The Application Of Membrane Computing In Function Optimization

Membrane computing knows as P system, is a new branch of natural computing. P system with its special distributed and parallel, non deterministic characteristics has been widely used in many fields.In order to make the function more efficient and shorten the execution time, we called function optimization algorithm. Many problems in real life can be transformed into function optimization problems, and these functions are usually exhibit nonlinear, high dimension, non differentiated and non convex. Because of the fast convergence speed, high accuracy in optimization problems, optimization algorithm is widely used. The traditional optimization algorithm has the following drawbacks:their mechanism and the single structure of the constraints caused by sensitivity to initial values and falling into local minimum condition.Due to the drawbacks of traditional functions optimization algorithm and the high dimension of the complex optimization problems. This paper proposed two optimization algorithms inspired by the genetic algorithm. The performance of the proposed algorithms test on the benchmark functions. The main research results are as follows:(1) A function optimization algorithm based on P systems and differential evolution mechanism (DE-MO). There are three kinds of evolution (mutation, crossover, the selection) as evolutionary rules of membrane frame. The use of transport mechanism of membrane computing put the best object in the current iteration into each basic membrane as the next iteration update to increase the convergence speed of the algorithm. The membrane structure of this experiment is cell type of P system. The DE-MO algorithm has tested in the benchmark functions, whose comparative test algorithms are DE variants.(2) A function optimization algorithm based on P systems and artificial bee colony mechanism (ABC-MO). It is a kind of intelligent algorithm evolution mechanism, which is introduced in the framework of the P system and the evolution is artificial bee colony algorithm. The transport rule of membrane computing improves the speed of convergence of algorithm. The model structure is cell P system used in experiment, the ABC-MO algorithm has tested in the benchmark functions, whose comparative test algorithms are ABC variants.Considering these algorithms contain random factors, each algorithm in the experiment uses the mean and standard deviation as the optimization measure. The average reflects an algorithm getting the optimal solution of the average performance, and the standard deviation is used to evaluate the stability of the algorithm. Comparing the experimental results show that the proposed optimization algorithms are superior to other algorithms on the whole.