摘要
系统建立了基于神经网络的电池荷电状态(SoC)的预测模型,可用于对电池电量有精确预测需求的设备中。首先,基于自适应神经网络模糊推理系统(ANFIS)的预测模型确定了网络学习算法,采用MATLAB仿真程序用不同方法构造初始ANFIS模型,利用实验数据对模型网络进行训练,分析ANFIS系统结构和参数的变化。其次,将模型值与实际测得的结果进行对比,对网络的各个参数进行调整后再次用仿真比对预测效果。最后,设计了嵌入式系统硬件和软件的结构,用正弦波注入法解决了电池内阻测量这一难点。
The system establishes a model based on adaptive neural network for predicting the State of Charge（SoC）,which can be applied to systems that need accurate control of battery power. Firstly, the ANFIS-based prediction model defines the net- work learning algorithm. The MATLAB simulate program generates several initiations to train the ANFIS model, and trains the net- work model with experimental data, analyses the change of ANFIS structure and values of parameters. Then, by comparing the val- ues between ANFIS model and the real experiment, this paper adjusts the parameters and contrests experiment again. Finally, the design of hardware and software is performed, and an AC injection method is obtained in this paper to deal with the problem of measure battery internal resistance.
出处
《电子技术应用》
北大核心
2014年第9期58-61,共4页
Application of Electronic Technique
