Recently Software defined networking (SDN) becomes the hottest area of study for network researchers. This new architecture allows the network to be centralized and manageable due to the flexibility of SDN controllers. On the other hand, the centralized nature of a controller is becoming a risk as a single point of failure. Other networks like vehicular ad hoc networks are able to attach into the SDN environment and called Software defined vehicular networks. Both technologies are combined to improve programmability, reliability, scalability, and network management efficiency. But the openness nature of SDN controller raises questions about the security issues. As a solution, Intrusion detection systems are proposed in many approaches to fill the gap in SDN controller vulnerability. In this Study we proposed an Intrusion detection system using a machine learning algorithms for software defined vehicular networks. R andom Forest and K-Nearest Neighbors are used as a classifier under different conditions to increase the performance of our model. The NSL-KDD dataset is used to train and test our proposed models, and it is a benchmark dataset for a number of state-of-the-art Network Intrusion Detection System algorithms. Our models outperform other works by utilizing one hot encoder preprocessing and Recursive Feature Elimination feature selection techniques. We train and test the first model using 13 features selected by the RFE feature selection method and the second model using 7 SDN basic features purposely selected. Random Forest Classifier scores outstanding performance with accuracy of 99.14%. Key words: Software Defined Networks, Vehicular Ad Hoc Network, Intrusion Detection System, Machine Learning Algorithms, Random Forest, K nearest neighbors