Optimisation of Structured Neural Controller Based on Continuous-Time Policy Gradient