Integrated machine learning algorithms for predictive maintenance of construction equipment

In the domain of hydraulic excavator predictive maintenance (PdM), the introduction of machine learning (ML) has been seen to revolutionize the field of construction project management to a great extent. In this study, the impact of ML on the maintenance of the equipment has been examined. The literature studied for this paper includes the domain of PdM for the reduction of overheating and leakage failures, estimation of oil and fuel consumption, and vibration reduction. The effectiveness of the algorithms, including Decision Tree, Support Vector Regression, LightGBM, and XGBoost, on the augmentation of the working of the hydraulic excavator has been highlighted. One of the main advantages of ML is its ability to efficiently go through the large database, effectively identifying the causes of failure. This helps to reduce the instances of unforeseen shutdowns of the equipment. The accuracy of the LightGBM algorithm, which has been utilized for the predictions, has been high, as indicated by the R2 value of 0.998 for the predictions. The proposed approach has been investigated for the five potential causes of failures, including Overheating, Hydraulic Oil, Fuel Consumption, Vibration, and Leakage. The feature importance technique has been utilized to compare the effects of the potential causes on the maintenance needs of the equipment. One of the interesting findings of this study is the identification of Hydraulic Oil as the most important factor for the determination of the time for the maintenance of the equipment. The findings of this study are significant for the construction management field, as the timely intervention for the maintenance of the equipment can result in a reduction of costs and time, along with an improvement in the quality of the work.