Load Decomposition Method Based on Improved Dpc-Igwo-Elman
Aiming at the problems of single load characteristics and low decomposition accuracy in existing decomposition methods,a non-intrusive load decomposition method combining the improved density peak clustering algorithm and Elman neural network optimized by the improved gray wolf optimization algo-rithm was proposed.Firstly,the calculation method of local density was improved for the lack of adaptive ability of Clustering by fast search and find of density peaks(DPC)when dealing with complex data sets,and the improved DPC was applied to the clustering analysis of electrical load data,then the working sta-tus labels of electrical appliances were obtained and coded.Subsequently,Elman neural network was used to construct the decomposition model and improved grey wolf optimizer(IGWO)was applied to optimize the network parameters.Finally,according to the network output code,the working state labels of the e-lectrical appliance were obtained,and the active power was fitted according to the corresponding load char-acteristic information,then the load decomposition was completed.The test and experimental comparison on public data sets proved that the load identification accuracy and active power fitting effect of IGWO El-man model were better than other models.
non-intrusive load decompositiondensity peak clustering algorithmgrey wolf optimizerEl-man neural network