铝电解电容器加速退化试验设计与寿命预测研究  

Research on Accelerated Degradation Test Design and Life Prediction for Aluminum Electrolytic Capacitor

作　　者：杨涛 汪旭 肖江林 YANG Tao;WANG Xu;XIAO Jianglin(CRRC Zhuzhou Institute Co.,Ltd.,Zhuzhou,Hunan 412001,China)

机构地区：[1]中车株洲电力机车研究所有限公司,湖南株洲412001

出　　处：《控制与信息技术》2022年第1期114-120,共7页CONTROL AND INFORMATION TECHNOLOGY

基　　金：省科技创新团队(2020RC4054)。

摘　　要：铝电解电容器是寿命敏感器件,随着时间的推移,其参数退化到一定程度时,必然会影响电路板的寿命,故而研究其退化规律与寿命特征是至关重要的。文章从铝电解电容器结构和退化机理出发,将温度作为加速敏感应力设计了加速退化试验,并采用加速退化数据进行寿命预测,给出了延长铝电解电容器使用寿命的正向设计方向;从状态修的需求出发,提出了基于BP神经网络的铝电解电容器剩余寿命预测方法,其预测数据来源既可以是现场实测数据,也可以是加速试验数据。将BP神经网络所预测的电容量退化值与退化试验的实测数据以及试验数据的最小二乘线性拟合预测值进行对比分析,结果表明,基于BP神经网络的电容值的预测误差在3%以内,而通过最小二乘线性拟合的预测误差在6%左右,从而验证了BP神经网络寿命预计算法的优越性,为后续电路板级电容器在线监测技术开发与应用提供有力支撑。As a life sensitive device, aluminum electrolytic capacitor degrades its parameters to a certain extent with the passage of time, which will inevitably affect the life of electronic board, so it is very important to study its degradation law and life characteristics. Starting from the structure and degradation mechanism of aluminum electrolytic capacitor, this paper designs an accelerated degradation test with temperature as the accelerated sensitive stress, uses the accelerated degradation data for life prediction, and gives a top-down design direction for extending the life of aluminum electrolytic capacitors. From the perspective of the demand for repair according to condition, a method for predicting the remaining life of aluminum electrolytic capacitors based on BP neural network is proposed. The source of prediction data can be field measured data or accelerated test data. By comparing the degradated capacitance predicted by BP neural network with the measured data of the degradation test and the predicted value of the least squares linear fitting of the experimental data, the results show that the prediction error of capacitance based on the BP neural network is within 3%, while the predicted value of the least squares linear fitting is around 6%, which verifies the superiority of the BP neural network life prediction algorithm, and provides strong support for development and application of the subsequent on-line monitoring technology of board level capacitors.

关 键 词：铝电解电容器 加速试验 寿命预测 BP神经网络 状态修 正向设计 

分 类 号：TP202.1[自动化与计算机技术—检测技术与自动化装置]