Abstract: To solve the problem that the output power of the magnetically coupled resonant wireless power transmission (MCRWPT) system drops sharply when the coils are over coupled, we propose an adaptive particle swarm optimization (APSO) algorithm that meassures the algorithm process by particle variance. Using frequency splitting to consider the power and efficiency characteristics of the system, the tracking target point is selected as the maximum power point on the right side of the inherent resonance frequency. The proposed variance APSO algorithm dynamically adjusts the parameters according to its variance process factor, which improves its global performance in the early stage and its convergence in the later stage. Simulations verifiy that the proposed variance APSO algorithm has higher steady-state accuracy and fewer convergence algebras when implementing the maximum power point tracking of the MCRWPT system, and the advantages of the algorithm are statistically significant.