Electricity is a symbol of modern industrial civilization,and a necessity of social production and life.The share of electric energy in China’s energy structure is increasing year by year.Therefore,the rational and standardized use of power energy has become the focus of social attention.In the power system,non-intrusive load decomposition and monitoring as one of the important components of the intelligent power technology system.It can not only help to solve the increasingly serious energy problems and increase economic benefits,but also monitor the dangerous electricity use behaviors in real time to avoid the occurrence of dangerous accidents.This is of great significance to individuals,power companies and other parties.However,the current research in this field at home and abroad is more about demonstrating the reliability of various load identification methods,less attention has been paid to practical application.At the same time,the nowadays smart meters can only read the total amount of electricity used in the building,but cannot analyze the internal load components of users in-depth,and cannot accurately measure the electricity consumption of single device,so it is impossible to improve the electricity information collection system and smart electricity system,let alone to realize the two-way interaction of intelligent electricity consumption.This paper effectively uses the existing smart meter hardware foundation,combined with pattern recognition,data analysis and other methods to form a complete set of “smart meter + load identification + energy consumption monitoring” power consumption submetering system.This system does not require additional cost of hardware,and meets the installation requirements of no power failure and no home entry.A meter on the home bus can realize real-time processing and analysis of users’ load data,and provide a detailed electricity bill,electricity consultant,and electricity safety.The specific works are as follow.1.The domestic and international research status of the three main parts of nonintrusive power consumption sub-metering(non-intrusive load identification,power submetering,and smart meter billing system)are described and analyzed.2.System architecture design and requirement analysis.Technical framework is designed based on the overall structure of the system.The requirements in functions and performance for the system are described and key supporting technologies of the system are proposed.3.It achieves the non-intrusive load identification based on smart meters.The main content includes load identification feature selection,collection devices,communication links and communication modules.The wavelet denoising analysis of current is performed.The Fast Fourier transform is used for harmonic calculation.Realize load decomposition and feature extraction based on harmonic additivity.The event detection method based on the harmonic difference between the front and back steady-state current is established.The multi-class support vector machine model is established.4.The sub-metering strategy is designed based on load identification records to solve the running time of every single electrical appliance in 5 categories,and then to obtain the online status of electrical appliances at each reporting time point.To improve the accuracy of sub-item measurement,the total active power reported by the second level of the meter end is decomposed proportionally to obtain the discrete point diagram of the actual operating power of various single loads,and then the power curve function is obtained by spline interpolation,and then the energy consumption of various single loads is obtained by decomposition.5.According to the key system performance proposed by the demand analysis,the corresponding solutions are put forward,which ultimately improve the system performance.6.Experiments verify the measurement accuracy of the sub-metering system,and perform error analysis and data visualization. |