基于神经网络的接收机宽带非线性行为建模

Wideband Nonlinear Behavior Modeling of Receiver with Neural Network

为了预测复杂电磁环境下接收机的非线性效应,文中基于实数时延径向基函数神经网络,构建了具有记忆效应的接收机非线性神经网络模型。分别采用K-均值聚类算法和正交最小二乘法对模型的隐含层中心和权值进行选取和学习,并用接收机的输入输出实测数据对模型进行训练。通过宽带信号的同相和正交两个分量对模型进行验证。模型仿真结果与实测数据相吻合,模型的归一化均方误差可达-41.88 dB。该结果表明,所构建的神经网络模型具有较快的收敛速度、良好的建模精度和泛化能力。

In order to predict the nonlinear effect of receiver in complex electromagnetic environment,a nonlinear behavior model of receiver with memory effect is constructed based on real-value time-delay radial basis function neural network.The K-means clustering algorithm and the orthogonal least square method are respectively used to select and learn the center of the hidden layer and weight of the model,and the model is trained with the input and output measured data of the receiver.The model is verified by the in-phase and quadrature components of wideband signals.The simulation results are in good agreement with the measured data,and the normalized mean square errors of the model reaches-41.88 dB.The verification results show that the neural network model has fast convergence speed,good modeling accuracy and generalization ability.