Design And Implementation Of Vehicle Hybrid Network Gateway Based On Bidirectional LSTM-SVM Hybrid Model

As the basis of automotive communication technology,vehicle-mounted bus technology is widely used in automotive electronic unit communications.With the intelligent development of modern cars and the popularization of in-vehicle entertainment systems and advanced driver assistance systems(ADAS),traditional vehicle bus technologies are increasingly unable to meet the needs of the current environment.As an alternative to the traditional vehicle bus,vehicle Ethernet technology is gradually advancing.However,the development of in-vehicle Ethernet technology is immature,and the traditional in-vehicle bus cannot be completely replaced.The current industry is more inclined to use hybrid in-vehicle network systems,and use appropriate networks for areas with different bandwidth requirements.As a device connecting different networks,the vehicle gateway system needs to convert different communication protocols while ensuring its own stable operation.At the same time,the gateway needs to dispatch internal information queues in a timely and accurate manner.For this reason,this article aims at the related problems faced by the gateway system,and based on the functions of the hybrid network gateway,designed an auxiliary system based on the bidirectional long and short-term memory neural network(Bi-LSTM)and support vector machine(SVM)as the hybrid model.The main function of the system is to assist the gateway to dynamically adjust its own operating status to ensure the stable operation of the gateway system.The main research content of this paper is as follows.(1)This article designs a vehicle hybrid network gateway hardware system based on multiple controller area networks(CAN)and 1 vehicle Ethernet.(2)This paper designs the main functions of the hybrid network gateway system to ensure the basis for the mutual transformation of different networks.By referring to the traditional information queue scheduling algorithm,this paper designs the information queue scheduling algorithm in line with this topic.(3)The gateway load prediction node is designed and constructed to provide auxiliary support for the gateway system.Based on LSTM network,this paper designs structural optimization network models,such as temporal attention mechanism,spatial attention mechanism and recent enhancement network,to improve the prediction accuracy of gateway load.At the same time,in order to eliminate the influence of fuzzy samples,SVM model was introduced in this paper to build a mixed model,in order to divide the fuzzy samples to the determined boundary and further improve the prediction accuracy.(4)Finally,the whole system is tested to verify the stability of the operation and the integrity of the function of the gateway system.The test and experiment show that the gateway system with auxiliary equipment has the same functionality as the traditional gateway,and has the stability and real-time dispatching data that the traditional gateway does not have.