Non-intrusive electrical fingerprint identification method based on GMM-UBM

Abstract

The invention discloses a non-intrusive electrical fingerprint identification method based on GMM-UBM, which is characterized by comprising the steps of construction and training of a general background model, construction and training of a Gaussian mixture model, extraction of electrical data features, acquisition of real-time electrical data and the like. According to the method, different states of different devices are identified, and after general background model training is completed, for any new target identification electric appliance, a high-accuracy identification effect on the electrical fingerprint of the electric appliance can be achieved only by using training of a very small amount of data training. Compared with the prior art, the method has the advantages that the change of the working state of the electrical equipment in the circuit is identified with high accuracy, so that the functions of monitoring the circuit, warning dangerous electricity utilization and the like are achieved, the method is simple and convenient, the requirement for training data volume is low, training is easy and efficient, the distinguishing difficulty of feature categories is greatly reduced, and the training efficiency is improved. The method has a wide application scene.

Description

technical field [0001] The invention relates to the technical field of electrical fingerprint identification, in particular to a non-invasive electrical fingerprint identification method based on a Gaussian mixture model-general background model. Background technique [0002] Just like everyone has unique fingerprint information, different electrical appliances have different electrical characteristics such as current and voltage when they are connected to the circuit to work due to different internal capacitance, inductance, resistance and other electrical components. These electrical characteristics are also different. The different electrical characteristics are the "electrical fingerprints" of each appliance. Through the method of non-intrusive monitoring, real-time monitoring of the electricity consumption in the circuit, identification and recording of the working conditions of the electrical appliances in the circuit, and even identification of dangerous electricity c...

AI Technical Summary

Problems solved by technology

This need for a large amount of preparatory data makes the use of the model much more difficult

On the contrary, if the data is insufficient, it will not only make it difficult to distinguish the feature categories, but also the accuracy of classification and discrimination will be low, which will greatly affect the safe use of electricity for electrical work and the application significance of real-time monitoring of circuits.

Images