Severe defluidization occurs when a reaction that is accompanied by a decrease in gas volume is performed in a fluidized catalyst bed. Defluidization occurs because the gas velocity in the emulsion phase decreases to less than the minimum fluidization velocity. Because numerous useful industrial reactions involve a decrease in gas volume, the development of a method for rapid detection of defluidization is important for maintaining good fluidization. In this study, the inside of a column was visually observed during CO2 hydrogenation in a fluidized bed. When the reaction rate was low, good fluidization was always observed. However, defluidization, such as channeling, occurred occasionally when the reaction rate was increased. These reaction conditions were classifled as the defluidization region in this study. The time-change characteristics for the frequency of pressure drop fluctuations were studied using continuous wavelet transfer (CWT). Additionally, the autocorrelation function was used to determine the dominant frequency. Using these functions, the decrease in the dominant frequency and the periodicity of the pressure fluctuations were detected in the defluidization region, even when the bed was observed to be satisfactorily fluidized. On the other hand, the prediction of defluidization itself several seconds before it occurred proved difficult.