An adsorption isotherm identification method based on CNN-LSTM neural network

Abstract

Abstract Context: Adsorption isotherm is integral to comprehending the adsorption mechanism and catalytic processes. Despite the well-established research on isotherm classification methods, existing techniques for isotherm identification are hampered by inefficiency, human interference, insufficient feature information extracting and the fact that specific types isotherm of identification can be achieved. To overcome these limitations, an end-to-end isotherm identification method based on a CNN-LSTM neural network is proposed, which employs a combination of Convolutional Neural Network (CNN) and Long Short-Term Memory network (LSTM) to extract the features of adsorption isotherm. Additionally, dataset is constructed for training and validating the neural network using various theoretical adsorption isotherm equations, thereby eliminating the need for time-consuming and expensive repetitive experiments. The results indicate that the model achieves identification accuracy of 100% on both the training and validation sets constructed using theoretical adsorption equations. The model's F1-score on the test set, which consists of five categories actual adsorption isotherms, is 88.85%, and there has been a 20% enhancement in the mean precision of isotherm identification compared to the traditional method. All these results demonstrate that the proposed method can accurately identify adsorption isotherm. Method: Pycharm was used as the experimental and testing platform, python 3.9 was used as the programming language, Tensorflow 2.11.0 and Keras 2.10.0 were used to train and test the CNN-LSTM model, numpy 1.21.5 and scipy 1.81 were used to generate the train and validation dataset.