Högskolan i Skövde

Industrial robots are central to modern manufacturing, yet their high energy consumption presents significant economic and environmental challenges. Traditional optimization approaches often prioritize cycle time reduction without adequately addressing power efficiency or workcell layout. This project integrates a knowledge mining process through Mimer-based Flexible Pattern Mining (FPM) into a robot simulation-based optimization framework based on ABB RobotStudio to more efficiently minimize cycle time, peak power, energy consumption, and cell area simultaneously with using NSGA-II algorithm. A simulation model of a robotic cell was modified to model trajectories and equipment placements, while a RobotStudio optimization (RSOpt) add-in was used to generate and evaluate thousands of configurations. By mining high-performing solutions in Mimer, interpretable rules were derived and applied to tighten decision bounds, reducing wasted evaluations and computational cost. The results demonstrate that Pareto fronts comparable to those in reference studies were reproduced with up to an order of magnitude fewer evaluations, achieving reductions in total energy and peak power at similar cycle times. The findings confirm that knowledge-guided optimization enhances both efficiency and sustainability while lowering computational burden, providing a practical and transferable methodology for energy-aware robotic simulation and layout design.