An Efficient Algorithm for Optimal Design of Diagnostics

Abstract

Even though much work has been done in developing models to design diagnostic tests for fault isolation in electronic systems, there is still a lack of efficient, fast procedures. A new approach to the cost-effective design of fault isolation procedures is investigated. It is oriented specifically toward built-in-test (BIT) diagnostic subsystems for modular electronic equipment. The problem is formulated as a search tree. A branch and bound technique is used to find the optimal sequence of tests for the automatic BIT diagnostic to isolate a single malfunctioned unit among a group of Line Replaceable Units (LRUs). Dominance and branching rules are developed, and the algorithm is presented. An example problem is solved to illustrate the efficiency of the branch and bound algorithm. Computational results are given and discussed. The following conclusions can be drawn. 1. The branch and bound approach can be used successfully in designing a cost-effective fault-isolation procedure. Because of the branching and dominance rules, many of the non-optimal solutions are eliminated early in the solution procedure which could efficiently reduce the size of the required search tree, as well as the time and storage needed to find the optimal solution. 2. The proposed branch and bound algorithm is efficient in solving problems of size up to eight LRUs. 3. Even though the size of problems solved efficiently by the algorithm is limited to eight LRUs, this size is still greater than the size of any problem reported to be solved optimally in any previous work.