Demonstrate the Predictive Model of Machine Learning in Smart Grids to Process Large Electricity Data According to Power Demand and Supply
Jitendra Managre, Namit Gupta · Procedia Computer Science · 2025
Smart grids are the improved and featured version of traditional power grids associated with the internet of things and machine learning adoption. Forecasting electric consumption using smart meter data is one of the most prevalent applications of machine learning. Detail examination of customers’ smart meter data is required to discover essential variables and the source of variation between appliance consumption levels and customer demand. With the ML adoption in smart grids, power load prediction has become possible for a particular client or sector. This paper presented an ML based prediction model of electricity load with time series data taken from the smart meter for understanding the power demand and supply for different consumers. In this regard, two machine learning models LSTM and CNN have been performed for prediction accuracy terms of performance improvement, and it has been found that CNN based model is more accurate than LSTM. The accuracy of the prediction of consumers’ electric consumption using CNN is almost 84%, which is significantly higher than previous research in the same field. The model’s performance is measured in terms of MSE which is the loss function for the predictive models. The performance of prediction by comparing the predicted data and original data is also demonstrated in this paper.