Prediction of Degradation Process of IC Performance Attributes by the Method of Windows with Using of Neural Networks

Application of neural networks for prediction of IC performance attributes by the example of degradation process of low level output voltage () TTL IC such as 106ЛБ1 by results of life test is examined. Prediction is performed in system Matlab/Simulink by a method of windows. Dependences of predicted values from window width and from prediction depth are investigated. Modelling has shown that it is possible to achieve the good agreement of forecasts with experimental data by selection of neural networks architectures and by selection of entrance window width.