The time-series prediction Group Method of Data Handling (GMDH) method is proposed to predict the friction coefficient of YN9X/Si3N4 tribopair of seawater hydraulic axial piston pump based on wear experiment. The effects of surface roughness, YN nickel content, load, sliding speed and temperature on the friction and wear interaction and sensitivity of YN/Si3N4 tribopair in seawater were studied by orthogonal test. The influence of corrosion wear on the tribological properties of YN material was quantitatively analyzed. The experimental results show that YN9X has better tribological properties than other materials in artificial seawater. Raman spectrum analysis proves that there is tribological reaction between the friction pairs, which can effectively improve the tribological properties of the materials.