AI-Driven Intelligent Control: Paradigm Evolution, Key Technologies, and Future Prospects
Sicheng Jiang, Hongzheng Fan, Shuqing Yin · Knowledge Commons (Lakehead University) · 2026
As an important branch of automatic control, intelligent control has undergone a profound paradigm shift from classical control theory to deep integration with artificial intelligence since its emergence in the second half of the 20th century. Grounded in K.S. Fu's classic definition that "intelligent control is the combination of artificial intelligence, control theory, and operations research," this paper systematically reviews the paradigm evolution of AI-driven intelligent control, dividing it into three stages: knowledge-based intelligent control, data-based intelligent control, and knowledge-data-fused intelligent control. On this basis, it provides an in-depth analysis of the theoretical frameworks and research progress of key technologies, including deep reinforcement learning, adaptive dynamic programming, fuzzy-neural control, and multi-model intelligent control. Subsequently, it elaborates on the practical achievements of AI-empowered intelligent control through four typical application scenarios: industrial automation, unmanned systems, electric power and energy systems, and process control. Finally, in response to current core challenges such as interpretability, safety and reliability, and real-time performance, it envisions future directions including large model-driven control, generative AI integration, and human-machine hybrid augmented intelligence. This paper argues that artificial intelligence is moving from "instrumental empowerment" toward "paradigmatic restructuring," driving intelligent control toward higher levels of autonomy, generality, and intelligence.