Autonomous gait is a fundamental element for access to independent life and avoiding the de-socialization of people with motor disabilities. In this context, this research is part of the EXOKID project which aims at designing a personalized exoskeleton for children with cerebral palsy. For such personalisation, two 9 years old twin sisters, one with spastic cerebral palsy (C) and a healthy one (H) without any impairments, performed several walks with electromyography (EMG), kinematics and force acquisitions. This paper presents a model of the knee and hip's spastic angular displacement of C during a walk using mechanical differential equations. Two models were designed: one based on the timing of the muscular activation, the other where the timing is defined using a genetic algorithm (GA). These models highlight the spastic contributions of the muscles involved in the walk (agonists and antagonists of the joints studied) and their activations. The amplitude of the activations for both models was carried out using GA. Gait cycles were modeled with a determination coefficient (R 2) higher than 84% for both models.