| The power batteries for electric vehicles (EV) are considered to be the green pro-ject in the 21st century, which can solve the problems of pollution and energy shortage caused by combustion engines. But the research and development of the technique and the relevant investigation of power batteries is a barrier for popularization and application of the electric vehicles.Having analyzed the electronic mechanism of the MH/Ni batteries during charge and discharge, compared the advantages of MH/Ni batteries with other kinds of bat-teries, the results is that MH/Ni batteries is the promising power battery for EV due to its advantages such as high power lever and long life.Principle of the reaction of the MH/Ni battery, the characteristics and existed problems of different kinds of SOC measuring methods are compared, such as use voltage or resistance of the batteries to get the SOC, then put forward the way of ANN to estimate SOC of the batteries in this research.In order to estimate SOC effectively, based on the principle of the charge and discharge of the MH/Ni batteries, analyzed the traditional ways of estimating SOC, in this research paper a neural network model of MH/Ni batteries on LM network was built up. The artificial neural network (ANN) method, which can be used to approach nonlinear function with many input/output parameters, is originally used to estimate the discharge capacity of MH/Ni batteries when changing the discharge current and voltage. The satisfactory result has been obtained from comparison of simulation re-sults.The artificial neural network has an ability of numerical approximation any nonlinear function. Simple error back prorogation neural network and ELM neural network could estimate SOC with big error. Combine those two together, speeds up the training convergence. At last design a neural network resembling multi-layer feed-back neural network, the results shows that it could estimate the SOC with the state of constant current and non-constant current.
|
PDF Full Text Request