针对基于雷达目标结构特征的识别方法对目标姿态角变化敏感的问题，利用模糊极大－极小（ＦｕｚｚｙＭａｘ－Ｍｉｎ，ＦＭＭ）神经网络作为模式分类器，提出抽取稳健性目标结构特征的方法，构造出能较好地减少目标姿态变化敏感性的ＦＭＭ神经网络雷达目标识别系统。利用实测的五种飞机目标数据对其进行了性能检验，证明该识别方法能在较大的姿态模区内取得优良的识别效果。 Aiming at the problem that the recognition method based on radar target structure is sensitive to the change of target attitude angle, a method of extracting robust target structure features is proposed by using FuzzyMax-Min (FMM) neural network as a pattern classifier. The FMM neural network radar target recognition system can reduce the sensitivity of target attitude change well. The measured performance of the five aircraft target data was used to prove that the identification method can achieve good recognition results in the larger pose model.