Search

The basic sine cosine algorithm (SCA) has some shortcomings such as low accuracy, slow convergence and easy to fall into local optima when using it solve high-dimensional complex optimization problems. An improved SCA (iSCA) was proposed. Firstly, a nonlinear conversion parameter rule based on inverted sigmoid function was designed to replace the original linear strategy, and a good transition from global search to local search was achieved. Secondly, the personal historical best information was embedded into the position search equation to guide the optimization process. The solution precision was further improved and convergence was accelerated. Finally, a somersault foraging mechanism was introduced to generate a new position to increase the population diversity, thereby reducing the probability of falling into local optima. 10 high-dimensional benchmark test functions,10 high-dimensional UCI datasets and two wind turbine fault datasets were selected for experiments, and the results of iSCA were compared with the basic SCA, memory-guided SCA(MSCA), and improved grey wolf optimizer(I-GWO) algorithms. The results show that iSCA has better performance than other methods in terms of solution precision and convergence speed. The proposed algorithm is effective for solving the high-dimensional optimization problems. [ABSTRACT FROM AUTHOR]