Unlock the Intermittent Control Ability of Model Free Reinforcement Learning