A Data-Driven Health Indicator-Based Predictive Maintenance Framework for Enhancing Sustainability in Smart Logistics Systems: A Case Study of Strategic Equipment at Imam Khomeini Port

Predictive maintenance is considered one of the key pillars of sustainability in smart logistics systems. The sudden failure of strategic equipment in ports, in addition to causing operational downtime and increasing costs, has negative effects on energy consumption and environmental sustainability. This study aims to enhance operational and economic sustainability in the port logistics chain by proposing a data-driven framework for health monitoring and Remaining Useful Life (RUL) prediction of Neuero grain-suction motors at Imam Khomeini Port. In this framework, a native Health Indicator (HI) is first extracted from real multi-sensor data using Principal Component Analysis (PCA). Subsequently, an LSTM neural network is developed to model the degradation trend, and an Extended Kalman Filter (EKF) is employed to reduce noise and increase prediction stability. The proposed hybrid LSTM-EKF model achieves a compared with the baseline LSTM-only model. Experimental results on ۲۸۸۰ real hourly samples show that the model, with an RMSE of ۲.۰۶ h and MAE of ۱.۶۴h, improves accuracy by more than ۵۰% relative to base models. Consequently, unplanned downtime can be reduced by up to, maintenance costs by, and energy consumption is expected to decrease, further enhancing the overall sustainability of the port logistics system.

کلیدواژه ها:

Predictive Maintenance ، Health Indicator ، Deep Learning ، Smart Logistics ، Sustainability