Chinese Journal of Computers   Full Text
Title	Applying Improved AO* Based on DPSO Algorithm in the Optimal Test-Sequencing Problem of Large-scale Complicated Electronic System
Authors	JIANG Rong-Hua WANG Hou-Jun LONG Bin
Address	(School of Automation, University of Electronics Science and Technology of China, Chengdu 610054)
Year	2008
Issue	No.10(1835—1840)
Abstract &   Background	Abstract An algorithm of improved AO*based on discrete binary particle swarm optimization (DPSO) is proposed, which can solve the Optimal Test-sequencing problem in large-scale complicated electron system. DPSO optimizes the test sets which can isolate the expanded node in AO*algorithm to decrease the number of node; The improved AO*limits the test cost range of node and lessens the traces. The result of real operation show that this algorithm not only reduces the computational complexity, cuts down the test cost, shorten the test time; but also avoids the "computational explosion" when the test set is too large. Keywords discrete binary particle swarm optimization； AO* Algorithm； test sequence； Huffman coding； design for testability Background The research in this paper is concentrated on optimal test sequencing problem, which is taken account into in design for testability and fault diagnosis field. The existing methods for test sequential problem can easily cause ‘computational explosion' and not be fit for large-scale system. This paper is supported in part by the National Defense Basic Research under grant No.A1420061264 and General Armament Department pre-inquest Foundation under grant No.51317040102. The first grant is for system design for testability analysis, and the second grant is to development an integration software environment for system design for testability. The paper proposes a DPSO-AO*algorithm for test sequential problem, which utilizes the DPSO algorithm to select the test point at each nodes of fault isolation tree and improves AO* algorithm to adjust the test sequence. The proposed method highly reduced the number of created test nodes than AO* algorithm, saved 10% test cost and shorten four times test time.