A Sensitivity Matrix-Based Temperature-Augmented Probabilistic Load Flow Study

A spatiotemporal probabilistic model‐based temperature‐augmented probabilistic load flow considering PV generations

“…The primary requirements for PSSA include (1) system model development, (2) forecasting model for input uncertainties and correlations, and (3) application of a suitable solution method. The system model that considers the electro‐thermal coupling effect of transmission branches is more realistic for PSSA . Further, Gaussian mixture approximation method or its improved versions are more accurate in approximating multimodal probability distributions of power system variables as compared with other methods.…”

“…In view of the fact that the other probabilistic load flow (PLF) inputs such as load powers and ambient temperatures, the time series have complex pattern, the extraction of their concealed seasonalities is strenuous, and the development of a generic model with regressors possessing a theoretically formulated physical relevance to the predictable varying periodic patterns as in Prusty and Jena is difficult. On this line, several preprocessing methods as applied to load power and ambient temperature is witnessed. In Do Coutto Filho et al, stationarity is ensured by fitting a curve using a set of mathematical functions.…”

“…In Prusty and Jena, a regression model consisting of time trend and Fourier terms, respectively, of order 3 and 5 are used to trace the periodic variation and then is removed from the raw data. Further, the frequency component of one cycle/year is filtered out from the ambient temperature data in Prusty and Jena, while in Prusty and Jena, a regression model with second‐order time trend terms and first‐order Fourier terms is proposed for preprocessing. The keen observations from the residuals after preprocessing in Prusty and Jena indicate that although stationarity in mean is attained, there is a volatility effect on the variance of the residual.…”

“…The application of any of the preprocessing methods mentioned above perhaps is inadequate owing to their incapability in predicting the varying periodic patterns in accordance with the location. Nevertheless, a regression‐based method stands out with a befitting performance as it extracts the true contribution of a set of chosen regressors by operating holistically on a complete dataset. However, such a model that amalgamates a single order or a randomly selected order of Fourier terms with time trend terms as in Prusty and Jena often fails to capture the real‐time complex periodic patterns; as they either underfit or overfit the model.…”

Quantile regression averaging‐based probabilistic forecasting of daily ambient temperature

“…Further, it is vital to incorporate the time of the year into the modeling. 4,11 There is a plethora of research attention devoted to the probabilistic modeling of daily time stepped ambient temperature. Daily ambient-adjusted DLR used by utilities considers forecasted daily peak ambient temperature while leaving the wind speed and maximum solar heating used for SLR.…”