Sensitivity Estimation And Its Application In Smart Distribution Networks With Diverse PMU Conditions

The high penetration and integration of large-scale renewable resources and flexible loads are changing the traditional operation mode of distribution networks with radial power flow.Smart distribution network(SDN)is introduced to address the issues in quality and reliability of power supply caused by the volatility of the renewable resources and flexible loads,which is also challenging the current measurement infrastructure.The application of advanced measurement devices,represented by the synchronized phasor measurement unit(PMU),will provide powerful data support and new decision-making method to the operation control and energy management of distribution networks.However,limited by the higher installation cost,distribution networks will vary in the measurement level when configuring and utilizing PMU.This thesis focus on voltage-to-power sensitivity estimation and voltage control methods of distribution networks with diverse PMU conditions.The main contributions of this thesis are summarized as follows:Voltage-to-power sensitivity estimation method is developed for totally PMUobservable distribution networks without relying on the model parameters.A robust estimation method of voltage-to-power sensitivity for distribution networks is proposed considering the measurement noise and bad data.A modified sparse-recovery method using the voltage relevance is proposed to exploit the sparsity of power flow Jacobian matrix,with which the required amount of measurement data is reduced.The estimation accuracy and computation speed have been also improved.The voltage-to-power sensitivity estimation method by simultaneously considering the sparsity of the Jacobian matrix and the robustness is further proposed and applied to the topology identification of distribution network.Optimal PMU placement method for distribution networks with incomplete model parameters and various conventional measurements is proposed.Topology changes,unknown line parameters,zero injections,conventional measurements,existing PMUs,and their relevance are all considered in the model.Generalized optimal PMU placement model based on integer linear programming(ILP)is constructed.The observability of the distribution network is ensured under all of the possible operation topologies and the required PMU number is also reduced.Network equivalent method for voltage-to-power sensitivity estimation of PMUunobservable distribution networks using PMU measurements is proposed.PMU-based equivalent model of the distribution network is proposed to simplify the PMUunobservable distribution network into an observable network without relying on model parameters.A modified Kalman filter is exploited to guarantee the accuracy and robustness of the equivalent parameter estimation.With the equivalent parameter estimated,the voltage-to-power sensitivity of the simplified distribution network is finally obtained with high precision even under measurement noise and bad data conditions.Augmented sensitivity estimation based voltage control methods are presented for PMU-unobservable distribution networks.Augmented nonlinear fitting model of voltage-to-power sensitivity is proposed to address the inaccuracy caused by the operation status change and nonlinearity of the voltage-to-power relation.Sensitivity parameters are estimated using PMU measurements.Measurement-based voltage control methods are proposed to boost the computation speed of solving the voltage problem and to reduce the requirement of the measurements.The adverse effects of the incomplete and inaccuracy line parameters are also avoided.