Two-level hierarchical model predictive control with an optimised cost function for energy management in building microgrids

Hybrid energy storage systems have been increasingly envisaged for building microgrids to soften the drawbacks arising from the unpredictability of renewable energy resources and dwelling occupancy. The combination of long- and short-term energy storage systems can enhance the building microgrid capacity of shifting the demand peak toward periods of power generation, increasing the marks of self-consumption rate. However, the design of energy management systems for hybrid energy storage microgrids is more complex than single ones due to a greater number possible solutions. Faced with this issue, this paper proposes a two-level Hierarchical Model Predictive Controller (HMPC) enhanced by two data-driven modules to improve the performance of building microgrids equipped with hybrid energy storage continuously and automatically. With minimum pre-design steps, the two data-driven algorithms improve the accuracy of Li-ion batteries and hydrogen storage models and determine adequate parameters for the HMPC cost function. Relying predominantly on data measurements, the proposed hierarchical controller determines which energy storage device must be run on a daily basis based on the estimation of the annual self-consumption rate and the annual microgrid operation costs. This real-time analysis decreases microgrid expenditure because it avoids grid penalisation regarding the energy autonomy index and reduces the degradation and maintenance of energy storage devices. Compared to a standard rule-based strategy, the proposed controller reduces annual costs up to 5% in residential buildings and 9% in non-residential ones. In contrast, compared to a conventional HMPC the annual expenditure is reduced from 1% to 7% in both types of buildings.