Reference prediction in optimal control of smart-grid with asynchronous RDG's

This paper proposes an efficient interaction infrastructure between the Utility and distributed customers in a future smart grid. Detailed steps of a complete interaction cycle are presented, including demand response (DR), distributed customer mode-switching decision making and stochastic tracking control for the Utility-maintained conventional generation facilities. We further extend our previous work by considering the realistic scenario of asynchronous load demand signals from different customer loads. To compensate for different delays seen by different customer/load signals, we design a Kalman filter (KF) based prediction scheme to construct the correct reference signal. The prediction problem is essentially transformed into a problem which can be solved by the standard Kalman filtering technique. Due to the system linearity and the fact that different customer/loads are decoupled with each other, we show that the centralized reference prediction can equivalently be implemented distributively, where a distributed Kalman filter is implemented for each of the delayed load signal. In addition, we establish that the separation of the reference prediction and tracking design is still optimal for the original objective function. Simulation results are presented to show the performances of the proposed prediction and tracking schemes.