The Method Of Power Grid Load Modeling And Feature Extraction Based On Big Data

Today,with the rapid economic development,the urbanization process is also gradually accelerating.Various urban industrial parks and development zones are constantly emerging.The problem that comes with it is that the requirements for power quality are gradually increasing.Due to the launch of a new economic development strategy of green,low-carbon,environmentally friendly and renewable energy,the introduction of various green renewable energy sources has provided a variety of energy supply options for smart parks,and brought a profound impact on the future development of industrial park power distribution systems.change.The grid side and the user side need more in-depth information exchange,in which it is very important to identify and analyze user power consumption information.Therefore,this paper proposes a data processing method,namely IVMD,and uses this algorithm as a basis to implement grid load data modeling and feature extraction to prove the effectiveness of the proposed algorithm in data processing.First of all,in terms of load data preprocessing,because the data contains noise,it needs to be preprocessed for subsequent operations.The improved variational mode decomposition(VMD)proposed in this paper optimizes the needs of traditional VMD The problem of relying on subjective judgment of the input K value and the problem of edge effects affecting the accuracy of the reconstruction curve,and then using the time-frequency conversion function of IVMD to convert the grid load data into frequency domain signals,judge according to the signal frequency,remove the noise contained in it,improve The stability and practicability of the data provide data support for the subsequent steps and generate model training data.Secondly,in terms of load modeling,the IVMD algorithm set LSTM algorithm proposed in this paper is used to fit the collected measured data of the power grid to complete the modeling of the grid data,and the built load model is verified by comparison with the existing load modeling method.Accuracy.Finally,the IVMD algorithm proposed in this paper is also used to process the collected industrial park load data,and then use the Convolutional Neural Network(CNN)to extract the features of the preprocessed data,and use the simple cycle unit Simple Recurrent Unit(SRU)trains it to form a two-layer network training model,and compares with other feature extraction methods to verify the accuracy of the proposed algorithm in extracting data features.