Abstract:
In a sensor system, the selection of appropriate sensors is very important to obtain a better classification performance. An optimized set of sensors is necessary for accurate analysis of different analytes. Adding many sensors to sensing systems does not improve the accuracy of the classification. On the contrary, the noise generated from the redundant sensors negatively affect the accuracy of the classification. In this research, robust and reliability-based multi-objective sensor array optimization models are proposed to optimize the sensor arrays under uncertainty. Both selectivity and diversity criteria have been considered for constructing the objectives functions. A sensor system prototype capable of detecting analytes like smokes and volatile organic compounds has been designed and used to demonstrate the proposed model. A statistical criterion, general resolution factor (GRF) and Principal Component Analysis (PCA) are used to evaluate the optimization results. The experimental results indicate that the proposed methods can successfully identify the Pareto optimal solutions and an optimized set of sensor array, providing improved input quality for the pattern recognition.
