The rapid development of intelligent industrialization has driven continuous innovation in technological equipment, resulting in the generation of large amounts of real-time data streams. Within these data streams, the statistical characteristics of the data may change over time, a phenomenon known as concept drift. Concept drift significantly impacts the performance of machine learning models. Failure to detect and address it in a timely manner can lead to a gradual decline in model performance, resulting in erroneous decisions and potentially causing substantial losses in industrial applications. This paper reviews the current research progress on concept drift detection and adaptation from the perspective of industrial applications. First, the paper focuses on supervised methods for industrial concept drift detection, providing a detailed overview of the development of relevant techniques, including performance-based methods, windowing techniques, and ensemble approaches. Second, to address the prevalent issue of label scarcity in industrial scenarios, the application of semi-supervised and unsupervised learning techniques in concept drift detection is systematically discussed. Furthermore, the paper discusses the impact of prevalent class imbalance challenge in industrial environments on concept drift detection and reviews strategies for addressing this issue. Finally, the paper summarizes concept drift adaptation methods in industrial settings and outlines potential directions for future research to enhance the performance of concept drift detection methods in complex and dynamic environments.