The signal processing of pulse coherent wind lidar usually uses fixed range-gate to divide the time domain signal, and perform frequency spectrum calculation for each range-gate to obtain wind speed information. The time domain signal division of the fixed range-gate has the problem of non-periodic truncation of the intermediate frequency signal, which leads to spectrum leakage during spectrum calculation, resulting in errors and reduced signal-to-noise ratio. An adaptive range-gate division method based on full cycle search was proposed. The length of the range-gate was adaptive to the frequency of the intermediate frequency signal, which could realize the full cycle division of the signal, avoide the problem of spectrum leakage and improve frequency estimation accuracy. The two processing methods were simulated and analyzed by adding noise signal. The results show that the adaptive range-gate method can realize the adaptation of the range-gate length and the intermediate frequency signal. When the signal-to-noise ratio was less than 1 dB, the intermediate frequency estimation error obtained by this method was 38%-62% of the fixed range-gate method. The adaptive range-gate division method was used to process the turntable and wind field echo signals obtained by the laser wind measurement radar system, and the results were compared with the wind lidar equipped with the fixed range-gate method. The results show that the root mean square error of the adaptive range-gate method for the speed measurement of the turntable is 0.19 m/s, and the range resolution of atmospheric wind speed measurement varies from 7 to 11 m, which is better than that of fixed range-gate method, and improves the range resolution and measurement accuracy of the wind lidar.