论文部分内容阅读
This study uses an electronic nose including 20 sensors to recognize the pathogen species of male rats, and each rat is injected with three kinds of pathogens respectively.How to remove the redundancy and correlation among sensors is the key to wound diagnosis.Traditional feature selection algorithm is complex especially in medical electronic nose.In this paper, a feature selection method based on binary quantum-behaved particle swarm optimization (BQPSO) algorithm is proposed, and after the feature selection, genetic quantum-behaved particle swarm optimization (GQPSO) is used to develop a synchronous optimization of sensor array and classifier parameters.Radial basis function neural network (RBFNN) combined with BQPSO and GQPSO is used to select proper sensors and optimize the RBF parameters and sensor array synchronously.Experimental results indicate that this method is effective and remarkable on the recognition of rats wound infection based on electronic nose.