Research On Membrane Computing Optimization Methods

Membrane computing models are theoretic ones of distributed parallel computation which are some features of living cells. In a membrane computing system, the multiset of objects evolves according to the reaction rules associated with the membrane and applied in a maximally parallel, nondeterministic manner. Most membrane computing systems are computationally universal. However, there are few of practical applications. In order to develop the membrane computing, especially improve the significance of practical applications of membrane computing, several membrane computing optimization models, algorithms and a strategy are put forward.After bringing out the basic framework, the elements of membrane computing optimization algorithms are investigated. The single objective and multiobjective optimization methods are put forward. The dynamic membrane computing optimization methods are investigated. Based on the research work in theory, the membrane computing optimization methods are applied in control systems and chemical engineering. The main researches topics and contributions are as follows:Firstly, the basic framework of membrane computing for devising compartmentalized models is extended. The idea of a class of the membrane computing for optimization problems is put forward. Inspired by different functions of membranes in biology, three algorithms with different characteristics of single objective optimization are constructed. They are the cell-like membrane computing optimization algorithm, the membrane computing optimization algorithm with information redundancies, and the membrane computing optimization algorithm based on rules of shrinkage and stretch. They not only have the basic features of the standard membrane computing, but also explore the current research results of evolutionary computing. The contrastive trials with other optimization algorithms have illustrated the excellent performance of the membrane computing optimizations methods.Secondly, the multiobjective membrane computing optimization methods are investigated based on the research of the single ones. They are the multiobjectiveoptimization algorithm based on P systems (PMOA) and the multiobjective optimization algorithm with tissue-like structure (TPS). Their unique structure divides the whole population into several subpopulations, which decrease the computational complexity. The structure of PMOA is dynamic and its membranes merge and divide at different stages. Almost a dozen of popular algorithms are compared using several test problems. Simulation results illustrate that the multiobjective membrane computing optimization methods have the best performance. Their solutions are closer to the true Pareto-optimal front and distribute well. Moreover, the multiobjective membrane computing optimization methods converge fast.Based on the previously proposed static multiobjective membrane computing optimization methods, the dynamic multiobjective optimization problems are discussed and the dynamic multiobjective membrane computing optimization algorithm is developed. Moreover, an example of a dynamic multiobjective optimization problem arising from a dynamic control loop is investigated. The "membrane control strategy" is proposed as a solution scheme.Based on the research work above, the membrane computing optimization methods are applied in chemical engineering and control systems. In the systematically optimization of the binaphthol enantiomers separation process using simulation moving bed technology, membrane computing optimization method maximizes synchronously several conflicting objectives, purities of different product and productivity. The membrane computing optimization methods are used in design of controller. The objectives and constraints are analyzed and improved for the membrane computing optimization methods. The relationship between the performances and designing parameters are analyzed. The comparisons with other algorithms illustrate the superiority of membrane computing optimization methods.