An Optimal Equipment Replacement Model Using Logical Analysis of Data

An Optimal Equipment Replacement Model Using Logical Analysis of Data

Alireza Ghasemi (Industrial Engineering Department, Dalhousie University, Halifax, Nova Scotia, Canada) and Sasan Esmaeili (Industrial Engineering Department, Dalhousie University, Halifax, Nova Scotia, Canada)
Copyright: © 2015 |Pages: 14
DOI: 10.4018/ijsds.2015040105
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

In this study, Logical Analysis of Data (LAD) is used to propose an optimal equipment replacement model. Unlike most classification techniques, LAD has the advantage of not relying on any statistical theory which enables it to overcome the conventional problems concerning the statistical properties of datasets. LAD is employed to estimate the equipment's survival and failure probabilities. These probabilities are then used to build a dynamic programming model to minimize the average long-term replacement cost of the equipment. The proposed method is successfully applied on Prognostics and Health Management challenge dataset provided by NASA Ames Prognostics Data Repository. The performance of the model is compared to that of the well-known Proportional Hazards Model.
Article Preview

1. Introduction

Condition Based Maintenance (CBM) is a maintenance program that takes into account equipment’s health condition indicators and age to optimize or improve the maintenance decisions (Jardine et al. 2006). CBM uses the equipment age and collected information concerning its health, in order to identify hidden failure, to prevent failure and/or to determine the optimal maintenance actions. Optimal maintenance action is identified by considering the cost and risk of failure and predictive replacements.

Logical Analysis of Data (LAD), first introduced in (Crama et al. 1988), is a combinatorics, optimization and Boolean logic based methodology for analysis of datasets. The typical objective of LAD is to extract knowledge hidden in records of a dataset in order to detect the sets of causes that would lead to certain effects. LAD has been broadly applied for the analysis of datasets from different fields such as medicine, biotechnology, economics, finance, politics, properties, oil exploration, manufacturing and maintenance. Several researchers applied LAD in medical fields such as cell growth, breast cancer, coronary risk, and electrocardiography in order to predict behavior of medical models (Abramson et al. 2005, G. Alexe et al. 2006, S. Alexe et al. 2003, and Lauer et al. 2002). Alexe et al. (2004 and 2005) used LAD on medical data such as B-cell lymphoma, and ovarian cancer in order to diagnose medical diseases. LAD is applied in various other fields such as voting, credit card scoring, housing, labor productivity, country risk, composition of soil in the oil, genotyping, and psychometric in order to discover knowledge from the data and estimate the behavior of the models (G. Alexe et al. 2008, Boros et al. 2000, A. B. Hammer et al. 1999, P. L. Hammer et al. 2006, P. L. Hammer et al. 2004, and Kim et al. 2008). Yacout et al. (2010), Bennane et al. (2012), and Mortada et al. (2011 and 2012) applied LAD to diagnose equipment failure in case studies of power transformer, oil transformer, aircraft, and rotor bearing. LAD proved to be a promising technique that provides interpretable results and its performance is comparable to pioneer techniques in equipment failure diagnostics.

Unlike the earlier applications of LAD on industrial equipment data, our focus is not on diagnostics, or predicting an imminent failure, i.e. prognostic but on developing an optimal replacement model to minimize the average long-term cost of the maintenance system. We do so by integrating the outcome of LAD prognostic and a dynamic programming model. To the knowledge of the authors, this application of LAD is untested in the literature.

In the following section, we will briefly explain LAD methodology and its application to predict equipment’s chance of survival using the age and health conditions of the equipment. Then wewill introduce a replacement policy to minimize the long-term replacement cost of the equipment using a dynamic programing approach.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing