Particle Filter For Dynamic State Estimation With Hybrid Measurements

Power system state estimation is the key part of energy management system (EMS). Dynamic state estimation filter out the real state of power system from the measurement data, and do safety assessment through state prediction. Current power system measurement includes traditional SCADA data and a small amount of PMU data. Particle filter for dynamic state estimation with hybrid measurements is studied in the article.SCADA and PMU are introduced as the main kinds of the measurement in the power system. The difference between these two kinds of measurement are compared and analyzed. Matching method of the mixed measurement is also studied in the paper. In addition, aiming at the system observability, PMU measurement optimal configuration is discussed, and a configuration method is proposed in the paper.Particle filter are introduced in the state estimation, the basic thought of the particle filter is using the random weighted samples to represent the system state. Weights are adjusted by observed variables. System state are estimated by the particle state plus their weight. Power system dynamic model is proposed in the paper, and system element like power line and transformer are mathematically modeled, forming the power network model.Static state estimation and dynamic estimation are combined in the paper. SCADA measurement are estimated by P-Q decomposition state estimation method, the result are regarded as one of the measurement for dynamic state estimation. The nodes equipped with PMU can obtain PMU voltage phasor measurement, and its neighbor nodes’ pseudo voltage phasor measurement can be conversed by the current phasor measurement. These three kinds of measurement constitute power system hybrid measurement. The updating method of particle weight is designed for the hybrid measurement. The system state is estimated by the particle state and particle weight. And the particle are resampling during the computation for remaining the number of effective particles. The simulation is carried on IEEE 14 and IEEE30 bus systems, which result shows that the method is effective.