Planning maintenance in manufacturing systems is challenging, especially when dealing with unpredictable machine breakdowns. This paper presents a Deep Reinforcement Learning (DRL) framework to automate and optimise maintenance decisions. Our approach uniquely models a realistic industrial environment where machine failures are stochastic and can occur in succession, a critical factor often simplified in traditional methods. The DRL agents learn to make decisions using local machine data combined with key system-wide performance metrics, enabling a modular yet globally-aware strategy. We benchmarked our DRL policy against a state-of-the-art Metaheuristic Genetic Algorithm (MGA). The results demonstrate that our DRL approach achieves two key advantages. First, it matches the production throughput of the benchmark, particularly under moderate operational stress. Second, it significantly reduces the overall maintenance workload and enhances system robustness against the variability of machine failures. This work highlights the potential of DRL to create intelligent, decentralized maintenance strategies that improve both efficiency and resilience in complex, high-variability industrial environments.

Information asymmetry and the misreporting of operational data severely challenge contemporary supply chains (SCs), leading to inefficiencies, disruptions and performance deterioration. Traditional anomaly detection techniques typically focus on detecting downstream consequences rather than proactively identifying misreporting behaviors. To address this gap, this study introduces an interpretable AI-driven framework, employing the Isolation Forest algorithm, for the early detection of misreported data in multi-tier SCs. Central to this approach is the introduction of Engineered Cross-Stage Features (ECSFs), designed to capture relational discrepancies across different SC echelons. The framework was evaluated using a synthetic three-tier SC dataset, comprising both standard operational conditions and artificially perturbed data points representing misreporting behaviors and different scenarios of information asymmetry among SC actors. In a comparative assessment against Hotelling’s T2 and an Autoencoder, our ECSF-based Isolation Forest framework demonstrated superior efficacy, highlighting its potential as a practical solution for proactively identifying and mitigating misreporting in complex SC environments.

Maintenance scheduling is critical function across numerous industries, where the dynamic complexity of operations often challenge the systems efficiency. This study explores the application of Deep Reinforcement Learning (DRL) to refine scheduling decisions by integrating a simulation tool that replicates an industrial production line. This integration eases the modelling and simulation in real-time of machine operations, job flows, and maintenance activities, capturing the dynamics and complexities of a typical production environment. Our developed tool assesses various maintenance strategies and their direct impacts on productivity, leveraging DRL to enhance decision-making capabilities. We introduce an innovative job-sequencing rule that complements the DRL framework, systematically analysing its effectiveness against traditional heuristic methods. The comparative analysis confirms that our DRL-based approach, coupled with the job sequencing rule, significantly optimises maintenance timing and resource allocation in a flow shop setting. By synthesizing simulation with intelligent algorithms, our method not only optimize maintenance tasks but also boosts overall production efficiency.

Deep Reinforcement Learning (DRL) has been included into the production system for multiple objectives, including control, scheduling, and maintenance planning. Maintenance must be planned sensibly and economically in order to preserve the usable life of the production systems while not sacrificing productivity and so minimising costs and losses. In this work a hybrid simulation-based and DRL approach is employed to develop an agent that can autonomously determine when to do preventative maintenance by considering the failure probability at a particular instant and the length of time since the last maintenance operation has been performed. The novelty of this approach is the configuration of the DRL setting, in particular the reward function. Results are promising comparing the approach with a heuristic from the literature, as they show that the frequency of machine failures is dramatically reduced.

Considering a Flow Shop production line in an Industry 4.0 setting where the Cyber-Physical System (CPS) and Internet of Things (IoTs) can be deployed, a newly Performance-based Decentralised Dispatching Rule (PDDR) is proposed. It combines known dispatching rules with the knowledge of the monitored production system state. The goal is to provide a novel dispatching rule based on production line performance oversight. The governance system considers the machine condition in terms of machine utilisation. Regarding the assessment scenario, the proposed rule has been tested and compared with the well-known Short Processing Time (SPT) and the First-In-First-Out (FIFO) rule in a higher generality way by taking into account unforeseen events that may occur in production (such as breakdowns, potential rework, micro-stops, and unplanned machine setups). The simulation results showed interesting results where the flexibility of this rule, as well as its practical use with real hypotheses are its main advantages.