Studies On Synthesis Load Modeling Based On Muti-source Measurements

Informatization, digitization, automation, interaction are the main features of smart grid and the goal for most grid. Many sorts of big data emerged with the development of smart grid and sensors and brought new opportunity for load modeling.First, data preprocessing is taken including outlier detection with Gaussian mixture model, the standardization formula amending, dimension reduction with locality preserving projections. The paper uses Gaussian mixture model to cluster the big data from SCADA system, and cluster again with strategy of different time span, finally find the time-varying of load. The AMI big data is used to execute energy disaggregation, and four kinds of load are identified so we can compile statistics of load component. The big data from WAMS aroused by disturb is used to identified the dynamic load.This paper uses three kinds of grid big data to perform the load modeling, and the main jobs are:(1)The paper proposes a novel energy disaggregation method based on multi-output extreme learning machine (MO-ELM) to analysis low-frequency monitoring data gathered by meters distributed in a building. The MO-ELM whose parameters of feature mapping need not be tuned and can be fixed once randomly generated requires fewer optimization constraints with the objective function and results in simpler implementation. The evaluation results using a dataset of power traces collected in real-world home setting shows that proposed method have a satisfied classification accuracy, training speed and generalization performance and less computational time. Statistics of classified load can provided the axact parameters to the synthesis load modeling.(2)The time-variant characteristic of load are studied with load curve from SCADA, Gaussian mixture model can cluster the curve by computing probability, adjust time segment will lead a different period of variation. Two time of clustering with different strategy lead to a time varying pattern, and give the advice of load model changes.(3)A load modeling method was proposed based on the Hardy space theory and Caratheodory-Fejer interpolation (CFI). The load model was mapped into linear load model set with prior information in Hardy space. Consistency problem between measured data and model set was formulated to a linear matrix inequality. Simulation results show that the variations of load composition in a certain scale have little effect on model parameters. Simulations using measured data from the phase measurement unit in a power station demonstrate the practicality and validation of the proposed method when exiting UBB error.(4) Locality preserving projections is proposed for dimensional reduction, the method is capable of handling grid data with high dimension, multimoding, and preserving information locality and important. Oulier detection is also implemented to remove the bad data.The muti-source measurements of power system contain valuble information. It is possiple to apply data mining and machine learning method to got the hiding pattern of synthesis load modeling.