Bayesian Physics-Informed Neural Networks With MIQPSO-Backstepping Control for Vibration Suppression in Nonuniform...

Area of Science:

• Robotics and Control Systems
• Artificial Intelligence in Engineering
• Mechanical Engineering

Background:

• Quay cranes face challenges with flexible cable vibrations, payload swing, and rotation, impacting tracking accuracy and transport efficiency.
• Modeling these systems involves complex, time-varying, and spatially distributed partial differential equations.
• Underactuated dynamics present significant control design hurdles for achieving precise trajectory tracking.

Purpose of the Study:

• To develop an advanced trajectory tracking strategy for nonuniform quay cranes.
• To suppress flexible cable vibrations and reduce payload motion (swing and rotation).
• To enhance overall tracking accuracy and maritime transport efficiency.

Main Methods:

• A Bayesian physics-informed neural network (BPINN) framework was proposed, integrating tension constraints into the loss function.
• Hamiltonian Monte Carlo (HMC) sampling was used for inferring system states within the Bayesian framework.
• Differential flatness and an adaptive backstepping controller were employed to manage underactuated dynamics, coupled with a multistrategy improved quantum-behaved particle swarm optimization (MIQPSO) for parameter tuning.

Main Results:

• The BPINN effectively suppressed flexible cable vibrations by incorporating tension constraints.
• The adaptive backstepping controller, utilizing differential flatness, ensured global uniform ultimate boundedness of the system states.
• The MIQPSO scheme optimized control parameters, balancing exploration and convergence for robust performance.
• Simulations and experiments validated the strategy's ability to achieve fast, accurate tracking and significant vibration reduction.

Conclusions:

• The proposed BPINN-based trajectory tracking strategy offers a robust solution for controlling quay cranes.
• The integration of physics-informed neural networks and advanced control techniques significantly improves operational efficiency and stability.
• The method demonstrates effectiveness in suppressing vibrations and enhancing tracking accuracy even under external disturbances.