Barrel Life Based On Performance Degradation Data Reliability Analysis

The barrel is a key component of artillery and automatic weapons, which length of life often directly determines that of the entire weapon system.The main objective of this thesis is the prediction of barrel's life based on the performance degradation data, including the following main contents:(1) Based on the failure mechanism study of the barrel, this thesis has determined the performance degradation parameters of weapon barrels and failure threshold, and discussed several methods for barrel life prediction by reviewing documents and counting test data.(2) Considering the complexity of barrel's failure mechanism and the uncertainty of the degradation track, this thesis takes the degradation process as a 'black box' and takes advantage of the BP neural network to predict the future degradation data, thus completing the prediction of barrel's life. Moreover, this thesis forms corresponding BP neural network by using Matlab to realize the life and reliability prediction of the barrel.(3) This thesis optimizes the program of barrel's degradation test, which takes the minimum p-order sub-optimal variance as the objective, sample volume, monitoring frequency and time interval (terminal time) as optimization parameters and a fixed test budget as a constraint. This thesis adopts the genetic algorithm to optimize the program, which uses less time for calculation and strong ability for search. It also programs corresponding procedures by Matlab, realizing the optimization of the program.(4) This thesis has finished the real-time prediction on barrel's performance reliability on basis of time series analysis, and finished the prediction of future performance degradation volume, calculated the mean and variance of the predicted value and established the reliability equation by exponential smoothing. The optimization goal of this thesis is the minimum error sum of squares between prediction value and the real value with smoothing constant as the optimization parameter. This thesis obtains the best smoothing constant by optimization of genetic algorithm and programs corresponding procedures on basis of Matlab.(5) This thesis has developed the software of barrel's life prediction in view of the above theoretical research and designed the prediction program of barrel's life by Matlab and the users'interface of software program by using Excel, which has connected communication between Matlab and Excel through Excel Link Interface.