船用固体氧化物燃料电池电堆建模与仿真  

Modeling and Simulation of Marine Solid Oxide Fuel Cell Stack

作　　者：靳方圆 陈鑫 周海峰[1,2] 罗成汉 黄元庆[3] JIN Fangyuan;CHEN Xin;ZHOU Haifeng;LUO Chenghan;HUANG Yuanqing(School of Marine Engineering Institute,Jimei University,Xiamen 361021,Fujian,China;Key Laboratory of Naval Architecture and Ocean Marine Engineering of Fujian Province,Xiamen 361021,Fujian,China;School of Aerospace Engineering,Xiamen University,Xiamen 361102,Fujian,China)

机构地区：[1]集美大学轮机工程学院,福建厦门361021 [2]福建省船舶与海洋工程重点试验室,福建厦门361021 [3]厦门大学航空航天学院,福建厦门361102

出　　处：《船舶工程》2021年第10期104-110,138,共8页Ship Engineering

基　　金：国家自然科学基金项目(51179074);福建省自然科学基金项目(2021J01839,2018J01495);现代精密测量与激光无损检测福建省高校重点试验室项目(B17119);集美大学科研启动金(ZQ2013007);集美大学横向课题项目(S20127);福建省教育厅项目(JAT200242,JAT170318)。

摘　　要：由于船用固体氧化物燃料电池复杂的内部环境不利于对其直接建模,针对船舶设备负载变化导致燃料利用率变化,提出利用径向基函数神经网络的“基”函数。结合宽度学习网络结构和恒燃料利用率控制策略,构造了一种识别恒燃料利用率的固体氧化物燃料电池电堆特性的新方法。根据恒燃料利用率控制策略经研究获得电池的输入输出数据组;在输入输出数据组的基础上,利用Lipschitz商准则确定非线性模型的最佳输入变量阶数;利用粒子群优化算法来估算基于径向基函数宽度学习网络参数,包括增强层参数、增强层和输入层到输出层参数。该方法得到的辨识结果精度高且计算量小。试验结果验证了该方法的准确性和有效性。Because the complex internal environment of solid oxide fuel cells is not conducive to its direct modeling,the radial basis function neural network is proposed to use the‘basis’function of the radial basis function neural network in view of the change of the fuel utilization rate caused by the change of the ship’s equipment load.Combined with the broad learning network structure and constant fuel utilization control strategy,a new method to identify the characteristics of solid oxide fuel cell stack with constant fuel utilization is constructed.According to the constant fuel utilization control strategy,the input and output data sets of the battery are obtained through research.Based on the input and output dates,the Lipschitz quotient criterion is used to determine the optimal input variable order of the nonlinear autoregressive with external input model.The particle swarm optimization algorithm is used to estimate the parameters of the quasi-broad learning system,including the enhancement layer,the enhancement layer and the input layer to the output layer parameters.The identification results obtained by this method are highly accurate and small amount of calculation.The experimental results show the accuracy and effectiveness of the method.

关 键 词：固体氧化物燃料电池 宽度学习网络 粒子群算法 电堆特性 辨识模型 径向基函数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]