Subjects: Chemistry >> Applied Chemistry submitted time 2024-01-23
Abstract: Addressing the drawbacks of conventional formula design methods, such as being cumbersome, time-consuming, resulting in material wastage, and yielding poor optimization outcomes, a novel approach for optimizing the formula of aluminum alloy cleaners is proposed by combining the Latin Hypercube Design (LHD) algorithmwith the Multilayer Perceptron (MLP) neural network. Leveraging the comprehensive exploration of parameter space using the LHD algorithm and the efficient modeling capability of the MLP neural network, the formula for aluminum alloy cleaners was successfully optimized. With the optimized formula, the cleaning efficiency increased from 87.9% to 98.24%, and the corrosion rate decreased from 4.2 mg to 0.3 mg. Regression analysis of the model's predicted data against experimental data yielded a correlation coefficient greater than 0.98, demonstrating consistency between predicted and experimental data.
Peer Review Status:Awaiting Review