本文主要介绍了一种针对自适应阵盲算法:多目标变模算法-MT-VMA(下面用MT-VMA表示)。这种算法可以在强大的信道干扰情况下盲分离重叠的多路单边带语音信号,而且不用对波达方向(DOA)的预先估算或者对接收机数据校准而分离出有用的信号(SOIs)。本文通过仿真通过一个6单元的天线阵列,盲分离在HF频带6组重叠的单边带语音信号实例结果,介绍MT-VMA的主要功能结构和性能。 This article mainly introduces a kind of adaptive blind algorithm: multi-objective variable modulus algorithm-MT-VMA (MT-VMA below). This algorithm blindly separates overlapping multiple single-sideband speech signals under strong channel interference without the need to isolate the useful signals (SOIs) prior to DOA estimation or receiver data calibration . In this paper, six groups of single sideband speech signals in the HF band are blindly separated by simulation through a 6-element antenna array. The main functional structure and performance of MT-VMA are introduced.