To reduce the influence of thermal error on the machining accuracy of NC machine tools,the selfbuilt five axis precision CNC machine tool was taken as the research object.The change rules between position error and temperature change was found out.The thermal error model of Y axes was established based on least square support vector machine（LS-SVM）.The parameters of the least square support vector machine model were optimized.According to least square support vector machine thermal error model,The prediction data comparing curves of the linear axes positioning error in the thermal equilibrium state were calculated.Through analysis,the prediction accuracy of the thermal error model based on the least square support vector machine is better,the fitting bandwidth of the linear axes value is narrower,and the mean square deviation is smaller.According to the LS-SVM model,the error compensation experiment was carried out,and the error was reduced by 87.3%.The experimental results demonstrate that the least square support vector machine modeling method has high prediction accuracy and precision.