An approach to implement electricity metering in real-time using artificial neural networks

Abstract

As many utilities move toward deregulation, the research focus on spot pricing of electricity has led to the development of complex spot pricing-based electricity rate models. As research matures to implementation stages, approaches to meter the actual power consumption in real time are required. In this work, the authors model a real-time electric power metering approach based on neural networks. A carefully designed artificial neural network (ANN) is trained to recognize the complex optimal operating point of an all-thermal electricity generating utility. A real-time rate is allocated to each bus for a given power system's loading pattern and the recall process is instantaneous. The proposed approach is tested using a spot pricing model on five- and 14-bus electric power systems. Different loading levels are used for each bus.