气体传感器阵列信号处理的混合神经网络

A mixed neural network for signal processing of gas sens or array

:讨论了基于气体传感器阵列的混合气体识别的信号处理方法 ,将自组织特征映射神经网络与BP神经网络相结合 ,采用先进行气体分类后识别气体组份的方式 ,将传统方法中的全程拟合改为分段拟合 ,降低了算法的复杂性 ,提高了识别率。

A mixed neural network for signal processing of gas sen sor array is discussed. This paper combines self-organizing feature map(SOM) neu ral network with back-propagation(BP) network first to classify the mixed gases and then to specify the components of the mixed gases. This actually amounts to a piecewise approximation, rather than a traditional whole approximation, of th e recognition procedure,so the algorithm is simpler and more efficient.