Research On Quantitative Modeling Of Testability Evaluation

Testability is a crucial characteristic of wherther system can response of its working startus timely and accurately. As the system is becoming more and more complex and integrated, the difficulty of fault detection and isolation is growing and the cost of maintenance is increasing. This requires the products have good testability. Now, the product testability has aroused widespread concern and attention. Testability evaluation is needed in every stage of product design. At present, general testability evaluation model is mostly qualitative model, which is lack of quantitative information, inaccurate, only applicable to testability expection and can not be used during process of product stereotypes and acceptance. Besides, physical testability verification has the disadvantages of great danger, high coast and fault injection difficult. So after deep study of testability evaluation techniques, the quantitative model of testability evaluation is proposed, the algorithm for solving quantitative model is designed and testability evaluation software is developed. The main contents are as follows:Firstly, the principle and mathematical description of multi-signal flow testability evaluation model is analyzed. There is necessity for testability evaluation model developed from the qualitative to the quantitative, because multi-signal flow model requires more experienced modelers and its error is large. So the quantitative model of testability evaluation is proposed and its modeling method is analyzed.Secondly, the algorithm process of solving quantitative model is designed. The method of fault sample set generation based on improvement Failure Criticality is proposed, in which both the failure rate and severity is considered. For fault dictionary is difficult to establish or establish inaccurate ly, the method how to establish fault dictionary automatically is proposed. For the two-dimensional correlation matrix can not describe the relationship among fault, fuzzy field and test, the three-dimensional correlation matrix is proposed and the method of static testability analysis and testability index calculation is studied.Then, the Open Modelica software is extended according to the requirements of the quantitative model and the algorithm process, so that the software has the ability of building testability evaluation quantitative models, generating fault samples, injecting fault model, state simulation, test function simulation, generating fault dictionary, test decision simulation and testability analysis. The design of testability quantitative evaluation software is completed.Finally, a low-pass filter circuit quantitative model of testability evaluation is built to verify the function of the quantitative evaluation software, including fault sample generation, state function simulation, test function simulation, fault dictionary generation, test decision simulation and testability analysis. Besides the quantitative evaluation software is used to build a military digital circuit quantitative model of testability evaluation, and the results are compared with those of physical verification which proves the accuracy and usefulness of the quantitative modeling techniques in testability evaluation.