Design Of Non-intrusive Load Monitoring System Based On Deep Learning

Non-intrusive load monitoring(NILM)is a non-contact monitoring technology based on power system.Compared with the traditional sensor-based load monitoring technology,non-intrusive load monitoring can realize the identification of power system load without installing equipment for power failure.This paper mainly combines deep learning technology to research and design the collection of load data,load identification method,and load monitoring system.The specific work is as follows:1)By analyzing the principle of event detection and filtering denoising,in order to solve the difficulty of multi-load modeling and the problem of easy false detection of traditional event detection methods,it is improved on the basis of CUSUM,combined with the characteristics of electrical quantity changes when household appliances are switched,The CUSUM algorithm and Gaussian filtering algorithm considering the data changes before and after the switching event are designed.2)To solve the problem of feature overlap caused by similar electrical characteristics in load identification,the power trajectory of the load is used as the load imprint.On the basis of convolutional neural networks,the K-means algorithm is combined to identify the load,and the effectiveness of the algorithm is verified based on actual test data.3)Aiming at the problem of low recognition rate and slow speed when running under multiple loads simultaneously,a DBN network identification method based on improved sparrow algorithm is proposed.The improved sparrow algorithm is used to optimize the parameters of each layer in the DBN network,improve the accuracy of the network in load identification,extract load data through a built experimental platform,verify the effectiveness of the algorithm,and compare traditional identification algorithms.4)Design the architecture and modules of the non-intrusive load monitoring system,and elaborate on the design of each module of the system in combination with the performance of the system and the functional requirements of users,including the data acquisition module and control module of the non-invasive monitoring system.module,cloud data module,etc.