Scholarlink

<p>Quadrotor unmanned aerial vehicles (Q-UAVs) face potential challenges especially when motors fail during flight. This paper presents an adaptive fault tolerant control system strategy that is designed to maintain stable operation of quadrotors under such failures. The proposed approach begins with a fault detection scheme that combines two complementary methods: a physics-based residual test which compares the predicted thrust to measured behavior for flagging if any anomaly exists, and a wavelet neural network trained to differentiate failures from disturbances like wind. To ensure stable operation after detection of fault, a reconfigured model predictive control scheme (MPC) augmented with an adaptive thrust correction is employed. In this framework, two different update strategies were considered and tested: a <i>dense</i> update strategy, which updates thrust commands at a high frequency (e.g., 50 Hz) and a <i>sparse</i> update strategy that commands only once per second (1 Hz), allowing more computational time but introduces latency. High-fidelity simulation results demonstrate that the dense strategy is clearly preferable as it achieves approximately 76% reduction in altitude root-mean-square error (RMSE) and a 74% reduction in peak error, and faster recovery than the sparse mode. This outcome reveals that the control-update frequency plays a dominant role in fault tolerant operation. The proposed approach offers a robust and computationally efficient framework suitable for fault-tolerant quadrotor control, emphasizing the need for appropriate update-rate selection in the safe operation of quadrotors.</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

Adaptive fault tolerant control of quadrotors: effects of sparse and dense sampling

  • 期刊论文

  • 发表日期：

    2026年5月11日

查看全文
  • G. Jithendra Sudheer,
  • Santhosh Kumar Varanasi

摘要

Quadrotor unmanned aerial vehicles (Q-UAVs) face potential challenges especially when motors fail during flight. This paper presents an adaptive fault tolerant control system strategy that is designed to maintain stable operation of quadrotors under such failures. The proposed approach begins with a fault detection scheme that combines two complementary methods: a physics-based residual test which compares the predicted thrust to measured behavior for flagging if any anomaly exists, and a wavelet neural network trained to differentiate failures from disturbances like wind. To ensure stable operation after detection of fault, a reconfigured model predictive control scheme (MPC) augmented with an adaptive thrust correction is employed. In this framework, two different update strategies were considered and tested: a dense update strategy, which updates thrust commands at a high frequency (e.g., 50 Hz) and a sparse update strategy that commands only once per second (1 Hz), allowing more computational time but introduces latency. High-fidelity simulation results demonstrate that the dense strategy is clearly preferable as it achieves approximately 76% reduction in altitude root-mean-square error (RMSE) and a 74% reduction in peak error, and faster recovery than the sparse mode. This outcome reveals that the control-update frequency plays a dominant role in fault tolerant operation. The proposed approach offers a robust and computationally efficient framework suitable for fault-tolerant quadrotor control, emphasizing the need for appropriate update-rate selection in the safe operation of quadrotors.