Artificial intelligence and machine learning applications in drying systems: A comprehensive review

The industrial drying sector faces persistent challenges regarding energy inefficiency, nutrient degradation, and complex heat-mass transfer interactions. To mitigate these constraints, a systemic paradigm shift is transitioning dehydration technologies from rigid empirical models toward adaptive machine learning frameworks. This comprehensive review evaluates the technical evolution of artificial intelligence (AI), machine learning (ML), and deep learning (DL) topologies in modern drying operations. Traditional mechanistic formulations suffer from high computational costs and estimation barriers, which severely limit online process control. Data-driven machine learning architectures, including artificial neural network (ANN) and deep learning models such as recurrent neural network (RNN) and long short-term memory (LSTM) networks, as well as deep-learning-enhanced model predictive control (MPC) strategies, offer superior accuracy in predicting process kinetics and reducing specific energy consumption. To address the data-dependency bottlenecks of pure black-box models, this paper critically examines the integration of physics-informed machine learning (PIML) and physics-informed neural networks (PINNs), which mathematically enforce governing transport equations and boundary conditions directly within neural networks. Furthermore, the implementation of high-fidelity digital twins and real-time surrogate models is explored as a method to circumvent heavy computational overhead during complex porous-media and microwave interactions. Finally, this review addresses critical industrial challenges, emphasizing the urgent need for dataset standardization across research institutions and the strategic alignment of smart drying platforms with global carbon footprint reduction targets and United Nations Sustainable Development Goals (UN SDGs). Through this structured analysis, a structured computational roadmap is provided to guide the drying industry toward an autonomous, high-fidelity, and net-zero engineering future.