Abstract: For the parallel distributed processing,associative memory,self organization,self learning and strong mapping abilities,artificial neural networks have shown broad application prospects in many fields. The method and procedure of single variable time series forecasting based on neural networks are described.The forecasting model for the wear trend of tribosystems is proposed,in which the quantitative ferrographic parameter Q is used as the characteristic parameter for forecasting wear trend.The single step and multi step forecasting method for the wear trend forecasting based on the neural network are discussed.The model is applied to the wear trend of the gear pump.The results are very satisfactory.