Optimizing electric vehicle charging in distribution networks: A dynamic pricing approach using internet of things and Bi-directional LSTM model

Nobody wants to invest in worldwide transmission and distribution corridor upgrades. Thermal, solar, and other sources are added as needed. We're thrilled about using more power for air conditioning, heating, and Electric Vehicles (EVs). The fast rise in EVs has dramatically raised inhabitants' power usage. Furthermore, because EV charging times overlap with peak periods of customer Electric power Consumption (EC), disorganized charging of EVs would result in a utility system overload. Traditional control methods could be more reliable and must consider all EVs' uncertainty. The control could be more reliable because it's hard to determine how many EVs use Vehicle to Grid (V2G). This article presents Deep Learning (DL)-based Bidirectional Long Short Term (B-LSTM) models for predicting feeder-wise EC for distribution substations. It compares them with two more additional models using DL techniques. Dynamic pricing computation considers for the feeder-wise EC forecasts. The proposed Demand Response Program (DRP) contains the Day-ahead Dynamic electricity Pricing (DADP) to optimize the maximum demand at a peak time in a distribution substation. The next day electricity price is sent to the previous day; this will assist the end users are optimally scheduling their Sensitive Price Appliances (PSA) and EV charging. This can assist EV owners with varying time sensitivity in making optimal charging decisions. This DRP helps the scheduling process to postpone the transmission and distribution for a few years. •The dynamic price of energy is examined individually for each distribution line.•Deep learning methods predict how much electricity each feeder will use.•The peak-to-average ratio in the distribution network goes down when EVs are added.•IoT devices handle price-sensitive loads, like in EVs, when prices or demand are high.•A dynamic price for the day ahead was made using deep learning to predict how much electricity would be used.