This Paper successfully develops the method of Generalized Likelihood Ratio(G.L.R.)in detectingfaults of the system.It has been found very imPOrtant to pay more attention to the validation of measuringdata which must ...This Paper successfully develops the method of Generalized Likelihood Ratio(G.L.R.)in detectingfaults of the system.It has been found very imPOrtant to pay more attention to the validation of measuringdata which must contain richer information relating to the fault.The foil measuringpoints,partial measuring points, single observation and multi-observation for detection of single fault and multi-fault have allbeen considered in our algorithm which basically depends on the theory of G.L.R.,the new concept of calculating residues is also put up as the base of this detecting algorithm. It can not only accurately detect thefaulted branches of the System, but also locate the slanted measures and even determine amplitude of thefault.By simulation with the software MATLAB, the simulation results of some application examples showthat our detection is much better than that by [1]. HOwever, for our method, the precision of sensors shouldbe known and a threshold of confidence for decision should be determined by user.展开更多
文摘This Paper successfully develops the method of Generalized Likelihood Ratio(G.L.R.)in detectingfaults of the system.It has been found very imPOrtant to pay more attention to the validation of measuringdata which must contain richer information relating to the fault.The foil measuringpoints,partial measuring points, single observation and multi-observation for detection of single fault and multi-fault have allbeen considered in our algorithm which basically depends on the theory of G.L.R.,the new concept of calculating residues is also put up as the base of this detecting algorithm. It can not only accurately detect thefaulted branches of the System, but also locate the slanted measures and even determine amplitude of thefault.By simulation with the software MATLAB, the simulation results of some application examples showthat our detection is much better than that by [1]. HOwever, for our method, the precision of sensors shouldbe known and a threshold of confidence for decision should be determined by user.
