Multi-sensor Fuzzy Data Fusion In The Fluidized Bed Flow Pattern Recognition Applications

Gas-solid fluidized beds have been assumed as important process units in the production process of chemical and energy industry. Information of flow regime is needed in the model building, design, research and development and operation of gas-solid fluidized bed, which is extremely important to improve the performance of the fluidized bed and increase the heat transfer rate and mass transfer rate. Since pressure fluctuation time series in gas-solid fluidized beds which are easy measure by normal pressure instruments, contain many dynamic information, such as particle characteristics, operating conditions, bubble behaviors, pressure fluctuation time series are analyzed to get the information of flow regimes in this paper. The complexity of pressure fluctuation its the reflection of nonlinear two-phase movement in the fluidized beds. Since complexity analysis has simple but effective algorithm with few changeable parameters, complexity parameters, including algorithm complexity (C(n)), fluctuate complexity (Cf) and Shannon entropy (Entropy), are adopted to indicate the flow regimes in the gas-solid fluidized beds. The three complexity parameters reflect the complexity degree in the fluidized beds from different aspects and are able to indicate some flow regimes independently. For the first time, it is found that the Shannon entropy is able to identify bubble and turbulent in the fluidized bed, which are difficult to identify by the former nonlinear parameters, such as chaos parameters or C(n) or Cf.Based on observation and analysis of large amount of experimental phenomena, regime division is modified combined with morphological variation of the chaos attractor and numeric variation of the complexity parameters. It is figured out that there are transitions between the deterministic flow regimes and the adjacent flow regimes do not vary suddenly from the previous regime to the posterior regime. The complexity parameters vary continuously in the transition flow regimes. It is found that there is an obvious distinction between the starting bubble state and the fully bubble state. Transitions are quantified with a fuzzy variable named "membership degree" to indicate the transition degree.Fusion theory and the complexity theory are associated and applied to the flow regime identification in gas-solid fluidized beds for the first time. Two-level fusion model is established. In the character fusion level, three complexity parameters, which are C(n), Cf and Entropy of the pressure fluctuation time series of the same pressure transducer, are combined to identify the flow regimes. Membership functions are established to indicate the transition flow regimes. In the decision level, the identification result of different pressure transducer in the different spatial location in the gas-solid fluidized beds are combined to get the final identification result. With the two-level data fusion, flow regime information in the gas-solid beds is reflected from multi aspect and full complementary of all the available information is realized. The experimental result shows that the identification result of the multi-parameter and multi-transducer data fusion is better than that of the single parameter or single transducer. In a word, fuzzy fusion theory supplies a valuable new method to identify the flow regimes in the gas-solid fluidized bed.