Colored Petri Net Based Modeling Of Parallel Automatic Test Systems

The performance of automatic test systems (ATS) is an important research subject of modern measurement field. Along with the progress of test equipments and test methodology, conventional ways of sequential task scheduling and system design become the bottleneck of increasing performance. Recent years, the academic field has made a study on parallel task scheduling and methods of ATS integration and has achieved firstfruits. Several key problems related to parallel task scheduling and modeling ATS with Colored Petri Nets (CPN) are systematically studied in this dissertation. Two new algorithms of parallel task scheduling are proposed, and CPN is applied in the modeling of parallel ATS. Combined with a practical research project, task scheduling and design verification of radar receiver ATS are theoretically and experimentally studied, and valuable achievements are gained. The detail discussion and main contribution are listed as follows:Two new algorithms of parallel task scheduling named TaskScheduler and TaskScheduler-T are proposed, which are based on the data correlation, control correlation and resource correlation relationship. TaskScheduler is used when the test time of tasks is unknown, and TaskScheduler-T is used in the condition that the test time of tasks is already known. It is proved that task sequences automatically generated by above algorithms have the highest degree of parallelism or have the shortest test time. The simulation results illustrate TaskScheduler and TaskScheduler-T algorithms efficiently improve the resource utilization and cut down overall test time.A novel ATS design method based on formalization is presented, and ATS using parallel task scheduling algorithms is modeled with CPN. Because the CPN analytical procedure can be applied to verify properties of the ATS CPN models, the correctness of system design is guaranteed from early stage of system integration. Reachability tree and linear algebra methods are utilized in the verification of ATS CPN model. Boundness, liveness, fairness, persistence, conservation, structuralboundness and structural conservation are proved for automatic test systems which employ TaskScheduler and TaskScheduler-T algorithms. Therefore the correctness of TaskScheduler and TaskScheduler-T is verified.Combined with a project of developing radar ATS, a CPN model of ATS using TaskScheduler-T is proposed for the radar receiver. The simulation result indicates the ATS is efficient and practically valuable.All the algorithms in this dissertation have been implemented on a uniprocessor computer. The correctness and efficiency of parallel task scheduling algorithms are proved by the simulation results.