Non-invasive Real-time Electrical Identification Based On Sliding Window

Household electricity has become an important part of the high proportion of social electricity consumption.With the large-scale popularization of household appliances,saving electricity,safe electricity and other issues have been more and more attention.The existing load monitoring methods are all for off-line data,these methods may have some help with the analysis of the laws of using electricity.However,because of the lag of these methods,so many demands cannot be met,such as dynamic planning electricity demand analysis,the construction of smart home,parents planning the needs of adults’ entertainment time.Aiming at these needs,this paper presents a non-invasive real-time electrical identification method based on sliding window.Based on the existing methods and ideas of load decomposition,we choose the features model and recognition algorithms which are suitable for non-invasive real-time electrical identification and eliminate the ones which are not suitable.In this paper,we first use the transient events to construct the feature model for the real-time electrical identification,and then propose the traditional feature model based on the waveform trajectory and the improved sliding window-based feature model.For analyzing the performance and shortcomings of the models and algorithm,this paper uses the REDD database to carry out experiments.Furthermore,this paper compares the recognition results with other literature using REDD data,and confirms that the recognition effect of the model and algorithm used has been improved.By using the REDD database to verify the feasibility of the algorithm,this paper further builds the test system installed in the laboratory.We use this system to collect working electrical data of the laboratory electrical appliances and verify the long-term running performance of the algorithm in the real electricity environment.Based on the experimental results,it is shown that the non-invasive real-time electrical recognition algorithm based on sliding window can be applied with the real environment,and the recognition effect is ideal.