Bacterial Foraging Optimization Based on Multi-colony Cooperation Strategy
Churong Zhang, Jun Yu, Ben Niu · 2020
We introduce two effective strategies into the conventional bacterial foraging optimization (BFO) to improve information utilization and promote information exchange, and propose a powerful multi-colony cooperation BFO (McCBFO). The first strategy randomly divides the whole population into multiple non-overlapping colonies, and the number of colonies in each generation is determined randomly rather than fixed. In other words, the same individual has the opportunity to be assigned to different colonies in different generations to maximize the diversity within the colony and avoid concentrated distribution. The second strategy aims to accelerate the search efficiency by improving information sharing within the same colony and strengthening the information exchange between different colonies. Specifically, the information of the best individual in each colony is shared with other individuals from the same colony and guides them to evolve towards the potential area. Besides, the poorer half of each colony exchange genetic information with randomly selected individuals from other colonies to generate diverse offspring individuals. To analyze the performance of two proposed strategies, we run BFO, (BFO + the second strategy), and McCBFO on three different dimensions (i.e., 2-D, 10-D, and 30-D) of 28 benchmark function from the CEC2013 test suite, and each function is independently run 30 times. The experimental results confirmed that our proposed McCBFO can accelerate the convergence speed and improve the convergence accuracy significantly, and we recommend using the second strategy only for simple problems while their combination for complex problems.