Design And Implementation Of Non-intrusive Load Identification System Based On Steady-state Characteristics

Electric energy is an important energy carrier in the digital information society.It is cleaner,more efficient and more convenient secondary energy converted by other forms of energy by the power sector.Since the second industrial revolution,electric energy has always been supporting the rapid improvement of the world's scientific technological level and the prosperity of the world economy.Electric energy is widely used in various fields related to industrial production and people's lives.It has become an indispensable“energy element”for society.However,as a secondary energy that is most important to the industrial society,electric energy is also facing various problems like shortages and waste.In order to use electric energy more rationally,it is necessary to perform load monitoring and identification to the area electricity use.The traditional method of load monitoring and identification obtains the electrical behavior of each electrical device through the sensors and other hardware devices on each monitored load.This invasive monitoring method requires a lot of time and money for installation and maintenance,nor can it meet the needs of an evolving power system.On the contrary,non-intrusive load identification?NILM?can effectively solve the problem of inadequacy of adaptability caused by intrusive identification.To solve the problem of low accuracy and low efficiency for electrical equipment by non-intrusive load identification,this paper proposes a comprehensive HMM?Hidden Markov chain mode?with non-intrusive load identification based on L1 norm.This method improves the identification accuracy and efficiency of the electrical equipment effectively.By establishing a characteristic database with characteristic parameters including the steady-state harmonic current and steady state active power,this method obtains an accurate steady-state state segment by edge detection through iteratively computing with L1 norm minimization algorithm.Then the trained Markov chain model was used to correct it,which effectively solved the defects of shortcoming of stability and recognition performance for traditional pattern recognition algorithms.Through simulation and experimental analysis,the proposed method and the traditional identification method is analyzed and compared.The simulation results show that this method can effectively improve the accuracy and efficiency of the NILM identification system.The research work done in this paper includes the following three parts:?1?Analyze the theoretical knowledge related to non-intrusive load identification,including the analysis and selection of power load feature quantities;the edge detection algorithm used to obtain a stable and steady state segment,and the correlation of the establishment of the feature database analysis.?2?Focusing on the in-depth analysis of the load current harmonic characteristics,the non-intrusive load identification core algorithm is designed in detail using the harmonic parameters to establish a preliminary sparse domain identification model.Various norm solution methods are analyzed and the ₁l norm identification algorithm is proposed.By analyzing and comparing the characteristic parameters of the electric load and combining the state time parameters of the electric load,this paper completes the comprehensive identification of the working state of each electric equipment combining with the improved hidden Markov model.?3?This paper completes the overall platform design for the non-intrusive load identification model system.Based on the advanced RT-Lab experimental platform,simulation verification of the proposed algorithm was performed;through the establishment of peripheral acquisition and filtering circuits,a non-intrusive load identification real-time monitoring platform was constructed to perform real-time monitoring and identification of the electricity load in the measured area.And horizontal comparison and analysis of the similar algorithms are carried out.