基于证据异常度的水下多源信息预处理方法

Underwater Information Preprocessing Method Based on Evidence Anomaly Factor

摘要:

Dempster-Shafer证据理论在水下多源目标识别领域有着广泛而重要的应用, 但经典的证据理论在融合高度冲突的证据时往往会导致一些反常理的结果, 如Zadeh冲突悖论;针对这一问题,综合考虑证据体之间的冲突程度和支持程度,提出一种证据异常度的概念并对原始证据集进行异常检测,基于检测结果对原始证据体进行权重分配,引入全集项,修正证据源;在保持Dempster组合规则不变的前提下,进行有效的证据预处理,实验仿真结果验证了算法的有效性;证明对证据体进行有效的修正,可以改进经典证据理论的缺点,达到更好的融合结果。

Abstract:

Dempster-Shafer evidence theory has been widely used in many strategic fields, such as underwater target recognition. But combination of the DS evidence theory always brings some paradoxical behaviors, e.g., the Zadeh paradox problem. Aim at this problem, overall consideration with the degree of conflict and the degree of support between evidence bodies is required. A definition called anomaly factor was proposed, which can be used to detect the abnormal evidence. The weight was assigned to multi-source evidences based on the detection results. Experiment results prove that the proposed method is effective, and it can get better results than the typical DS method.