A Fog Based Customer Demand Forecasting (CDF) Strategy for Smart Grids Using IoT Technology

In this paper, a novel framework for effective energy management of residential customer is provided to reduce electricity consumption. Advanced Smart Grids (ASGs) can assist a variety of functions thanks to Internet of Things (IoT). These smart devices generate big data, which can be uploaded to the cloud for additional analysis. Fog computing tier operates as a bridge between the IoT devices integrated in Smart Electrical Grid(SEG) and the cloud to overcome cloud issues. Based on the indicated three-tier design, a novel Customer Demand Forecasting (CDF) strategy has been introduced. CDF strategy consists of (i) Feature Selection (FS) stage and (ii) Demand Forecasting (DF) stage. FS stage identifying the most important features that allow the demand forecasting model to produce quick and accurate results. A Hybrid Feature Selection (HFS) approach is used to pick the effective features, which integrates evidence from two feature selectors;(i) Information Gain (IG) as a filter method and (ii) Binary Particle Swarm (BPS)optimization is used as a wrapper method. Then, an Improved KN³ B (IKN³ B) predictor has been used in DF stage trying to provide accurate demand forecasts based on the selected subset of features from the previous stage. In fact, IKN³ B combines both K-Nearest Neighbors (KNN) classifier and Naïve Bayes (NB)classifier and then it improves the characteristics of them to provide the best demand forecasts as possible. Based on experimental results, It is conduced that CDF strategy is demonstrated to have a positive influence on system reliability, resilience, and stability by introducing accurate demand predictions.