Abstract: Accurate underground gas concentration change prediction is essential for achieving safe and efficient production. Actual downhole systems often have time lags in the causal effects between variables, which may lead to poor performance of the traditional grey prediction model and affect the subsequent production and optimization operations. Under the time lag characteristics of the traditional grey prediction model, to solve the problems of the unclear mechanism of the driving term, as well as the deficiency of introduction rule. A new multivariate grey prediction DOGM(1,N) model with time lag characteristics is proposed in this paper. Based on the traditional OGM(1,N) model, the time-delay parameter is introduced into the driving term sequence. In order to solve the lack of analysis of the complete process of identifying the driving term sequence in the existing multi-variable grey model with time delay, this paper proposes a method for identifying the time-delay parameters and related factors sequence of driving term based on grey correlation analysis. Finally, the effectiveness of the method proposed in this paper has been verified by the simulation study of downhole gas concentration prediction. The results show that the DOGM(1,N) model has high prediction accuracy for the prediction problem of a small sample multivariable system with time-delay characteristics.