Voltage-Current Density Characteristics for Precipitators Using Artificial Neural Networks

Abstract

Unlike previous numerical techniques, this study utilizes artificial neural networks (ANNs) to predict optimal ion mobility at specific applied voltages. The finite difference method (FDM) integrated with the full multigrid method (FMG) is employed to train the ANN model. This model is designed to predict ion mobility and optimize computational grids for a given precipitator design. Traditionally, the FDM-FMG method required several iterations to achieve convergence of the potential error, with the computed current density aligning with experimental data. This approach, however, proved computationally expensive. The incorporation of ANNs significantly reduces the computational efforts of the FDM-FMG method, enhancing overall performance by minimizing the time required for convergence. This leads to a more efficient numerical process, particularly in large-scale simulations. The proposed method has been rigorously validated against previously published experimental results, demonstrating excellent agreement and showcasing its potential for improving computational efficiency in ion mobility prediction within electrostatic precipitators.