A FastMethodforProbabilistic Reliability Assessment ofBulk Power SystemUsing FSOM NeuralNetworkasSystemStates Filters

For solving theproblemthatMonte-Carlo sampling technique normallyusedin powersystem probabilistic simulation haslowefficiency, thispaper proposes a fastmethodusingfuzzyselforganizing map (FSOM)neural network assystemstates filter toevaluate thereliability ofbulkpowersystemforthefirst time. SOM isespecially appropriate toestimate thereliability ofpower systembecause it's training timeshorter thanotherneural network. Invalid system states canbefiltered byfuzzy SOM neuralnetwork, itreduces significantly thenumberof systemstates shouldbeevaluated. Thenew methodof FSOM neuralnetworkcombined withsequential Monte- Carlosimulation results ina significant reduction inthe computational effort required tocomputecomplex power systemreliability indices. CasestudyoftheIEEE-RTStest systemandapractical large-scale systemarepresented to demonstrate theeffectiveness and feasibility of the developed algorithm. V Vx V~~~1+ /~~~~~a