Research About Algorithm and Application of Tunable Activation Function Neural Network

摘要：

　　The paper improved the traditional transferring pattern and function about the BP neural network,it replaced the fixed transfer function with the mutative activation function to construct the tunable activation function neural network; it also gave the improved network algorithm and the conclusion.The actual diagnosis was imitated on the engine lubricating oil fault; the simulation indicated the method had the characters that the convergence speed was quick; the precision was high,and the generalization capability was best.

关键词： Tunable activation function Neural network Algorithm Diagnosis

作者: Zhengwu Wang Hang Cui

作者单位: The First Aeronautic Institute of the Air Force, Xinyang, Henan, China Zhengzhou University, Zhengzhou, Henan, China

会议类型: 国际会议

会议名称: 2013西安国际航空维修与管理学术会议

会议地点: 西安

会议语种:英文

页码: 55-61

在线出版日期: 2013-11-25（万方平台首次上网日期，不代表论文的发表时间）