基于改进的FCM聚类算法对温度测点的优化和建模

Research on Optimization and Modeling of Temperature Measurement Points Based on Improved Fuzzy C Means Clustering Algorithm

在分析了国内外热误差建模方法的基础上,提出了一种基于改进的模糊C均值聚类算法,从而基于多元线性回归理论建立教学型复合机床主轴热误差模型。使用温度传感器对机床主轴不同位置进行温度测量,并采用改进的模糊C均值聚类法对所测量数据进行分组,筛选出每组的最优温度值进行迭代。采用优选出的温度数据,采用多元线性回归建模法对机床主轴热误差进行预测建模。通过实验验证多元线性回归理论创建的预测建模分析可得:补偿后,教学型复合机床的主轴Y、Z方向受温度影响的热误差降低到了5.4μm以内,通过对改进的模糊C均值聚类法和多元线性回归模型相结合,使机床主轴在Y、Z方向误差有所降低,能更好的预测主轴热误差,从而提高机床加工精度。

On the basis of analyzing the modeling method of thermal error at home and abroad, a fuzzy C mean clustering algorithm based on the improved method is proposed, which is based on the theory of multivariate linear regression to establish the thermal error model of the spindle of the teaching type machine tool. The temperature sensor is used to measure the temperature of different position of machine tool spindle, and the improved fuzzy C mean clustering method is used to group the measured data, and the optimal temperature value of each group is iterated. The thermal error of machine tool spindle is predicted and modeled by multiple linear regression modeling method based on the optimized temperature data. Through experimental verification, the prediction modeling analysis created by multiple linear regression theory can be obtained: after compensation, the thermal error of the direction of Y and Z of the spindle of the teaching type machine tool is reduced to less than 5.4 μ m. By combining the improved fuzzy C mean clustering method with the multiple linear regression model, the spindle of the machine tool is in the Y and Z direction error. It can reduce the thermal error of spindle and improve the accuracy of machine tool.