Abstract
The machining noise produced in honing operation is honing sound. This study attempts to investigate experimentally the relation between the honing sound and the finishing performances of stones and to construct the neural network system to discriminate the finishing conditions. The main results obtained are as follows. When the stone pressure Pn is larger than the critical pressure Pc, the stone is on the condition of self-dressing and keeps the superior cutting ability. Then the honing sound with high sound pressure level is produced. The sound pressure becomes higher under the severer honing conditions. The honing conditions are grouped into three classes based on the magnitude of stone pressure; the low pressure condition (Pn is lower than Pc), the normal pressure condition (Pn is larger than Pc) and the severe pressure condition (Pn is much higher than Pc). The neural network system is constructed to recognize the frequency spectrum of the honing sound produced under these three conditions. The effects of cell number of hidden layer and of learning rate on prediction results are investigated. It is experimentally proved that the proposed neural network system has a high discrimination rate and the honing sound is applicable as a signal to monitor the finishing conditions of stones.
