Enhanced Principal Component Analysis And Its Application To Online Flooding Monitoring In Packed Towers

Flooding is the limit of the operating conditions in packed towers. However, with the demand of safety and efficiency, it is desirable to operate close to flooding point without actually reaching flooding and causing instability. Therefore, there is a constant need for improved flooding detection methods, particularly those adaptable to a range of different processes.After an overview of the traditional methods of flooding monitoring, PCA-based(Principal Component Analysis) methods are presented for monitoring flooding in this thesis. In order to improve the monitoring performance, an improved model, i.e., Enhanced-DPCA(Enhanced Dynamic Principal Component Analysis), based on the ensemble learning strategy is proposed.The main contributions in this thesis are as follows:(1) Traditional approaches for flooding monitoring are far from ready to be used in commercial systems because of their shortages, while a large amount of operational data can be utilized for operation analysis. PCA and its improvements are proposed for online flooding monitoring in this work. It finds that a single global model is often difficult to achieve a satisfied monitoring performance because of the diversity and imbalance of sample data. To overcome the problem, an ensemble modeling method combined with FCM(Fuzzy c-Means) clustering algorithm and Bayesian inference is proposed. The monitoring results of an air-water packed tower indicate that the proposed enhanced method obtains better and more reliable monitoring performance.(2) Enhanced-DPCA is introduced to the data acquisition software to develop a real-time monitoring software for the air-water packed tower. Before applied to experiments, the parameters of Enhanced-DPCA are adjusted in view of the data fluctuation and process noise. Then, the monitoring software is used for monitoring the operation conditions of different spray densities. The monitoring results demonstrate that the packed tower can be operated under a safe and efficient gas velocity. Finally, the monitoring software is improved to realize the automatic control function. Under normal condition, gas velocity can increase automatically to reach a more efficient working condition. Once flooding, the software can decrease the frequency of the centrifugal blower to return to normal condition.