Offline Neural Network-Assisted RCBA for Solving Large-Scale Economic Dispatch

Economic dispatch (ED) problem which aims to reduce cost of generations is a traditional but great important topic in power system research. However, the evaluation of original function in execution of intelligent optimization algorithm is gradually become an economically/computationally expensive problem as increasing of scale of power system. This problem poses great challenges to intelligent optimization algorithms which need a lot of time to evaluate function. Thus, this paper presents a developed method to solving ED problem of large-scale complex power system using off-line neural network-assisted hybrid bat algorithm (RCBA). Specifically, neural network model is used to replace the expensive calculation process based on fitness function in RCBA which combines chaotic map and random black hole model. Chaotic maps are applied to avoiding premature convergence. Random black hole model is used to prevent premature convergence, enlarge exploitation area and accelerate convergence. Experimental results on IEEE 118-bus system and IEEE 300-bus system with different constraints have illustrated the effectiveness of approach to solve ED problem.