Dynamic Prediction Algorithm of User Power Consumption Based on Key Points and Transfer Learning
Due to the time-varying characteristics of user power consumption data,the accuracy of user power consumption dy-namic prediction is low.Therefore,a user power consumption dynamic prediction algorithm based on key points and migration learning is designed.The transfer learning model is used to mine the data,and the key point method is used to calculate the co-sine similarity of the two data.After the calculation,the data are fused and denoised,and are processed by dimensionless.A variational modal is established,the collected signal is decomposed into several simple signal components of wave motion law.Combined with support vector regression method and PSO algorithm,the dynamic prediction of user power consumption is real-ized.The experimental results show that the proposed prediction method has high accuracy in the prediction of electric energy consumption of electric boiler,air conditioning,working and non-working days.
万方数据
维普
