Abstract: In order to ensure the safety production of the fully-mechanized coal mining,an evaluation and prediction were conducted o the reliability of the hydraulic powered suppor.Based on the reliablity engineering of the cal machinery industry late started and a lot of falure-fre work time data was lost,the paper provided that the fault data from the timing censored test were under the condition of the smal sample.The BP neural network model was aplied to the study and simulation on the failure samples and experience reliability to enlarge the sample volume.Acording to the reliability of the mechanical products generally would obey Weibuldistribution features,the paper had a study of the volume enlargedsample applied to the estimation method of the Wleibul distribution parameters. The study showed that the Weibul distribution of the two parameters would be suitable to deseribe the reliablity of the hydraulic powered support and according to thereliability function and the failure rate functon,the reliability and failure rate of the hydraulic powered support could be obtained at any time.When the sample data were small.the BP neuralnetwork could be applied to the enlargement of the sample volume and would have the high accurate estimated parameters.