Finite horizon optimal tracking control of partially unknown linear continuous-time systems using policy iteration