This paper presents a study on fluid flow modeling around an oscillating object in a channel using the Lattice Boltzmann Method (LBM) on GPU devices. The research aims to evaluate fluid behavior when an object oscillates in a channel, using the LBM with the TensorFlow framework for implementation on both CPU and GPU. The study models fluid flow over a circular cylinder with radius R = 0.00625 at various Reynolds numbers to verify the computational algorithm. The LBM is a computational method that models fluid as a collection of fictive particles on a discrete lattice, allowing for efficient simulation of fluid dynamics. The method is particularly suitable for discrete space and has been widely used in fluid mechanics due to its efficiency and accuracy. The paper discusses the implementation of LBM on GPU devices, highlighting the benefits of parallel processing and the use of modern frameworks like TensorFlow for cross-device compatibility. The study includes experiments on fluid flow around a stationary and oscillating circular cylinder. Results show that the flow around a stationary cylinder exhibits symmetric patterns with Karman vortex streets at higher Reynolds numbers. When the cylinder oscillates, the flow patterns change, with vortex structures breaking symmetry and increasing in size. The simulations also show that the scale of these structures decreases with increasing oscillation frequency, leading to larger-scale formations. Performance analysis indicates that GPU devices offer significant speedups over CPUs, especially for larger scales. The study concludes that the LBM is well-suited for simulating various fluid dynamics scenarios, and that GPU utilization can greatly enhance computational efficiency. The implementation was validated using known experiments and different Reynolds numbers, demonstrating the method's effectiveness for both fixed and oscillating objects.This paper presents a study on fluid flow modeling around an oscillating object in a channel using the Lattice Boltzmann Method (LBM) on GPU devices. The research aims to evaluate fluid behavior when an object oscillates in a channel, using the LBM with the TensorFlow framework for implementation on both CPU and GPU. The study models fluid flow over a circular cylinder with radius R = 0.00625 at various Reynolds numbers to verify the computational algorithm. The LBM is a computational method that models fluid as a collection of fictive particles on a discrete lattice, allowing for efficient simulation of fluid dynamics. The method is particularly suitable for discrete space and has been widely used in fluid mechanics due to its efficiency and accuracy. The paper discusses the implementation of LBM on GPU devices, highlighting the benefits of parallel processing and the use of modern frameworks like TensorFlow for cross-device compatibility. The study includes experiments on fluid flow around a stationary and oscillating circular cylinder. Results show that the flow around a stationary cylinder exhibits symmetric patterns with Karman vortex streets at higher Reynolds numbers. When the cylinder oscillates, the flow patterns change, with vortex structures breaking symmetry and increasing in size. The simulations also show that the scale of these structures decreases with increasing oscillation frequency, leading to larger-scale formations. Performance analysis indicates that GPU devices offer significant speedups over CPUs, especially for larger scales. The study concludes that the LBM is well-suited for simulating various fluid dynamics scenarios, and that GPU utilization can greatly enhance computational efficiency. The implementation was validated using known experiments and different Reynolds numbers, demonstrating the method's effectiveness for both fixed and oscillating objects.