Multi-objective optimization of the environmental-economic dispatch with reinforcement learning based on non-dominated sorting genetic algorithm

This paper presents an improved non-dominated sorting genetic algorithm II (NSGA-II) approach incorporating a parameter-free self-tuning by reinforcement learning technique called learner non-dominated sorting genetic algorithm (NSGA-RL) for the multi-objective optimization of the environmental/economic dispatch (EED) problem. To evaluate the performance features, the proposed NSGA-RL approach is investigated on ten multi-objective benchmark functions. Besides, to evaluate the effectiveness of the proposed approach, the standard IEEE (Institute of Electrical and Electronics Engineers) of 30-bus network with six generating units (with/without considering losses) is adopted, with operating cost (fuel cost) and pollutant emission as two conflicting objectives to be optimized at the same time. In comparison to literature, it was observed that the proposed approach provides a better satisfaction level in conflicting objectives with well distributed Pareto front, in comparison with the classical NSGA-II method, and to other existing methods reported in the literature. The NSGA-RL was found to be comparable to them considering the quality of the solutions obtained, with the advantage of non-time spent for parameters tuning.