Abstract: The daily production data of the Zhangheng satellite can reach up to 20 GB, rendering manual methods inadequate for handling such massive data demands. This paper proposes a rapid and robust Lightning Whistle Scattering Coefficient Automatic Extraction Method (LWSC-AEM). Firstly, detailed data from the Search Coil Magnetometer (SCM) of the Zhangheng satellite is extracted using a sliding window of 0.8 seconds, which is then transformed into time-frequency plots and audio files. Secondly, a YOLOV5 neural network is employed to automatically locate LW in the time-frequency plots and output their time-frequency position information. Subsequently, the corresponding audio data containing Lightning Whistlers is extracted based on this time-frequency position information from the files, and zero-padded to form audio segments of 0.8 seconds. Finally, the Mel Frequency Cepstral Coefficients (MFCCs) of the audio segments are extracted and fed into a Gate Recurrent Unit (GRU) improved with a multi-head attention mechanism to automatically extract the LW scattering coefficient. Applying this method to the data from the VLF band of the SCM payload of the Zhangheng satellite in February 2020 yields the following results: the average absolute error and average absolute percentage error are 0.453 and 0.176 respectively. Compared to the method by Ref. 1, the average absolute error is reduced by 1.079, a decrease of 70%, and the average absolute percentage error is reduced by 0.148, a decrease of 46%. The average processing time per data segment is 0.074 seconds, which is a reduction of 0.826 seconds, or 92%, compared to the method by Ref. 1, which processed each data segment on average. The automatic extraction method for lightning whistler wave scattering coefficients proposed in this paper can quickly and accurately extract these coefficients.