An approximate algorithm for optimization problems based on Membrane Computing

Membrane computing is one of the youngest branches of natural computing. Inspired by membrane computing theory, a full binary tree membrane computing algorithm is presented. It divides tentative solutions into some sub-solutions and puts them into membrane region, respectively. The sub-solutions are modified simultaneously using the sub-algorithm in every region. Using evolution rule in every membrane region, FBTMC communicates between the brothers membrane region. Then the best and worst solutions in a region are sent to the parent and child regions, respectively. FBT-Membrane algorithm repeats updating and transporting solutions until a terminate condition is satisfied. Thus, the best solutions are output from the skin membrane region. The numerical experiment indicates that the new algorithm has the features of stability and strong global exploration ability. Compared with GA and PSO, FBTMC has the faster convergence and its solution is more accurate, does not fall into local optimum, and can get global optimal approximate solution on the multi-peak and high-dimensional function.