The　traditional　signal　detection　method　based　on　Fourier　transform　has　no　time　resolution　in　the　frequency　domain　and　the　detection　method　based　on　wavelet　transform　cannot　characterize　the　slow-varied　signal.　Focusing　on　the　problem　mentioned　above,　a　new　detection　method　based　on　normalized　quadratic　spectrum　(NQS)　in　the　time-frequency　domain　is　proposed.　Firstly,　Fourier　transform　of　the　original　signal＇s　amplitude　spectrum　is　performed　in　the　frequency　domain,　and　then　NQS　is　obtained　by　normalizing　the　transformation　of　amplitude.　Afterwards,　the　signal　detection　is　performed　based　on　the　obtained　NQS.　The　supplementary　zero　or　adding　window　methods,　etc.　can　be　used　simultaneously　to　prevent　from　spectrum　leakage.　It　is　proved　that　NQS　can　reflect　both　the　frequency　and　the　time　characters　of　the　signal.　This　merit　of　NQS　can　overcome　some　pitfalls　of　traditional　detection　methods.　The　validity　of　the　proposed　method　is　verified　when　it　is　applied　to　the　character　expression　of　the　pole　and　oil　pipeline＇　s　conductive　signal.