A prediction method of hydraulic oil performance degradation based on extreme learning machine (ELM) with particle swarm optimization (PSO) was proposed.Taking L-HM46 anti-wear hydraulic oil as an example,the oil performance degradation experiment was designed to detect the viscosity,opening angle,moisture content and decline degree of the hydraulic oil.Based on the proposed performance degradation prediction method of hydraulic oil,the external and internal parameters of ELM were optimized by ergodic search and PSO algorithm respectively,and the optimal performance degradation prediction model was established.Taking viscosity,opening angle and,moisture content as the input eigenvector and decline degree of hydraulic oil as the output of the model,the performance of hydraulic oil was simulated and analyzed by PSO-ELM performance degradation prediction model.The results show that the calculation results of PSO-ELM algorithm are in good agreement with the experimental data.The prediction accuracy of PSO-ELM algorithm reaches 98.47%,which is higher than that of elm algorithm.It shows that PSO-ELM algorithm can more accurately predict the decline of hydraulic oil,and provide a reference for determining the oil change time.