The distribution of lithium in lithium-ion batteries is critical to their performance and lifespan. Studies have shown that thermal neutron imaging can effectively perform non-destructive testing of this distribution. However, most current thermal neutron imaging systems rely on expensive, bulky, and immobile nuclear reactors or spallation neutron sources, significantly limiting their application scenarios. Therefore, this study adopts a portable D-T neutron generator as the neutron source and uses Monte Carlo simulation methods to optimize the design of key components in the thermal neutron imaging system. Simulation results indicate that the optimized thermal neutron imaging system achieves a thermal neutron flux of 8.36 × 104 n·cm−2·s−1 and a thermal neutron content (TNC) of 14.4% at a collimation ratio of 50. Within an imaging range of Φ100 mm, the non-uniformity of the thermal neutron fluence is approximately 2.07%. Simulation imaging results of lithium and iron step wedges demonstrate that the compact thermal neutron imaging system designed in this study can effectively distinguish between iron and lithium of the same thickness. This finding provides an important reference for subsequent detection of lithium distribution within lithium-ion batteries.