Abstract
Abstract
Thermal deformation of high-speed motorized spindles is one of the primary factors limiting the machining accuracy of precision CNC machine tools. Improving the robustness of axial thermal drift prediction under multi-speed operating conditions is of great significance for achieving high-precision thermal error compensation. In this study, a thermal error prediction model based on a subtractive clustering adaptive neuro-fuzzy inference system (SC-ANFIS) was proposed for a high-speed motorized spindle supported by ceramic ball bearings. First, the spatial distribution characteristics of the spindle steady-state temperature field were analyzed using the finite element method, and temperature and thermal drift data under different discrete spindle speeds were obtained through a thermal characteristics experimental platform. To reduce multicollinearity among variables, a hybrid method combining fuzzy c-means clustering (FCM) and grey relational analysis (GRA) was employed to optimize ten candidate measurement points into three temperature-sensitive points. Considering that both thermal error characterization parameters and thermal displacement are relative quantities, the SC-ANFIS model adaptively extracted high-density data clustering centers in a multidimensional feature space to determine fuzzy inference rules, thereby alleviating the structural redundancy caused by the conventional grid partitioning method. Experimental results under multiple spindle speeds demonstrated that the proposed SC-ANFIS model achieved improved prediction accuracy and robustness compared with conventional ANFIS, backpropagation neural network (BPNN), and multiple linear regression (MLR) models. The proposed approach provides a feasible modeling strategy for thermal error prediction of high-speed motorized spindles under multi-speed operating conditions.
Graphical Abstract