TCN multivariate time series prediction method based on parallel space-time attention mechanism

The invention discloses a TCN multivariate time series prediction method based on a parallel space-time attention mechanism, and the method comprises the following steps: firstly defining a formula, then constructing a multivariate time series prediction model which comprises two parallel network trunks, wherein the spatial attention branch trunk extracts spatial correlation between an exogenous sequence and a target sequence through a spatial attention module, the time attention branch trunk captures time dependence among all time steps in a window through a time attention module, and the space attention module and the time attention module are respectively connected with the two same stacked TCN trunks and the full connection layer; and finally, inputting the multivariable time sequenceinto the multivariable time sequence prediction model to obtain a final prediction result. According to the method, a space-time attention mechanism and the TCN are combined, so that higher accuracy is achieved compared with