Fuel cell health state estimation method and system based on compensation GPR model

The invention belongs to the technical field of fuel cell health state estimation, and particularly relates to a fuel cell health state estimation method and system based on a compensation GPR model, and the method comprises the steps: obtaining feature parameters of a fuel cell under a discharge condition, and employing a self-sparse self-encoder to extract target features in the feature parameters; taking the target feature as the input of a pre-constructed Gaussian process regression (GPR) model, taking the battery SOH as the output of the GPR model, performing hyper-parameter optimization on the GPR model by adopting a Kepler planetary optimization algorithm, and outputting an SOH first predicted value by the GPR model after optimization; setting a compensation function to carry out compensation optimization on the SOH first predicted value to obtain an SOH second predicted value; and selecting an average absolute percentage error, a mean square error and a maximum error as evaluation indexes to evaluate