Testability modeling method covering uncertainty of test point

Abstract

The invention discloses a testability modeling method covering the uncertainty of a test point. The method comprises four steps of 1, after describing a system by using a dependency graphical model, finding out the fault rate of each component and the fault rate of each test point according to the historical data of the system; 2, computing the undetectable rate, second-type false alarm rate and untrustworthy rate of the system; 3, working out the corresponding detectable influence degree, false alarm preventing influence degree, trustworthy influence degree and risk-free influence degree of the system when each test point is uncertain; and 4, establishing test dependency matrixes (a detectable influence degree dependency matrix, a false alarm preventing influence degree dependency matrix, a trustworthy influence degree dependency matrix and a risk-free influence degree dependency matrix) when the test point is uncertain. After the test dependency matrixes are obtained via the above four steps, the optimization of the test point can be performed by using an intelligent algorithm, and thus the testability index is specifically improved, and the testability design is optimized. Therefore, the method fills in the blank in the technical field of testability.

Description

technical field [0001] The invention belongs to the technical field of testability, in particular to a testability modeling method considering the uncertainty of test points. Background technique [0002] Existing test modeling is often carried out under deterministic assumptions, without considering the actual uncertainties. This kind of model is often based on a large number of positive assumptions. For example, as long as there is a fault, the test point that the signal flows upward can be measured, that is, the reliability of the test point and the influence of other uncertain factors on the test result are not considered. But in fact, because a large number of uncertain factors are artificially ignored, the predicted effect of the test design is better than the actual effect. [0003] Aiming at the uncertainty of testability, many studies have been carried out at home and abroad, and many new testability modeling methods have been proposed. Wang Baolong, Huang Kaoli a...

AI Technical Summary

Problems solved by technology

In the actual system, the test point is not completely reliable, because it will fail, its test capacity is not unlimited, and the test capacity of each test point is also different

Images