航空雷达故障诊断的β上近似属性约简及规则融合

β Upper Approximation Reduction and Rule Fusion of Airborne Radar Fault Diagnosis

航空雷达故障诊断的一个主要目标是求出某种程度上可能发生的故障类型,为此,首先引入β上近似属性约简,定义了β上近似决策辨识集和辨识矩阵,给出了求取β上近似约简集的方法;其次,提出了一种基于包含度的规则融合方法,该方法充分挖掘属性值信息,可以得到全部属性条件下的带可信度的决策规则,有效解决了在各类型故障样本数目相差较大的情形下的规则融合问题;最后,以某型航空雷达的整机测试为例,在基于β上近似约简对其测试参数进行约简的基础上,应用基于包含度的规则融合方法,得到了全部属性取值条件下的航空雷达故障诊断规则。结果表明:该方法直观有效、计算简便,为航空雷达故障诊断提供了新的思路和途径。

The acquisition of decisions with some extent reliability is one of important aids in airborne radar fault diagnosis. The β upper approximation reduction is firstly introduced, moreover β upper approximation discernibility sets and discernibility matrix are defined and the acquirement approach of βupper approxima-tion reduction is given. Secondly,the rule fusion method is offered to sufficiently mine the attribute-value in-formation and the decision-making rules with reliability cover all values of the attributes are obtained, which efficiently solve the rule fusion problem in the condition of the very different number among all fault kinds. Finally, taking a test of an airborne radar as an example, this paper cuts test parameter based onβupper ap-proximation reduction and applies the rule fusion method to obtain the airborne radar fault diagnosis rules which cover all values of the attributes. Results show that the methods in this paper are visually effective and easy calculation, which offer a new though and approach for airborne radar fault diagnosis.