Hydrogen fuel cell life prediction method based on ensemble learning

The invention discloses a hydrogen fuel cell life prediction method based on integrated learning, and the method comprises the steps: obtaining multivariable historical monitoring data in the working process of a hydrogen fuel cell, carrying out the feature selection of the multivariable historical monitoring data through a limit gradient lifting algorithm, determining a health index and an optimal input variable set, and obtaining an optimal input matrix; a Blending integration-based life prediction model is constructed, a base learning layer is composed of a DBN, a GRU and a TCN, and a meta learning layer is composed of a GPR; optimizing the key hyper-parameters of the integrated model by using an improved artificial rabbit optimization algorithm IARO; the constructed integrated model is trained in combination with the optimal input matrix and IARO, and a Blending integrated prediction model after optimization training is obtained; and a final prediction result of the residual life is obtained by combining