This paper proposes a possibility walking state analysis model for legged robots based on the feedback of acceleration sensor processed by discrete Fourier transform the dynamic analysis of the walking of legged robot. According to Mahalanobis distance, this model can express the legged robot walking status quantitatively. Using quadrupled robots as an example for evaluation, the experimental results show that this model is able to describe the walking of legged robots and express the walking status in real time, and it can help legged robots adjust their locomotion with the change of environment to ensure their walking stability.