Abstract: To enhance the ride comfort of semi-active suspension vehicles in diverse driving scenarios, a clustering-based operating condition identification method was proposed. Combined with a switchable skyhook strategy, a novel semi-active suspension control method was developed. The proposed method constructed a neural network between test metrics and model parameters based on the full-vehicle seven-degree-of-freedom (7-DOF) dynamic equations, enabling estimation of unknown vehicle parameters and dynamic model modification. Using weighted sprung and unsprung mass root-mean-square acceleration as features, K-means++ clustering analysis was performed on vehicle driving conditions based on measured data. The optimized current control strategies for each cluster were determined through simulations using the modified model, generating a current strategy map for comprehensive driving conditions. Real-vehicle tests demonstrate that the clustered control strategy significantly enhanced ride comfort over on-off skyhook algorithms, and with rapid cluster transitions, pitch suppression was effectively enhanced.