Nonlinear equalizer based on function chain neural network

The invention provides a nonlinear equalizer based on a function chain neural network. The function chain neural network is improved and reconstructed into an equivalent complex single-layer perceptron nonlinear equalizer. Due to randomness of a function chain neural network mapping process, in order to enable feature extraction to be more sufficient, high-dimensional random mapping is carried outagain on the basis of features obtained through high-dimensional nonlinear mapping. Finally, the features input into the single-layer perceptron comprise initial input features, first-stage mapping features and second-stage mapping features, the features are expanded into a new input matrix in a column form, and output signals are obtained according to network weights. The calculation complexityis low, and the equalization effect of a deep neural network level on signal nonlinear damage is achieved. 本发明提供一种基于函数链神经网络的非线性均衡器，将函数链神经网络改进并重构为一个等效的复数单层感知机非线性均衡器。由于函数链神经网络映射过程的随机性，为了使特征提取更为充分，我们对于经过高维非线性的映射得到的特征基础上再次进行高维随机映射