Abstract
In this study, a new technique of in-process characterization of the grinding wheel surface is proposed. Some reference topographies of the wheel surface are formed by different dressing conditions. and the condition of the wheel surface is discriminated where the dynamic fre-quency spectrum signals of grinding sound and/or vibration are analyzed by a neural network technique. In the case of conventional vitrified-bonded alumina wheel, both grinding sound and vibration can be identified under the optimum network configuration in such that learning rate is 0.003 and number of hidden layer is 160. Accordingly this system can recognize the differ-ence of the wheel surface in a good degree of accuracy insofar as the micro-topography of abrasive grains are relatively widely changed.
