Improvement of vibration isolation performance of multi-mode control seat suspension system through road recognition using wavelet-LSTM approach

Extensive research has significantly improved the vibration isolation performance of off-road vehicle seat suspensions, effectively addressing issues such as driver fatigue, and low-back pain. However, variations caused by road roughness, vehicle speed, and load can lead to stability switches and sudden events in seat vibration excitation. Ignoring these factors in semi-active and active seat suspensions can cause insufficient robustness and excessive costs. To address this, we propose an intelligent damping switching method that optimizes seat suspension damping mode by considering the stability and suddenness of excitation. Utilizing a long short-term memory (LSTM) network, we accurately identify stability, while the multi-input and multi-output optimization (MIMO) method labels the network input through system identification of the seat dynamics model. Event trigger (ET) handles suddenness effectively. By combining these techniques, our approach achieves effective vibration isolation while maintaining the desired suspension deflection. Comparative analysis validates our novel seat suspension control system design approach.