Surface roughness monitoring and dimensional error control in turning by quasi-sensor fusion

A quasi-sensor fusion approach is described which uses a single sensor (strain gauge) instead of multiple sensors. A neural network is used as a decision-making to estimate surface roughness and dimensional errors on a lathe. Five processed data signals from the strain gauge are fed to the neural network. Experimental results show that the proposed approach can estimate the quality of workpieces to within five micrometers for surface roughness and ten micrometers for dimensional errors