Quantitative Uncertainty Analysis For Load Model Based On The Stochastic Response Surface Method

Dynamic simulation of power system is an important tool for power system dynamic stability analysis, network planning and operation, the reliability of which is directly related to safe and stable operation of power system. Mathematical models of power system include generator model, excitation system model, prime mover model and load model. The accuracy of these models and parameters is the basis of simulation reliability. However, it is necessary to notice that there always is difference between mathematical models and real power system. Among the models, with features of randomness and dispersity, load model brings lots of uncertainties into real system, such as approximation, algorithm error etc. This dissertation applies Stochastic Response Surface Method(SRSM) to quantitatively analyze uncertainties of power system influenced by fluctuant fault time and load model. The method solves limitations of traditional method and shows better robustness and curve fitting than recently proposed method, which is called Probabilistic Collocation Method. The case of IEEE39-bus system concludes that uncertainties from input parameters have great impacts on generator angle and bus voltage. Moreover, the uncertainty can be very different when input parameters have different distributions. To increase practicability of SRSM, the dissertation builds load model based on measured data of PMU in a real large-scale system at first, then detects key load areas and chooses uncertain parameters and their range using sensitivity analysis, analyzes generaor angles, bus voltage and active power uncertainties with SRSM under different operation conditions at last.