A modified model for calculating delay at high saturation intersections based on Maximal Information Coefficient is proposed in this paper. Firstly, UAVs were used to investigate measured data for all variables in the HCM delay model, and the maximum mutual information correlation analysis of variables and combinations in HCM model was carried out by using rjava to call MIC algorithm coding, accordingly?, the correlation of mutual information between traffic flow, saturation, green signal?ratio and delay value is compared and tested through the statistical software R, thus to choose a nonlinear relation to modify the existing model. Then, the modified model with the best interpretation data containing and the least free parameters was selected as the optimal choice according to Akaike information criterion. Finally, by using the modified model to calculate the delay value of high(excessive) saturated intersection cases, comparing with results from HCM2010 and traffic simulation, it is concluded that the new model proposed in this paper has higher accuracy.