Study of estimation of battery state of charge (SoC) and state of health (SoH) using neural network techniques

Nowadays the electric vehicle(EV) has become the main development direction for all OEMs, to meet the challenges of environment and energy saving. The core factor which leads the development and performance of electric vehicle is the use of Lithium battery, that varies the available range of vehicle and its power performance. Various Lithium-ion battery has been introduced into market leading different performance characteristics with increased energy density. From development point of view, different battery development direction requires a reliable system for control and monitoring the status of the battery. From this point the usual challenges of battery condition accuracy occurs •??Long voltage relaxation time to reach its open circuit voltage (OCV) after a current pulse •??Temperature and SOC dependence with Flat OCV-SOC curve for most of the SOC range •??Precise measurement of initial condition from unknown condition due to internal resistance loss This thesis addresses to compare via the modelling approach using an equivalent circuit battery plant with extended Kalman filter and neural network method from machine learning approach to present a robustness result of battery condition estimation comparison. The simplified implementation of the Extended Kalman filter offers a computationally efficient option for runtime SOC evaluation on-board vehicles, where the Neural network method represents a new solution with less complex modelling requirement for having more efficient development effort. The objective of the varies measurements or monitoring methods are all intended to be able to have the battery condition accurately calculated. Where the following 2 factors are generally introduced as key characteristics of battery: •??State of charge (SOC) is a crucial index used in the assessment of battery storage system; it indicates the charge level of a battery cell. Li-iron technology introduce a flat discharge curve, long-life cycle, and high energy density characteristics keeping the voltage stable up to 80% of discharge time, which makes it very difficult to estimate SoC on a simple voltage measurement for its non-linearity. The voltage difference between two SoC values may be so small that it is not possible to estimate the state of charge with good precision. (1) •??State of Health (SoH) describes the difference between battery capacity on used and fresh battery that leads to the cell aging. It is defined as the ratio of the maximum battery charge to its rated capacity with expression as a percentage: SoH (%) =100*Qmax/Cr Cr = The rated capacity Qmax(Ah) = The maximum charge available of the battery