TL;DR
This paper introduces a novel fault-tolerant control strategy for unmanned marine vehicles (UMVs) that can handle unknown system parameters and actuator failures. By combining integral sliding mode control with a Takagi-Sugeno fuzzy model and a radial basis function neural network, the system can maintain stable operation even when critical components malfunction.
What Problem Does This Paper Address?
Unmanned marine vehicles (UMVs) are increasingly used for various tasks, from research to logistics. However, their operation in complex marine environments is challenged by unpredictable disturbances and the potential for actuator failures. Traditional control methods struggle with these issues, especially when the internal state variables (premise variables) of the UMV's fuzzy model are unknown, which is common due to factors like yaw angle uncertainty. This paper tackles the problem of designing a robust control system that can maintain stability and performance for UMVs despite these uncertainties and faults, using only measurable outputs.
Methods
The core of this research lies in a sophisticated control strategy that integrates several advanced techniques:
- Takagi-Sugeno (T-S) Fuzzy Model: UMVs are modeled using a T-S fuzzy model, which is adept at approximating complex nonlinear systems. A key challenge addressed is that the "premise variables" (inputs to the fuzzy rules) are unknown.
- Radial Basis Function Neural Network (RBFNN): To handle the unknown nonlinear dynamics within the T-S fuzzy model, an adaptive RBFNN is employed. This allows the system to learn and approximate these nonlinearities online.
- Integral Sliding Mode (ISM) Control: This is the primary control paradigm. ISM is known for its robustness to uncertainties and disturbances. The paper proposes an output feedback ISM approach, meaning it relies only on measured outputs, not full state information.
- Compensator with Switching Mechanism: To deal with the unknown premise variables and their impact on the fuzzy model's membership functions, a compensator is designed. This compensator utilizes a switching mechanism based on the upper and lower bounds of these membership functions, which helps to reduce conservatism in the control design.
- Fault Estimation and Fault-Tolerant Control (FTC): The system incorporates mechanisms to estimate actuator faults and adapt the control strategy accordingly. This fault-tolerant aspect ensures that the UMV can continue operating safely even if thrusters fail or lose effectiveness.
- H∞ Performance: The control design aims to satisfy a prescribed H∞ performance index, which guarantees a certain level of disturbance attenuation.
Datasets
Not discussed in the paper. The paper relies on a simulated UMV model and parameters derived from existing literature.
Key Insights and Ideas
- Output Feedback for Unknown Premise Variables: A significant contribution is the development of an output feedback ISM control strategy that can handle unknown premise variables in T-S fuzzy models. This is crucial because obtaining full state information in real-world marine environments is often difficult.
- Bridging Fuzzy Logic and Sliding Mode: The paper effectively bridges the gap between fuzzy logic control (for modeling unknown nonlinearities) and sliding mode control (for robustness and fault tolerance), particularly in an output feedback setting.
- Reduced Conservatism: The proposed switching mechanism, which leverages the bounds of unknown membership functions, is highlighted as a way to reduce conservatism compared to previous methods.
- Adaptive RBFNN for Nonlinearities: Using an adaptive RBFNN to approximate nonlinear terms simplifies stability analysis and enhances the system's adaptability to unmodeled dynamics.
Key Results
The paper demonstrates the effectiveness of its proposed control strategy through simulations. Key findings include:
- Stable Operation Despite Faults: The controller successfully maintains asymptotic stability and achieves the desired H∞ performance even when actuator faults (e.g., 40% loss of effectiveness, stuck thruster) are introduced.
- Robustness to Disturbances: The system shows good performance in rejecting both random noise and sine wave ocean disturbances.
- Comparison with Existing Methods: Simulations show that the proposed method outperforms a previously published scheme [22] that does not account for faults, where system states and control signals diverge after fault occurrence.
- Quantitative Performance: The paper presents steady-state error values for different disturbance types, indicating good accuracy. For instance, steady-state errors for y1, y2, and y3 are in the order of 10⁻⁵. The H∞ performance index achieved is 0.1, which is better than a reference value of 0.2.
Limitations and Future Work
- Limitations:
- The study is based on simulation results, and real-world experimental validation is not presented.
- The paper focuses on specific types of actuator faults (loss of effectiveness, stuck faults). More complex fault scenarios might require further investigation.
- Future Work: The authors suggest that future research will focus on investigating delay and event-triggering issues in T-S fuzzy UMV dynamic positioning (DP) systems. This implies extending the current framework to handle time delays and more efficient communication protocols.
Summary updated