Soft dielectric elastomer actuators (SDEAs) have manifested tremendous potentials in the field of soft robots. However, the tracking control of the SDEA still confronts enormous challenges because of its complicated nonlinear characteristics. In this paper, a nonlinear proportional-integral-derivative (NPID) controller is devised to fulfil the tracking control of the SDEA. A dynamic model of the SDEA is established, and then applied to construct a simulation system to coarsely tune the parameters of the NPID controller based on the iterative optimisation algorithm. Subsequently, according to the results of the coarse tuning, the parameters of the NPID controller are finely tuned in practical experiments. Finally, the NPID controller of the SDEA is proved to be effective via several tracking control experiments with different reference trajectories.