Dynamic compensator-based near-optimal control for unknown nonaffine systems via integral reinforcement learning.