The Improvement of Information Neural Network and Its Applications
Fu Liu · Keisou gijutsu · 2010
The fault information importance of feature vector isn't taken into account in the Traditional Information Neural Network(INN),so its recognition efficiency decreases sharply.Aiming at above problems,an improved INN is proposed in this paper considering importance weight value,which optimizes network structure and presents optimization arithmetic of structure parameters.Finally,the improved INN is applied to the pattern recognition and fault diagnosis and its availability is shown by results.