Prediction and Sensitivity Analysis of CO2 Capture by Amine Solvent Scrubbing Technique Based on BP Neural Network

Abstract

With the rapid development of artificial intelligence, bionic algorithm has been gradually applied in various fields, and neural network has become an important and hot issue in the field of scientific research and engineering in recent years. This article proposes a BP neural network model to predict the capture ability and sensitivity of CO2 in monoethanolamine (MEA) aqueous scrubbing technique from a 2 × 1,000 MW coal-fired power plant expansion project in eastern China. The predicted values agree well with the experimental data with a satisfactory mean square root error (MSRE) ranging from 0.001945 to 0.002372, when the change in the circulation amount of MEA and the accuracy of prediction results of the back propagation neural network (BPNN) algorithm is as high as 96.6%. The sensitivity analysis results suggested that the flue gas amount has a marginal effect on the system performance, while further attention should be paid to the MEA circulation amount, which is crucial to the CO2 capture amount. The temperature profiles show the typical behavior of the reactive absorption column where a temperature bulge can be seen at the bottom of the column due to the high L/G ratio of the experimental and prediction results. The coefficients of correlation R2 with the change of MEA circulation amount, change of CO2 concentration, and steam consumption are 0.97722, 0.99801, and 0.98258, respectively. These results have demonstrated that the present study has established the BPNN algorithm as a consistent, reliable, and robust system identification tool for CO2 capture by the amine solvent scrubbing technique of operation in coal-fired power plants.

Non-Destructive Prediction of Anthocyanin Content of Rosa chinensis Petals Using Digital Images and Machine Learning Algorithms

“…BPNN and RF algorithms are two commonly used machine learning algorithms. BPNN is a multilayer feed-forward neural network that corrects errors using a back-propagation algorithm [19]. It has strong nonlinear mapping, self-learning, self-adaptive, and generalization capabilities and high fault tolerance [19,20] and is suitable for regression or classification problems.…”

“…BPNN is a multilayer feed-forward neural network that corrects errors using a back-propagation algorithm [19]. It has strong nonlinear mapping, self-learning, self-adaptive, and generalization capabilities and high fault tolerance [19,20] and is suitable for regression or classification problems. Therefore, it is the most widely used neural network model and is particularly useful for solving nonlinear problems [21].…”