基于自适应动态无偏最小二乘支持向量机的刀具磨损预测建模

Modeling Method for Tool Wear Prediction Based on ADNLSSVM

由于训练样本数量有限,滑动时间窗长度以及监测模型不能自适应调整和更新等因素,传统基于机器学习的刀具磨损预测模型存在精度和效率较低等问题,因此提出了一种基于自适应动态无偏最小二乘支持向量机(ADNLSSVM)的刀具磨损预测模型。采用公开数据库中的铣削加工数据集,通过时频域分析和小波包分解等手段从振动信号中提取特征量,并进一步利用相关性分析从中选择有效特征量作为模型输入。试验结果表明该方法所建模型具有较高的建模效率和预测精度。

In the building process of a tool condition prediction model with traditional machine learning methods,the limited number of available training samples and the fixed length of the sliding time window and prediction model resulted in lower modeling accuracy and efficiency.Dynamic model was set up to monitor the tool wear states by using an ADNLSSVM.Feature vectors were extracted by time-frequency-domain analysis from data set of open database of milling processes,and parts of them were selected by correlation analysis as model inputs.The experimental results shows better modeling efficiency and prediction accuracy.