Development of a fuzzy-nets-based in-process surface roughness adaptive control system in turning operations

This paper discusses the development of an in-process surface roughness adaptive control system for a CNC turning operation, using fuzzy-nets modeling and tool vibrations measured with an accelerometer. The goal of this system is to predict the surface roughness of a surface being turned, determine if the surface roughness being generated is higher than the desired specification, and if so to adapt the feed rate of the turning operation in order to obtain a surface roughness no higher than that specified. Fuzzy-nets models for prediction of surface roughness and adapted feed rate were trained using feed rate, spindle speed, tangential vibration and measured surface roughness data collected during experimental runs. A series of validation runs indicated that the system could successfully meet its goal both in detecting out-of-spec surface roughness conditions, and adapting the machine tool to obtain a final surface roughness at or slightly below the desired surface roughness.