Research On Inverter Reliability Based On Neural Network

Inverter is a DC/AC converter of outputting alternating current voltage and stability frequency. The technology of inverter is the modern power electronics. Therefore, studying the reliability of the inverter is important meaning. In this paper, DND-[48]220/4.5 the inverter as the research object is studied by combination of artificial neural networks, probability theory, mathematical statistics, and reliability engineering.First of all, a BP neural network model with the reliability prediction of the inverter is proposed. The appropriate neurons are chose by analyzing the function and structure of the inverter. It is proved that the predicted results using this method are much more precise and credible than the results getting by the tradition prediction means.Secondly, a few reliability model distinguish methods are compared and analyzed, and used of neural networks in reliability failure on the distribution of selection strengths, and designed the recognizing models of reliability based on neural network. And the model is applied to the inverter first failure time distribution.Finally, basis of getting the failure distribution type of the the inverter, a reliability parameter estimation model is set up based on BP neural network, the curve of reliability and parameter is draw. The reliability of the inverter can easily to be analyzed and evaluated.Through the research of this subject, the data of reliability analysis based on artificial neural networks, probability theory and mathematical statistics improved the reliability of data analysis automation and intelligent level of reliability theory to solve engineering technical problems and deepen the existing methods of reliability engineering.