Journal of Frontiers of Computer Science and Technology ›› 2016, Vol. 10 ›› Issue (8): 1122-1132.DOI: 10.3778/j.issn.1673-9418.1602015

Previous Articles     Next Articles

Exploration of High-Precision Adaptive Wavelet Neural Network Artificial Intelligence Method

LIU Jingwei1,2+, ZHAO Hui3, ZHOU Rui2, WANG Pu2   

  1. 1. College of Information, Capital University of Economics and Business, Beijing 100070, China
    2. College of Electronic Information and Control Engineering, Beijing University of Technology, Beijing 100124, China
    3. Research Institute of Information Technology, Tsinghua University, Beijing 100084, China
  • Online:2016-08-01 Published:2016-08-09

高精度自适应小波神经网络人工智能方法探索

刘经纬1,2+,赵  辉3,周  瑞2,王  普2   

  1. 1. 首都经济贸易大学 信息学院,北京 100070
    2. 北京工业大学 电子信息与控制工程学院,北京 100124
    3. 清华大学 信息技术研究院,北京 100084

Abstract: In order to solve the accuracy problem and the requirements of artificial intelligence systems, this paper proposes an adaptive wavelet neural network (AWNN) method, which is combined with the advantages of wavelet analysis method, BP (back propagation) and RBF (radial basis function) neural network. Then, this paper uses AWNN into intelligent video analysis and intelligent control system, and verifies better convergence and high accuracy of AWNN. Firstly, AWNN is analyzed theoretically and a group of comparative simulation experiments are implemented to verify that AWNN can improve speed and precision. Secondly, AWNN based intelligent video analysis systems are implemented to verify that AWNN has more accurate content classification ability. Finally, AWNN based intelligent control systems are implemented to verify that AWNN has better control features.

Key words: wavelet neural network, intelligent system, video analysis, intelligent control system, neural network control

摘要: 针对当前人工智能方法存在的训练精度瓶颈问题和智能系统对高精度人工智能方法的迫切需求问题,结合小波分析和BP(back propagation)、RBF(radial basis function)神经网络的优点,提出了自适应小波神经网络(adaptive wavelet neural network,AWNN)方法,将其应用于智能视频分析系统和智能控制系统,并验证了AWNN方法可以取得更好的收敛性、准确性、精度等。通过对AWNN方法与经典的神经网络进行理论分析,并与计算机仿真进行对比分析,验证了该方法可以提升经典神经网络的速度和精度;进而通过将AWNN方法植入真实的视频分析系统进行实验,验证了AWNN方法与现有的视频分析技术相比具有更准确的内容分类能力;最终将AWNN方法与经典控制方法相结合,通过与两种现有的神经网络控制方法进行对比分析,验证了AWNN控制方法具有更好的控制性能。

关键词: 小波神经网络, 智能系统, 视频分析, 智能控制系统, 神经网络控制