论文部分内容阅读
根据实际机械生产过程中设备的全寿命数据集不易获取以及性能退化数据难以在线监测的情况,提出了一种在只有少子样的截尾数据时利用设备运行状态信息来对单个设备的运行可靠性进行评估的方法。在该方法中,首先从设备的振动、电流信号中提取并优选出几项退化特征量;其次针对截尾数据,采用了高斯过程机械学习法拟合预测出设备失效时的各项退化特征指标值;然后以这些特征指标值为输入,借助于分布假设、贝叶斯推理,构建了一个运行可靠性的评估模型;最后通过数控刀具的失效实验对该方法进行了验证。结果表明:预测精度可以达到3%。此结果证明了该方法的可行性和有效性。
According to the fact that the whole life data set of equipment is not easy to be acquired and the performance degradation data is difficult to monitor online, a method is proposed to use the equipment operating status information to run the reliability of a single equipment with only a few samples of censored data How to evaluate. In this method, firstly, several degradation features are extracted and optimized from the vibration and current signals of the equipment. Secondly, for the censored data, the Gaussian process mechanical learning method is adopted to fit the degradation index Then, based on the distributional assumptions and Bayesian inference, an evaluation model of operational reliability is constructed based on these characteristic index values. Finally, the method is validated by failure of numerical control tools. The results show that the prediction accuracy can reach 3%. This result proves the feasibility and effectiveness of this method.