Electric vehicle charging station demand prediction model deploying data slotting

Abstract

Accurate prediction of energy requirement at charging station is essential for optimizing infrastructure usage, ensuring grid stability, and minimizing operational cost. Literatures suggest deployment of machine learning techniques to forecast the station demand. One major challenge associated with development of machine learning models is the inherent uncertainty in electric vehicle charging behaviour that includes variations in charging patterns, user preferences, and vehicle types. The conventional pre-processing techniques fail to dislodge nonlinearities and highly random patterns that include very low or zero-charging. Employing such techniques affects the model’s forecast accuracy. This article performs data-slotting during pre-processing stage and then selects the best among 1-hour, 2-hour, 3-hour and 4-hour slots, to frame the feature vectors. The 4-hour data with minimum variance is suggested to frame the dataset. Four distinct datasets, comprising different combination of average and total demands as predictor and response respectively are considered. The created dataset is deployed in Random Forest, Categorical Boosting, Extreme Gradient Boosting and Light Gradient Boosting models. The article recommends Categorical Boosting Regression model with least mean absolute error, mean square error and root mean square error of 0.0726, 0.0112, and 0.1059 respectively. Furthermore, the use of feature vector comprising of aggregated load for prescribed slots and the response representing the aggregated demand is observed to provide the least prediction error by the suggested model. The suggested model fed by the proposed feature vector offers significant advantage to charging station operator by enhancing the operational efficiency while performing resource and cost management with strategic planning. Develop a suitable ML model to forecast energy demand of EVCS based on a prescribed feature vector deploying the following procedures: • Data collection of EVCS sourced from Perth & Kinross Council's EV charging stations for the period spanning from 01 September 2017 to 08 December 2019. • Preprocessing of the collected data including slotting into various time frames to address the data non-linearity and determination of best slot based on the variance. • Preparation of 4 distinct following dataset using slotted data from stage 2: i) Date, day of the week, slot number, average demand at same slot on previous day, average demand at same slot on two days prior and average demand at previous slot on the previous day are predictor variable. The output variable represents the average energy demand for a prescribed slot and date. ii) Predictor variables are the date, day of the week, slot number, total demand at same slot for the previous day, 2 days prior total demand at same slot, previous day previous slot total demand. The output variable indicates the total energy demand for a prescribed time slot and date. iii) The date, day of the week, slot number, average demand at the same slot on the previous day, average demand at the same slot for two days ahead, and the average demand at the preceding slot on the previous day as predictor variables. The response represents the total energy demand for a specified slot and date. iv) Predictor variables feature the date, day of the week, slot number, total demand at the same slot on the previous day, total demand at the same slot two days prior, and the total demand at the preceding slot on the previous day, where as the output variable represents the average energy demand for a specified slot and date. • Development of ML models such as RFR, CBR, XGBR and LGBMR algorithms deploying various dataset created in 3. • Performance analysis of each ML model under different case studies based on the performance indices, mean absolute error (MAE), mean square error (MSE) and root mean square error (RMSE). • Selection of the best EVCS energy forecast model among RFR, CBR, XGBR, and LGBMR algorithms and the most suitable dataset that provides accurate result. The selected model is expected to enhance the operational efficiency of charging station.

Topics & Concepts