基于红外的SSA-CNN-GRU电路板芯片故障诊断  

Fault diagnosis of SSA-CNN-GRU circuit board chip based on infrared

作　　者：王力 李志新 张亦弛 WANG Li;LI Zhi-xin;ZHANG Yi-chi(Airborne Electronic Systems Deep Maintenance Laboratory,College of Vocational Technology,Civil Aviation University of China,Tianjin 300300,China)

机构地区：[1]中国民航大学电子信息与自动化学院机载电子系统深度维修实验室,天津300300

出　　处：《激光与红外》2023年第4期556-565,共10页Laser & Infrared

基　　金：国家自然科学基金委员会与中国民用航空局联合资助基金项目(No.U1733119);中央高校基本业务费项目(No.3122017107);基于红外技术与数据驱动的机载电路板卡故障诊断与预测研究项目(No.2021YJS018)资助。

摘　　要：针对电路板温度数据诊断率不佳的问题,本文提出了基于红外的SSA-CNN-GRU电路板芯片故障诊断模型。首先,根据红外热像仪采集芯片温度数据,建立多维特征模型;然后,在故障诊断模型输入端和CNN-GRU通道分别添加注意力机制,构建双注意力结构,自适应识别有效数据段和提取红外图像有效特征;接着,利用麻雀搜索算法优化注意力机制权值分配,获取全局最优超参数;最后搭建SSA-CNN-GRU故障诊断模型,实现芯片故障模式的高精度诊断。实验采用电源电路板进行可靠性分析,实验结果表明,本文算法在诊断精度可达98.73%,且稳定性、可靠性方面均优于对比算法。To solve the problem of poor diagnosis rate of circuit board temperature data,a SSA-CNN-GRU circuit board chip fault diagnosis model based on infrared is proposed in this paper.Firstly,a multi-dimensional feature model is established based on the chip temperature data collected by infrared thermal imager;secondly,an attention mechanism is added to the input of the fault diagnosis model and the CNN-GRU channel respectively to construct a dual attention structure to adaptively identify effective data segments and extract effective features of infrared images;then,the sparrow search algorithm is used to optimize the weight distribution of attention mechanism to obtain the global optimal super parameters;finally,the SSA-CNN-GRU fault diagnosis model is built to realize high-precision diagnosis of chip fault mode.The experimental results show that the diagnosis accuracy of this algorithm can reach 98.73%,and outperforms the comparison algorithm in terms of stability and reliability.

关 键 词：红外技术 注意力机制 麻雀搜索算法 门控循环单元 故障诊断 

分 类 号：TN219[电子电信—物理电子学]