Fault Prediction of CNC Machine Tools Based on Gutenberg-Richter Law and Fuzzy Neural Networks

Abstract

The magnitude-frequency relationship is one of the most cited empirical equations by seismologists for studying seismic activity, and it has been widely used in earthquake forecasting and earthquake hazard analysis. Applying this equation to fit and analyze a large amount of CNC machine failure data, similar conclusions to those of seismology were obtained. The adaptive neural network ANFIS model was used to predict the b-value as an output quantity, associated with the fault level and number of faults. The sample data were tested separately using the ANFIS toolbox of MATLAB software and the Neural Net Fitting APP function. The ANFIS model has better accuracy in b-value prediction, which shows that the ANFIS model has certain ability to predict the parameters of the G-R Law. b-value prediction reflects the stability of CNC machine operation to a certain extent, and the change of b-value can speculate the possibility of CNC machine failure in continuous operation, which has certain reference significance for the normal operation of production.