Sliding Time Window Electricity Consumption Optimization Algorithm for Communities in the Context of Big Data Processing

Abstract

Big data frameworks enable companies from various fields to build models that allow them to increase profit margins by improving decision making at different levels (middle management, senior management, and board) or by attempting to boost sales by customizing consumers' experience based on their history and feedback. Institutions and other entities also use big data coming from all kinds of sensors, data that can be used to detect, in real time or in retrospect, possible problems (e.g., frauds, malfunctions, and supply shortages), or to identify patterns and trends. In this paper, we organize large volumes of community electricity consumption data coming from smart meters, smart plugs, and other sensors, but also data regarding consumers' preferences in order to assist them to dynamically optimize their electricity consumption. In this regard, we develop a novel optimization approach that re-schedules every fifteen min the appliances for residential consumers to reduce both the consumption peaks and the payments at the community level. The consumers send their day-ahead schedule that is optimized and further implemented to some extent. Thus, we monitor the electricity consumption via sensors and smart meters and dynamically adjust the schedule in case the real consumption deviates from the optimized plan, considering appliances constraints and consumers' preferences. Every fifteen min, the algorithm evaluates the differences between the optimized schedule and the actual consumption and controls the operation of the interruptible appliances to stick with the day-ahead schedule as much as possible.

Keywords

• Day-ahead electricity and real-time consumption optimization algorithms
• big data
• stream processing
• smart meters
• sensors