Assessment of Demand Side New Energy Consumption Capacity based on Convolutional Neural Network

Demand-side new energy consumption evaluation is crucial for smart grid planning but difficult because it is highly variable and has complicated consumption patterns. Traditional forecasting methods are usually unable to capture spatial and temporal correlations. This paper introduces a Convolutional Neural Network (CNN) to analyse the smart meter data to forecast consumption capacity at the household and industrial levels. CNN’s hierarchical architecture allows multivariate time-series data such as energy consumption, temperature, and time-of-day to be extracted for feature extraction. User profiles are classified and load patterns are predicted well. Training involves historic energy data gathered from different systems. Accuracy of 98.2 %, 98.2 % Precision, and 98.2 % F1score are used as evaluation metrics. With respect to ML and conventional methods, CNN demonstrates an improvement in predictive accuracy. This solution enables more effective energy distribution and demand-side management and increases the reliability of smart grid infrastructure.