Bimodal fruit fly optimization algorithm based on cloud model learning

The Fruit Fly Optimization Algorithm (FOA) is one of the latest swarm intelligence-based methods inspired by the foraging behavior of fruit fly swarm. To improve the global search ability and solution accuracy of the FOA, a bimodal adaptive fruit fly optimization algorithm using normal cloud learning (BCMFOA) is proposed in this paper. Based on the labor allocation characteristics of the swarm foraging behavior, the fruit fly population is divided into two groups in the optimization process according to their duties of searching or capturing. The search group is mainly based on the fruit fly’s olfactory sensors to find possible global optima in a large range, while the capture group makes use of their keen visions to exploit neighborhood of the current best food source found by the search group. Moreover, the randomness and fuzziness of the foraging behavior of fruit fly swarm during the olfactory phase are described by a normal cloud model. Using a normal cloud generator and an adaptive parameter updation strategy, the search range of the fruit fly population is adaptively adjusted. Therefore, the ability of FOA to avoid local optima is enhanced greatly. Twenty-three benchmark functions are used to test the performance of the proposed BCMFOA algorithm. Numerical results show that the proposed method can significantly improve the global search ability and solution accuracy of FOA. Compared with existing methods such as PSO, DE, AFAS, the experimental results indicate that BCMFOA has better or comparative convergence performance and accuracy. The application to the multi-parameter estimation of a permanent magnet synchronous motor further confirms its good performance.