基于IVYA-GRU的燃料电池汽车的电池寿命预测OA
Battery life prediction for fuel cell vehicles based on IVYA-GRU
为了解决燃料电池汽车的燃料电池在运行过程中剩余使用寿命预测的可视化和精确性问题,提出了一种常春藤优化算法(Ivy algorithm,IVYA)结合门控循环单元(gated recurrent unit,GRU)神经网络的预测模型,对车用燃料电池的剩余使用寿命进行预测.首先,对使用的静态工况下的老化实验数据集进行数据预处理,分别是间隔采样和数据平滑,然后采用IVYA获得GRU的最优超参数组,再利用GRU准确预测燃料电池电压.将提出的方法与长短时记忆网络、GRU和北方苍鹰优化算法优化门控循环单元相比较,在该方法的预测下,均方根误差、平均绝对误差、预测寿命的相对误差和决定系数分别为0.000 535 55V、0.000 420 6 V、0.035 9%和0.998 7,所提方法在相同条件下具有最高的老化预测和RUL估计精度.
To improve the visualization and accuracy of the remaining useful life(RUL)prediction of fuel cell vehicles during operation,this paper proposes an Ivy algorithm(IVYA)combined with gated recurrent unit(GRU).The predictive model of GRU neural network is employed to predict the remaining service life of fuel cells.First,the aging experimental data set under static conditions is preprocessed by interval sampling and data smoothing.Then,IVYA is introduced to obtain the optimal hyperparameter set of GRU.Next,GRU is used to predict the fuel cell voltage.Finally,the proposed method is compared with the short-time memory network,the GRU and the Northern Goshawk optimization algorithm.With the proposed one,the root-mean-square error is 0.000 535 55 V,the average absolute error 0.000 420 6 V,the relative error 0.035 9%and the coefficient of determination of the predicted life 0.998 7.It has the highest aging prediction and RUL estimation accuracy under the same conditions.
朱楠;王靖岳;何宇亭;丁建明
沈阳理工大学 汽车与交通学院,沈阳 110159沈阳理工大学 汽车与交通学院,沈阳 110159沈阳理工大学 汽车与交通学院,沈阳 110159西南交通大学 牵引力国家重点实验室,成都 610031
信息技术与安全科学
燃料汽车氢燃料电池常春藤优化算法门控循环单元数据驱动
fuel cell vehicleproton exchange membrane fuel cellIVYAGRUdata driven
《重庆理工大学学报》 2026 (3)
28-34,7
国家自然科学基金项目(51875096)辽宁省自然科学基金项目(2020-MS-216)
评论