论文部分内容阅读
针对传统的多向主元分析(Multiway Principal component Analysis,MPCA)常会导致误诊断,且对批生产过程难以保证在线状态监测和故障诊断的实时性,提出了一种改进的MPCA与动态时间错位(Dynamic Time Warping,DTW)方法,该方法采用多模型非线性结构代替传统的MPCA单模型线性化结构,并利用对称式DTW算法解决了多元轨迹同步化的问题。将该方法应用到青霉素发酵批过程的在线故障监测中,结果表明它克服了MPCA不能处理非线性过程和实时性问题,并避免了MPCA 在线应用时预报未来测量值带来的误差,提高了批过程性能监测和故障诊断的准确性。
In order to overcome the shortcomings of MPCA, MPCA often leads to misdiagnosis, and it is difficult to guarantee the real-time performance of on-line condition monitoring and fault diagnosis for batch process. An improved MPCA and dynamic time misalignment Dynamic Time Warping (DTW) method, which uses the multi-model nonlinear structure instead of the traditional MPCA single-model linearization structure and solves the problem of multiple locus synchronization by using the symmetric DTW algorithm. The method was applied to the on-line fault monitoring of penicillin fermentation batch process. The results showed that it overcomes the problem that MPCA can not handle the nonlinear process and real-time problems and avoids the error brought by future measured value when MPCA is online, Process performance monitoring and troubleshooting accuracy.