Research On Non-Intrusive Industrial Load Decomposition Algorithm Based On Steady State Characteristics

Non-intrusive load monitoring(NILM)technology refers to the electrical signals collected at the power inlet combined with a series of algorithms to analyze the working conditions of each electrical load within the system.NILM plays an important role in regulating the electricity consumption behavior of production enterprises and in the process of grid dispatching.Traditional non-intrusive load decomposition algorithm,the decomposition process includes data collection,data preprocessing,event detection,feature extraction and load identification,each process has its corresponding algorithm implementation.However,the existing algorithms for NILM have the problems of lack of rationality in load feature selection,single application scenario,complex decomposition process and low decomposition accuracy.This paper proposes an improved load decomposition model based on the combination of correlation analysis and fully connected neural network.By calculating the Pearson correlation coefficient between each load characteristic and the operating state of the electrical equipment,the load characteristics are selected to improve the accuracy of load decomposition.The fully-connected neural network model is adopted,and the load feature with a high correlation coefficient with the load state is used as the input of the network,and the working state of the load is used as the output of the network,which saves the steps of event detection and feature extraction in the traditional NILM decomposition process.using a deep fully connected network for direct load identification.The improved model has the advantages of simplifying the NILM decomposition process,improving the robustness of the model and the applicability of the model.In addition,considering the obvious time-series correlation between various production equipment in the production activities of industrial enterprises,a combination of recurrent neural network is proposed to identify the load of industrial enterprises.The effectiveness of the algorithm is verified by a large number of simulation experiments on enterprise-level equipment data sets.The experimental results show that the algorithm has a high decomposition accuracy in the face of different industrial production equipment,which has certain significance for the production monitoring and management of industrial enterprises.