The error state evaluation method of CVT based on spatiotemporal information fusion

In order to address the problems of the existing CVT online evaluation methods,which only utilize secondary output data,resulting in low data dimensions,poor reliability,and low accuracy,this paper proposes an error state evaluation method of the CVT based on spatiotemporal information fusion.First,the spatiotemporal correlation law of the secondary output data under the CVT operating state is analyzed;Secondly,based on this law,the multi-dimensional spatiotemporal characteristics are decoupled from the coupling relationship between the CVT monomer-independence and group-correlation;Finally,a GMU-Attention-RBF algorithm model is constructed to realize the dynamic evaluation of the CVT operation error state with spatiotemporal interaction and unity,and the state evaluation accuracy is more than 99%. Compared with the prior art,this method has better robustness and generalization.