Based On Artificial Neural Network Study On Application Reliability For Tractor

Bad reliability and more failures are cciimon problem in current agriculture mechanize. In this dissertation, trace tractor (uxxlel: Lkngfanghong 1002) being research object, theory analyzing and apply studying of tractor reliability were conduct with subjects combine including reliability project way, Artificial Neural Network, probability theory & mathematical statistics, stochastic process and confuse mathematics.(1)It is important to select reasonable reliability model. After analyzed a few reliability model distinguish methods, author first put forward method of selecting reliability model which base on Artificial Neural Network. On basis of theory analysis, he proved BP algorithm and wrote procedure with MKI1AB language. The method was applied to analyze first tractor failure data. Estimating reliability model parameters is the base to analyze reliability. On basis of analyzing a few complex reliability model parameter estimating method, Artificial Neural Network method was put forward. It was tested by computer simulated. Its character is simple principle, easily calculating and convenient appliance.(2)Analyzed tractor failure process, quantity definition of improve and inferior of tractor failure process was established. The relations between failure improve and inferior and failure intensity function and accumulating failure intensity function was discussed. Especially, for Weibull process, the relations between failure inferior and improve and shape function of Weibull process. Tractor failure process was firstly described with nonven poisn processeibull process and simulated data of failure tracking experiment.(3)Large quantity and importance scenic failure data were obtained by tractor failure tracking experiment in one work year. Failure intensity function and accumulate failure intensity function of 1002 tractor were set up with weibull process imitating those data and Artificial Neural Network estimating parameter which all describe tractor failure regulation at initial work stage. That is bath basin curve of initial failure. The estimating value of the amount of *ork of man failure interval and work hour of mean failure interval were obtained.(4)Characteristic observe value of reliability ,repair and efficacious was obtained with data of failure tracking experiment. On the basis of confuse complex assessing and analyzing 1002 model tractor reliability. Its problem and reason were analyzednd put forward improve propose to factory and consumer.