Artificial Plants Algorithm Design

The existing computational intelligence algorithm mainly simulated the chemical and physical laws and animal social behavior, but the growth modes of the plants were not paid enough attention. Because the adaptability of the plants to environment was strong and growth rate was slow, the simulation to the plant growing process provided a new idea for solving the multi-model optimizing problems.According to the biological background and some plant growing mechanisms, such as photosynthesis and phototropism mechanism, the article proposes a new type of random algorithm called artificial plant optimization algorithm. This algorithm makes use of the simulation technology to research photosynthesis and phototropism mechanism on the impact of plant growth. The paper builds the framework of artificial plant algorithm and gives the concrete ways of photosynthesis and phototropism operator on basis of botany research achievements.The artificial plant optimization algorithm is used for the control of chaotic systems and achieves good optimization effect with low-dimension. But the experiment results of numerical optimization with high-dimension simulation show that the effect is not satisfactory. So, considering the plant physiology characteristics again, the artificial plant optimization algorithm with apical dominance mechanism is put forward. The apical dominance mechanism is an internal part of adjusting plant growth itself and removed by human. It is designed for artificial plant optimization algorithm as the operator to avoid the trouble of the local optimum.Due to many parameters that the artificial optimization plant algorithm with apical dominance mechanism contains, the influence of those parameters on the algorithm performance is obvious. In order to determine which group of parameters can make the algorithm achieve better performance, experimental design can be used the parameter selection. The orthogonal experiment design is used to filter six parameters which have a great influence on the algorithm performance and uniform experimental design is used to choose better parameters combination to achieve the goal of improving the algorithm performance.