Gast ::
| Hauptseite > Publikationsdatenbank > Investigation on the Influence of Real-World Weather and Demand Forecasts on the Model Predictive Control of a Hydrogen-Battery Energy Storage System |
| Hauptseite > Publikationsdatenbank > Investigation on the Influence of Real-World Weather and Demand Forecasts on the Model Predictive Control of a Hydrogen-Battery Energy Storage System |
| Contribution to a conference proceedings | FZJ-2023-01337 |
; ; ;
2022
This record in other databases:
Please use a persistent id in citations: doi:10.11581/dtu.00000267
Abstract: Model predictive control is used in many applications due to its capability to incorporate complex system dynamics and long forecast horizons. Complex hybrid energy storage systems using hydrogen and battery storage systems have vastly different storage dynamics that necessitate a detailed system model in combination with a long forecast horizon. The integration of renewable energy sources increases the importance of energy storage over long time periods due to their inherent dependency on highly volatile environmental energy sources. Model predictive control determines the ideal scheduling of the different storage systems using mathematical optimization and therefore utilizes volatile renewable energy sources more efficiently while decreasing the strain on the energy grid. However, the actual energy production and demand are not available under real-world conditions. Therefore, appropriate forecasts are needed to ensure a high quality of the optimization results. This work investigates the influence of open-source weather forecasts and demand forecasting on the quality of model predictive control of a hybrid hydrogen-battery energy storage system with a photovoltaics (PV) parkas renewable energy source. We use weather forecasts and real-world measurements from the Deutsche Wetter Dienst (dwd) to predict and calculate the energy production of a PV-park. Moreover, we utilize data of real-world office buildings to deduce a realistic electricity demand in combination with autoregressive models for the demand forecast. Using the forecast inside a simulative sensitivity analysis, we show that the solar power forecast has the most significant impact on the operation of the hybrid energy system. The impact of short-time fluctuations in the PV-power drastically decreases the performance of the model predictive control algorithm. Furthermore, we show the influence of the control horizon length for model predictive control since the prediction uncertainty increases over time. Finally, we demonstrate the mitigation of effects from short-term fluctuation in solar power by adapting a rule-based controller to adjust the results of the scheduling optimization with regard to the operating points of the storage components.
|
The record appears in these collections: |