In order to improve the accuracy of the support vector machine algorithm applied to the driving condition recognition, an optimization strategy of genetic algorithm was put forward. Based on the principal component analysis theory, the characteristic parameters were extracted from the load spectrum data collected in 4 typical urban working conditions and used as the input parameters of the recognition model. Then the parameter optimization space was determined by the grid search method and the optimal value within the range could be found by the genetic algorithm. The simulation experiment results show that the recognition accuracy can improve by 3.44% based on the recognition model through the optimization of support vector machine with the genetic algorithm.