柱效关系用于RBF神经网络色谱重叠峰解析

将径向基函数(RBF)神经网络引入色谱重叠峰解析领域·为了使RBF神经网络能适应于色谱重叠峰解析的需要,先在RBF神经网络学习算法中引入了基于可行域约束和共享小生境技术的遗传算法,而后又用两阶段遗传学习算法训练该神经网络以使其具有了结构自学习和参数优化的能力,适应于组分数未知的色谱重叠峰解析的需要,最后又将柱效关系引入至遗传算法的适应度函数中,极大地限制了解的空间,减少了病态解发生的概率·实验表明本方法解析精度较高,很适用于多组分色谱重叠峰解析,并且具有不需人为干预,自动确定网络结构即组分数的优点

Application Relation Between Column and its Efficiency in the Resolution of Overlapping Chromatographic Peaks Using RBF Neural Network

Radial basis function neural network (RBFNN) was introduced to resolute overlapping chromatographic peaks. A genetic algorithm based on sharing niche and possibility solution domain restrain is introduced,and then the two -phase genetic algorithm is used to train RBFNN so that it has the capacities of self learning structure and self learning parameters. The relation between column and its effciency is used properly in fitness function of the genetic algorithm for restricting solutions and reducing probability of illness solutions greatly. High resolution accuracy can be gotten by this method,and it is very suitable for the analysis of multicomponent overlapping chromatographic peaks. The network structure can be automatically confirmed without human interposition.