Life Prediction of Hybrid Supercapacitor Based on Improved Model-Extreme Learning Machine
Life Prediction of Hybrid Supercapacitor Based on Improved Model-Extreme Learning Machine
复制标题
DOI：
10.1109/pedg.2019.8807678
复制
发表时间：
2019-06
期刊：
影响因子：
--
通讯作者：
Y. Zhou;Shuo Li;Kai Wang
中科院分区：
文献类型：
--
作者：
Y. Zhou;Shuo Li;Kai Wang
混合超级电容器既具有传统电容器的大电流快速充放电特性，又具有电池的储能特性。它们广泛应用于各个领域，是能源技术研究的热点。本文以五氧化二锑/氧化铈复合电极为研究对象，深入研究了温度、充放电电压、电流、放电深度和振动对其降解的影响机理。提出了一种新的混合超级电容器性能退化机理模型。模型极限学习机（Model-Extreme Learning Machine， MELM）用于在线识别上述模型。具体步骤如下：将上述模型集成到极值学习机的隐藏层中，然后利用递推最小二乘法对极值学习机中隐藏层的权值进行识别，然后即可完成上述模型的在线参数识别。通过ESR和C监测超级电容器的老化程度，准确估计剩余使用寿命。实验结果表明，该模型具有较高的识别精度和较好的预测效果。
Hybrid supercapacitors not only have the characteristics of high current fast charging and discharging of traditional capacitors, but also have the energy storage characteristics of batteries. They are widely used in various fields and are hotspots in energy technology research. In this paper, the hybrid supercapacitor with antimony pentoxide/cerium oxide composite electrode is studied, and the influence mechanism of temperature, charge and discharge voltage, current, depth of discharge and vibration on its degradation is studied in depth. A new model of performance degradation mechanism of hybrid supercapacitors is proposed. The Model-Extreme Learning Machine (MELM) is used to identify the above models online. The specific steps are as follows: the above model is integrated into the hidden layer in the extreme learning machine, and then the weight of the hidden layer in the extreme learning machine is identified by using the recursive least squares method, and then the online parameter identification of the above model can be completed. The aging degree of the supercapacitor is monitored by ESR and C to accurately estimate the remaining service life. The experimental results show that the model has higher recognition accuracy and better prediction effect.