Neural ordinary differential gray algorithm to forecasting models of controlled systems

Due to the feasibility of gray model for predicting time series with small samples, theory is well investigated since it presented and currently evolved in an important manner forecasting samples. This study proposes a new prediction criterion based on neural ordinary differential equation, which named mode. mode permits be learned by training process contains whitening equation. It needed prepare structure series, compared other models, according regularity actual specimens advance, therefore this can provide comprehensive applications as learning properties distinct data specimens. To acquire better has highly predictive efficiency, afterward, trains using Runge–Kutta method obtain sequence solve model. The controller establishes advantageous theoretical foundation adapting novel wheels comprehensively spreads utilize extent mechanical elastic vehicle wheel.