摘要
对短波信道进行效能评估有助于选择合适的通信链路以提高通信质量。利用云理论处理模糊性和随机性等不确定性问题的优势,并结合人工神经网络可以很好地逼近一个具有复杂非线性输入与输出关系的未知效能模型的特点,我们采用云神经网络进行短波信道通信效能的评估。同时,我们加入了η学习率和α动量因子以避免局部最小值、收敛速度慢以及发生震荡等问题。实验结果表明,采用云神经网络进行短波通信效能评估,可以完全基于短波通信的历史数据勿须添加人为主观因素得到客观的评估结果,并可根据需要灵活地调整云神经网络的输入进行短波通信效能评估,使得评估结果更加准确。
The effectiveness evaluation of HF channel could contribute to selecting the appropriate communi- cation links and improving the communication quality. According to the advantages of uncertainties, such as the randomness and fuzziness of cloud theory, and in combination with the feature that artificial neural net- work could be approximate to a unknown efficiency model of complex nonlinear relationship between input and output, cloud neural network is adopted to evaluate the effectiveness of HF communication channel. Meanwhile the eta and alpha momentum factors are added to avoid local minimum, low-speed convergence and oscillation problems. Experimental result indicates that based on history data, the efficiency evaluation of HF channel with cloud neural network could achieve objective result without adding any artificial subjec- tive factors, and the input of cloud neural network could be flexibly adjusted in accordance with the require- ment of HF communication effectiveness evaluation, thus to achieve more accurate evaluation result.
出处
《通信技术》
2014年第2期195-199,共5页
Communications Technology
关键词
短波通信效能评估云神经网络
effectiveness evaluate, HF communication, cloud neural network
