Parameter estimation of dynamic load model using field measurement data performed by OLTC operation

Abstract

An accurate representation of the dynamic behavior of an electrical load is very important for the analysis of power system operation regarding its stability characteristics, especially for voltage stability analysis. This paper presents the practical experience of dynamic load modeling using the real system measurement data. In order to capture the dynamic behavior of the power system on a load bus, a perturbation was introduced into the system by adjusting the transformer OLTC (on-load tap changer). That caused voltage changes in the network, and the resultant dynamic system response could be recorded by the power system monitoring devices. Before starting the load modeling process, the measurement data were preprocessed by a proposed DWT-based algorithm (Discrete Wavelet Transformation) to suppress the signal noises of the measurement data and, moreover, to make it more feasible to estimate the load model parameters. By performing the parameter optimization process, the parameters were then identified with the measurement data by NONLSQ technique (non-linear least square)

Estado del Arte y Tendencias en el Modelamiento de Carga

“…En la literatura se han utilizado una gran cantidad de técnicas y metodologías de optimización. Algunas de las más utilizadas son: mínimos cuadrados, mínimos cuadrados no lineales y similares [23]- [31]; Levenberg-Marquardt [32], [33]; algoritmos genéticos [34]- [38]; optimización por enjambre de partículas [39], [40]; redes neuronales artificiales [41]- [43]; filtro de Kalman [44]; ajuste vectorial [45]; máquinas de soporte vectorial [46] y; árbol de decisiones [47].…”

“…La estimación del modelo dinámico ERL con registros de eventos ha sido estudiado en [26], [31], [33], [44], [58]. Las técnicas utilizadas son: mínimos cuadrados, mínimos cuadrados no lineales [26], [31], [58], "Levenberg -Marquardt" [33] y, "Unscented Kalman Filter" [44].…”

“…La estimación del modelo dinámico ERL con registros de eventos ha sido estudiado en [26], [31], [33], [44], [58]. Las técnicas utilizadas son: mínimos cuadrados, mínimos cuadrados no lineales [26], [31], [58], "Levenberg -Marquardt" [33] y, "Unscented Kalman Filter" [44]. La precisión alcanzada es satisfactoria, salvo en [31] que resalta que el método de mínimos cuadrados tiene limitaciones para estimar el modelo ERL de potencia reactiva.…”

Enhanced Contingency Analysis—A Power System Operator Tool

“…In [52] the authors demonstrated the procedure of dynamic load modeling by using real-world measurements. To capture, i.e., record load dynamics, voltage perturbations were introduced by altering tap settings on the transformer ULTC, thus changing the voltage on the secondary (load) side.…”

Online PMU-Based Method for Estimating Dynamic Load Parameters in Ambient Conditions

“…The weighted least squares method has been applied in [5] to derive the load model. In [6], the nonlinear least square (NLS) technique is used to identify the parameters of a dynamic recovery load in the presence of voltage changes due to load tap changer operation. An improved NLS method is proposed in [7], however, an initial guess on parameter values is required for the estimation.…”