Mechanical System Reliability Study On Neural Network

Along with the increasingly vigorous competition of the global market,the quantity of all kinds of products is becoming increasingly important, andimmediately the reliability of product is required more and more highly. So it isan important task to advance the rationality and applying quality that moreperfecting and improving reliability research theory. In this paper, authorcombined Artificial Intelligence,pattern recognition, reliability theory withstochastic process tightly, then put forward a new,quick and accurate methodto estimate probability model, solved the problems of reliability data analysisand experience accumulating in traditional reliability distribution identification.In the beginning of this paper, a few reliability model distinguish methodsare compared and analyzed, and the merits and the demerits are concluded.After analyzing in-depth the feature of Artificial Neural Networks, a new andrespective method of estimating mechanical reliability distribution based on BPnetwork is proposed.According to the theory of intelligent pattern recognition, statisticalfeatures of random data that are produced through MATLAB are selected. Onthe foundation of these characteristic parameters, the procedure of BP networkthat is used in reliability model identification is set up.Besides, on basis of analyzing a few complex reliability model parametersestimating method, a new parameters estimating method using adaline neuralnetwork is put forward, and it is tested by computer simulation. This methodwas developed by means of mathematical analysis and theory of ArtificialNeural Network. Its character is simple principle, easily calculating andconvenient appliance.At the end of this paper, aiming at Normal distribution, weibulldistribution, Logarithmic normal distribution and Exponential distribution, a lotof random data time series are testing identified in the trained model. Thesimulations proved that this kind of reliability model distinguish method canimprove the reliability analyze efficiency and objective, owe higher theorymeaning and wider engineering practical value.