1 1.2. Main concepts 13 where both costs and benefits are quantified, and an optimal balance between the two is obtained (Dekker, 1996). This may result in reduced maintenance costs, extended asset life, maximised availability, and ensured workplace safety (Ogunfowora and Najjaran, 2023). The set of rules that provide guidance for e!ective maintenance management is the aimof a maintenance policy, and the framework to optimise such policy is called Maintenance Policy Optimisation (MPO). There are di!erent types of maintenance policies, such as reactive, proactive, and aggressive (for more, see Tinga, 2013), which are tailored to address varying operational demands and asset conditions. Applications of MO include manufacturing (Chin, Varbanov, Kleme#, et al., 2020; X. An, G. Si, T. Xia, et al., 2022), energy (Shafiee and Sørensen, 2019; Bermejo, Gomez Fernandez, Pino, et al., 2019), and civil infrastructure, such as wind farms (J. Xia and Zou, 2023), pavements (W. Chen and Zheng, 2021; Pourgholamali, Labi, and K. C. Sinha, 2023), bridges (Frangopol and Bocchini, 2012), railways (Guler, 2016), sewer mains (Obradovi$, %perac, and Marenjak, 2019), and nuclear power plants (Lapa, Pereira, and A. Mol, 2000). MO also extends beyond engineering to applications in the healthcare domain (Mahfoud and Biyaali, 2018). Dealing with MO is a complex optimisation problem due to factors such as uncertainty, scalability, dynamic environments, high-dimensional search spaces, multiple and conflicting objectives. Consequently, numerous approaches have been explored to address MO, and the literature extensively covers these methods. The most relevant reviews include: Deshmukh, Sharma, and Yadava, 2011, which distinguishes between qualitative and quantitative (also noted in Y. Sinha and Steel, 2015), as well as discrete- and continuous-time MO, providing 13 types of model classifications. Ding and Kamaruddin, 2015 categorises models into three types: certainty, risk, and uncertainty, each with further sub-categories. The degree of certainty corresponds to the information available regarding the states of nature that influence the system under optimisation analysis. Goyal, Pabla, Dhami, et al., 2017 discusses soft-computing as an approach that mimics the human mind’s ability to handle uncertainty and imprecision, aiming for adaptability, robustness, and cost-e!ectiveness in reaching near-optimal solutions. Shafiee and Sørensen, 2019 enlists di!erent solution techniques, including Markov models, operations research models (e.g., dynamic programming), Petri nets, simulations (e.g., using Monte Carlo sampling), Bayesian networks, fuzzy models, and Intelligent-based models (e.g., employing ML and Expert Systems). Also discusses group maintenance and opportunistic replacement. Syan and Ramsoobag, 2019 focuses on multi-criteria optimisation, often dealing with resource constraints and conflicting objectives, and discusses non-evolutionary and evolutionary algorithms.
RkJQdWJsaXNoZXIy MjY0ODMw