### Abstract

Original language | English |
---|---|

Journal | I E E E Transactions on Smart Grid |

Volume | 9 |

Issue number | 4 |

Pages (from-to) | 3419-3429 |

Number of pages | 11 |

ISSN | 1949-3053 |

DOIs | |

Publication status | Published - 2018 |

### Keywords

- Distributed optimisation
- Energy management system
- Energy storage systems
- Microgrid
- Model predictive control
- Multi-agent systems
- Dynamic optimal power flow

### Cite this

*I E E E Transactions on Smart Grid*,

*9*(4), 3419-3429. https://doi.org/10.1109/TSG.2016.2631600

}

*I E E E Transactions on Smart Grid*, vol. 9, no. 4, pp. 3419-3429. https://doi.org/10.1109/TSG.2016.2631600

**Network Topology Independent Multi-Agent Dynamic Optimal Power Flow for Microgrids with Distributed Energy Storage Systems.** / Morstyn, Thomas; Hredzak, Branislav; Agelidis, Vassilios G.

Research output: Contribution to journal › Journal article › Research › peer-review

TY - JOUR

T1 - Network Topology Independent Multi-Agent Dynamic Optimal Power Flow for Microgrids with Distributed Energy Storage Systems

AU - Morstyn, Thomas

AU - Hredzak, Branislav

AU - Agelidis, Vassilios G.

PY - 2018

Y1 - 2018

N2 - This paper proposes a multi-agent dynamic optimal power flow (DOPF) strategy for microgrids with distributed energy storage systems. The proposed control strategy uses a convex formulation of the ac DOPF problem developed from a d-q reference frame voltage-current model and linear power flow approximations. The convex DOPF problem is divided between autonomous agents and solved based on local information and neighbour-to-neighbour communication over a sparse communication network, using a distributed primal subgradient algorithm. Each agent is only required to solve convex quadratic sub-problems, for which robust and efficient solvers exist, making the control strategy suitable for receding horizon model predictive control. Also, the agent sub-problems require limited power network information and include only a subset of the centralised optimisation problem decision variables and constraints, providing scalability and data privacy. Unlike existing distributed optimal power flow methods, such as alternating direction method of multipliers, under the proposed control strategy the information required by each agent is independent of the communication network topology, providing increased flexibility and robustness. The performance of the proposed control strategy was verified for an ac microgrid with distributed lead-acid batteries and intermittent photovoltaic generation, using an RTDS Technologies real-time digital simulator.

AB - This paper proposes a multi-agent dynamic optimal power flow (DOPF) strategy for microgrids with distributed energy storage systems. The proposed control strategy uses a convex formulation of the ac DOPF problem developed from a d-q reference frame voltage-current model and linear power flow approximations. The convex DOPF problem is divided between autonomous agents and solved based on local information and neighbour-to-neighbour communication over a sparse communication network, using a distributed primal subgradient algorithm. Each agent is only required to solve convex quadratic sub-problems, for which robust and efficient solvers exist, making the control strategy suitable for receding horizon model predictive control. Also, the agent sub-problems require limited power network information and include only a subset of the centralised optimisation problem decision variables and constraints, providing scalability and data privacy. Unlike existing distributed optimal power flow methods, such as alternating direction method of multipliers, under the proposed control strategy the information required by each agent is independent of the communication network topology, providing increased flexibility and robustness. The performance of the proposed control strategy was verified for an ac microgrid with distributed lead-acid batteries and intermittent photovoltaic generation, using an RTDS Technologies real-time digital simulator.

KW - Distributed optimisation

KW - Energy management system

KW - Energy storage systems

KW - Microgrid

KW - Model predictive control

KW - Multi-agent systems

KW - Dynamic optimal power flow

U2 - 10.1109/TSG.2016.2631600

DO - 10.1109/TSG.2016.2631600

M3 - Journal article

VL - 9

SP - 3419

EP - 3429

JO - I E E E Transactions on Smart Grid

JF - I E E E Transactions on Smart Grid

SN - 1949-3053

IS - 4

ER -