Aiming at the demand of energy-saving technology development, this study establishes a road condition and driving style recognition model, focus on evaluating the fuel consumption differences of different driving styles, and establishes a global optimization energy-saving driving strategy model. Firstly, based on the selected feature parameters, principal component analysis, clustering and neural network training, a high-precision recognition model is established, and the influence of different recognition periods on the rationality of recognition results is studied. Secondly, the driving style recognition is carried out for the big data of a truck road, and the fuel consumption difference caused by different driving styles is quantitatively evaluated. Finally, a global optimization energy-saving driving strategy model is established based on dynamic programming algorithm. The results show that the average driving energy consumption of the vehicle with sport driving style decreases by 12.76%, as well as the driving power demand under medium and low speed road condition is reduced.