【摘要】 提出了一种基于压缩传递函数的神经网络盲均衡算法。利用压缩传递函数,可以避免神经元的净输入落入传递函数的饱和区,从而避免权值调整进入代价函数性能曲面的平坦区,有效提高了算法的收敛速度。利用计算机仿真将文中算法、传统前馈神经网络盲均衡与目前海上靶场遥测信号接收所用的判决反馈均衡方法进行了比较,结果证明文中提出的算法具有更好的均衡性能,更具实用性。更多还原

【Abstract】 Blind equalization by neural network based on compressed transfer function is proposed in this paper.Adding steep-coefficient to transfer function can make the input of nerve cell avoid falling into the saturation area,thus plainness area introduced by adjusting of weight of neural network is prevented,thereby,convergence rate is improved.Simulations show that the proposed scheme has better performance than traditional blind equalization by FNN and DFE which used in navy range for receiving signal of telemetry.更多还原