In order to design a passive lower limb exoskeleton with good assisting effect, an optimal design method of lower limb exoskeleton was proposed based on the analysis of the motion and mechanical characteristics of human walking and the mechanical performance of the relevant major muscle groups. Through the human walking experiment, the kinematics information and plantar reaction force were obtained to drive the simulation of Anybody, and the mechanical data of lower limb muscles during walking were obtained. With the help of Hill muscle model, a simplified model of muscle?tendon?bone of hip joint in the sagittal plane of the human body was established, and a virtual torsion spring was added to simulate the role of exoskeleton, forming an integrated model of human body and exoskeleton. On this basis, the human-computer interaction force and the muscle activation of the wearer were quantitatively analyzed when wearing the assisted exoskeleton. The calculation models of muscle activation degree and metabolizable energy with torsion spring stiffness as variable were established, and the stiffness of virtual torsion spring was optimized by particle swarm optimization to obtain the optimal value with the goal of minimum metabolizable energy. Based on this, the design scheme of hip joint assisted exoskeleton mechanism was proposed, and the difference between the auxiliary torque of the mechanism and the virtual torsion spring torque was minimized as the goal to optimize, and the optimal values of the tension spring stiffness and the size of each connecting rod were obtained as the design parameters of the exoskeleton mechanism. At the same time, the prototype of hip joint assisted exoskeleton was made and the experiment of assisted walking was carried out. The results showed that the metabolic energy of human body was significantly reduced when wearing the assisted exoskeleton. The research method can provide reference for the design and analysis of other lower limb exoskeletons.