A method for multi-channel phase internal calibration of meteor radar based on adaptive differential evolution and collaborative position estimation is presented. This approach constructs a composite objective function based on statistical properties, specifically the proportion of angle-of-arrival (AOA) estimation "outliers" and the characteristic altitude distribution of meteors. By employing the JADE search strategy with an adaptive mutation factor and optional external archive, the method synchronously conducts iterative estimation the AOA and altitude information of meteor trails. This framework efficiently facilitates the high-precision multi-channel phase offsets and meteor trail positions. Primarily simulation results, fulfilled with a 24-hour observational data set containing 18,318 meteor trails, demonstrate that the search algorithm typically achieves stable convergence within 150 generations. The root mean square error of the multi-channel phase offsets is less than 1°. The proposed method substantially enhances the data quality of meteor radar and is highly applicable for the routine on-line self phase calibration measurements.