【作者】 王瑞敏；

【导师】 曹广益；

【作者基本信息】 上海交通大学 ， 控制理论与控制工程， 2008， 博士

【摘要】 燃料电池被认为是发展最好的能源和未来最理想的氢能利用方式之一,因此,燃料电池发电技术的研究和开发成为目前全世界的研究热点。质子交换膜燃料电池(PEMFC)是继碱性燃料电池、磷酸燃料电池、熔融碳酸盐燃料电池和固体氧化物燃料电池之后发展起来的一种新型燃料电池。随着PEMFC技术的逐渐成熟和接近商业化,PEMFC系统的发电性能必须得到可靠有力的保证。输出电压是表征PEMFC性能最重要的参数。PEMFC工作温度对其性能有很大的影响。本文参考了国内外研究质子交换膜燃料电池的经验,利用公开文献和本所的实验数据,建立了PEMFC电堆的动态参数模型。在此参数模型的基础上,利用智能方法进行系统辨识,建立起电压和温度的辨识模型。最后基于该辨识模型,设计了智能控制器对PEMFC的输出电压和温度进行控制研究。本文的主要工作内容和成果包括:(1)建立了比较全面的PEMFC仿真模型。本文结合机理和经验的建模方法,吸收了前人对PEMFC输出特性模型的研究成果,将电堆的主要物理结构参数作为变量,建立起PEMFC仿真模型。所建立的动态模型包括双电层动态、流道动态和温度动态,是较以往的模型更全面更系统的PEMFC动态模型。然后利用Matlab/Simulink对该模型进行了仿真分析。仿真结果表明该仿真模型能够正确反映各种操作参数的动态变化,可以作为系统设计、分析和控制的有效工具。(2)应用径向基神经网络建立了PEMFC电压辨识模型。本文采用了一种改进的自适应最简结构算法对神经网络进行训练,使径向基神经网络在较短的时间内达到较高的精度。基于上述电压辨识模型,分别设计了基于正交最小二乘算法的模糊神经网络内模控制器和模糊PID控制器对PEMFC的输出电压进行控制,并对两种控制器进行了比较分析。PEMFC电压控制方案是将电流密度作为干扰量,采用调节阳极气体流量和阴极气体流量将输出电压控制在期望工作点。仿真结果表明,第一种控制器的控制效果更好一些。(3)应用自适应模糊神经网络建立了PEMFC温度辨识模型。基于上述温度辨识模型,设计了基于Rough集理论的自适应模糊神经网络控制器和基于BP神经网络的PID控制器对PEMFC的温度进行控制研究,并对两种控制器进行了比较分析。控制器的目的是保证质子交换膜燃料电池系统工作在一个合适的温度范围内,并尽可能的减少波动范围。仿真结果表明两种控制器都可以达到预期目的,但是在进入稳态时间、超调量等控制性能指标方面,基于Rough集理论的自适应模糊神经网络控制器都明显的比基于BP神经网络的PID控制器要好。更多还原

【Abstract】 Proton exchange membrane fuel cell is considered as one of the best energy and a method of hydrogen use. So, research and development of fuel cell power technology has become a study hot.Proton exchange membrane fuel cell (PEMFC) is a new fuel cell that develops after the alkaline fuel cell, the phosphoric acid fuel cell, the molten carbonate fuel cell and the solid oxide fuel cell.As the PEMFC technology become more and more mature and commercialize, PEMFC system generate electricity performance must be strongly assurance. Output voltage is the most important parameter of PEMFC and operation temperature has vital influence on PEMFC performance. Based on PEMFC experimental and published data, we developed a PEMFC stack dynamic parameter model. Based on said parameter model, we obtain voltage identification model and temperature identification model. Based on identification models, PEMFC output voltage and temperature control problems are studied. The main achievements and contributions are summarized as follows:(1) Based on opened literatures and research results, using PEMFC stack major parameters as variables, combining mechanism model and experimental model, we develop a blocking parameter model of PEMFC which includes charge double layer capacitance dynamics submodel, cathode channel dynamics submodel, and stack temperature dynamics submodel. A great deal simulation results show that the model is enough to reflect the PEMFC stack performance.(2) We develop a PEMFC output voltage identification model based on RBF neural-network. Using an improved self-adaptive simplest structure algorithm train RBFNN, we get a high accuracy network with rather short time. With current density as disturber, the control systems control the output voltage at an expected operating point by adjust anode gas value and cathode gas value. We design an adaptive fuzzy neural-network controller and a fuzzy PID controller to have a control study on the PEMFC output voltage. Simulation results show that the two controllers both can stabilize the PEMFC’s output voltage in an expectant value. Compare the two controllers, the first one has better performance.(3) PEMFC temperature control problem is studied. We develop a PEMFC temperature identification model based on fuzzy-neural network. Based on said temperature identification model, we design an adaptive fuzzy neural-network controller and a neural-network PID controller to have a control study on PEMFC temperature. The objective of the controller is to ensure the PEMFC system work in a proper temperature extent, and reduce fluctuate range as large as possible. The simulation results of the identification models show that the models can reflect PEMFC’s characteristics correctly and the first controller has higher performance.更多还原