The Design Of The Artificial Plant Algorithm Geotropism Operator

Artificial plant optimization algorithm (APOA) is inspired by natural plantgrowing mechanism. In the standard version, it includes photosynthesis operator,phototropism operator and apical dominance operator. In this article, the maincontritions are listed as follows:(1) Population diversity is a key factor for evolutionary algorithms. For thestandard APOA, population diversity is small when it falls into local optima, andin this case, the different growing period is omitted because the difference oflight intensities are nearly the same. Therefore, a random perturbation strategy isdesigned to overcome this shortcoming. When the population diversity is slow,one predefined exponential form will be used to adjust the differences amonglight intensities. To test the performance, several famous benchmarks are used,and simulation results show this strategy can escape from local optimaefficiently.(2) Geotropism is one important character. As the gravity results,it increasethe efficiency of photosynthesis. Inspired by this phenomenon, we suppose onevirtual centroid defined by the light and gravity, and a new operator is designedto guide the search pattern. Simulation results show APOA with geotropismoperator is better than the standard version significantly.(3) Cluster structural optimization problem is a NP problem. The maincharacter is the number of local optima increases exponentially. In this article,APOA with geotropism operator is used to solve this problem. To furtherinvestigate the performance, seeding technique and L-BFGS are also employed.Simulation results show the validity.