This research area focuses on developing theoretical, computational, and data-driven digital twins of energy systems with a high share of renewable generation and distributed energy storage facilities. To this end, principles will be developed for the construction, adaptation, and operation of a digital twin that integrates physical models, data flows, forecasting, state estimation, quantitative uncertainty assessment, and control algorithms into a single computational framework. Such a digital twin must reproduce the current state of the power system, update parameters based on data, forecast future states, assess the model’s confidence limits, and support analyses of stability, controllability, and reliability.
The methodological core of this research area consists of physically grounded modeling, parameter identification, state estimation, data-driven model adaptation, forecasting, quantitative uncertainty assessment, reduced-order models, optimization, and control theory. Special attention will be given to parameter identifiability, state observability, model sensitivity to uncertain parameters, and model-data consistency.
Key approaches:
- The research will cover four levels of representation of the power system. The component level includes models of renewable generation, power converters, energy storage devices, and dynamic loads. The system level covers hybrid microgrids, distributed energy storage architectures, and systems with a high proportion of power electronics. The information level includes data flows, parameter identification, state estimation, model updates, and the detection of discrepancies between predicted and actual system behavior. The control level covers adaptive control, energy flow optimization, and the management of storage charging and discharging modes.
- Model development will combine physical principles with data analysis methods: models of generation from renewable sources with variable resource availability, battery models accounting for state of charge, technical condition, thermal behavior, and degradation, as well as load models with stochastic or temporal structures.
- The algorithmic component will include methods for parameter identification, model adaptation in real time or near real time, state estimation, generation and load forecasting, anomaly detection, uncertainty analysis, and control optimization. For distributed storage systems, a digital twin will be used to synthesize charging, discharging, and power reserve modes, taking into account forecasted generation, demand, the technical condition of the storage systems, temperature constraints, and degradation processes.
- Quality and reliability criteria for digital twins of power systems will be developed. These criteria will evaluate the accuracy of state reproduction, forecasting error, sensitivity to uncertain parameters, stability of control decisions, computational efficiency, model consistency with data, and the ability to reproduce transient, emergency, and extreme operating modes.
- The methods and algorithms will be tested in the MATLAB/Simulink environment and related computational modeling tools using reference models of renewable generation, energy storage systems, power converters, hybrid microgrids, dynamic loads, and control algorithms.
- Test scenarios will include variations in renewable energy resources, step changes in load, grid-connected and off-grid modes, temperature limitations of energy storage systems, degradation processes, forecasting errors, and fault conditions.