METHOD OF FAST SUPER DECOUPLED LOADFLOW COMPUTATION FOR ELECTRICAL POWER SYSTEM

摘要

The loadflow computation method of the present invention called Fast SuperDecoupledLoadflow (FSDL) is characterized by the use of 1) the slack-start whereby thesame voltagemagnitude and angle as those of the slack-node are used for all the othernodes as an initial guesssolution, and thereby reducing almost all effort of mismatch calculation inthe first P-.theta. iteration,and PV-nodes voltage magnitudes are assigned their known values while updatingvoltage anglesafter P-.theta. iteration, 2) the restricted rotation angles to the maximum of-36 degrees for improvedconvergence in the situations of low voltage solution, 3) the gain matrices[Y.theta.], and [YV] whoseelements are defined independent of rotation angle, 4) the same indexing andaddressinginformation for storing both the constant gain matrices [Y.theta.], and [YV]factorized using the sameordering regardless of node types as they are of the same dimension andsparsity structure.In addition, the FSDL method is characterized in the modification of the realpowermismatch at PV-nodes by a factor K p= Absolute (B pp/Y.theta.pp) such that RPp = .DELTA.P p /(K p V p) in orderto correct the Super Decoupled Loadflow model at PV-nodes in P-.theta. sub-problem therebyachieving improved convergence and consequent processing acceleration to agreat extent. Alsowhile solving for [.DELTA.V] in Q-V sub-problem, all the rows and columnscorresponding to PV-nodesin factorized gain matrix [YV] are skipped from calculation.In addition to listed four characteristics in the beginning, an alternateNovel Fast SuperDecoupled Loadflow (NFSDL) method is characterized in that it employs overlapupdate of thevoltage angle corrections at PV-nodes. It involves calculation of anglecorrections at PV-nodesalong with voltage magnitude corrections at PQ-nodes in Q-V sub-problem inaddition to regularangle corrections in P-.theta. sub-problem as illustrated in Fig.4. Thisimproves convergence andachieves consequent processing acceleration of the same order as that of theFSDL method.