根据日益复杂的空战格局以及基于微分对策理论设计的空空导弹导引律所存在的缺陷与不足,提出了基于预见学习算法的微分对策导引律改进方案。首先讨论了微分对策导引律的缺陷,并针对其不足研究了基于最优控制理论的预见学习方法;其次设计了基于CMAC(小脑关节模型)的预见学习运算过程,将其应用于空空导弹预见逃逸目标的未来逃逸轨迹,并令导弹据此进行最优决策实现预见学习制导;最后根据逃逸目标飞行轨迹的恶意性突防程度,总结专家经验,将预见学习与微分对策相结合实现优势互补,形成了基于预见学习的微分对策导引律。其仿真结果较经典导引律有明显的优势,能以较小的控制量、较短的拦截时间获得很小的脱靶量,尤其适合拦截具有大机动恶意突防行为的目标。

According to the vigorous pattern of the air combat and the drawback of the air-to-air missiles guidance law designed on the basis of differential game theory,a new method was proposed for improving the differential game guidance law,which is based on predictive learning algorithm. Firstly,the drawbacks of differential game guidance law were discussed,and according to which,the predictive learning guidance based on the optimal control theory was studied. Secondly,the predictive learning process was designed based on calculation of Cerebellar Model Articulation Controller (CMAC),and was used to predict the future trace of the escaping targets for missiles,then the missile could make an optimum decision according to the predicted trace and realize predictive learning guidance law. At last,based on both penetration degree of escaping targets and the experience of experts,the differential game and the predictive learning were combined,formed a differential game guidance law based on predictive learning. Compared with classical guidance,the simulation results showed that the new guidance can get small miss distance with minor control variable and shorter time of intercepting,which is suitable for intercepting maneuvering targets.