虚拟基准设计及在设备技术状态监测中的应用分析

Virtual datum design and application analyse for equipment technical state monitoring

针对测试设备在周期校准期间性能参数漂移的监测问题,提出了采用虚拟基准代替实物基准进行设备状态监测的方法,给出了虚拟基准系统实现原理和结构框架。采用神经网络算法(NN)构建虚拟基准预测模型,多项式回归算法(MR)构建参考模型并作为比较基模型,获取虚拟基准的NN状态误差输出值和MR状态误差参考值。设计了可信度指标RI(reliance index)和相似度指标SI(similarity index)作为虚拟基准可信度和测量过程数据符合性评价标准,对虚拟基准的可靠性和设备技术状态数据进行分析。实际应用表明,设计的虚拟基准方法与实物基准方法相比,有较高的监测准确性。其平均绝对百分比误差(MAPE)和最大相对估计误差(MREE)分别为0.12%和1.06%。

Deal with the question of detect the equipment-performance drift happening in-between the scheduled measurements,a novel virtual datum scheme （VDS） is proposed for monitoring process quality without taking actual measurements, the VDS implementation principle and framework was presented. This work adopts a neural-network （NN） algorithm as the conjecture algorithm for establishing the VD conjecture model and uses a multiregression （MR） algorithm as the reference algorithm for establishing the reference model that serves as a comparison base for the conjecture model, the virtual datum status error of NN conjecture and MR reference is procured. The RI（reliance index） was defined for estimate the reliance level of the virtual measurement value. The SI （similarity index） was defined for evaluate the degree of similarity of the input-set process data. Both RI and SI are applied to determine the reliability of the virtual datum results and analyzing the process data of production equipment. Test results show that the proposed method can achieve high prediction accuracy with mean absolute percentage error being 0. 12% and maximum error being 1.06%.