Water outburst accident is the main threat to the safe production of coal mine roadway excavation and working face mining. The hidden water-bearing abnormal bodies such as goaf water and high-pressure Ordovician limestone water are the core sources of frequent water disasters in deep mining, causing serious safety hazards and economic losses. Transient electromagnetic technology, as the mainstream method for detecting abnormal water abundance bodies in coal mines, is strongly electromagnetically interfered by underground rails, metal supports, cables and other metals and charged bodies, resulting in serious distortion of the electromagnetic response received by the detection equipment. The traditional response model based on the ideal environment is difficult to adapt to the processing and interpretation algorithm, which is easy to cause misjudgment. In order to solve the above problems, based on physical simulation and numerical simulation, combined with the forward and inverse algorithm of transient electromagnetic data, this paper aimed to improve the interpretation accuracy of water-bearing abnormal bodies in coal mines. The results showed that the research had important practical significance for reducing production cost, improving safety production level and realizing efficient mining in coal mines. It had broad application prospects and could produce significant social and economic benefits.