Multi-objective optimization of sensor array using genetic algorithm |
| |
Authors: | Zhe Xu Susan Lu [Author vitae] |
| |
Affiliation: | Department of Systems Science and Industrial Engineering Binghamton University, New York 13902, USA |
| |
Abstract: | A multi-objective optimization method using genetic algorithm was proposed for sensor array optimization. Based on information theory, selectivity and diversity were used as the criteria for constructing two objective functions. A statistic measurement of resolving power, general resolution factor, and visual inspection were used to evaluate the optimization results with the aid of principal component analysis. In each Pareto set, most nondominated solutions had better statistics than the combination using all potential sensors. Also the principal component plots showed that different vapor classes were generally better separated after optimization. The experiment results indicated that the proposed method could successfully identify a set of Pareto optimal solutions of small size; and most optimized sensor arrays provided input with improved quality, i.e. better separation of target analytes. The running time for implementing the multi-objective optimization was satisfactory. |
| |
Keywords: | Multi-objective optimization Entropy Information gain Selectivity Diversity Genetic algorithms Sensor array |
本文献已被 ScienceDirect 等数据库收录! |