Research On Testability Modeling And Analysis Of System Based On Modelica

Testability is a characteristic of system; it reflects the ability of detecting workstatus of system quickly and accurately. The level of testability directly determinesthe difficulty of acquiring system information. The system testing and diagnosisbecomes more difficult as the complexity of system increases. However, people putforward higher requirements on the testability of system.The method of testability modeling and analysis can quantitatively evaluate thetestability index of system, guide developers to improve the design for testability,improve the level of testability, shorten the development cycle of system, save thedevelopment costs and help maintenance personnel to develop fault diagnosis andisolation strategies. Therefore, studying the method of testability modeling andanalysis is of great significance. This paper puts testability index distribution,testability modeling and analysis, fault diagnosis strategy optimization together tobuild a testability system. The software of system testability modeling and analysisis developed based on studying the theories above seriously. The main contents ofthe paper are as follows:Firstly, the modeling principle, mathematical model and modeling steps ofmulti-signal flow are studied, and then the methods of testability analysis based onmulti-signal flow model are studied. The advantages and disadvantages ofmulti-signal flow model are analyzed; the multi-signal flow model doesn’t have theability to describe the uncertain relation between faults and tests, so a new methodof testability modeling and analysis based on Bayesian network is proposed in thispaper.Secondly, study the principle, mathematical model and classical algorithm oftestability index distribution, and then analyze and compare the advantages anddisadvantages of different algorithms. Make a research on optimization algorithm offault diagnosis strategy. Combine the failure rate and the cost of testing as themeasure of choosing tests and optimize test sequence with weight algorithm on thebasis of testability analysis.Thirdly, make a research on syntax and compiler theory of Modelica which is alanguage for multi-domain modeling. Improve the modeling mechanism of theModelica language and make it not only can build functional model, but also canbuild testability model, and then develop the component libraries of testability.Finally, system testability modeling and analysis software is developed, and theoverall scheme and data structures of software are designed. The modeling platform,simulation module, testability analysis module, index distribution module and fault diagnosis strategy optimization module are designed. The functions and accuracy ofthe software which is developed in this paper are verified with specific examplesand the results of testability analysis are compared with the results got by TEAMSwhich is the commercial software for testability.