Optimization of ethylene glycol electrodialysis desalination based on artificial intelligence hybrid model

Utilizing wavelet neural network and simulated annealing-particle swarm optimization algorithm to optimize the parameters of electrodialysis desalination process. Firstly, a single factor experiment was conducted to preliminarily explore the influence of electrodialysis operation voltage, operation time, electrode plate spacing, and electrode liquid concentration on desalination efficiency. Then, a wavelet neural network model was used to train and predict the data samples, and Sobol sensitivity analysis was conducted on the experimental influencing factors. Finally, the wavelet neural network model was coupled with the simulated annealing-particle swarm optimization algorithm. The optimized electrodialysis conditions and corresponding desalination rates in this system were obtained. The trial-and-error method results indicated that the 4-10-8-1 wavelet double hidden layer neural network model was a suitable prediction model. The degree of influence of various factors on the desalination effect was in descending order: operating voltage, operating time, electrode liquid concentration, and electrode plate spacing. When the unit membrane voltage was 0.42V/cm2, the operating time was 13.85 hours, the electrode spacing was 12.11cm, and the electrode liquid concentration was 0.21mol/L, the predicted optimized desalination rate reached 97.13%. After t-test, this value was highly consistent with the validation experimental results. This study can provide accurate and reliable theoretical support and data sources for the comprehensive promotion and deep application of ethylene glycol electrodialysis desalination process. © 2024 Chemical Industry Press Co., Ltd.. All rights reserved.