基于机器分类判决把握度自评估的人机结合

Human Machine Cooperating Based on Confidence Self Evaluation for Machine Auto Classification Decision

随着机器自动分类判决在电子对抗中的广泛应用,使得侦察干扰设备的自动化水平也不断提高,但在某些场合下机器判决的准确性却始终难以达到预定要求。通过对该问题的分析,指出了机器自动判决优势的发挥对低维特征空间的依赖,分析了判决空间维度的扩展对判决准确度提升的正面影响,提出对机器自动判决结果进行把握度自评估,并以此为参考来实施人工二次辅助分类判决,从而在保持机器自动快速判决的条件下,通过人机结合的方式有效提升整个系统的判决性能。最后以信号情报侦察中的信号检测与调制识别两类典型任务为例,阐述了该方法的实施流程与应用优势,从而为电子对抗中人机工程的理论研究和人机结合的实际应用提供了重要参考。

The automation of reconnaissance and jamming equipments has been greatly enhanced in recent years,because machine auto classification decision is widely utilized in electronic warfare. But its accuracy cannot satisfy some applications. After the problem is analyzed,the machine auto decisiondependency on characteristic space withfew dimensions is disclosed. The more dimensions are necessary for more accurate judgment. Then the method about confidence self evaluation for machine auto classification decision is put forward,and the combination with human assisting is implemented according to confidence values. The integrated performance for the whole system becomes better by human machine cooperating under condition that the judgment is as fast as before. The typical applications about signal detection and modulation type recognition in signalintelligence are considered as examples,in which the approach steps and advantages are explained in detail. It is an important reference for theoreticresearch and engineering application of human machine engineering in electronic warfare.