Study On The Cellular Differentiation Optimization Algorithm And Its Applications

Optimization problems are ubiquitous in every department of national economy andvarious fields of science and technology, such as engineering design, production planning,pattern recognition and artificial intelligence, etc. Therefore，it is very important how to findthe optimal scheme in all feasible schemes as soon as possible.In view of more and more optimization problems with high complexity, this paperstudies deeply on the process of cell differentiation and multi-agent theory, and proposes anew biomimetic optimization algorithm--Cellular Differentiation Optimization Algorithm(CDOA). Specific study results are as follows:1. Biomimetic intelligent optimization algorithm. Deeply analyse of the theory andapplication of several commonly used intelligent optimization algorithms, and sum up theirexisting problems and shortcomings. At the same time point out their great deficiencies in theaspect of high-dimensional optimization problems.2. Cellular differentiation optimization algorithm modeling. The cellular differentiationprocess is deeply studied in cell biology. Inspired by the cellular differentiation, by simulatingthe cell differentiation behaviors, such as division, growth, migration, adhesion and apoptosis,the operators of cellular differentiation are designed. Each cell represents a solution in thesearching space, several cellular differentiation behaviors are exhibited for finding the optimalsolution according to the activity value of a cell. Finally, the model of cellular differentiationoptimization algorithm is built.3.Hgh-dimensional function optimization based on CDOA. The proposed CDOA isapplied to several benchmark complex functions optimization with20~1000dimensions.Experimental results show that CDOA can converge to the optimum of complex numericalfunctions with high dimensions rapidly in spite of its simple procedure and effortlessimplementation. Compared with the multi-agent genetic algorithm, experimental results showthat the proposed CDOA outperforms some of the state-of-art in high-dimension numerical function optimization.