Study on automatic detection of doped proportions of polyimide matrix inorganic nanocomposite films based on wavelet energy distribution proportion features and extreme learning machine

Abstract

In order to test the component of nanocomposite films, this article presents a model for automatically detecting the doping ratio of polyimide matrix inorganic nanocomposite film based on wavelet energy distribution proportion features and extreme learning machine. In-situ polymerisation was used to prepare nanocomposite films doped with 15 wt%, 20 wt% and 25 wt% BaTiO3, respectively, while SEM was used to obtain the surface texture image. Experimental results show that the average TP rate obtained by this model is 0.922; average FP rate is 0.039 and average precision 0.924, indicating that the model is efficient in detecting and recognising films that have different doped proportions. Compared with the traditional model, predictive ability of this model is more predictive performance.