The invention discloses an unmanned bicycle balance control method based on uncertain reasoning learning. According to the method, a lateral inclination angle (inclination angle speed), a handlebar rotation angle (rotation angle speed) and handlebar control torque of an unmanned bicycle body are used as input, increments of moments before and after the lateral inclination angle (inclination angle speed) and the handlebar rotation angle (rotation angle speed) are used as output, and a probability dynamic model (PDM) of a system is constructed by utilizing a GPR (Gaussian process regression) method; performing parameterization processing on the control strategy, and performing subsequent state sequence prediction by using the established PDM; a cost function is constructed through a vehicle body lateral inclination angle (inclination angle speed) and a handlebar rotation angle (rotation angle speed), and finally a control strategy of the system is optimized in combination with a gradient search method. According to the method, a prior dynamic model of the unmanned bicycle does not need to be established through a complicated physical principle, a probability dynamic model is learned from interactive empirical data, the uncertainty of the unmanned bicycle during operation is fully considered, the influence of model deviation on a control strategy is reduced, the robustness and generalization ability of an unmanned bicycle system are improved, and the method is suitable for popularization and application. And the method has very strong engineering significance.
本发明公开一种基于不确定推理学习的无人自行车平衡控制方法。该方法以无人自行车车体侧向倾角(倾角速度)、车把转角(转角速度)以及车把控制力矩为输入,以车体侧向倾角(倾角速度)、车把转角(转角速度)前后时刻的增量为输出,利用高斯过程回归(GPR)方法构建系统的概率动态模型(PDM);将控制策略做参数化处理,利用建立的PDM进行后续的状态序列预测;通过车体侧向倾角(倾角速度)、车把转角(转角速度)构造代价函数,最后结合梯度搜索方法优化系统的控制策略。该方法无需通过繁杂的物理原理建立无人自行车的先验动力学模型,从交互的经验数据中学习概率动态模型,充分考虑了无人自行车运行时的不确定性,降低了模型偏差对控制策略的影响,提高了无人自行车系统的鲁棒性及泛化能力,具有很强的工程意义。
Uncertain reasoning learning control method for self-balancing of unmanned bicycle
一种无人自行车自平衡的不确定性推理学习控制方法
2025-04-22
Patent
Electronic Resource
Chinese
IPC: | B62M RIDER PROPULSION OF WHEELED VEHICLES OR SLEDGES , Antrieb von Radfahrzeugen oder Schlitten durch den Fahrer / B62J Fahrrad- oder Motorradsättel oder -sitze , CYCLE SADDLES OR SEATS / G05B Steuer- oder Regelsysteme allgemein , CONTROL OR REGULATING SYSTEMS IN GENERAL / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung |
Vector torque balancing device of high-speed unmanned bicycle and control method
European Patent Office | 2023
|