Distribution Network Parameter Estimation Based On Innovation Graph Method

Distribution state estimation is one of the important components of distribution management system.In order to ensure its accuracy,it is necessary to obtain more accurate the parameters of branch network.However,in actual operation,for various reasons,sometimes there are errors in the parameters of individual branch network.The existence of branch parameter errors will lead to a decrease in the accuracy of distribution state estimation results.Especially for weak-loop distribution networks,parameter errors will cause more obvious calculation errors,which makes the state estimation results deviate from reality.Considering that the innovation graph method has good identification effect for branch parameter errors in transmission network,this paper decides to study the identification and estimation of branch parameter errors in distribution network based on the innovation graph method.This paper proposes a parameter estimation method for distribution network based on innovation graph.For radial distribution network,the branch where the parameter error occurs is identified according to the innovation difference path exceeding the threshold value.And the branch power loss is calculated by the measurement at the head and the end,then the true value of the branch parameters can be estimated.For the weak-loop distribution network,the parameter errors of a branch in the weak-loop are first determined according to the bad innovation circuit.Then by estimating the impedance values of all branches in the loop and comparing those values with the impedance data stored in the database,it is identified that the branch with the largest parameter deviation is the branch with parameter error.The example of IEEE33 shows that the above method can identify the branch impedance parameter errors accurately and obtain more accurate parameter estimates.This paper further analyses the case in distribution network when parameter errors,bad data and topological errors exist simultaneously.Firstly,the parameter errors of radiation branch and the bad data can be identified according to the innovation difference vector.For the case that the characteristics of the two kinds of abnormal events overlap and make it difficult to identify,the enumeration method is used to identify.Then the topological errors can be identified according to the modified prediction ratio.Finally,the parameter errors on weak loops can be identified according to the bad innovation loops.The example of IEEE33 shows that the above method is effective in identifying parameter errors,bad data and topological errors.This paper also analyses the identification effect of the innovation graph method when the state estimation is started,if there is an unrecognized abnormal event in the distribution network,including the topology error that the loop closes but not report,the loop disconnects but not report and bad data injected into nodes,which makes the prediction state inaccurate.The example of IEEE33 proves that the innovation graph method has strong adaptability to the incorrect prediction state,and it can still get the correct identification results.