基于遗传级联神经网络的化学溶液浓度预测

在化学实验中经常需要对化学溶液各成分的浓度给出比较准确的预测。通过遗传算法来优化级联神经网络，利用神经网络的学习预测能力来预测化学溶液的浓度。首先用小波网络对混合溶液测出的极谱信号进行拟合并提取特征；然后用神经网络对提取的信号特征学习训练到一定程度后，把要预测浓度的化学溶液的极谱信号经小波网络提取的特征输入该网络后，给出预测值。计算结果表明，预测结果基本符合要求。

Prediction of Concentration of Chemical Substance in Solution Based On Cascade Neural Networks Optimized by Genetic Algorithm

In chemical experiments,we often need to give proper prediction of concentration of chemical substance in solution.This paper presents a kind of method to predict concentration by using the neural networks' learning and prediction ability,and the cascade networks is optimized by genetic algorithm.First we use wavelet network one to extract features from the chromatographic spectra signals;second we train the neural network two through the extracted features;last we can get the result by inputting the predicating signals to the trained network above.Computational result shows the method is effective.