To address the issues of insufficient accuracy and susceptibility to interference under non-linear conditions in unmanned aerial vehicles (UAV) power line inspections within global navigation satellite system (GNSS) denied conditions, a high-precision positioning technology integrating electromagnetic field and inertial navigation is proposed. A spatial distribution model of the electromagnetic field around transmission lines was established using the finite element method. Multi-feature electromagnetic observations including field intensity amplitude and body-frame components were constructed. A robust error-state cubature Kalman filter (RESCKF) fusion framework was designed. A normalized innovation squared (NIS) test was incorporated to achieve high-precision integrated navigation. From simulation results, the mean horizontal and vertical positioning errors are reduced to 0.3 569 m and 0.0 091 m, respectively, by the proposed algorithm under complex conditions. Compared with the extended Kalman filter (EKF), the accuracy is improved by 69.5% and 72.3%, respectively. Compared with the standard cubature Kalman filter (CKF), the accuracy is improved by 55.0% and 39.2%, respectively. High-precision positioning with centimeter-level vertical accuracy and decimeter-level horizontal accuracy is achieved under complex GNSS-denied operating conditions. The robustness and continuous positioning reliability of UAV power inspections are significantly enhanced.