Abstract: The concept of "Machining Error Flow" is introduced in order to adapt to the development of the machining processes monitoring, controlling and quality assurance in modern manufacturing system. Two kinds of dynamic models of machining error flow are developed on the iterated mapping and catastrophe theory. The principle, strategy and method of real-time machining error control are proposed. Meanwhile, the experimental results of the least square regression of the machining errors in multiple operation and the error prediction using neural networks are also presented.

Key words: Catastrophe theory, Error flow, Iterated mapping, Least square regression, Neural networks, Prediction