Because of the opening of In-vehicle network, there are several important problems to be dealt with, such as the security and validity of data. Firstly, the article builds a construct model based on driving behavior and speed. Secondly, it makes an analysis of preventing data injection by using the construct model above and the naive Bayesian network classifier, so as to take effective measures to guarantee the vehicle security. In the end, an experimental simulation is carried out to prove that the proposed method can effectively improve the accuracy of data quality analysis and lower the false rate as well.