In order to improve the tracking accuracy of 3D dynamic acoustic array for 2D maneuvering target in colored noise environment, the adaptive interacting multiple model unscented particle filter algorithm based on measured residual was proposed. The 3D motion acoustic array tracking system dynamic model was established, and initial probability density function was also defined based on unscented transformation, after that, the measured covariance and state covariance were online adjusted by measured residual and adaptive factor. Finally, the Matlab simulation results between different algorithms show the validity and superiority of the presented algorithm in tracking accuracy, stability and real-time capability.