Variable-weight Combination Prediction of Thermal Error Modeling on CNC Machine Tools

Abstract

Due to the thermal error modeling of CNC machine tools has characters of small sample and discrete data, the variable-weight combined modeling method was presented by integrating time series analysis and least squares support vector machines. Taking minimum sum of error square of prediction model as the optimization criterion, optimal weights in different time were calculated. Using grey GM (1, 1) model to predict the variable weights, the prediction result of thermal error was obtained as well. Application of the grey variable-weight combined model on a five axis vertical machining center indicated that it can get higher prediction accuracy than single modeling method. Therefore online error compensation to CNC machine tool becomes more effective.