Integrated Energy Use Sensingtechnology For Energy Internet At Parklevel

The integrated energy system of the park based on the Energy Internet mainly includes various energy systems such as electricity,gas,and thermal energy.Aiming at the integrated energy consumption perception of the energy Internet,it mainly studies the collection,correction and transmission of energy consumption data,and then predicts the user’s energy consumption to realize the optimization of the user’s energy consumption.Accurate and real-time perception of load data is fundamental.The accuracy of load data will directly affect the planning,operation and forecasting of the integrated energy system in the region.Therefore,the research of integrated energy load energy consumption perception is of great significance.The integrated energy consumption perception is mainly composed of the optimized layout of the user-side energy consumption data collection and monitoring points,the correction and denoising of the abnormal load data,and the short-term load prediction.The layout of monitoring points and the correction and denoising of abnormal load data are all to improve the quality of the collected load data,to ensure that the historical load data used in load forecasting is more reliable,and to reduce the error of load forecasting results.In terms of collecting monitoring points,this thesis will study the location,quantity and effect of monitoring points;in collecting abnormal data in the monitored load data,this paper will identify,correct and de-noise abnormal data loads.Research;in terms of multi-energy load forecasting,this thesis uses the deep learning method for load forecasting,and obtains better results.The main research contents of this thesis are:1.Analyzed the significance of the research background of this subject,and defined the main research direction and methods.2.This paper analyzes and compares the optimal layout methods of user-side energy consumption data collection and monitoring,and finally determines the PUM as the monitoring equipment,and optimizes the layout of the monitoring points three times,so that the location of the monitoring points is better,and the number of monitoring equipment is more.less,the monitoring effect is better.3.Classify and identify abnormal load data collected by monitoring.For the abnormal data types,this thesis mainly uses the density estimation method to identify the abnormal data;for the distorted data types,the wavelet threshold method is mainly used for denoising.4.This thesis introduces and analyzes the basic theory of load forecasting,and then introduces the multivariate load forecasting method based on deep learning.According to the actual case analysis,the prediction effect of the shallow learning algorithm and the deep learning algorithm is compared,and the advantages of multi-task learning over single-task learning are also analyzed,and the prediction results meet the prediction error range.