As the demand for measurement accuracy intensifies in modern industry and scientific research, the evaluation of measurement uncertainty has become a crucial component in ensuring product quality and optimizing production processes. Traditional methods for assessing measurement uncertainty are widely applied in static and linear systems but have shown limitations when dealing with high-dimensional, nonlinear, or dynamic systems. In recent years, non-statistical methods have effectively complemented uncertainty evaluation for complex systems. In particular, neural network techniques, with their powerful data processing capabilities and nonlinear modeling capabilities, have become essential tools for uncertainty evaluation. This study provides a comprehensive review of statistical methods, including the Guide to the Expression of Uncertainty in Measurement and the Monte Carlo method, as well as non-statistical approaches such as Bayesian inference, grey evaluation, fuzzy evaluation, and maximum entropy evaluation. The fundamental principles, advantages, limitations, and typical application scenarios of each method are systematically analyzed. In addition, the recent trend of integrating these uncertainty evaluation methods with machine learning is discussed. Particular attention is given to emerging neural network-based indirect evaluation methods, including deterministic models, Bayesian neural networks, and ensemble learning frameworks. The modeling and evaluation strategies of these approaches are examined in the context of complex nonlinear systems, highlighting their potential and current limitations. Finally, the applicability of various uncertainty evaluation methods is summarized, and future research directions are outlined. The study suggests that the integration of multiple evaluation paradigms can enhance the modeling capability and reliability of uncertainty estimation in complex measurement systems, reduce dependence on large data samples, and better address the increasingly intricate measurement requirements in modern industry and scientific research.