Research And Application Of Non-invasive Load Monitoring Method Based On Residual Network

Along with the development of industrial Internet,the proportion of industrial energy consumption is increasing,and industrial production equipment is gradually exposed to great danger on the Internet.How to reasonably plan the use of electrical energy,avoid unnecessary energy waste,save energy,and how to ensure the safety of production equipment has become an important issue for the sustainable development of the current society.Non-intrusive load monitoring technology can efficiently and accurately grasp the energy consumption of power-using equipment and equipment operation status,and has the advantages of convenient information collection,simple installation,easy maintenance and low equipment costAt present,although non-intrusive load monitoring has achieved certain results in household electricity consumption,it has not yet been promoted in the industrial field,and the accuracy of load decomposition needs to be further improved.In this paper,we address these problems by collecting industrial equipment loads and using an improved residual network structure to perform non-intrusive load decomposition on industrial loads,which fills the gap of load decomposition on industrial loads and applies it to the problem of safety of industrial equipment.The specific work is as follows:(1)In this paper,a load data acquisition system is built to collect data from industrial production equipment through a data acquisition module,and the collected data are stored in a My SQL database,from which the extracted data are used as experimental data for subsequent experiments.The load characteristics of the industrial production equipment were also analyzed.(2)The extraction performance of load activation is largely improved by the power-time load activation extraction method.The residual network model proposed in this paper is used to train the non-intrusive load decomposition model on five industrial production equipment in the load dataset of industrial production equipment proposed in this paper,and the validation results are derived on the test set,finally with the accuracy,recall and F1 score higher than 87% for the checker,and the accuracy,recall and F1 score higher than The results make the load decomposition model achieve a good decomposition effect.(3)For the current security problems faced by the industrial Internet,industrial production enterprise equipment exposed to the Internet faces great hazards,this paper proposes to compare the similarity of power curves by calculating the Frey interval distance to judge the identity of the equipment,and after the arithmetic experiment verification,the identity of the production equipment can be identified very effectively through the method of this paper,and the identity of the experimental equipment is verified.