Pillar Industry Judgment Based On Multi-source Power Data And Its Energy Efficiency Prediction

With the deep integration of new information technologies such as the Internet of Things and big data with the power system,the power system will enter a smart era.Meanwhile,more and more attention paid to energy conservation and sustainable development of industry in China.Reasonable judgement of pillar industries based on multi-source power data is of great significance for pillar industries and even the entire region’s energy consumption,reflect abnormal energy efficiency,and energy efficiency optimization.Firstly,this paper analyzes and extracts several indicators from the multi-source power data,and classifies the indicators into economic contribution,energy consumption,energy consumption structure,energy saving,environmental protection and water saving to discuss.And these indicators are divided into extremely large indicators,extremely small indicators and interval indicators.And the electricity consumption indicator in the multi-source power indicators is selected to analyze the optimal interval of the interval type index and its solution process.Then data mining technology is applied to a large number of multi-source power data.Aiming at the characteristic that the clustering algorithm can divide the data into different data clusters according to a certain similarity metric,this paper proposes to cluster all industries by clustering method to classify industries with similar pillar properties into the same cluster.Considering the accuracy of a single clustering algorithm,using the idea of integrated learning method for reference,the integrated clustering method is proposed to judge the pillar industries.Using the secondary learner DBSCAN density clustering to integrate the judgment results of each base learner(common clustering algorithm).Meanwhile,we can get industrial clustering results with higher quality and robustness.Then,the entropy weight method is used to give weight to all industry characteristic indicators,and the characteristic indicator data of each cluster cluster industry is combined to determine the pillar industry clusters.Display the clustering results through data visualization technologyFinally,an energy efficiency model is established,and it is proposed to reflect the change of industrial energy efficiency by predicting the value of the industrial energy efficiency model and the value of the past same period,reflecting the level of industrial energy efficiency.Further more,the LSTM time series neural network model is used to predict the energy efficiency of pillar industries,and the industrial energy efficiency model sequence data is used as the input of the LSTM time series neural network to predict the industry’s future energy efficiency model data.Provide scientific basis for regional typical industries and important industries to save energy.