Abstract: In view of the complex harmonic pollution caused by the high proportion of new energy being integrated into the distribution network, the existing detection methods are difficult to effectively deal with the dynamic changes of sub-synchronous, super-synchronous harmonics and non-stationary harmonics. Therefore, the research on harmonic current detection technology of new energy distribution network filters is carried out. The adaptive phase tracking algorithm is introduced to replace the traditional PLL, and the definition of instantaneous reactive power is improved according to the operational characteristics of the new energy distribution network. Meanwhile, in view of the shortcomings of the fixed parameter filtering method in harmonic current detection, wavelet transform is adopted for detection. However, there are problems in the selection of wavelet basis functions and the setting of wavelet parameters in wavelet transform. Furthermore, a parameter optimization method for the wavelet transform of intelligent algorithms is proposed. Taking the genetic algorithm as an example, the selection of the wavelet basis function and parameter optimization are regarded as a combined optimization problem, and the harmonic detection accuracy is used as the fitness function for solution. Practical applications show that this method can maintain high detection accuracy and dynamic tracking capability under both steady-state and non-steady-state operating conditions. It can provide reliable harmonic compensation instructions for active power filters, effectively enhancing the harmonic control level of distribution networks and the safety of system operation.