论文部分内容阅读
在高斯基径向基函数神经网络 (RBFNN)学习算法中引入了鲁棒性和随机全局寻优的两阶段遗传算法 :结构学习和参数优化。通过两阶段学习算法的交替使用 ,使网络具有结构自学习和参数优化的能力 ,而后将网络应用于组分数未知的重叠色谱峰解析。该方法具有不需人为干预 ,可自动确定网络结构即组分数的优点 ;并且解析精度较高 ,适用于多组分重叠色谱峰的解析 ;对完全重叠色谱峰也具有良好的解析能力
Two-stage genetic algorithm, robust and stochastic global optimization, is introduced in the Gaussian basis RBF neural network (RBFNN) learning algorithm: structure learning and parameter optimization. Through the alternating use of two-stage learning algorithm, the network has the ability of structure self-learning and parameter optimization, and then the network is applied to the analysis of overlapping chromatographic peaks of unknown composition. The method has the advantage of automatically determining the network structure, ie the number of components, without human intervention. The method has the advantages of high resolution and is suitable for the analysis of overlapping chromatographic peaks of multiple components. The method also has good analytical ability for completely overlapping chromatographic peaks