扑翼飞行器俯仰模糊PID控制研究

Research on Fuzzy PID Pitch Control of a Flapping-Wing Aircraft

摘要: 针对仿鸟扑翼飞行器俯仰单通道控制的离散实现与执行链路验证问题，本文在模型假设明确的条件下构建俯仰标称状态空间模型，设计基于STM32的模糊自整定PID控制器，并引入0.02 s采样保持、±15˚舵角软件限幅、50 Hz PWM映射和舵机一阶环节。在相同基础PID参数、相同采样周期和相同执行机构约束下，开展10˚阶跃响应、采样周期变化、等效输入扰动和关键参数敏感性仿真，并进行串口控制器在环、PWM映射和舵机空载回放验证。结果表明，在当前标称模型下，模糊PID将超调量由6.6347%降至5.6455%，调节时间由1.0760 s缩短至0.8570 s，RMSE与传统PID基本相当；在0.01~0.05 s采样周期及B (3)、A (3, 3) ±20%参数扰动范围内，其超调量和调节时间整体低于传统PID。三阶段STM32验证均获得101/101帧有效应答，PWM映射误差为0 μs，舵机空载回放未观察到异常抖动、卡滞或明显发热。

Abstract: To address the discrete implementation and actuator-chain validation of pitch single-channel control for a bird-like flapping-wing aircraft, a nominal pitch state-space model is established under explicit modeling assumptions. An STM32-based fuzzy self-tuning PID controller is designed with a 0.02 s sample-and-hold period, a ±15˚ software limit for the elevator angle, 50 Hz PWM mapping, and a first-order servo model. Under identical baseline PID gains, sampling conditions, and actuator constraints, 10˚ step response, sampling-period variation, equivalent input disturbance, and key-parameter sensitivity simulations are conducted. Serial controller-in-the-loop, PWM mapping, and no-load servo replay tests are also performed. The results show that, under the current nominal model, the fuzzy PID controller reduces the overshoot from 6.6347% to 5.6455% and the settling time from 1.0760 s to 0.8570 s, while the RMSE remains close to that of the conventional PID controller. Within the 0.01~0.05 s sampling range and ±20% perturbations of B (3) and A (3, 3), the fuzzy PID controller generally yields lower overshoot and shorter settling time. All three STM32 validation stages obtain 101/101 valid acknowledgements, the PWM mapping error is 0 μs, and no abnormal jitter, mechanical sticking, or obvious heating is observed during no-load servo replay.