Artificial Plant Algorithm Different Options Strategies

Artificial plant optimization algorithm (APOA) is a population-basedstochastic optimization algorithm by simulating the plant growing mechanism.There is only one virtual tree in APOA with constant branches. Each branchproduces the energy with photosynthesis, and then makes the growth motionaccording to the relative energy. Furthermore, there will be some probabilities tobe affected by natural and human. In this thesis, several improvements aredesigned to make APOA more effective.Firstly, in the standard version of APOA, there is no furcation in eachbranch. This phenomenon is confused with the natural tree because in naturaltree, each branch maintains many furcations to improve the efficiency ofphotosynthesis. To avoid this shortcoming, we incorporate the selection strategyinto the structure of APOA, and propose one new variant called artificial plantoptimization algorithm with selection strategy. To test the performance, fourfamous selection strategies: truncation selection (TS1), elitist-tournamentselection (ETS), tournament selection (TS2) and fitness uniform selectionstrategy (FUSS) are employed, and applied to solve the classic benchmarks,such as Rosenbrock, Rastrigin, Ackley and Griewank. Simulation results showelitist-tournament selection and tournament selection can improve theperformance significantly.Secondly, from the selection pressure viewpoint, it is easy to know theselection pressures order is TS1>ETS>TS2>FUSS. Inspired by this phenomenon,a new variant containing elitist-tournament selection and tournament selectionare designed. For this new variant, the selection pressure is between ETS andTS2, and simulation results show it is more effective than elitist-tournamentselection and tournament selection.Thirdly, the artificial plant optimization algorithm with double selectionstrategies is applied to solve the wireless sensor network location problem.Simulation results show it is more effective than the traditional DV-Hopalgorithms.