The paper contains an evaluation of possible application of laboratory studies of coronaeffect acoustic signal, generated by simulated damages and contaminations on the conductor surface, forassessment of technical condition of UHV transmission lines in real ambient conditions. The completedstudies have shown, that there is a possibility of extracting such a vector of distinctive features, which allowsa construction of a neural network model for the corona effect in laboratory conditions and its furtherapplication for recognition of possible sources of the intensified corona effect. Even if the number of correctrecognitions in laboratory conditions close to 90% should be treated as satisfactory, it should be also takeninto account that in real world conditions, where the line damages are not so frequent and their sources muchless obvious, there may be some problems with collecting a properly representative learning set. Thereforethe possibility of application of the results collected in laboratory conditions to the construction andverification of a neural model applicable in real conditions seems very attractive. The work presents theresults of the laboratory studies and preliminary verification of the ?laboratory” model for the set of datacollected in selected conditions of 400 and 220 kV UHV transmission line damages in Poland.
展开▼
