ContinuousDeepQ-LearninginOptimalControl Problems: NormalizedAdvantageFunctionsAnalysis