Research On Data-Driven Prediction Model Of Flooding Gas Velocity And Realtime Flooding Monitoring In Packed Column

The flooding gas velocity of packed column is the upper limit of its operating gas velocity. Research on the flooding of the packed column and forecasting its flooding gas velocity makes it possible that packed column can be operated steadily in the high capacity to avoid the loss and production halts of liquor due to the oversight or delayed judgment on flooding, which may lead to the tremendous economic loss. This research not only has great significance on the enhancement in safe operation of column, improvement in the economical efficiency, together with energy saving and emission reduction, but also has a wide prospect of industrial applications.Based on the summarization concerning the traditional experience model of flooding gas velocity and the method of online monitoring of flooding, this paper specifically introduces the original approach based on data-driven and builds an algorithm regarding the gas velocity prediction model of flooding, and the corresponding steps. Moreover, it also puts forward monitoring the acoustic wave change which formed due to flooding in the packed bed to realize the real-time monitor, which is a new theory as well as a method. Furthermore, these new methods have been verified via corresponding experiments.The encompassed innovative research of this paper is as follows:1. It makes compressive summary on the domestic literature data which have been published in recent decades with regard to experiment of random packing, so the corresponding data base have initially been established. In view of fact that the RBF (Radical Basis Function) network has launched the prediction model on the flooding gas velocity, the method of SVM (Support Vector Machine) regression is applied to this exploration in order to build the flooding gas velocity model of packed column. According to the available flooding gas velocity model of packed column, a prediction software under MATLAB is developed, which can estimate the flooding gas velocity of packed column in advance.2. Through the real-time monitoring experiment on the acoustic emission of the two-phase convection flow in packed columns and the analysis of the feature of acoustic wave signal above of the packed bed under the normal or flooding condition, it is found that the characteristic frequency of experimental column under flooding state is located at the range of180-200Hz.3. The exploration adopts the microphone array to collect the change of acoustic wave signal which is above the packing bed. Afterwards, NAH (Nearfield Acoustic Holography) is used for computing and then, the rebuilt diagram about change of sound pressure has been drawn. This method overcomes the flaw that average amount of acoustic wave in flooding area is the only thing the single microphone collected, while this method can better reflect the change of sound pressure in any positions within the area of flooding, and it can used to research the flooding in the certain part of packed column.Ultimately, this paper sums up the research findings, and presents the direction of research which needs to be explored and improved further.