In previous research, we have developed a high-dorsiflexion assistive robotic technology aiming for gait rehabilitation targeting on ankle dorsiflexion movement. A McKibben-type artificial muscle is applied to provide large dorsiflexion force while adding little weight to the device. This ensures the foot clearance before initial stance phase in gait. Meanwhile, a tension spring is deployed in series with the artificial muscle to support heel rocker function in loading response phase. Suitable spring coefficient for each individual differs according to ankle's dorsiflexion torque in loading response. An unsuitable spring would lead to knee deviation in this phase. In this study, we derived an identification equation to determine a suitable spring coefficient for individuals based on estimation of dorsiflexion torque required to support. An evaluation test on healthy objects was conducted, which shows no negative effects on participants' knee angles with the identified spring coefficient.