Terahertz　time-domain　spectroscopy　is　a　contactless　and　nondestructive　testing　technique　that　is　often　used　to　measure　the　thickness　of　layered　materials.　However,　the　technique　presents　limited　thickness　detection　resolution,　especially　in　the　thin　thermally　grown　oxide　(TGO)　of　thermal　barrier　coatings　whose　thickness　is　below　30　mu　m.　In　this　study,　an　SWT-BP　algorithm　combining　a　stationary　wavelet　transform　(SWT)　and　a　backpropagation　(BP)　neural　network　was　proposed,　and　the　regression　coefficient　of　SWT-detailed　results　was　0.92.　The　prediction　results　were　in　good　agreement　with　the　real-time　results;　it　demonstrated　that　the　proposed　algorithm　was　able　to　achieve　a　thickness　prediction　of　up　to　1-29　mu　m　of　the　TGO.　The　proposed　algorithm　is　suitable　for　thin　thickness　detection　of　the　TGO.　(C)　2020　Optical　Society　of　America