Construction and Application of Modifying Weights with Multi-Strategies Fruit Fly Optimization Algorithm Support Vector Regression

  • Yunlong Zhong
  • Qing Wu
  • Jianxin Chen
  • Yongli Niu
  • Shuchun Pang

Abstract

The present work constructs the fruit fly optimization algorithm integrated with multi-strategies modified weights support vector regression model (FOAAMWP-SVR). It is based on fruit fly optimization algorithm (FOA) by introducing multi-strategies modified searching weights and linear modified optimal fruit fly weight, and then combining it with support vector regression (SVR). To construct FOAAMWP-SVR model, the support vector regression model based on fruit fly optimization algorithm (FOA-SVR) is firstly formed to obtain the optimal parameter of support vector regression, and proceeds to integrate with multi-strategies to modify searching weight and the optimal fruit fly weight. In addition, a simulation experiment of activity product regression of solid solution in KCl-NH4Cl-H2O ternary system is applied to test FOA and FOAAMWP-SVR algorithms, and the calculation shows that the regression error of two models is small, while FOAAMWP-SVR is more accurate than FOA model.
Published
2017-09-01
How to Cite
Zhong Y., Wu Q., Chen J., Niu Y., Pang S., 2017, Construction and Application of Modifying Weights with Multi-Strategies Fruit Fly Optimization Algorithm Support Vector Regression , Chemical Engineering Transactions, 61, 925-930.