Animal locomotion requires dynamic interactions between neural circuits, muscles, and surrounding environments. In contrast to intensive studies on neural circuits, the neuromechanical basis for animal behaviour remains unclear due to the lack of information on the physical properties of animals. Here, taking Drosophila larvae as a model system, we proposed an integrated neuromechanical model based on physical measurements. The physical parameters were obtained by a stress-relaxation assay, and the neural circuit motif was extracted from a chain of excitatory and inhibitory interneurons, which was identified previously by connectomics. Based on the model, we systematically performed perturbation analyses on the parameters in the model to study their kinematic effects on locomotion performance. We found that modification of most of the parameters in the simulation could increase the speed of locomotion. Our physical measurement and modelling would provide a new framework for neural circuit studies and soft robot engineering.