基于衰退模式分类的电动汽车动力电池寿命预测

发布时间：2018-05-05 13:11

本文选题：动力电池 + 衰退点　；参考：《哈尔滨工业大学》2014年硕士论文

【摘要】：随着环境和能源问题日益严重，电动汽车逐渐进入了人们的日常生活中。动力电池作为能源系统的核心部分，其寿命是影响电动汽车性能的一个重要因素。本文围绕电动汽车动力电池寿命预测问题展开，基于衰退模式分类建立了动力电池寿命衰退模型，对工作在不断变化的工况下的动力电池的健康状态估计和剩余使用寿命预测进行了深入研究。本文通过研究锂离子电池衰退机理相关文献，并由美国国家航空航天局的锂离子电池充放电循环实验的数据进行验证，得到了影响电池寿命衰退的主要外部因素为放电电流、截止电压和环境温度。确认了具体研究目标为通过电池健康状态的估计进行电池剩余使用寿命的预测。提取了新的动力电池寿命衰退表征量衰退点，并验证了其有效性。建立了基于衰退模式分类的动力电池寿命衰退模型。在衰退模式这一概念的基础上，运用ART2神经网络完成动力电池的衰退模式分类，运用权马尔可夫链完成了动力电池的衰退模式预测，通过有理论模型的曲线拟合方法完成了单一模式下的衰退模型的建立，通过线性叠加方法完成了动力电池寿命预测。为了检验所建立的动力电池寿命衰退模型是否准确、有效，本文设计了动力电池充放电循环实验，，搭建了实验平台，完成了电池全生命周期的衰退数据采集工作。通过对采集到的衰退数据建立动力电池寿命衰退模型，完成了不同起点的电池健康状态估计和电池剩余使用寿命预测。将预测值与实测值、以及在未进行衰退模式分类条件下使用支持向量机所得预测值进行对比分析，发现基于衰退模式分类的预测结果整体上误差都在可接受范围内，同时，随着预测起点的向后推移预测准确性逐渐变好。本文以衰退点作为电池寿命衰退表征量，基于衰退模式分类，建立了新的动力电池寿命衰退模型，可以实现变工况条件下较为准确的电池剩余使用寿命预测，适合在线检测，对于电池管理系统的改善具有良好的现实意义。
[Abstract]:With the increasingly serious environmental and energy problems, electric vehicles have gradually entered into people's daily life. As the core part of energy system, power battery life is an important factor to affect the performance of electric vehicle. In this paper, the problem of battery life prediction for electric vehicles (EV) is discussed, and the model of battery life decline is established based on the classification of declining modes. The estimation of the health status and the prediction of the remaining service life of the power battery working under changing working conditions are studied in this paper. In this paper, the decay mechanism of lithium-ion battery is studied and verified by the data of the battery charge and discharge cycle experiment of NASA. The main external factor affecting the battery life decline is discharge current. Cutoff voltage and ambient temperature. The purpose of this study is to predict the remaining life of the battery by estimating the battery health status. A new power cell life decay feature is extracted and its validity is verified. A dynamic battery life degradation model based on decline mode classification is established. On the basis of the concept of decline mode, ART2 neural network is used to classify the decline mode of power battery, and the weight Markov chain is used to predict the decline mode of power cell. The regression model under a single model is established by the curve fitting method with a theoretical model, and the life prediction of the power cell is accomplished by the linear superposition method. In order to verify whether the model of battery life degradation is accurate and effective, this paper designs the battery charge and discharge cycle experiment, builds the experimental platform, and completes the whole battery life cycle data acquisition. Based on the data collected, the battery life degradation model is established, and the battery health state estimation and battery residual service life prediction are completed at different starting points. By comparing the predicted value with the measured value, and using the support vector machine (SVM), we find that the error of the prediction results based on the recession model classification is generally within the acceptable range, and at the same time, The accuracy of prediction improves gradually as the starting point of prediction goes backward. In this paper, the decay point is taken as the indicator of battery life decline. Based on the classification of degradation mode, a new power battery life degradation model is established, which can predict the battery remaining service life accurately under variable operating conditions and is suitable for on-line detection. It has good practical significance for the improvement of battery management system.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM912;U469.72

【相似文献】