"every family has a car, cars are more intelligent" is the trend of social and scientific and technological development. Intelligent network connected cars bring convenience to people, but there are also security risks. The data inside the car is transmitted through the CAN bus (controller area network bus). Once a hacker invades the electronic control unit of the car, the instructions in the CAN bus are at risk of being forged and monitored. The hacker can control the car and cause serious consequences. So we developed a vehicle network security monitoring system based on ML. The main features of this system are:

1. From the perspective of vehicle networking information communication, the frame structure of CAN bus, as well as security risks and possible problems are analyzed.

2. On the client, a driving information visual module is designed to vividly display the door status, vehicle speed, engine speed and other information converted from the CAN bus on the front-end interface, which is convenient for the car administrator to view in real time and can be used to judge whether the car is abnormal. Through the environmental index query interface, the environmental factors when the car is driving can be obtained, and the administrator can understand the general itinerary of the car owner and provide driving suggestions to improve the user experience.

3. On the server side, the ML algorithm is applied to the important functional modules of the system, the CAN bus data set is sorted, and the negative sample set is produced by simulating the injection attack, and the malicious message detection is realized for the CAN bus message with the specified ID.

4. Through many comparisons, the logical regression algorithm is selected as the recognition algorithm of the system, its accuracy rate is more than 93.66%, the running time is short, and it has significant advantages over other algorithms.

5. The system has good interactivity and friendly interface. The front end chooses to use elementUI and other frameworks and FullCalendar components for design. In addition to the functions of driving information viewing, simulated attack, attack forensics, etc., the system specially designs other practical functions such as schedule planning.

6. The system adopts a multi-factor analysis method to consider the car condition, environmental conditions, and the owner's trip, so that the system has complete functions and improved performance, thereby improving the level of safety monitoring.