumulative score (Cuscore) charts. A sine wave representing a bounded signal and a linear trend representing an unbounded signal are used as fault signatures for investigation purposes. Two different cases are analyzed, a known fault signature and parameter and a known fault signature but unknown parameter. Both the initial performances and steady-state performances of these charts are analyzed. The former case is somewhat artificial since it is synchronizing the control statistic and the fault signature; the latter one is more realistic since time of the occurrence of a fault is unknown. Simulation results show that GLR charts are robust to unknown signal start times and perform better than the alternatives for the cases compared.

show abstract

“…An example of estimating the change point is to use the generalized likelihood ratio test proposed by Apley and Shi (1999), Capizzi (2001), Lee and Park (2007), Pignatiello and Samuel (2001), and Runger and Testik (2003).…”

Section: Process Monitoring and Efficiencymentioning

confidence: 99%

Linear filter model representations for integrated process control with repeated adjustments and monitoring

Park

2010

Journal of the Korean Statistical Society

View full text Add to dashboard Cite

a b s t r a c tAn integrated process control (IPC) procedure is a scheme which combines the engineering process control (EPC) and the statistical process control (SPC) procedures for the process where the noise and a special cause are present. The most efficient way of reducing the effect of the noise is to adjust the process by its forecast, which is done by the EPC procedure. The special cause, which produces significant deviations of the process level from the target, can be detected by the monitoring scheme, which is done by the SPC procedure. The effects of special causes can be eliminated by a rectifying action. The performance of the IPC procedure is evaluated in terms of the average run length (ARL) or the expected cost per unit time (ECU). In designing the IPC procedure for practical use, it is essential to derive its properties constituting the ARL or ECU based on the proposed process model. The process is usually assumed as it starts only with noise, and special causes occur at random times afterwards. The special cause is assumed to change the mean as well as all the parameters of the in-control model. The linear filter models for the process level as well as the controller and the observed deviations for the IPC procedure are developed here.

show abstract

Design of change detection algorithms based on the generalized likelihood ratio test

Cited by 14 publications

References 6 publications

The impracticality of homogeneously weighted moving average and progressive mean control chart approaches

The impracticality of homogeneously weighted moving average and progressive mean control chart approaches

Control Charts for Monitoring Fault Signatures: Cuscore versus GLR

Linear filter model representations for integrated process control with repeated adjustments and monitoring

Contact Info

Product

Resources

About