Abstract: Aiming at the problem of multi-cause process quality diagnosis in manufacturing, an algorithm based on dichotomy is proposed. Quality cause set is divided by degrees to build its complete binary tree, and then every node in the complete binary tree is selected as non-control cause set to construct cause-selecting control chart of the intermediate quality index in its parent node. In quality diagnosis, control charts with same parent are compared to determine the scope of the abnormal quality causes until they are affirmed. Theoretical analysis and practice prove that the diagnosis algorithm based on dichotomy could make the procedure of quality diagnosis independent to the adjustment procedure, and the steps of control charts comparison can be reduced greatly, so the efficiency of diagnosis is highly improved.

Key words: processes quality diagnosis;dichotomy;binary tree;cause-selecting control chart