Research On Distribution Network State Estimation In Hybrid Measurement Environment

State estimation is an important part of the energy management systems.Its performance directly affects the monitoring,forecasting and accident analysis of the power system.Unlike the transmission networks,the measurement devices in the distribution networks are not fully configured.The measurement devices of the supervisory control and data acquisition(SCADA)system are only configured at important nodes of the main feeder.Therefore,the implementation of state estimation in traditional distribution networks is poor.With the rapid development of the phasor measurement unit(PMU)and the advanced measurement infrastructure(AMI),the types of measurement data are gradually increasing.It brings new opportunities to distribution network state estimation.However,due to price and technical reasons,it is unrealistic to configure PMUs at all nodes in the distribution networks.There will be a long-term coexistence of SCADA/PMU/AMI measurements in the distribution networks.Therefore,the distribution network state estimation in a hybrid measurement environment is studied.The main work is as follows:(1)The SCADA/PMU/AMI hybrid measurements in the distribution networks is studied,including the characteristics of different measurement devices in terms of measurement data,accuracy level,sampling period,and measurement delay.Multi-time hybrid measurement data fusion method for distribution networks is proposed.Observability of distribution network state estimation is studied,and a hybrid measurement state estimation model for distribution networks based on weighted least squares method is established.(2)Due to insufficient measurements in the distribution networks,pseudo-measurements are introduced to meet the observability requirements.Considering the measurement errors and the deviations between the pseudo-measurements and the actual values,the measurements and pseudo-measurements uncertainties will affect the performance of distribution network state estimation.The global sensitivity analysis(GSA)method of the distribution network state estimation is proposed,to identify the critical measurement uncertainties and their locations,which have an impact on the accuracy of state estimation.Based on the sparse polynomial chaos expansion,the global sensitivity indices are calculated to improve the computational efficiency of GSA.The metering placement method based on the importance ranking of uncertainties is established,and improves the accuracy of the distribution network state estimation.Compared with the traditional methods,the accuracy and efficiency of the proposed method are verified,as well as superiority in the placement of measuring devices.(3)In view of the lack of measurements in the distribution networks,a data-driven distribution network state estimation method is proposed,including offline learning and online state estimation.Using a large amount of marketing data in the distribution networks,a method for generating node injection power samples based on improved generation adversarial network is proposed to train offline state estimators.An online state estimation method based on deep neural network is proposed,which requires less measurements to estimate the system state.Compared with the state estimation method based on pseudo-measurements,the proposed method improves estimation accuracy.