Non-invasive load identification method based on MLCDTL

Abstract

The invention discloses a non-invasive load identification method based on an MLCDTL (Multi-Label Consistent Deep Dictionary Learning). The non-invasive load identification method based on the MLCDTLcomprises the following steps: 1) data processing; 2) event detection; 3) feature extraction; 4) load identification. Firstly, a peak in a power sampling sequence is filtered by adopting median filtering, so that pulse or fluctuation in an equipment operation process is favorably inhibited; then, according to the filtered power time sequence, the starting time and the ending time of the transientprocess are detected through an improved variable-length sliding window CUSUM bidirectional detection algorithm; then, according to the starting time and the ending time of the transient process, fastFourier transform (FFT) is conducted on the current sampling data in the steady state, feature vectors composed of current fundamental waves and harmonic components of all orders are obtained, and all the feature vectors are combined together to form a feature vector matrix; and finally, based on an MLCDTL algorithm, load identification is converted into a multi-label classification problem, anda target function is solved through iteration to realize non-intrusive load identification. The method can be comprehensive and objective, accords with actual conditions, and is higher in credibility.

Description

technical field [0001] The present invention relates to the field of non-intrusive load monitoring (Non-intrusive Load Monitoring, NILM), in particular to a non-intrusive load identification method based on MLCDTL (Multi-label Consistent Deep Dictionary Learning, that is, multi-label consistent deep conversion learning) . Background technique [0002] Non-intrusive load monitoring (Non-intrusive Load Monitoring, NILM) technology can obtain refined user internal load category and usage status data by decomposing and identifying the total load data of users, which is an effective solution to the problem of intelligent power load monitoring. way. Non-intrusive load monitoring does not require the installation of load monitoring devices on the user side or substantial upgrades to smart meters. It relies on the data acquisition devices and communication networks of the existing power consumption information acquisition system, and uses advanced data communication technology to o...

Images