Hyper-parameter optimization and post-processing method for non-intrusive load decomposition model

Abstract

The invention discloses a hyper-parameter optimization and post-processing method for a non-intrusive load decomposition model, and the method comprises the steps: S1, collecting the electrical operation data of a home bus and a target electric appliance, and enabling the electrical operation data to be used for building a data set of the model; s2, on the basis of a deep learning theory, constructing a non-intrusive load decomposition model based on a deep residual network for the target electric appliance; s3, performing optimization on hyper-parameters of the load decomposition model by using cluster Bayesian optimization; and S4, respectively establishing an optimal decomposition model for the target electric appliance, and training the trainable parameters in the model by using the training set, the verification set and the test set until convergence. According to the method, load decomposition can be realized, a Bayesian optimization method is introduced into load decomposition model hyper-parameter optimization, and the problems of poor effect and low efficiency caused by blind selection of hyper-parameters in a traditional method are solved; and the high efficiency of hyper-parameter optimization is realized through the search behavior of the group and information interaction in the group.

Description

technical field [0001] The invention relates to the field of non-invasive load decomposition, in particular to a hyperparameter optimization and post-processing method of a non-invasive load decomposition model. Background technique [0002] One of the urgent needs of smart grid and energy Internet is to obtain the power consumption data of individual appliances, so that users can understand the power consumption pattern of each appliance and reduce energy consumption accordingly. This is an important step towards a transparent and intelligent grid. The current measurement technology can only automatically read the total power consumption data, and it is difficult to further obtain the user's internal load information. Load splitting technology has become the main bottleneck in the development of smart grid. [0003] Non-intrusive load dis-aggregation (NILD) was first proposed by Professor Hart in the 1980s. It is a technology to estimate the power consumption of each elec...

AI Technical Summary

Problems solved by technology

Literature (Li Ruyi, Huang Mingshan, Zhou Dongguo, Zhou Hong, Hu Wenshan. Non-intrusive power load decomposition method based on particle swarm algorithm search[J]. Power System Protection and Control, 2016,44(08):30-36. ) adopts the method of combinatorial optimization, and uses particle swarms to search for the optimal matching of the harmonic current and power of each electrical appliance to the total harmonic current and total power to realize load decomposition; literature (Kolter J Z, Jaakkola T. Approximate inference in additive factorial HMMs with applicationto energy disaggregation[C].La Palma,Spain:Microtome Publishing,2012. The translation is Kolter J Z,Jaakkola T. Approximate inference of additive factorial HMM and its application in energy decomposition[C].La Palma,Spain:Microtome Publishing, 2012) Construct a hidden Markov model for load decomposition, but the performance of the algorithm will be affected when the number of electrical appliances increases; literature (Wu Xin, Han Xiao. Non-intrusive load decomposition algorithm for residential users based on signal sparse under

Images