基于Pareto蚁群算法的MVB周期轮询表优化设计

合理的多功能车辆总线(MVB)周期轮询表有助于均衡网络负荷、提高网络处理偶发信息的能力、保证实时通信的可靠性.为此提出一种有效的轮询表设计方法.将MVB周期轮询表的设计抽象成离散优化问题,根据IEC 61375-1国际标准和可调度性要求建立约束条件,将均匀度和相邻基本周期时间差作为优化目标,利用Pareto蚁群(Pareto ant colony,P-AC)算法求解.每个优化目标对应自己的信息素,信息素采用蚁群系统的规则更新,总信息素由两者加权得到,非劣解基于拥挤距离方法维护.与已有的优化算法相比,Pareto蚁群算法优化得到的轮询表均匀度更好,能够更有效地均衡网络负荷.

Optimization design of MVB period polling table based on Pareto ant colony algorithm

Good multifunction vehicle bus (MVB) period polling table contributes to balancing the network load and improving the ability of network processing sporadic messages, which can ensure the reliability of the real time communication. An effective polling table design method is proposed. The design of the MVB period polling table is abstracted into a discrete optimization problem. Constraints are obtained according to the IEC 61375-1 international standard and request of schedulability. The optimal objective consists of uniformity and adjacent basic period time interval. The solution is achieved by Pareto ant colony algorithm. In this algorithm, every objective has updated its own pheromone independently by rule of the ant colony system algorithm and the total pheromone is calculated by weighted summation of the two pheromones. The non-dominated solutions sets are maintained by the crowding distance method in this multi-objective problem. The experimental results show that the Pareto ant colony algorithm can perform better than the existing algorithms in uniformity and balancing the network load.