A Generative Model Based on Bootstrapping and Artificial Neural Nets for Transmission Gears Safety

Gear safety factors are primary indicators of current operational states of automobile transmissions, evaluated through a series of complicated calculations with numerous parameters of high dimensions. It is discovered that existing data of gear data from manufacturing processes are insufficient and high dense. Thus, based on Bootstrapping algorithm and generative adversarial network, a generative model is proposed to generate admissible data using existing data. The data generated by our model have wider spaces than other several oversampling methods, and outperform in safety identifications.