Random test generation for fault detection and diagnosis
Abstract
A method of test generation is developed which aims at: (1) eliminating the golden unit (a random test generation scheme in which the outputs of the golden unit and the circuit under test are compared); (2) minimizing the length of the test sequence; (3) reducing the complexity of deriving the tests; (4) minimizing the chances of making a wrong decision; and (5) minimizing the information required for reaching a decision about the state of the circuit. The testing procedure developed is analogous to the statistical hypothesis testing problem. Tests are derived for four different kinds of faults. The effectiveness of the tests is studied by randomly injecting faults into a circuit. A new testing procedure called the sequential probabilistic testing procedure is developed for the purpose of fault diagnosis. An algorithm is presented for computing the optimal sequential testing procedure for locating a fault.
 Publication:

Ph.D. Thesis
 Pub Date:
 1978
 Bibcode:
 1978PhDT........78G
 Keywords:

 Chips (Electronics);
 Electric Equipment Tests;
 Gates (Circuits);
 Probability Theory;
 Random Processes;
 Algorithms;
 Arithmetic And Logic Units;
 Pins;
 Sequential Analysis;
 Electronics and Electrical Engineering