基于红外技术的机载电路板故障诊断研究

发布时间：2018-05-24 01:16

本文选题：故障诊断 + 电路板　；参考：《中国民航大学》2017年硕士论文

【摘要】：机载电路板传统的故障检测是利用仪表手动测量电压、电流等信号,工程师依据经验判断故障原因。由于仪表检测无法处理高频电路实时故障,还会给精密的机载电路板带来二次故障,使得技术复杂、效率低。因此研究一种基于红外技术的机载电路板故障诊断方法和健康状态检测方法,提高检测的准确率和效率,具有通用性强与成本低的优势。首先,设计以红外热像仪为核心的机载电路板故障检测系统,完成采集数据、图像处理、数据提取等工作流程。高性能红外热像仪给系统融入先进的红外技术,从硬件角度提高故障诊断灵敏性。通过3个实验研究电路板发热模型,建立工作状态与红外特性的联系。然后,为采集到的电路板故障热图定义故障代码,构建包含81种故障的标准数据库。在热图差分法与序列分析法故障诊断实验的基础上,提出一种多级SVM故障诊断方法,建立温度信息与故障征兆之间的关系。提取8个特征值构成工作状态向量,从8个维度建立第一级故障诊断SVM模型和第二级数据融合SVM模型。使用遗传算法对参数c和g寻优,提高故障诊断准确率。最后,在灰色模型故障预测实验的基础上,提出使用SVM回归算法,预测电路板下一工作周期温度序列,并返回到第一级、第二级SVM模型预测故障代码。分析电路板失效机理,建立热效应随老化程度的变化关系。采集电路板19586飞行小时的温度序列,构建红外温度老化模型,划分为6个老化等级,训练SVM模型推算出实际飞行小时与剩余寿命。在横向课题研发的测试平台中,对WRT-2100型号机载气象雷达电路板进行实验验证。多级SVM故障诊断方法能够综合多维度信息实现故障的多分类,具有较高的准确率和效率。SVM故障预测均方误差较小,红外温度老化模型能够推算出老化程度与寿命。
[Abstract]:The traditional fault detection of airborne circuit board is to use the instrument to manually measure the signals such as voltage and current, and engineers judge the cause of the failure according to the experience. Because the instrument detection can not deal with the real-time fault of the high frequency circuit, it will bring secondary fault to the precise airborne circuit board, which makes the technology complex and the efficiency low. Therefore, a fault diagnosis method and a health condition detection method based on infrared technology for airborne circuit board are studied to improve the accuracy and efficiency of the detection, and have the advantages of strong generality and low cost. Firstly, an airborne circuit board fault detection system with infrared thermal imager as the core is designed to complete the work flow of collecting data, image processing, data extraction and so on. The high performance infrared thermal imager integrates advanced infrared technology to improve the sensitivity of fault diagnosis from the point of view of hardware. Through three experiments to study the heating model of circuit board, the relation between working state and infrared characteristic is established. Then, the fault code is defined for the collected circuit board fault thermal graph, and a standard database containing 81 kinds of faults is constructed. Based on the experiments of thermal graph difference method and sequence analysis method, a multistage SVM fault diagnosis method is proposed, which establishes the relationship between temperature information and fault symptoms. Eight eigenvalues are extracted to form the working state vector, and the first stage fault diagnosis SVM model and the second stage data fusion SVM model are established from the eight dimensions. Genetic algorithm is used to optimize the parameters c and g to improve the accuracy of fault diagnosis. Finally, based on the experiment of grey model fault prediction, SVM regression algorithm is proposed to predict the next working cycle temperature sequence of circuit board, and return to the first stage and the second stage SVM model to predict the fault code. The failure mechanism of circuit board is analyzed and the relationship between thermal effect and aging degree is established. The temperature sequence of 19586 flying hours of circuit board was collected, and the infrared temperature aging model was constructed, which was divided into 6 aging grades. The SVM model was trained to calculate the actual flying hours and the remaining life. In the horizontal research and development of the test platform, the WRT-2100 model of airborne weather radar circuit board is verified. Multi-level SVM fault diagnosis method can integrate multi-dimensional information to realize multi-classification of faults. It has high accuracy and efficiency. The mean square error of fault prediction is small. The infrared temperature aging model can calculate the aging degree and life span.
【学位授予单位】：中国民航大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V267;TN219

【参考文献】