Study on Testability Evaluation of Complex Equipment Based on Fault Injection Test Data

Abstract

Classic statistic method can not make use of the history test information and produce the evaluation conclusion with low confidence level and high risk under the condition of small sample, a new testability evaluation method based on test data in development stages is proposed in this paper. The result shows that this method can produce the evaluation conclusion with high confidence level in the same test condition and it is more rational with this new method than with classical statistical method and the traditional Bayes method. Introduction Testability is an important design feature of equipment. It describes the equipment's ability to detect and isolate faults. It can reduce cycle costs of equipment life and increase mission reliability, and enhance the comprehensive protection of equipment capability which has a good testability. In recent years, testability is becoming one of comprehensive security indicators as reliability and maintainability. The main indicators of testability is including fault detection rate (FDR), fault isolation rate (FIR), the false alarm rate (FAR), the fault detection time (FDT), fault isolation time (FIT) and retest OK rate (RTOKR), etc. The first three indicators are the most important among them. Therefore, assessment of testing complex equipment is actually assessing FDR, FIR and FAR. Complex equipment testability experiments In testing experiment, choose the pre-selected fault to be injected, and inject it into the unit of system or equipment, and detect its fault and implementation procedures for fault isolation and fault indication. Its corresponding results can only be two kinds of forms: success or failure; failure to be detected (detection success) or not detected (detection failure); fault is isolated to a specific unit (isolated success), or did not complete isolation (isolation failure); when indicates, test (or use) a failure instructions may be true (indicating success), or there is no actual fault (indicating failure), a false alarm. Testability Bayes evaluation of the complex equipment It’s the key of Bayes assessment that using a priori information to determine the prior distribution. To the overall test for complex equipment of success or failure type (binomial distribution), it commonly used conjugate prior distribution to determine the prior distribution in engineering, the parameter, FDR, conjugate prior is the Beta distribution. To facilitate the calculation, set the FDR for P, the experiment is divided into N stages and achieve prior distribution I ( , ) i i Be P a b . Its density function is 1 1 1 1 ( ) (1 ) ( ) (1 ) ( ) ( ) ( , ) i i i i a b a b i i i i i i i a b P P P P P a b a b π β − − − − Γ + − = − = Γ Γ (1) 0 1, P ≤ ≤ 1, 2, , i N =  Where i a >0, i b >0 are the ultrasonic parameters in development stage of prior distribution i. In the case of knowing the prior distribution, the value of i a and i b are keys of determining the prior distribution. Their value can be calculated used the method provided in literature [6]. 2155 2nd International Conference on Electronic & Mechanical Engineering and Information Technology (EMEIT-2012) Published by Atlantis Press, Paris, France. © the authors When determine i a and i b , then achieve prior distribution of I, ( , ) i i Be P a b . Development phase has been determined after is. kWith the prior distribution, combined with field testing experimental information (set the number of field test n, the number of fault detection failure f), and use Bayes Theorem to export the posterior distribution ,Its density function: 1 1 (1 ) ( ) ( , ) N N