Comparison of computational intelligence and statistical methods in condition monitoring for hard turning

Abstract

Hard turning is a critical machining operation that imposes strict requirements on both the cutting and machine tools. The condition of the machine in terms of its stability and ability to maintain proper operating conditions is very critical in a hard turning operation. Bearing wear and fixture alignment are two critical machine health factors that exert a great influence on the hard turning operation. This paper presents models to support hard turning processes. In particular, models have been developed to predict cutting tool flank wear and forces in hard turning based on experimental data. It has also artificially simulated bearing wear and fixture misalignment failures and developed models that can predict such failures in their incipient or propagating stages. These models were developed using regression as well as neuro-fuzzy techniques. Their performance was evaluated in two situations: distant future state predictions and predictions in the presence of noise. It was observed that neuro-fuzzy models perform better than regression models.