The efficiency, reliability and lifespan of the fuel cell are strongly affected by the dynamic load variation of its output, which renders it desperate to investigate the influence of the power change rate on the fuel cell. In this study, three data dimensionality reduction algorithms are applied to identify the power change rate of the fuel cell in a less time-consuming way. To achieve that, the cell voltages of the stack as high-dimensional data-set is instantly projected onto the one-dimensional eigenvector space by principal component analysis (PCA), Fisher discriminant analysis (FDA) and locally linear embedding (LLE), respectively. The eigenvalue of one-dimensional eigenvector has the potential for instantly identifying the power change rate of the fuel cell and can be used as a feedback parameter in the control, which can improve the dynamic response, reliability, and life-span of the fuel cell stack. According to the performance indicators of the algorithms including monotonicity, linearity and program execution time, the result shows that the PCA algorithm is the best-matched method for the real-time control of the fuel cell system. In the end, this study discussed some potential applications of this method in the fuel cell system, be it to be used alone or in a vehicular fuel cell hybrid system. (C) 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.