<p>This article investigates a new method for fault detection in transmission lines connected to wind turbines using a combination of Adaptive Fourier Decomposition (AFD) and deep learning. Due to the increase in the production power of wind turbines, the need to determine the adaptive threshold value for more accurate fault detection is felt. Initially, new signals are generated for all three phases using the data from both sides of the transmission line. Then, the AFD method is used to analyze these signals. AFD is an effective technique that helps extract the signal’s frequency components and provides essential information for fault diagnosis. In this process, the recognition accuracy is increased by identifying duplicate components and updating the residual signal. In addition, using a deep learning method, the threshold value for fault detection is adjusted adaptively, increasing the accuracy and reducing the detection time. The simulation results show that this method can effectively improve the performance of power transmission line protection systems. Finally, the proposed method has been compared with other relevant techniques to demonstrate its significant advantages and superiority.</p>