GA Based Real-time Algorithms For Multifunction Vehicle Bus Networks

With the rapid development of informationization for trains,real-time transmission of information is of great significance to train operation and safety.Generally,the Multifunction Vehicle Bus(MVB)is used for data communication among in-vehicle devices as one standard of the Train Communication Network(TCN).Therefore,it is necessary to study on the real-time performance of MVB in order to keep safe during operation.This paper mainly focuses on the problems of information uneven distribution and low bus utilization during the process of MVB periodic information transmission.Genetic Algorithm(GA)is used to optimize the periodic polling table(PPT)of period information transmission.The theory of train communication network is introduced,and the background significance of MVB research and the status quo of real-time research are expounded as well.Relevant studies carried out as follows:Firstly,the basic principle of MVB periodic information communication is expounded,including the data form of MVB and the specific process of transmission.The protocols and rules that must be followed to construct the periodic polling table are also elaborated,and the related parameters affecting the performance of MVB networks are explained.They lay a theoretical foundation for subsequent GA optimization research.Secondly,a traditional genetic algorithm is proposed to optimize the MVB periodic polling table.According to the international standard IEC61375-1,the constraints and optimization objectives are determined.After some relevant operations through genetic operators,the optimized PPT is obtained.Compared with the improved differential algorithm,the proposed method is validated to have a better performance.Thirdly,an adaptive genetic algorithm(AGA)is proposed.Based on the traditional GA,the setting method of the selection operator is preserved in AGA,and the crossover operator and mutation operator are adjusted by dynamic adaptive method.The simulation results show that the algorithm can improve the bus utilization.Finally,a hybrid algorithm combining GA and BP neural network(BP-GA)is proposed.In order to improve the performance of the algorithm,BP neural network is introduced to optimize the genetic operators.The hybrid algorithm combines the advantages of genetic algorithm in global search ability and the self-learning and trainingability of BP neural network.The simulation results show that the BP-GA is effective,and it can find the best solution faster.The methods proposed in this paper are validated to be effective to improve the real-time performance of MVB networks,which provides a theoretical basis for the customization of MVB networks control system and provides a guarantee for the safe operation of trains.