A new autoregression curve method based on energy gradient is proposed. It can improve the accuracy of measuring shear wave velocity with single-hole method. Firstly

the time-history curves of shear waves are transformed into a set of spatial energy gradient wave time-history curves. Secondly

short trigger average/long trigger average (STA/LTA) method is used to transform energy gradient wave time-history curves and pick the arrival times of P wave and S wave. Finally

a quadratic autoregression model is proposed to pick the accurate arrival times of P wave and S wave. The model is used to process the energy gradient curves near the arrival time. A comparing experiment is conducted to validate the theory. The shear wave velocities of one field tested with proposed method

cross-correlation function method and manual picking method are analyzed. The results indicate that in comparison with the cross-correlation function method

this method has a better inhibitory effect on noise

and not only can get accurately S shear wave velocity but also can get accurately P shear wave velocity. Meanwhile the proposed method has much higher accuracy and stability in practical test relative to the manual picking analysis. This method enriches the automatic-picking methods of shear wave velocity. On the basis of method proposed in this paper

cross-correlation function method and manual picking method

a shear wave measurement system is designed. It is easy to pick shear wave velocity automatically and produce a test report by this system.