Testability Test And Integrated Evaluation Technology With Virtual-Physical Test

Testability is an important quality assuring index in the equipment development and purchase process. It is paid more and more attentions from the producer, the purchaser and the consumer. Testability test and evaluation are key measurement to check the testability design and the manufacture level, and is the basis of the developing phase shift and the index verification. How to decrease effectively the test cost and risk, and shorten the test period, while assuring the confidence and accuracy of the testability test and evaluation is an emergent problem in the theoretical and practical fields.Aiming at the problems in physical testability test and evaluation such as large test sample size, difficulty in fault injection, high risk, long period and low confidence, an overall testability test and integrated evaluation scheme is proposed combining the research results in the virtual testability test technology. The sample size determination method, the sample size allocation and failure mode selection method, and the integrated testability index evaluation method are systematically studied.The major research point of the dissertation are as follows.1. Design for the overall testability test and integrated evaluation schemeFour classical sample size determination methods are analyzed and compared, the optimized test plan based on the single sampling method and sequential probability ratio test(SPRT) method is proposed. Aiming at the problem of insufficient prior information for the small sample test and incredibility of the virtual test data, the overall testability test and evaluation scheme based on virtual-physical test is proposed. The implementation flow of the scheme is described, and two key aspects are pointed out as the testability test planning based on Bayesian theory and the testability index integrated evaluation method using prior information of multiple sources.2. Research on the testability test plan based on Bayesian posterior risk criteria(1) Aiming at the incredibility of virtual testability test data, the current implementation pattern of virtual testability test and the features of the virtual testability test data are analyzed. A virtual-physical test data conversion method is proposed based on the information entropy theory.(2) Aiming at the problem that the single sampling method can not use the prior test data, the sample size determination method based on Bayesian posterior risk criteria is proposed. The equations and connotations of the posterior producer’s risk and the posterior consumer’s risk are given. The test plan solution flow for the posterior risk criteria method is given according to the rules between the posterior risks and the combination of test time and failed test time. The results show that the sample size is effectively decreased with the same test plan restriction parameters.(3) For the large error of random sampling and the irrational failure mode set due to the inaccurate failure rate during the sample size allocation, the equivalent effect of the virtual test data to the physical test data is analyzed, then the sample size allocation method which takes the environmental factors and the virtual test factor into account is proposed. The environment-failure rate factor calculation method based on grey relationship analysis and the virtual test factor calculation method based on the simulation credibility of failure mode are proposed. The factors are used to modify the sample size allocation model. The study shows that the sample size allocation method proposed effectively adjusts the sample size proportion among the components of the system. The failure mode set based on the method is much more rational, which can fully demonstrate the testability capacity of the equipment.3. Research on the testability test plan based on sequential posterior odds test(SPOT) method(1) Aiming at the problem that the actual sample size of SPRT method may be large, Bayes theory is taken to modify the SPRT method, then the testability test planning based on the SPOT method is proposed. The decision rules and the decision thresholds calculation method are given. The operational characteristic(OC) and average sampling number(ASN) of the single sampling method, the SPRT method and the SPOT method are compared. The OC and ASN varing rules under a range of different prior parameters are analyzed. The study shows that under the same test restriction parameters and test condition, the actual sample size of SPOT method is smaller than the actual sample size of SPRT method.(2) Aiming at the shortcomes of current truncated SPOT method on the definition of additional bilateral risk and the calculation of truncated test time, the testability test planning based on optimized truncated SPOT method is proposed. The bilateral risks split method, the calculation method of truncated test time and decision threshold are given. The OC and ASN of the optimized truncated SPOT method under different risks split types are compared and the determination principle for the optimal split type is confirmed. The OC and ASN of the SPOT method, the current truncated SPOT method and the optimized truncated SPOT method are compared, which shows that the OC similarity between the optimized truncated SPOT method and SPOT method is better than the current truncated SPOT method, and the ASN of the optimized truncated SPOT method is smaller. The result shows that the optimized truncated SPOT method can effectively supplement the SPOT method in the testability test planning.4. Research on the testability index integrated evaluation method using prior information of multiple sourcesA testability index evaluation model using prior information of multiple sources is proposed to solve the problem of low evaluation accuracy and confidence under small sample physical test data. Firstly, the types of the prior information such as testability prediction information, testability expert information and testability virtual test data are analyzed. The maximum entropy method is used for the conversion of testability prediction information and expert information, and the empirical Bayesian method is used for the conversion of testability virtual test data, which convert the prior information into the prior probability density functions(pdf) of the testability index. Secondly, a parametrical data consistency check method is used to check the compatibility between each source of prior information and physical testability test data. For the prior information passed the check, the prior credibility is calculated. A prior distribution is formed based on the prior pdfs and corresponding credibility. The Bayesian posterior distribution model is acquired with the mixed prior distribution and test data, based on which the point and interval estimates are calculated. Finally, examples are taken to verify the proposed method, and the impact of data amount of virtual test data and the VV&A result of the virtual prototype on the evaluation result are analyzed. The study shows that the proposed evaluation method can modify the accuracy of the evaluation, and both the increase of the data amount of virtual test data and the VV&A result of the virtual prototype can enhance the evaluation accuracy.5. Case studyA software named the testability integrated test and evaluation system is designed and a missile flying control system is used as the subject for method demonstration.