混合LSTM与卷积神经网络的氧气提取率预测方法

A hybrid LSTM and convolutional neural network method for predicting oxygen extraction rate

针对空分装置系统的运行参数量大、氧气提取率预测研究欠缺的问题,提出一种基于卷积神经网络(CNN)、长短期记忆网络(LSTM)以及混合LSTM和CNN的氧气提取率预测方法。将氧气提取率作为预测目标,基于卷积神经网络、LSTM、混合LSTM与卷积神经网络模型对其进行建模,并应用于空分装置系统运行采集的数据中。使用平均绝对百分比误差、均方根误差和平均绝对误差等指标来评价预测模型的精度,并使用模型训练时间以及模型收敛速度评估模型性能。实验结果表明,采用混合LSTM和卷积神经网络的氧气提取率预测方法的效果明显优于其他两种模型。

A method for predicting oxygen extraction rate based on convolutional neural network(CNN),long short term memory(LSTM),and a mixture of LSTM and CNN is proposed to address the problem of large operating parameters and insufficient research on predicting oxygen extraction rate in air separation plant systems.The oxygen extraction rate is used as the prediction target,and it′s modelling is conducted based on convolutional neural networks,LSTM,hybrid LSTM,and convolutional neural network models.It is applied to the data collected during the operation of the air separation device system.The accuracy of the prediction model is evaluated by means of average absolute percentage error,root mean square error,and average absolute error indicators.The model performance is evaluated by means of the model training time and model convergence speed.The experimental results show that the oxygen extraction rate prediction method using a hybrid LSTM and convolutional neural network has significantly better performance than the other two models.