淀粉薄膜的研究进展

针对黏稠食品灌装过程中高精度检测流量的难题,提出一种基于深度学习的高黏稠食品灌装流量实时检测方法。首先对采集到的流量相关过程变量进行序列化及归一化处理,转化为可供有监督学习网络处理的数据;然后对注意力机制下的长短时记忆神经网络(LSTM-Attention)进行训练和泛化,采用自适应矩估计优化算法(Adam),进而建立高黏稠食品灌装流量检测模型;最后将本模型检测所得流量值与其实际值进行对比,用均方误差函数(MSE)对该模型在灌装流量检测上的性能进行评价。通过与循环神经网络(RNN)、普通长短期记忆模型下流量检测的均方误差作比较,实验结果表明,本模型的流量检测精度较高,流量检测数据实时跟踪效果较好。

Research Progress on Starch Films

In recent years, with the development in environmental protection, the starch film have attracted much attention owing to its wide source, low price and complete biodegradability. The research significance and the purpose of starch films were introduced firstly. The preparation technology of starch films and the research progress of nano-reinforced starch films were reviewed, including the production principles of starch films, processing advantages and the effects of different nano-reinforced materials on starch films. Finally, the problems existing in the research of starch films were analyzed, and the development of starch films was prospected. Future research could focus on the thermoplastic film-forming technique of starch and the physicochemical modification of nano-fillers. Meanwhile, the synergistic enhancement of nanomaterials would be an important development direction in starch films.