Intelligent Maintenance Decision-Making for an Automated Leafy-Vegetable Production Equipment System in an Unmanned Plant Factory

Leafy-vegetable production in large unmanned plant factories relies on the coordinated operation of multiple equipment units. Equipment failures may cause system degradation or shutdown, thereby affecting production continuity and system performance. Appropriate maintenance is therefore important for improving system availability and maintaining continuous production. This study developed an equipment criticality-guided rolling maintenance decision framework integrating event-driven semi-Markov simulation, equipment criticality assessment, and Bayesian optimization. Then the system performance of different maintenance strategies was compared. The results showed that, with approximately 48 person·h of annual active maintenance resources, the strategy reduced annual shutdown time by 50.74 h compared with fixed-interval maintenance and increased capacity-weighted availability by 3.42 percentage points to 91.95%. When annual active maintenance resources increased to 144 person·h, annual shutdown time decreased by 54.56% compared with corrective maintenance, while capacity-weighted availability increased by 6.98 percentage points to 93.80%. Mean annual production loss and conditional value at risk at the 95% confidence level (CVaR95) decreased by 52.97% and 50.43%, respectively. As active maintenance resources increased, capacity-weighted availability continued to improve, but maintenance resource input exhibited diminishing marginal returns. This study improved capacity-weighted availability and reduced high production-loss risk in automated equipment systems for leafy-vegetable production in unmanned plant factories, providing a basis for rational active maintenance resource allocation.