A Caputo-α fractional order Takagi-Sugeno neural network based attitude adaptive control method for ship air spring floating raft vibration isolation device

• Established a CAE model for the rigid-flexibility and fluid-solid coupling characteristics of ASFR. • Constructed TSNN controller for adaptive posture control to enhance attitude control accuracy. • Utilized the fractional order optimization TSNN to accelerate its convergence speed. Air spring floating raft vibration isolation devices play a crucial role in reducing ship vibration, but issues such as insufficient air pressure or air leakage in the air spring can reduce their vibration damping effectiveness. To address the challenges of difficult multi-air-spring synchronization control, insufficient precision, and long control time, this paper proposes a Caputo-α fractional order Takagi-Sugeno neural network based attitude adaptive control method (FO-TSNN-AAC) for ship air spring floating raft vibration isolation device. Firstly, a comprehensive computer-aided engineering (CAE) model of the air spring floating raft vibration isolation device, incorporating rigid-flexibility and fluid-solid coupling, is established, with an elastic-plastic material model used to simulate the air spring’s deformation and provide a foundation for subsequent adaptive attitude control. Secondly, based on the offset of multiple air springs as the input and the control signals of solenoid valves as the output, a Takagi-Sugeno neural network (TSNN) controller is constructed. The adaptive adjustment property of the TSNN is utilized to dynamically adjust the parameters to face complex posture changes, thereby achieving synchronous control of multi-air-spring. Finally, by leveraging the non-local and memory properties of Caputo fractional order gradient descent to capture global gradient information, it achieves optimized gradient computation for the TSNN while significantly enhancing its convergence rate. The research results show that this method has improved control accuracy by 20%, while the required control time is only 2/3 of the original. The proposed method has significant engineering application value for reducing vibration and noise while enhancing comfort in ships.