Fouling prediction of heat exchanger surface under alternating magnetic field based on IGWO-SVR

Crystallization fouling of a heat exchanger surface under an alternating magnetic field was studied by using self-designed annular channel electromagnetic anti-fouling experiment platform to obtain experimental data including conductivity, induced current and fouling resistance in various magnetic induction intensities. The magnetic induction intensity of 300 Gs exhibited the best fouling inhibition effect, and the fouling inhibition rate was 78.89%. The induced current (first proposed in the research) was related to the change of magnetic in-duction intensity and salt concentration, which could directly reflect the characteristics of fouling resistance and alternating magnetic field. Based on the strong correlation between conductivity, induced current and fouling resistance, support vector regression (SVR) optimized by improved grey wolf algorithm (IGWO) was proposed to predict fouling resistance with conductivity and induced current as input variables, fouling resistance as output variable. Keeping other experimental conditions constant, the fouling resistance on the heat exchanger surface under the same and different magnetic induction intensities was predicted. Prediction results indicated that, the mean absolute percentage error was 3.24% for the former, 7.88% (300 Gs) and 4.04% (100 Gs) for the latter. IGWO-SVR had the highest prediction accuracy and the strongest generalization capability compared with support vector regression (SVR) and SVR optimized by genetic algorithm (GA-SVR), which demonstrated that IGWO-SVR was highly adaptable to predict fouling resistance in various situations.

Automated summary

This study investigated the crystallization fouling of a heat exchanger surface under an alternating magnetic field. Experimental data, including conductivity, induced current, and fouling resistance, were obtained at various magnetic induction intensities. The results showed that a magnetic induction intensity of 300 Gs had the best fouling inhibition effect with a fouling inhibition rate of 78.89%. The induced current was found to be related to magnetic induction intensity and salt concentration, providing insights into fouling resistance and the alternating magnetic field. A support vector regression (SVR) model optimized by improved grey wolf algorithm (IGWO) was proposed to predict fouling resistance based on the strong correlation with conductivity and induced current. The prediction results demonstrated the high accuracy and adaptability of the IGWO-SVR model compared to other methods.