Line Loss Prediction Method for Missing Dataset Based on Compressed Sensing

Abstract

The invention discloses a line loss prediction method under a missing data set based on compressed sensing. The invention provides a method for realizing power grid line loss prediction under a missing data set, which comprises the following steps of: clustering each characteristic quantity of historical load data by adopting a clustering method according to a principle that a contour coefficientis maximum according to three characteristic quantities of total load, a load rate and a peak-valley difference rate in historical data of a power grid; Arranging each type of historical line loss data into a time sequence matrix according to 24 integral points, and decomposing a line loss time sequence into intrinsic mode components; And selecting a proper dictionary matrix, and performing sparsetransformation on the modal componen; Performing data completion based on compressed sensing on the transformed sparse matrix; Performing inverse transformation of sparse transformation on the restored matrix to recover the complete component sequence in the time domain; And completing line loss prediction under the repair data set by using an elman neural network. The method can be used for complementing missing data of the power grid and predicting line loss of the power grid.

Description

technical field [0001] The invention relates to the technical field of power grid missing data recovery and line loss prediction, and in particular, relates to a line loss prediction method based on a compressed sensing missing data set. Background technique [0002] The line loss prediction provides a basis for finding out the problems in the power system structure and power consumption management, and then guides the loss reduction management work. The accuracy of line loss prediction is closely related to the effectiveness of loss reduction measures. In the modern power grid, the data collected by the smart meter is integrated on the information platform to form a real-time data set of power grid operation. When the smart meter has various software and hardware problems or the acquisition signal is suddenly interrupted during the data transmission process and the measurement is completed, the real-time data set of the power grid operation will become an incomplete data s...

AI Technical Summary

Problems solved by technology

Traditional missing data recovery methods, such as the difference method, have a relatively simple operation mechanism and a small amount of calculation. They take a short time and have good recovery effects when dealing with a small amount of missing data and data completion of small data sets. Due to its single operation mechanism in the data completion of the set, it is impossible to repair the missing data after analyzing the existing data, reflecting the changing trend of the missing data set

Images