Optimal control plays a central role in modern industrial, energy, transportation, and economic systems, where accurate decision-making and stability are essential. Traditional control approaches rely on analytical mathematical models; however, many real-world systems are highly nonlinear, time-varying, and influenced by uncertainties, making classical modeling insufficient. Recent advances in machine learning (ML) and artificial intelligence (AI) have enabled data-driven modeling, real-time forecasting, and adaptive decision-making. This paper presents an in-depth analysis of machine-learning-based optimal control methods, including neural network modeling, reinforcement learning (Q-learning, deep RL), and hybrid MPC-ML approaches. Practical applications in energy load control, industrial automation, robotics, transport optimization, and economic decision-making are discussed. A laboratory experiment on nonlinear system stabilization demonstrates that ML-based control achieves high prediction accuracy (97.8%) and effective system regulation. The results show that ML significantly enhances control efficiency in complex and uncertain environments. Future technological systems are expected to increasingly rely on intelligent, self-learning control strategies.