The gamma and inverse Gaussian processes are widely used to model monotonically increasing degradation phenomena. In many applications these models are treated as equivalent to each other, although this is not true. This makes the misspecification of these two processes a problem of concern. The point of this paper is to evaluate whether and how selecting the wrong model can impact on the performance of a condition-based maintenance policy recently proposed in the literature. The analyses are conducted by carrying out a large Monte Carlo study, where synthetic sets of degradation data are generated under three different gamma processes, which simulate as many experimental scenarios. The parameters of the competing models are estimated from the synthetic datasets and the resulting estimated models are used to optimize the considered maintenance policy. A misspecification is assumed to occur if the Akaike information criterion leads to prefer the wrong model. The effect of a misspecification is evaluated in terms of its impact on the long run average maintenance cost rate.
Impact on performances of a condition-based maintenance policy of misspecification of gamma with inverse Gaussian degradation process / Esposito, Nicola; Castanier, Bruno; Giorgio, Massimiliano. - (2022), pp. 800-807. (Intervento presentato al convegno 8th Intl. Symp. on Reliability Engineering and Risk Management (ISRERM 2022) tenutosi a Hannover Germany nel 4-7 September 2022) [10.3850/978-981-18-5184-1_GS-06-149-cd].
Impact on performances of a condition-based maintenance policy of misspecification of gamma with inverse Gaussian degradation process
Nicola esposito
;Massimiliano Giorgio
2022
Abstract
The gamma and inverse Gaussian processes are widely used to model monotonically increasing degradation phenomena. In many applications these models are treated as equivalent to each other, although this is not true. This makes the misspecification of these two processes a problem of concern. The point of this paper is to evaluate whether and how selecting the wrong model can impact on the performance of a condition-based maintenance policy recently proposed in the literature. The analyses are conducted by carrying out a large Monte Carlo study, where synthetic sets of degradation data are generated under three different gamma processes, which simulate as many experimental scenarios. The parameters of the competing models are estimated from the synthetic datasets and the resulting estimated models are used to optimize the considered maintenance policy. A misspecification is assumed to occur if the Akaike information criterion leads to prefer the wrong model. The effect of a misspecification is evaluated in terms of its impact on the long run average maintenance cost rate.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.