在主动测量中的基于灰色支持向量机的补调值预测

Fill Value Prediction Based on Grey Relational Support Vector Machine in Active Measurement

针对磨削加工中无法实现自动补调,且传统模型无法精确预测补偿值变化趋势这一问题,提出了将灰色关联系统与支持向量机相结合的预测模型,灰色关联系统通过分析比对关联度大小筛选出影响程度大的因素,并将对应的参数作为输入。以此训练出支持向量机预测模型,通过模型预测的补调值通过主动量仪实现自动增减。实验结果为通过交叉验证优化后的预测值平均相对对误差为MRE=0.385,均方根误差为MSE=0.266,实验证明了模型的可行性与可靠性。

In order to solve the problem that automatically changingfill value prediction can not be realized,and the traditional model can not accurately predict the change trend of compensation value in the grinding process.A prediction model which combines the grey correlation system with the support vector machine is proposed.The grey relational analysis system screened out the factors with great influence through analyzing and comparing the correlation coefficient.And the corresponding parameters are used as input to train the prediction model of support vector machine.The model to predict the compensation value is automatically increased by supplementing active measuring instrument.The experimental results are the average relative error of MRE=0.385 through cross validation,root mean square error MSE=0.266.The feasibility and reliability of the model are proved.