A PSO-SVM Method for Parameters and Sensor Array Optimization in Wound Infection Detection based on Electronic Nose

In this paper a new method based on the support vector machine (SVM) combined with particle swarm optimization (PSO) is proposed to analyze signals of wound infection detection based on electronic nose (enose). Owing to the strong impact of sensor array optimization and SVM parameters selection on the classification accuracy of SVM, PSO is used to realize a synchronization optimization of sensor array and SVM model parameters. The results show that PSO-SVM method combined with sensor array optimization greatly improves the classification accuracy of mice wound infection compared with radical basis function (RBF) network and genetic algorithms (GA) with/without sensor array optimization. Meanwhile, the proposed sensor array optimization method which weights sensor signals by importance factors also obtain better classification accuracy than that of weighting sensor signals by 0 and 1.