Autonomous underwater vehicles (AUVs) are an important equipment for ocean investigation. Actuator fault diagnosis is essential to ensure the sailing safety of AUVs. However, the lack of failure data for training due to unknown ocean environments and unpredictable failure occurrences is challenging for fault diagnosis. In this paper, a meta-self-attention multi-scale convolution neural network (MSAMS–CNN) is proposed for the actuator fault diagnosis of AUVs. Specifically, a two-dimensional spectrogram of the vibration signals obtained by a vibration sensor is used as the neural network’s inputs. The diagnostic model is fitted by executing a subtask-based gradient optimization procedure to generate more general degradation knowledge. A self-attentive multi-scale feature extraction approach is used to utilize both global and local features for learning important parameters autonomously. In addition, a meta-learning method is utilized to train the diagnostic model without a large amount of labeled data, which enhances the generalization ability and allows for cross-task training. Experimental studies with real AUV data collected by vibration sensors are conducted to validate the effectiveness of the MSAMS–CNN. The results show that the proposed method can diagnose the rudder and thruster faults of AUVs in the cases of few-shot diagnosis.

This paper presents a fault-tolerant model predictive control approach for cross-rudder autonomous underwater vehicles to achieve heading control, considering rudder stuck faults and unknown disturbances. Specifically, additive faults in the rudders are addressed, and an active fault-tolerant control strategy is employed. Fault models of autonomous underwater vehicles have been established to develop the fault-tolerant control method. In the controller design, the stuck faults of complete rudder failure are incorporated to ensure the heading angle control of the autonomous underwater vehicle in faulty conditions. Furthermore, the fault term is decoupled from the control input, and the decoupled control input, along with corresponding constraints, is incorporated into the model’s predictive controller design. This approach facilitates controller reconfiguration, thereby enhancing and optimizing control performance. Simulation results demonstrate that the proposed fault-tolerant model predictive control method can effectively achieve stable navigation and heading adjustment under rudder fault conditions in autonomous underwater vehicles.

An energy-efficient event-triggered heading control system (ETHCS) is proposed for an underwater gliding robot. The robot steers by adjusting the position of an internal axially moving mass, rather than with the actuators commonly used in traditional underwater vehicles. Based on the dynamic model, an energy-efficient ETHCS framework, which includes three linear velocity observers and an event-triggered high-order backstepping heading controller, is established. The three linear state observers estimate the horizontal and angular velocities of the robot using depth and attitude angles measured by a depth sensor and an attitude and heading reference system (AHRS). The controller is developed by using the dynamic surface control technique to avoid differentiating virtual control laws. A neural network is used to approximate recursive model uncertainties and disturbances. Relative threshold event-triggering conditions are designed to reduce the frequency of actuator action and the controller working time and to improve the energy-saving efficiency. The simulation and sea trial results show the effectiveness of the proposed control system.