Experimental study on the semi-quantitative measurement for ferromagnetic material based on extreme learning machine (ELM)

摘要

To study experimentally semi-quantitative measurement of ferromagnetic materials, this paper first proposes a modelling method to establish relationship between initial permeability of ferromagnetic material and material characteristics using extreme learning machine(ELM), and then empirical experiments are conducted on test pieces made of 45 steel. Its experimental results are compared with BP(back propagation)neural network, and we found that ELM is better than BP to describe the non-linear relationship between the characteristics parameters and the initial permeability, and can be used for sorting and semi-quantitative measurement of the characteristics parameters of ferromagnetic materials.