Study On The Petri Net Modeling And Performance Analysis Of A Testing System

As testing system evolving towards a more intelligent, automated and networked environment, the architecture and composition of such system become more and more complex. Since traditional system design methodologies are no longer able to keep up with the current requirements, there is an imminent need for a new set of system design theories and methodologies that can be applied to system modeling as well as accurate analyzing and evaluating system characteristics, so that designing, optimizing and controlling the system can be based on more sound rationales. Towards this goal, this paper applied the Petri net theories to the qualitative and quantitative study of a testing system, established several kinds of Petri net models for such system (including system architecture, hardware circuits, bus protocol and software etc.), studied the testing system modeling and analysis methodologies through analyzing and computation on those models and at the same time extended the Petri net theories and applications. Author's main contributions are listed in the following five areas:1. Applied modeling to typical circuitries, testing bus protocols and software design of the intelligent instruments, especially established a complete and hierarchical extended timed Petri net (ETPN) model for an automatic testing system of an actual aircraft engine upper air test platform parameters and summarized the general modeling methodologies of the testing system. The abstract ETPN model established is able to accurately, vividly, clearly and strictly represent the structure of the testing system as well as the complex relationships between different system components. The running of the ETPN model represents the dynamic workflows of the system. The objects of modeling are the commonly used components of the testing system, including analog circuits, digital circuits, A/D conversion circuits, memory circuits, bus, CPU, DSP and software etc. Especially worth mentioning is the original work of the hardware circuitry modeling and hardware software co-modeling. Thosemethodologies can be used in not only testing system but also other systems that consist of hardware circuitries, computer and software.2. Analyzed the system characteristics using the ETPN model. This includes qualitative analysis such as: reachability, activeness, boundary, security, conservation and fairness etc, as well as quantitative analysis such as: execution time of circuits, completion time of the acquisition tasks and working speed etc. Furthermore, through analysis, the relationships between time and resources constraint of the transitions (represent circuits or tasks) were obtained to guide the design of system architecture, circuits and software.3. In ETPN model, transitions' execution time are constant, but in real system they are stochastic due to various reasons, thus a general stochastic Petri net (GSPN) model that the execution time of the transitions satisfy negative exponential distribution has been proposed. Based on this, this paper utilized the GSPN to model and analyze the testing system. This paper proposed a computation reduction method for equivalent transition probabilities when simplifying the GSPN's reachable states through the elimination of instantaneous states, thus obtained the steady state probability through solving the equation and computed the resource utilization rate and testing tasks speed of the independent component.4. Based on the real situations of the system and circuitry testing that some transitions' execution time do not satisfy negative exponential distribution, author proposed a normal distribution general stochastic Petri net (NGSPN), i.e. execution time of the transitions conform to normal distribution, constant and instantaneous time. In order to solve the explosive expansion problem of the complex system's Petri net state space, author adopted a method that is different from the common stochastic Petri net analysis, studied the reduction methodologies of the basic structure of Petri net---cascading and parallel model, deducted an equivalent theoretical formula for cascading model reduction; proposed approximate estimation formulas for parallel models in common...