摘要
超级电容是储能系统中常用的储能元件,为了解决在对其使用寿命预测时,影响因子考虑不全面,预测精度不高的问题,提出基于支持向量机回归算法与改进的长短期记忆网络算法(Mogrifier LSTM-SVR)相结合的超级电容的使用寿命预测模型。通过引入温度、电压和电流三种超级电容寿命影响因子,建立更贴近实际的超级电容寿命模型,选取剩余容量作为特征参数。构建Mogrifier LSTM网络,在传统的LSTM中增加Mogrifier门机制,并利用支持向量机回归(SVR)对Mogrifier LSTM网络预测误差回归预测,修正误差。通过仿真实验和模型的预测结果对比分析表明,Mogrifier LSTM-SVR对超级电容寿命预测的准确性更高,误差波动量级更小。
Ultracapacitors are commonly used energy storage components in energy storage systems.In order to solve the problems of incomplete consideration of influence factors and low prediction accuracy in the service life prediction of ultracapacitors,a life prediction model of ultracapacitors based on support vector machine regression algorithm and improved Mogrifier LSTM-SVR was proposed.By introducing temperature,voltage and current,a more realistic supercapacitor life model was established,and the remaining capacity was selected as the characteristic parameter.The Mogrifier LSTM network was constructed,the Mogrifier door mechanism was added to the traditional LSTM,and support vector machine regression(SVR)was used to predict the error of the Mogrifier LSTM network and correct the error.Simulation experiments were carried out,and the comparison and analysis of the prediction results of the model show that Mogrifier LSTM-SVR haS higher accuracy in the prediction of ultracapacitor life,and the error fluctuation is smaller.
作者
王福忠
任淯琳
张丽
WANG Fuzhong;REN Yulin;ZHANG Li(School of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo Henan 454000,China)
出处
《电源技术》
CAS
北大核心
2023年第7期935-939,共5页
Chinese Journal of Power Sources
基金
国家自然科学基金项目(U1804143)
河南省科技攻关项目(212102210146)。
关键词
超级电容
寿命预测
长短期记忆网络
支持向量机回归
剩余容量
supercapacitor
life expectancy
long short-term memory network
support vector machine regression
surplus capacity
