| The state estimation of power system is an important part of the SCADA system’s data processing and it is of great significance for the control center to get the real-time operation state of the power grid.The innovation graph approach for state estimation has fast calculation and strong identification ability.In order to further improve its identification accuracy,this paper analyzes the possible errors in the process of identifying abnormal events in the innovation graph approach for state estimation.The main research includes:This paper studies the influence of the forecast state that contains unrecognized topological errors that cannot reflect the actual situation on the calculation of the innovation graph approach,deduces the relationship between the physical quantities in the innovation graph approach and the forecast error.This paper proves that if there is a topology error in which a certain line is in operation but the remote signaling display exits from operation,although the forecast state is affected,the topology error can still be identified by using the innovation graph method.When there is a topology error in which a certain line is out of operation and the remote display is still running,the paper points out that the results of power flow calculation using the injected power measurement at the current time and the wrong topology structure at the previous time can be used as the forecast state to perform the innovation graph approach and identify the topology error.The paper’s work shows that the innovation graph approach has strong adaptability to the inaccurate forecasting status.When the forecast is wrong,it can still obtain the correct identification result.When using the innovation difference to identify bad data,a fixed threshold is often used as the identification basis.However,this often leads to the problem of undetected bad data and false rejection of normal data.Therefore,this paper deduces the relationship between branch measurement error and its innovation difference.The identification method of standard innovation difference is proposed in the premise that the measurement error obeys the normal distribution,Monte Carlo method is used to prove that this method can improve the identification accurac y.When the number of samples is small,the confidence interval of innovation difference is given based on the normal distribution interval estimation.If the measurement error does not obey the normal distribution,a non-parametric bootstrap method is used to find the confidence interval of innovation difference.The effectiveness of the above methods is verified by IEEE examples. |
