Scholarlink

<p>This article proposes an active fault-tolerant scheme for advanced layout carrier-based UAVs, taking into account dynamic environments, system nonlinearity, and actuator faults. Unlike traditional nonlinear dynamic models, an actuator fault model is introduced that incorporates both rudder-effect cross-coupling and deflection nonlinearity. Subsequently, a hybrid multi-objective optimization&#xa0;(MOP) online reconstruction strategy is developed, to achieve rapid fault tolerance and accurate attitude response. This strategy integrates a multi-directional time-series model with a control-efficiency inverse model to form a Multi-Model Composite Trajectory Prediction&#xa0;(MMCTP) framework. Specifically, the multi-directional time-series model divides the control surfaces into time intervals based on fault characteristics, enabling the prediction of nonlinear variation trends across multiple evolutionary directions. Meanwhile, the control-efficiency inverse model, driven by control commands, collaboratively estimates the system’s response after faults occur. Furthermore, the Q-learning algorithm is embedded to adaptively update the optimization parameters in response to fault variations, thereby providing more accurate predictions. The closed-loop reconstruction of the rudder surfaces is proven to converge within a finite time. Finally, comparative simulations verify the scheme’s capability to precisely adjust control surface positions, suppress fault-induced oscillations, and enhance system stability and reliability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fault tolerance of advanced layout carrier-based UAV based on multi-model compound trajectory prediction strategy

  • 期刊论文

  • 发表日期：

    2026年2月6日

查看全文
  • Fengying Zheng,
  • Jianan Chen,
  • Mingxuan Xu,
  • Kaizhao Xu,
  • FanLiang Meng

摘要

This article proposes an active fault-tolerant scheme for advanced layout carrier-based UAVs, taking into account dynamic environments, system nonlinearity, and actuator faults. Unlike traditional nonlinear dynamic models, an actuator fault model is introduced that incorporates both rudder-effect cross-coupling and deflection nonlinearity. Subsequently, a hybrid multi-objective optimization (MOP) online reconstruction strategy is developed, to achieve rapid fault tolerance and accurate attitude response. This strategy integrates a multi-directional time-series model with a control-efficiency inverse model to form a Multi-Model Composite Trajectory Prediction (MMCTP) framework. Specifically, the multi-directional time-series model divides the control surfaces into time intervals based on fault characteristics, enabling the prediction of nonlinear variation trends across multiple evolutionary directions. Meanwhile, the control-efficiency inverse model, driven by control commands, collaboratively estimates the system’s response after faults occur. Furthermore, the Q-learning algorithm is embedded to adaptively update the optimization parameters in response to fault variations, thereby providing more accurate predictions. The closed-loop reconstruction of the rudder surfaces is proven to converge within a finite time. Finally, comparative simulations verify the scheme’s capability to precisely adjust control surface positions, suppress fault-induced oscillations, and enhance system stability and reliability.