Simulation Data Generation Method Based on Improved Cross-entropy Algorithm for the Prediction of User Electricity Consumption

Abstract

Aiming at the problem that the virtual energy meter needs to generate simulated power data in real time during the hybrid simulation of integrated energy measurement, this paper proposes a method for generating residential electricity simulation data based on a modified cross-entropy algorithm. On the basis of using non-intrusive load monitoring to decompose the actual electricity data load of resident users, this paper uses a modified cross-entropy algorithm to solve the model and generate load decomposition data in real time. Using the obtained decomposed data, this paper predicts the working state of the user's appliances, and uses the autoregressive time series model to generate simulated electrical data according to the electrical parameters and usage rules of the appliances. This paper improves the speed and accuracy of data generation while ensuring the authenticity of the generated data, thereby forming an automatic generation method for the electric energy measurement simulation data set for digital energy meters.