Large-Scale Metamodeling and Validation of Power Systems

• Published in: International Review on Modelling and Simulations Vol. 5; no. 2; pp. 546 - 560
• Main Authors: Shan, S, Zhang, W, Wang, G G
• Format: Journal Article
• Language: English
• Published: 01.04.2012
• Subjects: Computation; Computer simulation; Contingency; Disturbances; Electric power generation; Electric power systems; Representations
• ISSN: 1974-9821

Modeling the available transfer capacity of power systems is arduous due to contingencies and commonly performed by simulation. Meanwhile, both metamodeling and validating a large-scale computationally expensive simulation system is challenging due to "curse of dimensionality." This paper applies the recently developed Radial Basis Function-High Dimensional Model Representation (RBF-HDMR) to model the transfer capability of a 50-variable electric power system subject to contingency disturbances. The simulation-based transfer capability analysis has a combination of high-dimensional, computationally-expensive, and black-box features, which is called a HEB problem. The obtained RBF-HDMR is useful for power system analysis and operation planning. In addition, this work proposes two complementary sets of modified performance metrics for validation of metamodeling techniques in general. The proposed metrics overcome the deficiencies of some commonly-used metrics, and also the potential risk of using an individual or an arbitrary set of performance metrics, which is the common practice in the field of metamodeling. The accuracy and efficiency of RBF-HDMR, as well as the application of the two sets of performance metrics, are demonstrated through three practical cases under different operating scenarios for the given power system. The results show that RBF-HDMR is effective and efficient in modeling this large-scale HEB problem, and moreover, the two sets of performance metrics overcome the potential misjudgment might have been brought by using individual metrics.