基于模糊C均值改进算法和ANFIS的蓄电池SOC预测

Battery SOC Prediction Based on Improved Fuzzy C-means Clustering and ANFIS

蓄电池剩余电量预测作为蓄电池智能管理系统的核心部分,为合理控制蓄电池的充放电情况、延长蓄电池的使用寿命提供了判据。然而蓄电池剩余电量的影响因素复杂、预测难度较大。针对这一挑战性课题,提出一种基于改进的模糊C均值聚类和自适应模糊神经推理系统(ANFIS)的预测算法,采用减法聚类和加权模糊C均值聚类生成初始模糊推理系统,通过梯度下降法和最小二乘法混合算法对自适应模糊神经网络中的前件参数和后件参数进行训练,建立非线性预测模型。仿真结果表明,改进的聚类算法解决了传统模糊C均值聚类稳定性差以及对噪声点、错误点敏感的缺点,加快了收敛速度,在此基础上建立的蓄电池剩余电量预测模型也具有较高的预测精度。

The prediction of surplus capacity of batery can be used to reasonably tion and extend the batery life as the core of intelligent batery management system. Hof surplus capacity cause the difficulty of predicting accurately. To solve this challenging problem, a prediction algorithm basedon the improved fuzzy C-means clustering and Adaptive Network-based Fuzzy Inference System ii ference system is built by subtractive clustering and weighted fuzzy C-means clustering, then the hybrid algoritlim is used to trainthe parameters of fuzzy system to establish a nonlinear prediction model. The simulation results algorithm not only solves the shortcomings of traditional fuzzy C-means clustering but also batery surplus capacity prediction model has a high prediction accuracy.