Multi-Generator Control Strategies in MVDC System with Constant Power Loads
The main objective of this project is to develop, implement and validate in HIL testing, an innovative control method for generator side interface converters, which compensates for the negative resistance behavior of the loads in Medium Voltage DC systems (MVDC). The control must maintain system stability at every operating point in a multi-generator system and in presence of multiple constant power loads. This project is funded by the US Office of Naval Research for future application on board ships.
A major challenge of stability in presence of constant power loads is caused by their non-linear behavior. Hence, the main feature of the controllers is their linearization effect. In the current stage of the project, three control strategies, among which two newly developed, are compared and validated on the full-scale RTDS model of the MVDC distribution system on board the future all-electric ship. This comparison and validation process will lead to a final selection.
Nonlinear feedback: this central control approach, achieves linearization via non-linear feedback, and guarantees stability
via state feedback control of the capacitor currents, sum of generators current. For the non-linear feedback, the power of the individual constant power load has to be fed back. However, this centralized approach is adversely affected by the challenges of measuring and communicating in real-time the load powers. An alternative linearization approach, fit for decentralized structure, and based on Taylor expansion has been developed.
Linearization via Taylor expansion: this control method achieves the linear small signal approximation via Taylor expansion in the neighborhood of the normal operating condition. Stability is then achieved via state feedback control based on the small signal model. This control strategy could be implemented in a decentralized controller, preferable for survivability, in a shipboard power system.
Decentralized LQG: This control strategy achieves linearization via accurate estimation, by an Extend Kalman Filter, of the sum of the currents of constant power loads, and the current of the other generators. The non-linear part of the system is modeled as a virtual disturbance and it is considered as an additional state to be estimated. Stability is then achieved via optimal Linear Quadratic Regulator. The most challenging part of this approach consists in the synthesis of the dynamic model of the virtual disturbance. When losing one generator, this control approach can realize the system self-reconfiguration automatically. In fact, the of the individual virtual disturbance is based only on the local measurements and fixed local model.
The research work is carried out by ACS in collaboration with Prof. Sulligoi at the University of Trieste in Italy.