Prediction model and sensitivity analysis of shielding effectiveness of woven fabrics containing stainless steel fibers based on extreme learning machine

Abstract

Metal fiber blended fabrics, especially woven fabrics containing stainless steel (SS) fibers, are widely used in the field of electromagnetic shielding due to their advantages in shielding performance, wearability and cost. However, how to establish a simple, quick and high-precision model for predicting the electromagnetic shielding effectiveness (EMSE) of fabrics needs to be solved urgently. In this paper, different fabric structure parameters were combined to form thirteen combinations of input variables, and the EMSE values at six frequency points were taken as the output variables. Corresponding to each combination of input variables, a total of thirteen prediction models were established using extreme learning machine (ELM) algorithm. Furthermore, the optimal models were selected. The results show that the prediction model established by ELM algorithm has good predictive ability, and the two optimal prediction models show that the prediction accuracy of the model is lower at a relatively high frequency. In addition, the sensitivity analysis was carried out by using the improved Garson equation to obtain the relative contribution value of each fabric structural parameters. The result shows that the relative contribution of SS fiber content is the largest at all frequencies except 3000 MHz, but both the differences of relative contribution values between different variables and that between the same variable at different frequency points are not very significant. Therefore, the same prediction model can be used to predict EMSE at different frequency points, while the prediction model based on frequency range has higher accuracy.