Abstract: Aiming at the application of hybrid electric propulsion system of an UAV based on polymer exchange membrane fuel cell and lithium-ion battery， a power management system control technology based on adaptive neuro-fuzzy inference system was researched and developed to control the hybrid electric propulsion system and optimize the performance of the fuel cell gas supply system. Taking the mathematical model of the hybrid electric propulsion system of an UAV as the research object， the relationship between the fuel cell current and the compressor power of the fuel cell gas supply system was studied， and the reference model of the relationship between the fuel cell current and the optimal compressor power was established. On the basis of the reference model， an adaptive controller was introduced to optimize the performance of the fuel cell gas supply system. The controller based on adaptive neuro-fuzzy inference system dynamically adjusts the actual operating power of the compressor to the optimal value defined in the reference model. The on-line learning and training ability of the adaptive controller was used to identify the nonlinear variation of the fuel cell current and generate the control signal of the compressor motor voltage to optimize the performance of the fuel cell gas supply system. The proton exchange membrane fuel cell （PEMFC） and lithium-ion hybrid electric propulsion system model was developed in Matlab simulation environment， and the designed controller was simulated and analyzed. The results show that the controller based on adaptive neuro-fuzzy inference system provides a novel and comprehensive way to optimize the performance of the compressor in the fuel cell gas supply system， and enables the fuel cell gas supply system to obtain the maximum net power output. The net power output of the fuel cell system was compared with the optimal compressor power and the constant compressor power. The results show that the optimized compressor power configuration saves 2.62% more energy than the constant compressor power configuration. At the same time， the fuel cell adaptive neuro-fuzzy inference system controller optimizes the energy utilization of the fuel cell gas supply system.

Key words: Unmanned aerial vehicles;Adaptive controller;Neuro-fuzzy inference algorithm;Electric propulsion;Fuel cell;Gas supply system;Performance optimization

摘要： 针对无人机采用的聚合物交换膜燃料电池和锂离子电池的混合动力电推进系统，研究开发了一种基于自适应神经模糊推理系统的电源管理控制技术，以控制混合动力电推进系统，同时优化燃料电池供气系统的性能。用无人机混合电推进系统数学模型，研究了燃料电池电流与燃料电池供气系统压缩机功率之间的关系，建立了燃料电池电流与最佳压缩机功率关系的参考模型。在参考模型的基础上，引入自适应控制器来优化燃料电池供气系统的性能。基于自适应神经模糊推理系统的控制器将压缩机的实际运行功率动态调整到参考模型中定义的最佳值。自适应控制器的在线学习和训练能力用来辨识燃料电池电流的非线性变化，并产生压缩机电机电压的控制信号，以优化燃料电池供气系统的性能。在Matlab仿真环境中，开发了质子交换膜燃料电池和锂离子混合动力电推进系统模型，并对所设计的控制器进行了仿真分析。结果表明，基于自适应神经模糊推理系统的控制器为燃料电池供气系统压缩机性能优化提供了一种新颖而全面的途径，使燃料电池供气系统获得最大净功率输出。将燃料电池系统的净功率输出与最佳压缩机功率和恒定压缩机功率进行比较，发现优化的压缩机功率配置比恒定的压缩机功率配置节能2.62%。同时，燃料电池自适应神经模糊推理系统控制器优化了燃料电池供气系统的能量利用。

关键词: 无人机;自适应控制器;神经模糊推理算法;电推进;燃料电池;供气系统;性能优化