Distributed Nonlinear Model Predictive Control of Heterogeneous Battery Storage Systems

This paper deals with distributed control of variable efficiency heterogeneous battery storage systems connected in a microgrid. The nonlinear efficiency of battery charging is taken from converters’ datasheet and identified battery internal resistance, and a corresponding nonlinear model is derived dependent on state of energy. Individual battery control, as a local nonlinear problem, is solved by using a sequential linear program. Distributed control with the central supervisor is then used to coordinate individual batteries within the storage system and ensure improved overall system operation while respecting the joint constraint of the grid connection power capacity. Asymmetric projection algorithm is used for distributed control, based on iterative convergence of individual battery control actions towards the optimal operation of the whole system. The algorithm is implemented in a realistic simulation scenario of six different batteries with their corresponding characteristics, real electricity prices and measured load consumption. Presented results show that distributed control satisfies global constraint, while obtaining better cost than the decentralized approach with previously reported 7% of increased savings of nonlinear over constant efficiency.