粒子群优化的SVM算法在气体分析中的应用

提出一种用于电子鼻气体定量分析的基于粒子群参数优化的SVM算法。引入群能量守恒粒子群算法对SVM模型中的参数进行优化,获得最优参数组合,以减小定量分析中的平均相对误差。实验结果表明,引入群能量守恒的粒子群优化算法(SEC-PSO)对SVM回归模型中的参数进行优化,使参数的确定更为合理,与单纯基于SVM的方法相比平均相对误差由原来的2.17%降低到0.35%,所提出的算法有效地提高了拟合的精度,从而进一步提高预测精度。

Application of SVM algorithm for particle swarm optimization in quantitative analysis of gas

Electronic nose is an electronic mimic biological olfactory system,which is used to identification of qualitative and quantitative gas analysis.SVM algorithm based on particle swarm optimization(PSO) is applied to the electronic nose in the quantitative analysis of the gas.SVM is based on VC theory,using the structural risk of the principle of good nonlinear mapping ability and generalization ability.PSO is an algorithm based on swarm intelligence,and its main feature is the process of searching information in accordance with individual groups of the two types of information and information for decision-making.SEC-PSO algorithm is used in parameter optimization for SVM model.The experi-ments show that PSO introdued into support vector regression model parameters are optimized.Average relative error is reduced from 2.17% to 0.35%.Therefore,the proposed algolithn can improve the prediction performance effectively.