基于GA-BP神经网络的塑料X射线吸收光谱的辨识

摘要

This paper collects the X-ray absorption spectra of 15 kinds of plastic samples, carries on the data preprocessing, and then extracts principal component as the input of the subsequent model.The back propagation (BP) neural network model is established by using the training set to modify the weights and threshold values of network and obtains corresponding model.Finally, the neural network model is optimized by genetic algorithm, the optimal matrix of weights and threshold values are obtained.The experimental results show that BP neural network based on genetic algorithm optimization (GA-BP) can identify X-ray absorption spectra of plastic samples better and more stable than BP neural network, which has important guiding significance for the recycling of plastic.%采集15种塑料样本的X射线吸收光谱(XAS),对光谱数据进行预处理和主成分分析,建立误差反向传播(BP)神经网络和遗传算法优化的误差反向传播(GA-BP)神经网络模型,利用训练集进行网络训练,并通过测试集进行验证.结果表明,GA-BP神经网络相比于BP神经网络可以更好更稳定的对塑料样本的XAS进行识别,这对塑料的回收具有重要的指导意义.