TABLE OF CONTENTS Digital systems have two types of faults from the viewpoint of operational failures: permanent faults due to hardware failures or software errors, and intermittent faults due to transient failures [98], [99]. Intermittent faults are automatically detected by the error-correcting code and corrected by the error control [100], [101] or the restart [102], [103]. However, some faults occur repeatedly, and consequently, will be permanent faults. Some tests are applied to detect and isolate faults, but it would waste time and money to do more frequent tests.
Continuous and repetitive tests for a continuous Markov model with intermittent faults were considered in [48]. Redundant systems with independent modules were treated in [46]. Furthermore, they were extended for non-Markov models [98] and redundant systems with dependent modules [104].
This section applies the inspection policy to intermittent faults where the test is planned at periodic times kT( k= 1, 2, ) to detect these faults (see Figure 8.4). We obtain the mean time to detect a fault and the expected number of tests. In addition, we discuss optimum times T* that minimize the expected cost until fault detection, and maximize the probability of detecting the first fault. An imperfect test model where faults are detected with probability pwas treated in [50].
Suppose that faults occur intermittently; i.e., a unit repeats the operating state (State 0)...
