基于小波基的SVM多气体融合

为了提高气体传感器在多气体环境下的检测灵敏度,基于小波对偶框架和支持向量核函数的条件,提出了一种支持向量小波核函数.该核函数具备小波的多尺度插值特性和稀疏变化特性,提高了模型的精度和迭代的收敛速度,适用于信号的局部分析、信噪分离和突变信号的检测,从而能在提高支持向量机(SVM:Support Vector Machine)泛化能力的同时,提高辨识效果,减少计算量.基于该核函数和正则化理论提出的最小二乘小波支持向量机用于非线性系统辨识,对sin c函数的逼近.该小波核得到的绝对误差不超过0.004;在多气体分析中,比RBF(Radial Base Function)核所得的偏差小18.3%.这些表明SVM小波核具有更好的泛化能力.

Multi-Component Gas Data Fusion Based on Wavelet Support Vector Machine

A wavelet kernel for SVM(Support Vector Machine) based on wavelet dual frame theory and conditions of constructing SVM kernel is presented. It increases the precision and convergent rate of the model,and is especially suitable for local signal analysis, signal-noise separation and detection of jumping signals with the characteristics of multi-scale interpolation and sparse variation, thus enhances the generalization ability of the SVM, recognition efficiency and computation burden is alleviated. According to the wavelet kernel function and the regularization theory, a LS-WSVM (Least Square Wavelet Support Vector Machine) is proposed to simplify the solving process of SVM. The LS-WSVM is then applied to the nonlinear system identification to test the validity of the wavelet kernel function. In function sin c simulation, the Absolute error is no more than 0.004. In multi-gas analysis, the standard deviation is lower 18.3% than RBF(Radial Base Function) kernel.This hints that it has better generalization ability.