Data-driven multi-objective prediction and optimization of construction productivity and energy consumption in cutter suction dredging

Abstract

In dredging construction, cutter suction dredger (CSD) operation typically relies on manual experience with suboptimal control parameters, which can easily lead to low productivity and high energy consumption. This paper presents an intelligent decision-making approach for optimizing CSD control parameters based on multi-objective optimization (MOO). It employs high-dimensional feature selection techniques to identify key parameters affecting CSD performance, and develops a multi-output regressor chain-extreme gradient boosting (RC-XGBoost) model for concurrent prediction and an improved multi-objective gray wolf optimization algorithm to derive the decision-making solutions. Tian Jing Hao CSD operational data from the Pinglu Canal project in China is taken for case-study. Results show that RC-XGBoost model can effectively predict productivity and energy consumption with R² values of 0.961 and 0.989, respectively. The MOO framework demonstrates remarkable adaptability to diverse scenarios. It enables the adaptive acquisition of optimal control parameter combinations under various geological conditions, thus enhancing the operational reliability. Overall, productivity increases by 3.08 %, while energy consumption decreases by 2.40 %. This paper offers an approach for operators to optimize CSD productivity and energy consumption.