A robust predictive tool for estimating CO2 solubility in potassium based amino acid salt solutions

The acid gas absorption in four potassium based amino acid salt solutions was predicted using artificial neural network(ANN). Two hundred fifty-five experimental data points for CO_2 absorption in the four potassium based amino acid salt solutions containing potassium lysinate, potassium prolinate, potassium glycinate, and potassium taurate were used in this modeling. Amine salt solution's type, temperature, equilibrium partial pressure of acid gas, the molar concentration of the solution, molecular weight, and the boiling point were considered as inputs to ANN to prognosticate the capacity of amino acid salt solution to absorb acid gas. Regression analysis was employed to assess the performance of the network. Levenberg–Marquardt back-propagation algorithm was used to train the optimal ANN with 5:12:1 architecture. The model findings indicated that the proposed ANN has the capability to predict precisely the absorption of acid gases in various amino acid salt solutions with Mean Square Error(MSE) value of 0.0011, the Average Absolute Relative Deviation(AARD) percent of 5.54%,and the correlation coefficient(R^2) of 0.9828.

A Discontinuity and Cusp Capturing PINN for Stokes Interface Problems with Discontinuous Viscosity and Singular Forces

In this paper,we present a discontinuity and cusp capturing physicsinformed neural network(PINN)to solve Stokes equations with a piecewiseconstant viscosity and singular force along an interface.We first reformulate the governing equations in each fluid domain separately and replace the singular force effect with the traction balance equation between solutions in two sides along the interface.Since the pressure is discontinuous and the velocity has discontinuous derivatives across the interface,we hereby use a network consisting of two fully-connected sub-networks that approximate the pressure and velocity,respectively.The two sub-networks share the same primary coordinate input arguments but with different augmented feature inputs.These two augmented inputs provide the interface information,so we assume that a level set function is given and its zero level set indicates the position of the interface.The pressure sub-network uses an indicator function as an augmented input to capture the function discontinuity,while the velocity sub-network uses a cusp-enforced level set function to capture the derivative discontinuities via the traction balance equation.We perform a series of numerical experiments to solve two-and three-dimensional Stokes interface problems and perform an accuracy comparison with the augmented immersed interface methods in literature.Our results indicate that even a shallow network with a moderate number of neurons and sufficient training data points can achieve prediction accuracy comparable to that of immersed interface methods.