针对传统的接触式电参数故障检测方法 难以满足日益复杂的电子装备的检修需要问题，提出了一种基 于红外热像的可提升非接触式电子装备异常 检测性能的温升多特征方法。在红外与光学异源配准的基础上，获得电子 装备各元件的精确位置。然后基于红外观测图像获取 各个标准和待测元件中心区域的温升均值曲线。在对这些曲线进行全帧 程分段后，精细化地提取温升统计矢量。接着计算待测温升统计 矢量与标准温升统计矢量的差异并对其进行归一化处理，从而构建温升多特 征矢量。最后根据该矢量，通过分段表决和异常 量化实现电子装备热像异常的整体态势感知。与传统的热像 异常检测方法相比，本文所构建的温升多特征矢量可以更精细、更 稳健地描述温升变化。通过分段表决和异常量化实现了温升全程表征，为后续的电路 故障智能推理奠定了基础。实验结果表 明，本文方法具有检测准确率高、实时性好等优点。

To solve the problem that traditional contact electric parameter fault detection methods are difficult to meet the maintenance requirement of increasingly complex electronic equipment, a temperature rising multi-feature method based on infrared thermography which can improve noncontact electrionic equipment abnormality detection is proposed. Firstly, on the basis of infrared and optical heterologous registration, the accurate position of each element in an electronic equipment is obtained. Then, the mean value curves of temperature rising in the central areas of each standard element and each element to be detected are obtained on the basis of the infrared images observed. After the curves are divided into multiple subsections, the temperature rising statistical vectors are extracted precisely. Secondly, the difference between the temperature rising statistical vector to be detected and the standard temperature rising statistical vector is calculated and normalized so as to construct the temperature rising multi-feature vector. Finally, according to the vector, the whole situation awareness of thermal abnormality of the equipment to be detected is implemented by segmental judgment and abnormal quantification. Compared with the traditional thermal abnormality detection method, the temperature rising multi-feature vector constructed can more precisely and robustly describe the variation of temperature rising. The experimental result shows that this method has the advantages of high accuracy and good real-time performance. It lays a foundation for the subsequent intelligent reasoning of electric circuit fault.