基于ABC-SVM模型的固体氧化物燃料电池预测控制仿真研究  

Predictive control simulation of solid oxide fuel cells based on an artificial bee colony-support vector machine(ABC-SVM)model

作　　者：靳方圆 熊超 周海峰[1,2] 黄元庆 JIN FangYuan;XIONG Chao;ZHOU HaiFeng;HUANG YuanQing(School of Marine Engineering,Jimei University,Xiamen 361021;Fujian Provincial Key Laboratory of Marine and Offshore Engineering,Xiamen 361021;School of Aerospace Engineering,Xiamen University,Xiamen 361102,China)

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

出　　处：《北京化工大学学报（自然科学版）》2021年第4期96-104,共9页Journal of Beijing University of Chemical Technology(Natural Science Edition)

基　　金：国家自然科学基金(51179074);福建省自然科学基金(2018J01495);现代精密测量与激光无损检测福建省高校重点实验室项目(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(JAT190335/JAT180269)。

摘　　要：为了满足直流负载电压的稳定性及固体氧化物燃料电池(SOFC)的耐用性和安全性的要求,设计了两个控制回路分别对SOFC的输出电压和燃料利用率进行控制。通过设计一个简单的控制回路来使燃料利用率保持在恒定值,并在此基础上开发了一个非线性模型预测控制器以控制SOFC的输出电压。该非线性模型预测控制器基于改进的支持向量机(SVM)预测模型,首先利用Lipschitz quotients准则确定SVM预测模型的结构,然后通过人工蜂群算法(ABC)优化SVM参数。仿真结果表明,所提的基于ABC-SVM模型的SOFC预测控制算法可以很好地跟踪电压设定值,证明了ABC-SVM模型在SOFC非线性动态建模中的有效性。In order to meet the requirements of stable DC load voltage,durability and safety of solid oxide fuel cells(SOFCs),we have designed two control loops to control the output voltage and fuel utilization of SOFCs.A simple control loop is designed to maintain the fuel utilization at a constant value,and based on this,a nonlinear model predictive controller is developed in order to control the output voltage of the SOFC.This nonlinear model predictive controller is based on a modified support vector machine(SVM)predictive model.First,the SVM prediction model structure is determined using the Lipschitz quotients criterion,and then the SVM parameters are optimized by the artificial bee colony(ABC)algorithm.Simulation results show that the SOFC predictive control algorithm based on the ABC-SVM model proposed in this paper can effectively track the voltage setpoint,demonstrating the effectiveness of ABC-SVM in modeling the nonlinear dynamics of SOFCs.

关 键 词：固体氧化物燃料电池 支持向量机 人工蜂群算法 模型预测控制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]