SPIO-PLL-pshRNA分子探针转染卵巢癌SKOV3细胞及MR成像研究

针对间歇过程的多阶段问题，提出了一种新的划分阶段策略，可以将多阶段的数据进行准确的聚类，得到精确的阶段数目，避免陷入局部最优，提高了建模精度和故障检测准确度。首先采用不同展开方法相结合的方式对三维的间歇过程数据进行处理，消除了数据预估问题；其次结合图划分准则利用谱聚类方法将处理后的数据进行准确的阶段划分，并运用PCA建模方法分别对每一类进行建模；最后通过合适的模型对数据进行在线监控。通过青霉素仿真过程验证了本文方法的有效性。

Tansfection of ovarian cancer SKOV3 cells with SPIO-PLL-pshRNA molecular probes and MR imaging

A phase segmentation strategy to handle the batch processes with multiple operation phases is proposed in this paper. It can not only obtain an accurate stage classification and avoid falling into local optimum, but also improve the precision of modeling and the accurate of fault detection. Firstly, the three dimension batch process is handled by integrating several expansion methods such that the data estimation step may be removed. And then, the spectral clustering method together with graph partitioning criterion is utilized to divide the above data into several clusters which is further modeled via PCA. Finally, a suitable model is applied to monitor these data on line. The effectiveness and flexibility of the proposed method is validated through penicillin simulation process.