海量数据的高速传输促使着卫星通信向高频段、大带宽、高谱效率传输方向发展，对星载射频发射机的数字预失真补偿带来了很大困难。本课题旨在借助机器学习和压缩感知知识，提出一种自适应的星载超宽带预失真理论和方法，具备星载发射机失真不连续特性的统一表征辨识、强非线性和深度记忆效应补偿能力、低实现复杂度等特点。主要研究内容包括：对星载发射机的失真不连续非线性特性进行分析，研究将机器学习理论中的聚类分析算法，同最优分割算法相结合进行区域划分，建立基于分段仿射逼近理论的行为模型；通过稀疏系统辨识方法对分段仿射逼近模型进行简化，融合自适应信号处理算法，建立星载的超宽带自适应稀疏预失真器模型；利用带宽抑制算法，解决超宽带失真信号的频谱扩展难题。该项目的研究丰富和完善了预失真理论体系，为推动数字预失真在高分辨率遥感卫星以及宽带通信卫星星地馈电链路中的工程应用奠定理论基础。

The high-speed transmission of mass data impels the development of satellite communications toward high frequency band, wide bandwidth, high spectral efficiency transmission, which brought great difficulty for digital predistortion compensation of spaceborne radio frequency transmitter. The main purpose of the research is to propose an adaptive spaceborne ultra-wideband predistortion theory and method using machine learning and compressive sensing knowledge, which has the characteristics of the uniform representation identification of the distortion discontinuous characteristics of the spaceborne transmitter, compensation ability of strong nonlinearities and deep memory effects and lower implementation complexities. The research contents include, but are not limited to the followings: Analysis of the distortion discontinuous characteristics of spaceborne transmitter and research of combining the clustering analysis algorithm and the optimal segmentation algorithm in machine learning for regional division, and establishing the behavioral models based on the piecewise affine approximation theory; Based on the sparse system identification method, a fast and effective algorithm for the piecewise affine approximation model pruning is designed and implemented; The adaptive sparse predistorter suitable for spaceborne transmitter system can be constructed through the fusion of compressed sensing and adaptive signal processing theory; The spectrum expansion problem of ultra-wideband distortion signals are solved using bandwidth suppression algorithm. The research of this project enriches and improves the predistortion theory system, which lay the theoretical foundation of the actual application of the digital predistortion in the high resolution remote sensing satellite and the satellite to earth feed link of the broadband communication satellite.