论文部分内容阅读
将多向偏最小二乘(MPLS)方法应用于青霉素间歇生产过程的建模与故障诊断中。从青霉素反应过程的特点来看,数据具有多维性,应用传统的偏最小二乘方法会使过程的统计建模与故障诊断难以实现。MPLS可对间歇过程的多维数据沿变量方向进行分割,使得多批量的数据可以在过程的各操作阶段建立相应的PLS模型,从而完成对该反应过程的实时监视与故障诊断。运用T2统计、Q统计方法,结合贡献图对过程进行了仿真分析,从理论分析和仿真实验结果的一致性,证明了该方法在青霉素生产过程的故障检测与诊断方面是可行的。
The multi-directional partial least squares (MPLS) method is applied to the modeling and fault diagnosis of penicillin intermittent production process. The characteristics of penicillin reaction process, the data has multidimensional, the application of the traditional partial least squares method will make the process of statistical modeling and fault diagnosis difficult to achieve. MPLS can segment the multidimensional data of the intermittent process along the direction of variables so that multiple batches of data can establish corresponding PLS models in each operation stage of the process so as to complete the real-time monitoring and fault diagnosis of the reaction process. Using T2 statistics, Q statistical method and the contribution graph, the process is simulated and analyzed. The consistency between the theoretical analysis and simulation results proves that this method is feasible in the fault detection and diagnosis of penicillin production process.