基于LSTM神经网络的干燥含水量预测研究

Research on Dry Water Content Prediction Based on LSTM Neural Network

在PVC干燥中,产品的含水量受到温度、流量等多个参数的影响,鉴于参数数据之间的非线性以及序列间的相关性,传统的时间序列方法和传统的机器学习算法已经不能对未来干燥产品的含水量进行精确的预测。长短期记忆网络(LSTM)作为一种基于深度学习中的循环神经网络(RNN),它在RNN的基础上增加了输入门、输出门以及遗忘门,可以有效地处理RNN在运行大量数据时可能会带来的数据遗忘等问题,特别适合处理具有时间序列的数据。基于Pytorch深度学习框架构造长短期记忆网络模型,对产品含水量进行预测。结果表明:使用该模型对产品的含水量进行预测,其预测值和真实值的走向非常接近,精准度很高。

In PVC drying,the water content of products is affected by many parameters,such as temperature and flow rate.In view of the nonlinearity of parameter data and the correlation between sequences,traditional time series methods and traditional machine learning algorithms can no longer accurately predict the water content of future dried products.Long-term and short-term memory network(LSTM),as a kind of recurrent neural network(RNN)based on deep learning,adds an input gate,an output gate and a forgetting gate to RNN,which can effectively deal with the problems of data forgetting that may be brought by RNN when running a large amount of data,and is especially suitable for processing data with time series.Based on Pytorch deep learning framework,a long-term and short-term memory network model was constructed to predict the water content of products.The results showed that using this model to predict the water content of products,the trend of the predicted value and the real value was very close,and the accuracy was very high.