摘要
为了提高金属化膜电容器寿命预测精度,保证脉冲功率系统安全稳定的运行,提出一种新的基于粒子滤波算法的电容器寿命预测方法。该方法不需要大量寿命数据,能通过历史数据不断更新模型参数,准确地预测单个电容器寿命。对54个金属化膜电容器进行3种工况下的寿命退化实验,可以得到,相比于传统Weibull分布模型方法和Birnbaum-Saunders分布模型方法,所提方法的预测精度更高,3种工况下的平均预测偏差都〈100次,验证了所提出方法的有效性。
In order to improve the life prediction precision of the metallized membrane capacitor and ensure the safe and stable operation of the pulse power system, a new method of predicting the lifetime of capacitor based on particle filter algorithm is proposed. The method does not require a large number of life data, and can continuously update the model parameters through historical data, and accurately predict the life of a single capacitor. The life degradation experiments of 54 metallized membrane capacitors under 3 operating conditions can be obtained. Compared with the traditional Weibull distribution model method and the Birnbaum-Saunders distribution model method, the proposed method has higher prediction accuracy, the average prediction deviation under the 3 operating conditions is lower than 100 times, and the effectiveness of the proposed method is verified.
作者
熊诗成
鲁军勇
郑宇锋
曾德林
XIONG Shicheng;LU Junyong;ZHENG Yufeng;ZENG Delin(Navy Armament Academy,Beijing 100161,China;National Key Laboratory of Science and Technology on Vessel Integrated Power System,Naval University of Engineering,Wuhan 430033,China)
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2018年第7期2378-2384,共7页
High Voltage Engineering
基金
国家自然科学基金(51207162
51407191
51307176)
总装预研基金(9140C840409140C84026)~~