论文部分内容阅读
基于塔板色谱峰模型的径向基函数神经网络 (RBFNN)用于色谱 (含重叠 )峰解析是一种新方法。为了使 RBFNN具有结构重组能力 ,用于色谱峰解析的 RBFNN必须采用遗传算法 (GA)。虽然遗传算法具有鲁棒性和全局优化能力 ,但若种群过小 ,则陷于局部极小点的概率将增高。而塔板模型是一个低效模型 ,若选用过大的种群 ,必然使解析过程加长。为了提高算法效率 ,提出先用高效色谱峰近似模型标准高斯模型进行繁衍 ,而后再用塔板模型。
Radial basis function neural network (RBFNN) based on plate chromatographic peak model is a new method for chromatographic (including overlapping) peak resolution. For structural reorganization of RBFNN, the genetic algorithm (GA) must be used for RBFNN for chromatographic peak resolution. Although GA is robust and global optimization, if the population is too small, the probability of sinking to a local minimum will increase. The tray model is an inefficient model, if the choice of large population, the analytical process will inevitably lengthened. In order to improve the efficiency of the algorithm, it is proposed to multiply the standard Gaussian model of the high-efficiency chromatographic peak approximation model and then use the plate model.