Research on improved sparrow algorithm based on random walk

The optimization problem is a hot issue in today’s science and engineering research. The sparrow algorithm has the advantages of simple structure, few control parameters and high solution accuracy, and has been widely used in the research of optimization problems. Purposing at the problem that the sparrow search algorithm (SSA) can’t take into account the global and local optimization, an improved sparrow algorithm based on random walk strategy is proposed. After the sparrow search, the random walk is used to perturb the optimal sparrow to demonstrate its search-ability. At the original of the iteration, the random walk boundary is large, which is favourable to demonstrate the whole search-ability. After several iterations, the walk boundary becomes smaller, which improves the local search-ability of the best location of the algorithm. Taking the convergence speed, algorithm stability and convergence precision as evaluation indicators, the improved Sparrow Algorithm (RWSSA) is verified by 4 unimodal functions and 5 multimodal classical test functions, and compared with the traditional Sparrow algorithm. The experimental results show that the capacity of the improved sparrow algorithm based on random walk is significantly improved. At the same time, RWSSA is put into practice the power prediction problem, which checkouts the feasibility of RWSSA in actual engineering problems.