抄録
The paper demonstrates that normal power system operating data may be used to determine the power-frequency characteristics in power system. Data are processed as random signals using spectral analysis techniques and the results estimate the generator and load power-frequency characteristics as transfer functions. The generator characteristics Kg is determined as a transfer function of the system in which input signals are frequency fluctuation ΔF and generator power setting signal ΔPg and output signal is generator power output ΔPe. For estimating the load characteristics Kl, there is a difficulty. There is a relation of ΔPl=ΔLo+KlΔF, between actual load ΔPl, ideal load under the state of nominal frequency ΔLo and fluctuation of system frequency ΔF. While, because it is impossible to measure ΔLo directly, ΔLo and KlΔF can not be separated. As to a partial system, when its capacity is small enough compared with the total system capacity, the contribution of ΔLo in a partial system to ΔF of the total system may be ignored, that is, ΔF and ΔLo are considered as independent phenomena. So, the cross-spectrum of ΔF and ΔLo decreases to zero according to the increase of data in number and Kl of the partial system can be estimated.
