INTRODUCTION The development of a low-cost process for the removal of acid gases and solid particles in incinerator exhaust gases is desirable. A cyclone scrubber is one of the processes that can absorb gases, separate particles and decrease the temperature of the gas at the same time. To simultaneously understand the gas absorption and particle separation phenomena in a cyclone scrubber, it is very important to understand gas absorption and particle separation exclusively. In this article, gas-liquid sorption phenomena in a cyclone scrubber were studied. Despite the relatively simple design and wide use of these types of scrubbers, the processes occurring with simultaneous absorption and separation, their interactions and fluid dynamics are quite complex and give rise to rather complicated design and optimization problems. Consequently, its modeling is a complex task since it is necessary to solve a system of nonlinear differential equations and evaluate many chemical and transport parameters. Therefore, the development of simpler and more reliable models is the subject of much research. A process model is a functional relationship between variables that explains cause-and-effect relationships between inputs and outputs. Models can be developed from fundamental principles, such as the laws of conservation of mass, energy, and momentum and other principles of chemical engineering. Such models are able to explain the underlying physics of the system and are called phenomenological models. However, due to the complexity of the process in the cyclone scrubber system, it is very difficult to obtain accurate phenomenological models. Even if an accurate phenomenological model is obtained, ... half of the paper ... the proposed model is more reliable. Table 3 Average absolute relative errors between experimental and calculated values ​​Gas-liquid system Average absolute relative error, %CO2-Ca(OH)2-H2O:L/G = 0.16L/G = 0.1L/G = 0.07CO2 -NaOH-H2O:L/G = 0,160,3850,5400,3660,725CONCLUSIONThis paper studies the modeling strategy using artificial neural networks for the nonlinear dynamic processes of a cyclone scrubber. Three-layer feed-forward neural network (3-FFNN) was chosen for neural network modeling. Comparison between neural network simulation results and experimental data was discussed to demonstrate the validity of the proposed model. The comparison illustrates that the accuracy of 3-FFNN is satisfactory with the experimental one. In conclusion, the highly nonlinear behavior of the cyclone scrubber can be successfully modeled using the 3-FFNN.