Fuzzy observer design for local model network based battery state of charge estimation

The correct determination of the state of charge (SoC) is an important task in hybrid electrical vehicles. In this paper, a systematic approach to state of charge estimator design using the fuzzy observer architecture is presented. The nonlinear observer is based on a dynamic battery model which describes the nonlinear system behaviour of the cell terminal voltage in dependance of the charge/discharge current. In this paper, a purely data driven modelling approach is chosen in order to to enable the application for any type of battery chemistry. The proposed local model network (LMN) is shortly reviewed, an augmented state space representation of the LMN is derived and the nonlinear observer design and parametrisation is presented. In combination with LMNs, the architecture of the fuzzy observer helps to reduce the computational complexity of the global nonlinear filter significantly. Each of the local models in an LMN is a linear time-invariant dynamic system. The global filter is then derived from a linear combination of the local filters. The proposed concepts for SoC estimation are validated experimentally by means of a Lithium Ion power cell.