Liquid dangerous goods tank vehicles produce a large number of data in the operation process. If these data are well utilized, it can provide great help for judging whether the vehicle is in a safe operation state. Through preliminary research, the text has summarized multiple data elements that can reflect four typical hazardous working conditions. Based on the fusion method of multi-source safety–critical information, this paper establishes a vehicle multi-source information data model. A liquid dangerous goods tank vehicle information collection system has been developed to meet the needs of vehicle data collection. The system includes hardware sensors and software platforms, and has been installed on the vehicle for actual vehicle operation testing.