Physics-informed machine learning is a research paradigm and methodological approach that integrates fundamental physical laws, expressed typically as differential equations, conservation principles, or constraints, directly into the structure or training process of machine learning models, such as neural networks. It investigates methods for solving scientific and engineering problems governed by physical principles, particularly those characterized by limited, noisy, or sparse data. This approach enhances model accuracy, robustness, and generalization by ensuring predictions are physically consistent, thereby leveraging both data-driven insights and established domain knowledge.