Study On Scheduling Theory In Real-Time NC System And Its Applications

As numerical control (NC) technique is becoming prevalent in advanced manufacturing industry, there is an increasing demand for more complex, sophisticated, and dependable hard real-time computing systems, where periodic and aperiodic tasks generally co-exist. Both periodic tasks, such as cutter compensation, interpolation, position control, etc, and aperiodic tasks created by paroxysmal events must be completed within their deadlines. Meanwhile, timing constraints are essential for the correct running of the system even if hardware or software faults occur. Therefore, real-time computing requires jointly scheduling periodic and aperiodic tasks which are always correlative in execution ordering or resource contention, and needs fault-tolerance to behave correctly despite the presence of failures. In order to meet all above new requirements, we present a real-time scheduling theory for NC system, and the effectiveness and feasibility of theory is demonstrated with simulation results. The main work and contributions in this dissertation are as follows.Firstly, the art of the state of real-time scheduling theory is researched in this thesis. The principles and ranges of application of some typical returns are analyzed especially, and the main problems of existing research are put forward.Secondly, considering the characteristics of periodic and aperiodic tasks in NC systems, a new hybrid tasks system model is presented in the thesis and the heuristic best-first (BF) algorithm is introduced to schedule tasks of this model. The design principles and pseudocode describe of BF algorithm are presented. The computation complexity of algorithm is analyzed, and simulation experiment is carried out.Thirdly, considering the bad environment of industry field which tends to cause some unexpected faults of NC machine tool, fault-tolerant technique based on rollback recovery is introduced into BF algorithm to remove transient fault of system.Fourthly, the current research station of worst-case execute time (WCET) is analyzed in detail. Then, to improve the over pessimistic result and reduce the time cost of WCET prediction, a new scheme is proposed in the thesis. Through adjusting the structure of program, using symbolic execution technology and branch-and-bound method, the scheme eliminates infeasible paths and decreases the number of paths to be analyzed. The accuracy and efficiency of WCET analysis are raised effectively with this scheme.