摘要
将并行数控系统实时任务划分为周期性任务和数据依赖任务,针对数据依赖任务间数据流不平衡问题,提出了一种以数据流状态为调度依据的任务调度器,通过改造嵌入式实时操作系统e Cos,实现了对周期性任务和数据依赖任务进行统一调度的混合实时任务调度器。将GRBL串行数控系统划分为多任务,并基于嵌入式对称多核处理器S698PM及混合实时任务调度器搭建了并行数控系统实验平台。对混合实时任务调度器进行了调度实验,测试了负载平衡性、加速比,并通过加工实验验证了并行数控系统的正确性。
The real-time tasks of parallel CNC system were divided into periodic tasks and tasks with data dependencies.Aiming at solving the problem that data stream between tasks with dependencies was unbalanced,a task scheduler was proposed which consider data stream as scheduling basis.The embedded RTOS eCos was reconfigured into hybrid real-time task scheduler which could schedule periodic tasks and tasks with data dependencies simultaneously.Serial CNC system GRBL was divided into multiple tasks,and platform of parallel CNC system based on embedded symmetric multi-processor S698PM and hybrid real-time task scheduler was established.Scheduling experiment was conducted to test performance of load balance and speed-up ratio of hybrid real-time task scheduler,and the feasibility of parallel CNC system was validated by machining experiment.
作者
陈传伟
付铄
韩振宇
富宏亚
CHEN Chuan-wei;FU Shuo;HAN Zhen-yu;FU Hong-ya(School of Mechanical and Electrical Engineering,Harbin Institute of Technology,Harbin 150001,China)
出处
《组合机床与自动化加工技术》
北大核心
2018年第4期145-148,156,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家科技重大专项(2013ZX04013-011-09)