燃料电池电动汽车动力系统的多模型切换控制研究

Fuel Cell Electric Vehicle has been highlighted by the experts in the world due to the merits such as no pollution, high efficiency, long power supply time, low noise and etc. But, the time-dependent real operating requirements of the electric vehicle can not be satisfied because the dynamic response of Fuel Cell has a certain time delay. So, a small additional power supply device should be included for constructing a multi-energy power system for Fuel Cell Electric Vehicle. The power system is a highly interacted non-linear system having multiple inputs and outputs. Improving the dynamic response ability of Fuel Cell, controlling energy flow reasonably, and stably operating the power system with high efficiency are the key technologies. For solving the problems, this project will recommend an adaptive prediction, separate layer control structure based on multi-model. At first, the multi neural network models of sub-system such as Fuel Cell, the additional power supply device and etc will be constructed based on fuzzy cluster method, and the operating status of the system such as starting, accelerating, cruising, decelerating, stopping and etc will be analyzed. For each typical operating region in each status, the energy flow model of multi-energy power system will be constructed; In the following step, a prediction control strategy of the bottom layer model will be designed for each energy flow model. Finally, supervising adaptive controller for the top layer will be designed. The objective of stably operating the system with high efficiency can be obtained through adaptively switched to optimal energy flow for the control strategy of the bottom layer.

燃料电池电动汽车具有清洁无污染、效率高、供电时间长、噪音低等优点，已引起国内外专家的广泛关注。但燃料电池的动态响应具有一定的时滞，难以满足电动汽车实时变化的工况运行要求，需要配备小容量的辅助供电装置共同构成燃料电池电动汽车多能源动力系统。其动力系统是一个多输入多输出、强耦合的非线性系统，如何提高燃料电池的动态响应能力，并合理控制能量流，保证动力系统高效平稳运行是其中的难点与关键。为解决这个问题，本课题提出基于多模型的自适应预测分层控制结构。首先基于模糊聚类方法建立燃料电池、辅助供电装置等子系统的多神经网络模型，在此基础上分析系统的启动、加速、巡航行驶、减速制动等多个运行状态，在每个状态下针对各个典型的工作区分别建立多能源动力系统的能量流模型；然后为各个能量流模型设计底层模型预测控制策略，最后设计上层监督自适应控制器，通过对底层控制策略进行自适应切换优化能量流，以达到系统高效稳定运行的目的。

燃料电池电动汽车通常有启动、加速、巡航行驶、减速制动等多个运行状态，导致动力系统功率的波动比较频繁。燃料电池是一个典型的多输入/多输出、强耦合非线性系统，其输出电压与负载、氢气流量、空气流量、温度、湿度等参数呈严重的非线性关系。尤其是其动态特性随负载变化幅度较大，当负载频繁波动时，燃料电池的效率会大大下降。辅助供电装置蓄电池的输入输出特性也比较复杂，其荷电状态与电压、电流、温度等参数也呈复杂非线性关系，导致荷电状态难以测量与控制。另外，为减小体积和重量，蓄电池应尽可能小，这又增加了系统控制的困难。因此，设计控制器提高燃料电池的动态响应能力，并合理控制燃料电池、辅助供电装置及负载间的能量流，保证动力系统连续、稳定运行是其中的难点与关键.. 本课题在前期研究基础上，建立燃料电池、辅助供电装置等子系统的控制模型；分析燃料电池电动汽车动力系统的工作状态，然后分别设计燃料电池控制系统、DC-DC转换器的控制系统、包含散热模块的蓄电池管理系统以及燃料电池内部系统的多物理场耦合仿真分析。最后设计基于自适应切换控制策略的整车控制器，以达到系统平稳运行并节能的目的。