论文部分内容阅读
提出一种神经网络结合分离信号对功率放大器预失真建模的方法。将输入/输出信号的线性与非线性部分分开处理,利用神经网络良好的逼近能力,采用LM算法,拟合出功率放大器特性曲线,进而建立预失真模型,使非线性功率放大器的输入/输出曲线整体呈线性化。在保证输出幅度限制和输出功率最大化的前提下,与未作信号分离的神经网络建模方法、多项式建模方法以及Saleh函数模型方法相比较,发现信号分离神经网络建模方法能得到较小的归一化均方误差和误差矢量幅度。仿真结果表明,采用信号分离神经网络对功率放大器及其预失真建模,整体线性化误差较小、精度高、效果更佳。
A new method of modeling preamplifier for power amplifier based on neural network and separated signal is proposed. The input / output signal linear and non-linear part of the separate treatment, the use of good approximation ability of the neural network, the use of LM algorithm to fit the power amplifier characteristic curve, and then establish a pre-distortion model, the nonlinear power amplifier input / output curve The overall linear. Under the premise of ensuring the output amplitude limit and the maximum output power, compared with the neural network modeling method without signal separation, the polynomial modeling method and the Saleh function model method, it is found that the signal separation neural network modeling method can be smaller Normalized mean square error and error vector magnitude. The simulation results show that using the signal separation neural network to model the power amplifier and its predistortion, the overall linearization error is small, the precision is high and the effect is better.