Determining the Cost of Predictive Component Replacement in Order to Assist With Maintenance Decision-Making

Abstract

Asset and maintenance managers are often confronted with difficult decisions related to asset replacement or repair. Various analytical models, such as decision analysis and simulation, can assist a manager in making better decisions. This paper proposes that by combining renewal theory with decision analysis methods, the expected value (EV) of information for non-repairable components can be calculated. Subsequently, it is proposed that this method can be used to determine the expected replacement cost per unit time of predictive maintenance. It is argued that this predicted cost will give the maintenance decision-maker the ability to compare it to the cost of alternative maintenance strategies when choosing between strategies. Although this paper is limited to non-repairable components, the theory and methodology can also be applied to repairable systems. OPSOMMINGBate-en instandhoudingsbestuurders word dikwels gekonfronteer met moeilike besluite rakende die vervanging of herstel van fisiese bates. Verskeie analitiese modelle, soos besluitsanalise en simulasie, kan die bestuurder help om beter besluite te neem. Hierdie artikel stel voor dat deur hernubare teorieë te kombineer met besluitnemingsmetodes, die verwagte waarde van inligting vir nie-herstelbare komponente bereken kan word. Gevolglik word dit voorgestel dat hierdie metode gebruik kan word om die verwagte koste per tyd eenheid van voorspelbare instandhouding te bereken. Daar word geargumenteer dat hierdie beraamde koste die instandhoudings-besluitnemer die vermoë sal gee om die koste van verskeie instandhoudingstrategieë te vergelyk wanneer daar gekies word tussen strategieë. Hierdie artikel sal beperk word tot nie-herstelbare komponente, maar deur soortgelyke prosesse te volg, kan die teorie uitgebrei word na herstelbare stelsels.

Fuzzy ridgelet neural network prediction model trained by improved particle swarm algorithm for maintenance decision of polypropylene plant

“…However, the bacterial foraging algorithm has slow‐convergence speed; therefore, the bacterial foraging algorithm can be combined with the particle swarm algorithm to construct the novel intelligent searching algorithm, the searching ability, and convergence precision of algorithm can be improved greatly. The training effect of fuzzy ridgelet neural network can be improved based on the novel searching algorithm …”

“…The training effect of fuzzy ridgelet neural network can be improved based on the novel searching algorithm. 19 According to the variable and complex characteristics of maintenance process and cost of polypropylene plant, the fuzzy ridgelet neural network is constructed to predict the maintenance cost. The initial values of neural network are generated randomly, which reduce the prediction precision; in order to obtain the optimal initial parameters of neural network, the improved particle swarm algorithm is constructed to select the optimal initial parameters, and the proposed prediction model can have good stability and quick convergence speed.…”

Prediction of Radio Signal Failures of Communication Based Train Operating Systems by Machine Learning Methods

“…In recent years, a maintenance decision model has been established and successfully implemented in many areas. Grobellar S. and Visser J. K have combined a renewal theory and a decision analysis model to develop a model that predicts the frequency of equipment change [6]. O.F.…”

Prediction of railway switch point failures by artificial intelligence methods

“…In recent years, a maintenance decision model has been established and successfully implemented in many areas. Grobellar and Visser combined a renewal theory and a decision analysis model to develop a model that predicts the frequency of equipment change [8]. Bin Zhao et al developed a model that determines the most appropriate maintenance method with an optimal budget in petrochemical plants by the fuzzy neural network method [9].…”