Abstract: Zhenjing block Chang 8 oil layer is typical low porous,low permeable reservoir,fracture is main migration path,therefore the oil gas accumulation and reservoir development effect has important relations with fracture,the identification of fracture has very significant meaning to the oil and gas production.In this article,based on field and core fracture observation analysis,combined conventional log data and coring data,65 sand and mud rock sample have been built up.Using conventional log fracture identification mode to identify the reservoir fracture,which can identify effective fracture.Introducing probabilistic neutral network to optimal 35 samples,build up fracture identification fracture.Using this model to identify 19 samples,9 can be identified correctly,the forecast precision reach to 90%.The returning discriminant ratio is 100% on 27 model samples.According to coring data comparison,when the goal is under the multi-factor influence,and the nonlinear relation is rather strong,applying PNN to identify has advantage of high precision and efficiency.