Non-iotrusive Load Decomposition Based On Matrix Sparsity

The energy shortage problem has seriously hindered the sound development of China's economy and society,and power production is the main energy consuming field.In order to effectively improve energy efficiency and reduce energy consumption,the development of smart grid is the future goal and direction of China's power grid.Smart usement of electricity is a key component of the smart grid.It is built around the concept of intelligent service,which not only can greatly meet the power needs of users,but also optimize resource allocation.Smart usement of electricity makes users more closely connected with the power system,which helps to promote users' participation in the operation and management of the power system.By acquiring the user's electricity consumption information through intelligent power,the power supply company can understand the electricity consumption rules of the entire region,and then develop a more economical and energy-saving power generation,transmission and distribution scheme.In addition,users can consciously reduce unnecessary energy consumption after obtaining their own detailed information on electricity consumption,which is conducive to improving users'awareness of energy saving and emission reduction.Load decomposition is the main way to provide detailed electricity information,so it has become an important part of the development of intelligent electricity.The load resolver should be accurate,inexpensive,and low power.This paper presents a non-intrusive load decomposition method for home users according to the actual demand.The method is based on hidden Markov model and Viterbi algorithm.The hidden markov model can be created by collecting and analyzing the power consumption data of users to obtain the power consumption law of users.Since the matrix in the model has strong sparsity,the data compression algorithm is used to reduce the amount of data,and the corresponding data query algorithm is used to read the data required for the calculation from the compressed data.In addition,the Viterbi algorithm is improved in order to avoid unnecessary calculations in the decomposition process.The improved Viterbi algorithm can further improve the computational efficiency and reduce the energy consumption of the resolver.Finally,the probability of the combined state at the current moment and the accuracy of estimating the total power are used to infer the operating state of each load in the family.In evaluating the performance of the resolver,current evaluation methods are prone to problems of false high recognition accuracy.This paper takes the load state on-off event as the entry point,and evaluates the false high degree of recognition accuracy by measuring whether the resolver can accurately track the state on-off event.This method contributes to a more comprehensive analysis of the characteristics and scope of the decomposition algorithm.