Prediction about Missile Quality Based on Improved Tunable Activation Function Neural Network

摘要：

It was crux to improve the generalization capability with the neural network to predict the missile quality, which depended on two factors, one was structure and algorithm, another was quality, quantity and representation ability of the training samples. The paper constructed the tunable activation function neural network and replaced conventional transfer function with mutative function to extend its capability, used the improved algorithm to increase the precision, the application indicated the prediction effect was perfect.

关键词： tunable neural network missile quality prediction

作者: Wang Zhengwu He Shuaitian Zhang Fuliang

作者单位: The First Aeronautic Institute of the PLA Air Force, Henan, Xinyang, 464000

会议类型: 国际会议

会议名称: 第八届国际测试技术研讨会(8th International Symposium on Test and Measurement)

会议地点: 重庆

会议语种:英文

页码: 1504-1507

在线出版日期: 2009-08-01（万方平台首次上网日期，不代表论文的发表时间）