In the natural computation method, to solve the optimization problem of high-dimensional data, the population size needs to be increased to obtain higher accuracy, but at the same time, the time complexity is relatively large. If the population size is reduced, the algorithm will fall into local optimization due to the lack of population diversity. In order to solve the pro-blems such as difficult to balance population size, slow convergence rate and easy to fall into local optimum in optimization process, this paper proposed a natural calculation method based on MCE strategy, which was suitable for all kinds of optimization algorithms. It didn't depend on the specific steps of algorithm evolution and had universality. Firstly, the original solution space was divided into two types of large spaces with competitive relations, and each type of large space was decomposed into N-dim small space. Then, two different elimination methods of reverse learning and mixed mutation were carried out respectively in the two types of large spaces to eliminate the poor individuals. Finally, some better individuals in N-dim small space were selected to carry out competitive exchange across the two types of large spaces to maintain the diversity of the whole population. Thus, it improved the convergence speed and accuracy of the algorithm. It applied the proposed strategy to particle swarm optimization and genetic algorithm respectively, and compared with standard particle swarm optimization, genetic algorithm and current advanced improved swarm intelligence optimization algorithms, and verified the performance by high-dimensional classical test function. The experimental results show that the improved algorithm of multiple competition elimination has better optimization ability than other comparison algorithms and has universality.

[2023-03-20] Accepted Paper /doc/2023.01.0009/v1

[2023-08-05] Printed Article /doc/2023.01.0009/v2