针对铁道供电系统信息流处理中的延时问题,建立主备模式的通信预处理队列模型、并列服务器实时队列模型和服务器轮询队列模型.利用拉丁超立方抽样技术和蒙特卡洛方法对信息流服务过程进行研究,运用统计学方法得出排队参数概率分布、置信区间和模拟次数之间的关系.以2M光纤通信为例,取平均到达率1 024 kB·s-1和平均服务率1 500 kB· s-1进行蒙特卡洛模拟.结果表明:报文字节处理服务延时可控制在2～3 ms左右；验证了用拉丁超立方抽样代替随机抽样,可减少蒙特卡洛模拟的次数.比较并列服务器和单服务器的实时处理队列平均等待时间模拟结果可知:相同条件下单服务器的实时处理队列平均等待时间超出10 ms的概率为82.95％,超出13 ms的概率为56.08％,均大于并列服务器实时处理队列平均等待时间的概率.因此,采用并列服务器配置方式可减少铁道供电系统信息流实时处理中的平均等待延时.%Primary-backup communication preprocessing queue model, parallel server real-time queue model and server polling queue model are put forward for solving time-delay problem of information flow processing in railway power supply system. Latin hypercube sampling and Monte Carlo simulation are used to study the service process of information flow, and the relationships among the probability distribution of queuing parameters, confidence interval and simulation times are obtained by means of statistical method. Taking 2 M optical fiber communication for example, 1 024 and 1 500 kB ? S-1 is respectively adopted for the average arrival rate and average service rate to carry through Monte Carlo simulation. Results show that the time-delay for packet byte processing service can be controlled at about 2~3 ms, and it is verified that using Latin hypercube sampling instead of random sampling can reduce the number of Monte Carlo simulation. The simulation results of the average waiting time for real-time processing queue of parallel server and single server are compared. It is clear that, under the same conditions, the probability is 82. 95% for single server when the average waiting time of real-time processing queue exceeds 10 ms, and the probability is 56. 08% when the average waiting time exceeds 13 ms, both are greater than that of parallel server. Accordingly, parallel server configuration mode can reduce the average waiting time delay of information flow processing in real-time in railway power supply system.
展开▼
