In this paper, we introduce multi-objective deep centralized multi-agent actor-critic (MO-DCMAC), a multi-objective reinforcement learning method for infrastructural maintenance optimization, an area traditionally dominated by single-objective reinforcement learning (RL) approaches. Previous single-objective RL methods combine multiple objectives, such as probability of collapse and cost, into a singular reward signal through reward-shaping. In contrast, MO-DCMAC can optimize a policy for multiple objectives directly, even when the utility function is nonlinear. We evaluated MO-DCMAC using two utility functions, which use probability of collapse and cost as input. The first utility function is the threshold utility, in which MO-DCMAC should minimize cost so that the probability of collapse is never above the threshold. The second is based on the failure mode, effects, and criticality analysis methodology used by asset managers to assess maintenance plans. We evaluated MO-DCMAC, with both utility functions, in multiple maintenance environments, including ones based on a case study of the historical quay walls of Amsterdam. The performance of MO-DCMAC was compared against multiple rule-based policies based on heuristics currently used for constructing maintenance plans. Our results demonstrate that MO-DCMAC outperforms traditional rule-based policies across various environments and utility functions.

Aged earthworks constitute a major proportion of European rail infrastructures, the re-placement and remediation of which poses a serious problem. Considering the scale of the networks involved, it is infeasible both in terms of track downtime and money to replace all of these assets. It is, therefore, imperative to develop a rational means of managing slope infrastructure to determine the best use of available resources and plan maintenance in order of criticality. To do so, it is necessary to not just consider the structural performance of the asset but also to consider the safety and security of its users, the socioeconomic impact of remediation/failure and the relative importance of the asset to the network. This paper addresses this by looking at maintenance planning on a network level using multi‐attribute utility theory (MAUT). MAUT is a methodology that allows one to balance the priorities of different objectives in a harmonious fashion allowing for a holistic means of ranking assets and, subsequently, a rational means of investing in maintenance. In this situation, three different attributes are considered when examining the utility of different maintenance options, namely availability (the user cost), economy (the financial implications) and structural reliability (the structural performance and subsequent safety of the structure). The main impact of this paper is to showcase that network maintenance planning can be carried out proactively in a manner that is balanced against the needs of the organization.