Application Of The Teaching-learning Algorithm In Power Flow Optimization Of Power Systems

In recent years, artificial intelligence and computational intelligence technology have made rapid development, some new, efficient heuristic algorithms such as particle swarm optimization(PSO), genetic algorithm(GA), simulated annealing algorithm(SAA) and differential evolution algorithm(DE) have achieved remarkable results in solving the problem of power system optimization. However, there are some limitations of these algorithms, like easy to fall into local optimum and a lot of control parameters need to be set. For example, particle swarm optimization needs to set inertia constant ω, cognitive learning factor c1 and social learning factor c2 before the algorithm begins; genetic algorithm needs to set the crossover probability Px and mutation probability Pm.Simulated annealing algorithm needs to set cooling rate, the initial temperature, end temperature and chain length; differential evolution algorithm needs to set the scaling factor F and crossover probability Cr. Recently, there is a new algorithm called teaching learning based algorithm.It provides new ideas and method for solving nonlinear complex large-scale problems.Compared with other heuristic algorithms, its biggest advantage is free of control parameters, which provides a great convenience for researchers. This paper studies the application of the algorithm in power flow optimization.Optimal Power Flow(OPF) problem is a typical multi-objective nonlinear programming problem with scalability. With the continuous expansion of the system size, the number of variables and constraints surge, nonlinear function and numerous constraints between variables become more complex,making the large-scale OPF problems more difficult to be solved.This paper summarizes the existing state of the intelligent algorithms,and then introduces teaching learning based Optimization method(TLBO) for power flow optimization. Meanwhile, in order to improve the convergence performance,a wavelet mutation strategy is used to improve the algorithm, generating a large search space at the beginning of the algorithm to achieve global search, a small search space at the end of the algorithm in order to obtain a more accurate solution. In order to verify the effectiveness and robustness of the algorithm, a power flow optimization model is set up to realize the lowest network loss, generation cost, voltage deviation and atmospheric polllutants emission on IEEE30-bus and IEEE118-bus standard test systems,and the results are compared with a variety of intelligent algorithms and show that the teaching learning based algorithm and modified teaching learning based algorithm(MTLBO) are effective and practical.