Research Of The Design Of MVB Periodic Polling Table Based On Pareto Ant Colony Algorithm

Multifunction Vehicle Bus(MVB) connects the core components in train compartment. It is the foundation of traction, brake, supervision and service, and it guarantees safe operation of the train. The bus administrator(BA) needs to read the status information of those important modules periodically while the train is under operation, and the sequence is strictly arranged by the MVB periodic polling table(PPT). Excellent PPT can improve the MVB communication efficiency, balance the network load and enhance its ability to handle emergencies. The method provided in the international standard 1EC 61375-1 can’t guarantee the evenness, what’s more it prolongs the monitoring cycle of the modules, which goes against the real time administration. This paper presents the Pareto ant colony algorithm(P-AC) to optimize the PPT considering the protocol and the PPT feature.P-AC regards the mission as one scheduling problem. Firstly it gives model of the periodic information. There are two constraints:one is the demand of schedulability and the other is the upper limit of periodic time. It presents the concept "width" to illustrate the difference between the longest and shortest cycle, and introduce "fluctuation ratio" to express the difference between the adjacent cycle which reflects the complementary capability of basic cycle. It treats evenness and fluctuation ratio as the two optimization objectives in order to obtain the PPT that performs comprehensively. While coding, each objective corresponds one pheromone matrix, and the pheromone is modified following ant colony system algorithm(ACS). Total pheromone is the weighted sum and the weight is set randomly within the limit before the ant searches. To avoid falling into local optimum, the method take adaptive operation on pheromone evaporation rate and random selection probabilities. It manage the pareto optimal set using "the crowding distance" method. At last, it provides one directive formula to choose the final solution.The experiment results prove that this algorithm overwhelms the existing methods in both evenness and fluctuation ratio, which is to say that it can ensure the MVB network communication more reliable and effective.