In-vehicle communication has developed into a crucial element of today's driving environment as a result of the expanding additions of sensor-centric communication as well as computing devices inside a vehicle for a variety of purposes, consists of vehicle monitoring, physical wiring minimization as well as driving efficiency. The relevant literature on cyber security for in-vehicle communication methods does not, however, currently offer any certain solutions for in-vehicle cyber hazards. The existing solutions, which mostly rely on protocol-specific security approaches, do not provide a comprehensive security framework for in-vehicle communication. This study aims to develop an effective data transmission and intelligent machine learning technique for smart vehicle management in VANET breach detection. In this study, ensemble adversarial Boltzmann CNN architecture is used to detect breaches. The secure short hop opportunistic local routing protocol is then used to send the data. Throughput, QoS, training accuracy, validation accuracy, and network security analysis are all part of the experimental analysis for a variety of security-based datasets. the proposed technique attainedthroughput of 88%, QoS of 77%, training accuracy of 93%, validation accuracy of 96%, network security analysis of 63%, scalability of 75%.