Electric Power Load Forecasting Based on TPE-optimized Deep Forest Model
A deep forest model based on TPE tree structure optimization algorithm is proposed to predict the electric power load of three regions in a city.First,the data was cleaned to address the issues of data redundancy and large fluctuations in data characteristics.Secondly,according to the spatial-temporal correlation,the improved K-means++clustering method is used to classify the data of different regions with similar weather conditions.Fi-nally,the deep forest model optimized by TPE algorithm is used to predict the trend of electric power load of differ-ent types in the same area.The experimental results show that compared with the deep forest model without TPE al-gorithm optimization,the proposed prediction model has better performance,and it is also superior to the single ma-chine learning algorithm model and a simple integration between models,showing higher fitting degree and lower er-ror.
meteorological factortree structure optimization algorithmdeep forestelectric power load
NETL
NSTL
万方数据
