Developing Energy Forecasting Tools in Power Systems: Application to Microgrids
Abstract
In recent years, distributed energy resources and microgirds have attracted a great deal of interests in the power industry. The development of microgrids has required engineers and operators to enhance the efficiency and the energy management of such small-scale power systems. To do so, energy forecasting plays a key role in their optimal operation. In particular, Short-termWind Forecasting (STWF), Short-term Load Forecasting (STLF) and Short-term Price Forecasting (STPF) are important tools for reliable operation scheduling of grid-connected microgrids with renewable energy sources (e.g., wind). The generated energy forecasts are used in an optimization platform to ensure the most economical operation for microgrids as the end goal of this thesis.
The main focus of this thesis is the development of forecasting models that are tailored for the application to grid-connected microgrids. An STWF model is developed based on artificial intelligence and an evolutionary algorithm to provide wind forecasts. This model can be applied to generate wind predictions at the power system, wind farm and/or wind turbine levels. By statistically analyzing the behavior of electricity consumption at campus/building levels, an STLF methodology is developed based on neural networks to provide satisfactory forecasts for volatile electricity loads in microgrids. Further, an STPF is designed to improve the economics of grid-connected microgrids by taking advantage of energy arbitrage opportunities with the grid. It is noted that the microgrid is assumed to be large enough to have transactions based on market prices.
Numerical results in Chapters 2, 3, 4 and 5 of this thesis are provided based on Alberta, Ontario, British Columbia, California, Texas and NewYork power systems, and two campuses. The simulations show the effectiveness of the proposed neural networks for STWF and STLF. The statistical and economic evaluations show the satisfactory performance of the developed STPF model in scheduling a storage system in a microgrid. Moreover, deterministic and probabilistic optimization platforms are developed for the optimal operation of microgrids, which can help the operator apply the most effective approach under different scenarios of generation and market conditions.
Description
Keywords
Artificial Intelligence, Energy, Engineering, Engineering--Electronics and Electrical, Engineering--Environmental
Citation
Chitsaz, H. (2017). Developing Energy Forecasting Tools in Power Systems: Application to Microgrids (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/25623