Data-driven Power System Reliability Analysis Considering Uncertainty Of TCLs And Wind Power

Due to the low carbon emission and environment concerns,the development of clean energy such as wind power has drawn great attention and become a big trend in recent years.However,the intermittent and stochastic nature of wind power can have a negative impact on the power supply reliability of power systems.Recently,with the development of the Power Internet of Things and the power system marketization reform,the research and practice of demand response have been promoted greatly.Thermostatically controlled loads(TCLs)are one of the most important resources of demand response,which can significantly mitigate the power imbalance caused by wind power fluctuations.However uncertainty also exists in the response of TCLs.This paper measures the uncertainty of TCLs and wind power by data-driven algorithms,based on which the power system reliability is also analyzed.The main research content is as follows.(1)The reserve capacity evaluation method of aggregated TCLs is proposed.The operation mode of individual TCL is firstly analyzed and modelled based on the thermal and electrical model.Later considering the partial uncertainty of the model parameters and response characteristic of TCLs,the aggregation algorithm is proposed utilizing K-means clustering and Gaussian mixture model.Based on the aggregation algorithm the reserve capacity of aggregated TCLs is evaluated.(2)The wind speed probability distribution estimation method in multiple wind farms is proposed.Based on the framework of generative adversarial networks,the artificial neural networks with two-player game are trained utilizing historical wind power data.The probability distribution of wind speed in multiple wind farms as well as the spatial correlation among them are estimated without any prior knowledge.(3)The power system reliability analysis method is proposed considering the uncertainty of TCLs and wind power.Based on the uncertainty evaluation in both demand and supply sides proposed in the research above,the power system reliability analysis problem is reconstructed by data-driven method so that the Cross-entropy importance sampling algorithm can be utilized.Later with system state sampling,optimal load curtailment calculation and related reliability indices calculation,the reliability of power systems considering the uncertainty of TCLs and wind power is analyzed and evaluated with high efficiency.