Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Predictive analytics in smart power systems
With implementation of the smart power grid, availability of smart power meters, and integration of intermittent energy resources into the power system, power system agents will communicate in real time. Hence, they are required to be equipped with advanced predictive analytical tools which can help analyze the data for making appropriate predictive actions. This dissertation addresses the problem of an automatic energy management system (EMS) design for demand response and then expands its focus to time series modeling, the main component of any predictive model and which can be utilized for any smart power grid application. An EMS utilizes dynamic electricity prices received from a utility company to automatically schedule electric loads over a customer’s planning horizon. In this work, we propose an optimal shrinking horizon model for EMSs based on user preferences. The most distinguishing characteristics of the proposed model are its simplicity, generality, comprehensibility, and ease of implementation. The problem of an electricity price modeling for EMSs is then addressed and a price modeling methodology is proposed based on the information criteria approach that is well suited to the specific requirements of EMSs. Later, the problem of time series modeling is investigated in detail and a unique neurodynamic predictor network based on dynamic artificial neural networks is presented for time series modeling and prediction. The proposed model is an optimally structured recurrent neural network that is well suited to address the nonstationarity, nonlinearity, high volatility, and time-varying nature of a complex time series. Comparison of prediction results generated by the proposed model with recent approaches used for time series prediction demonstrates a significant improvement in prediction accuracy by the proposed model.^
Fakhrazari, Amin, "Predictive analytics in smart power systems" (2015). ETD collection for University of Nebraska - Lincoln. AAI3717953.