PEMFC Improved Fractional-order Subspace Identification Model Based on IECSDE Algorithm
To accurately describe the characteristics of a Proton Exchange Membrane Fuel Cell(PEMFC)during its power-generation process and the relationship among variable effects,an improved fractional-order subspace identification method based on the Information Exchange Cuckoo Search Differential Evolution(IECSDE)algorithm is proposed to establish a fractional-order state-space model for the PEMFC.In this study,a PEMFC model is established based on state-space equations.Considering the fractional order of PEMFCs,fractional differential theory is integrated into the subspace identification algorithm,and a Poisson filter is utilized to address the multi-order derivability of the experimental data.Additionally,the variable-step memory method is introduced to manage the weight coefficients in fractional-order differentiation,thus improving the identification accuracy and reducing the computational complexity.The parameters used in the identification significantly affect the modeling.Therefore,the IECSDE algorithm is proposed to optimize the parameters.The control parameters in the Cuckoo Search(CS)algorithm are adaptively adjusted to optimize the convergence precision and speed.Inspired by the updating method in the particle swarm optimization algorithm,the random-walk method in the conventional CS algorithm is improved,thereby reducing its randomness.The population is optimized separately using the Differential Evolution(DE)algorithm and the improved CS algorithm,with information exchange performed during the optimization to enhance the population diversity and algorithm robustness.The simulation results indicate that the optimization capability of the IECSDE algorithm is at least 10 times higher than that of the other five optimization algorithms for eight test functions.The model identification uses experimental data obtained from the PEMFC measurement and control platform.The model established in this study affords an error reduction of 20％by employing the short memory-based fractional subspace identification error.Its output-power error is controlled between 0 and 0.1,and its output-voltage error is controlled between 0 and 0.2.Additionally,it accurately simulates the power generation of a PEMFC.
万方数据
维普
