Wind turbine blade icing is an important challenge in wind power systems that can degrade turbine performance, increase maintenance costs, and impact safety. The purpose of this review is to provide an overview of currently available methods for predicting icing on wind turbine blades. First, the adverse effects of icing on wind turbine operation are described, including reduced power generation efficiency and increased risk of mechanical damage. Secondly, an overview of commonly used icing prediction methods, as well as prediction algorithms based on shallow machine learning and deep learning, is presented. Different data preprocessing methods are combined. The principles, advantages and disadvantages, and applications of each method are reviewed and compared. Finally, the limitations of the current methods and the directions for future research are pointed out, including improving prediction accuracy, reducing cost, and realizing real-time monitoring. These reviews are instructive for further research and development of new wind turbine blade icing prediction methods.