Table of contents

Article Summary:

Power system state estimation is an integral part of energy management systems (EM$). The data used by the power system dispatch center mainly comes from the data acquisition and monitoring system (SCADA) and the phasor measurement unit (PMU). Power system state estimation utilizes the redundancy of measurement information to estimate the voltage amplitude and phase angle in the power system, identify and eliminate errors caused by random disturbances, and obtain an estimated value of the system's operating state, realizing the current operating state of the system. Security monitoring plays an important role in modern power systems.

Extended Kalman filter (EKF) and unscented Kalman filter (UKF) are widely used methods in power system dynamic state estimation. This paper studies and analyzes the two methods, and realizes these two methods based on the Matlab platform. , the validity of the method is tested by using IEEE39 node examples.

Research Background:

Power system state estimation is an indispensable and important part of EMS. The network topology and measurement information data can be used to estimate the voltage amplitude and phase angle of the bus in the network and assist power flow calculation. State estimation can also be called filtering, which is a process of using the redundancy of measurement data to eliminate random errors caused by interference, identify bad data, and improve data accuracy. Its basic function is to combine the telemetry information uploaded by the data acquisition system to analyze the network topology and predict the state, and calculate the estimated value of the voltage amplitude and phase angle on each node of the power grid. In the case of limited hardware conditions, state estimation can fully explore the potential of existing hardware equipment on the original basis, identify bad data in measurement information and generate pseudo-quantity measurement instead, remove occasional error information, and improve data quality. Ensure the stability of the system. It ensures the reliability, accuracy and integrity of the grid data received by EMS, thus realizing the effective monitoring of the power system operation status. It provides a solid and reliable theoretical support for effective operating state analysis and lays a complete and accurate data foundation. Therefore, the research and analysis of practical models and high-precision algorithms for power system estimation has important theoretical and practical significance.

If you want to grasp the actual operating state of the changing system, such as the position of the driving car, the node voltage amplitude, voltage phase angle and other state quantities in the power system, there are two ways to choose from. One is to obtain the data of the state quantity through the measuring device, and the other is to establish a system model, and derive the state quantity at the current moment through the system state and change trend at the previous moment. However, in the actual measurement system, there will be random errors due to the influence of equipment, weather, human and other factors, and the established system model may also have defects, which may lead to deviations in calculation results, and cannot accurately describe the state of the system. Therefore, it is necessary to use statistical knowledge to obtain the state estimator of the system through an appropriate state estimation method. State estimation can also be called filtering. It can estimate the state quantity according to the measurement data and the system state transition model, and filter out bad data caused by random errors, so as to obtain a more accurate system operating state. State estimation is an effective means for computers to process real-time data, and it is widely used in automatic control, aerospace, industrial production and other fields.

Power system dynamic model:

In the power system, the measurement data collected by the SCADA system and the PMU system need to be transmitted to the dispatch center before they can be analyzed and utilized by dispatchers to assist in decision-making. The more and more accurate the information received by the dispatch center, the higher the guarantee of the smooth operation of the system. Due to the limitations of equipment or economic conditions, the communication device does not transmit all the collected data, but only the indispensable data that can fully describe the state of the system, that is, the state variables. Only with sufficient measurement data and network topology information can a complete system model be established.

The dynamic model of the power system usually adopts the second exponential smoothing method (Second exponential smoothing method), as shown in the formula (2-l), compared with the single exponential smoothing method, this method takes the trend of the data into account and has better smoothing Effect. Exponential smoothing method is a simple and effective short-term time series forecasting method, which has the advantages of less storage variables and fast calculation speed, and is very suitable for online calculation. For data with a long-term trend, it is more suitable to use the Hot two-parameter exponential smoothing method for processing.

Extended Kalman Filter (EKF):

The Kalman filter discussed earlier is a model that describes a stochastic linear system by combining the measurement equation and the state equation into a state-space model. In actual use, for many situations that cannot be simply expressed linearly, such as power system dynamic state estimation. The power system is a nonlinear system, and its measurement equation and state equation are also nonlinear. The Kalman filter is not applicable. It needs to be approximately linearized first, and the state space is linearized, and then filtered by the Kalman filter. . This is the basic idea of extended Kalman filter to solve nonlinear problems.

Unscented Kalman Filter (EKF):

EKF needs to use the Jacobian matrix in the linearization process, and there will be orthogonal items when the dimension is high, and errors are prone to occur during implementation. Aiming at the problem of poor nonlinear effect of Kalman filter, Julier adopted another approximation method different from EKF, and proposed Unscented Kalman Filter (UKF). ). The basic principle is to randomly distribute a group of Sigma points with weights around the mean point of the state quantity, and use the method of nonlinear function probability density approximation to achieve the effect of approximate linearization. The basic principle of UKF is as follows.

(1) Select the sampling method: