微型直接甲醇燃料电池(μDMFC)具有能量密度高、可便携使用、快速补能以及环境友好等优点.然而,由于膜电极在电化学反应中会退化,μDMFC的有效使用寿命有限,所以需要对其健康状态与剩余使用寿命进行评估,为燃料电池改性和控制策略提供决策支持.在结合数据驱动和机理模型预测方法的基础上,针对动态运行工况,提出一种基于等效电路模型(ECM)的μDMFC剩余使用寿命预测方法.在燃料电池的性能退化指标中,电池输出电压可以被实时监测从而获得电池的退化趋势,但这一指标无法单独提供精确的预测结果.通过测量电化学阻抗谱并结合ECM可以得到电池内部阻抗等深层信息,但这些深层信息不易被实时监测,通常只能低频离线测量.此外,燃料电池在实际应用中多处于变工况状态,其退化趋势和使用寿命受工作环境影响,传统基于电压退化趋势回归的预测方法无法应对工况的动态变化.因此,可通过定期离线获取内部退化参量建立预测模型.实验结果表明:与传统数据驱动的方法相比,基于内部退化参量的预测方法能更好地适应变工况环境,在燃料电池剩余使用寿命预测中具有更好的性能.

Micro direct methanol fuel cell (μDMFC) has the advantages of high energy density, portable use, fast replenishment, and eco-friendliness. However, the practical service life of μDMFC is often limited due to the deterioration of membrane electrode assembly in electrochemical reaction. Therefore, it is necessary to evaluate the health status and remaining useful life (RUL) of the cell to provide decision-making support for fuel cell characteristic modification and control strategy. Considering the pros and cons of data-driven and model-based methods, an RUL prediction method for μDMFC based on equivalent circuit model (ECM) is proposed. Among the degradation indicators of μDMFC, the cell output voltage can be monitored in real time to obtain the degradation trend. However, this indicator cannot provide accurate prediction results alone under dynamic operating conditions. Deeper-level information, such as the internal impedance, can be obtained by investigating the electrochemical impedance spectroscopy (EIS), but such in-depth information is difficult to be monitored in real time and can only be measured offline at low frequencies. Moreover, fuel cells are usually under dynamic operating conditions in practical applications, so their degradation and service life are affected by the operating conditions. Traditional output voltage regression-based prediction methods cannot cope with dynamic changes in operation. Therefore, the prediction model can be built through scheduled offline measurement of internal degradation indicators. The experimental results show that, compared with the traditional data-driven method, the prediction method based on the internal EIS characterization can better adapt to the variable operating conditions and has a superior performance in RUL predictions.