遗传神经网络在近红外光谱煤质分析中的应用研究

Application of Genetic Neural Network in Coal Quality Analysis with Near-infrared Spectroscopy

针对BP神经网络收敛速度慢及容易陷入局部最优解的缺点,结合遗传算法全局搜索的特点,提出了一种基于遗传算法和BP神经网络建立近红外光谱煤质分析模型的方法;并利用主成分分析法提取煤炭样品的主成分值,有效地压缩了数据。实验对比了BP模型与GA-BP模型,结果表明,GA-BP模型能有效地减小测试集的预测值与真实值之间的误差平方和,相关系数也得到了提高,有效地提高了预测精度和分析速度。

In view of the shortcomings of BP neural network,such as slow convergence,easily falling into local optimums,the paper put forward a method of establishment of model of coal quality analysis with near-infrared spectroscopy based on GA-BP neural network and characteristics of global searching method of neural network.The principal component analysis（PCA） was used to get principal component values and to compress data.The results of traditional BP neural network model and GA-BP model were compared,and the result showed that the GA-BP neural network model could not only reduce error sum squares between the predictive value and truth value,but also improve the correlation coefficient,which improves precision of prediction and speed of analysis effectively.