电子鼻模式识别算法的比较研究

文中比较了k-近邻法、线性判别分析、反向传播人工神经网络、概率神经网、学习向量量化以及自组织映射6种电子鼻模式识别算法的分类能力.采用了1个定量指标(识别精度)和4个定性指标(运算速度、训练速度、内存容量、抗干扰能力)对不同算法进行了系统比较.研究表明基于神经网络的模式识别算法比基于统计理论的模式识别算法具有更高的识别精度.如果同时考虑定性指标,当训练速度要求不高时,宜采用学习向量量化算法;能满足内存需求前提下,优先推荐采用概率神经网算法.对于选择性高的信号,采用线性判别分析可以达到最佳效果.

Comparison Study of Electronic Nose Pattern Recognition Algorithms

K-nearest neighbor(k-NN),linear discriminant analysis(LDA),back-propagation artificial neural networks(BP-ANN),probabilistic neural networks(PNN),learning vector quantization(LVQ) and self-organizing map(SOM) pattern recognition algorithms are compared for their ability to classify electronic nose data.Comparisons are made based on four qualitative criteria (calculating speed,training speed,memory requirements and robustness) and one quantitative criterion(classification accuracy).It is shown that the neural network based algorithms have higher classification accuracies than the statistical theory based algorithms.When considering the qualitative criteria,LVQ is suggested for applications where training speed is not required.PNN is recommended for applications where memory requirement is not critical.When high selective signals are applied,LDA should be the fittest algorithm because of its high classification accuracy,high speed and low memory requirements.