一种基于人工神经网络测定UF6中钼钛钨含量的新方法

传统的分光光度法在多组分分析中混合物诸组分光谱重叠,彼此干扰严重,需经复杂处理才能得到较为满意结果,而人工神经网络在实现非线性函数拟合方面具有独特优势.基于MATLAB,构建了一个三层BP网络,利用UF6混合溶液在不同波长点的吸光值实现多组分含量的同时测定,工程实践证明,结合分光光度法,基于人工神经网络的UF6中钼、钛和钨3种杂质含量的测量方法具有简单、可靠、准确、快速及易于掌握的特点.

A new method for determing the content of three elements Mo,Ti and W in UF6 based on artificial neural network

In allusion to the demerit of serious spectrum overlap in traditional spectrophotometry in analysis of the compound of multi-component(a relatively satisfied result can then be obtained after a complex disposal),from a new angle of special preponderant in no-linear function imitation with the artificial neural network(ANN),the possibility and the means to realize the multi-component simultaneous measure using the light-absorb quotiety of compound UF6 was dissertated.Based on MATLAB,a three layer BP network was constructed to actualize this conceit.The engineering practice proved that the method combined with spectrophotometry was simple,credible,accurate,rapid as well as easy holding for measuring the content of elements as Mo,Ti and W in UF6 based on ANN.