Abstract: Oil leakage detection technology is crucial for ensuring the safety of oil storage and transportation. Given the flammable and explosive nature of oil products, leaks from storage tanks or pipelines can lead to significant safety hazards, including fires and explosions. However, traditional infrared image target detection algorithms are prone to interference from ambient temperature, lack sensitivity to small leaks, and have limited detection capabilities under low-resolution conditions. To address these issues, an infrared thermal imaging algorithm for oil leakage detection based on Utility-Value (U-V) causal reasoning and energy-based model (EBM) was proposed. Advanced concepts such as causal reasoning and scene understanding from the U-V theory were integrated with EBM, enabling the algorithm to thoroughly analyze thermal imaging data. This approach facilitated an upgrade from perception to decision-making, thereby reducing the risk of safety incidents caused by delayed responses to oil leakage. The results indicate that this approach significantly enhances detection accuracy and anti-interference capability compared to traditional infrared image target detection methods. Verification using real scene data from an oil depot in Gansu revealed precision and recall rates exceeding 95%. The application of this approach in oil storage and transportation can provide greater reliability for the safe operation of oil pipelines.