To　diagnose　the　fault　in　the　Tennessee-Eastman　(TE)　process　more　accurately,　a　learning　pseudo　metric　(LPM)-based　case　retrival　method　is　proposed　to　replace　distance　measure　retrieval　method　and　a　case-based　reasoning　(CBR)　fault　diagnosis　model　of　TE　process　is　established.　Firstly,　the　LPM　metrics　are　established　to　train　the　LPM　model.　Then,　the　similarity　between　the　target　case　and　each　source　case　is　measured　to　find　the　same　type　of　cases　as　the　target　case.　Next,　the　solution　of　the　target　case　is　obtained　based　on　the　majority　of　reuse　principle.　Finally,　the　running　data　of　TE　process　are　used　to　carry　out　a　performance　test　and　a　comparison　experiment.　The　results　show　that　the　proposed　LPM-based　CBR　method　is　superior　to　traditional　CBR,　back-propagation　(BP)　neural　network　and　support　vector　machine　method　and　significantly　improves　the　accuracy　of　the　fault　diagnosis.　It　has　a　promotional　value　for　fault　diagnosis　in　the　actual　chemical　process.　©　2017,　Editorial　Department　of　Control　Theory　&　Applications　South　China　University　of　Technology.　All　right　reserved.