SOC and remaining range estimation for electric vehicles based on dual-model hybrid approach
To alleviate the problem of electric vehicle owners'mileage anxiety,this paper proposes a hybrid filter and neural network-based state of charge(SOC)estimation method for electric vehicles,which can accurately esti-mate the SOC and remaining mileage of electric vehicles.Firstly,a dimensionality reduction algorithm and a classi-fication algorithm are used to isolate five categories of driving behaviors that reflect vehicle energy consumption from the real vehicle dataset as part of the model input.Secondly,a hybrid model combining Kalman filtering and a two-layer bi-directional long short term memory neural network is built,which can reduce the noise of real-time data and combine with historical data to calculate EV SOC and remaining mileage.Finally,different model input fea-tures and model structures are compared to demonstrate the high accuracy of the proposed method.
electric vehiclesSOC estimatesremaining mileage estimatedriving behavior analysisdeep learn-ingKalman filtering
NETL
NSTL
万方数据
