Variable　decomposition　have　been　widely　applied　to　large-scale　multi-objective　problems.　Existing　studies　usually　consume　a　large　amount　of　computational　resources　in　the　decomposition　phase.　To　this　end,　this　paper　proposes　a　multi-objective　evolutionary　algorithm　based　on　hierarchical　grouping　(MOEA-HG).　The　algorithm　contains　a　decomposition　phase　and　an　optimization　phase.　In　the　decomposition　phase,　diversity　variables　and　convergence　variables　are　first　identified.　Secondly,　the　concept　of　contribution　is　introduced.　Convergence　variables　with　equal　contribution　to　the　optimization　objectives　constitute　a　subcomponent.　In　the　optimization　stage,　collaborative　optimization　is　proposed　to　deal　with　convergence　subcomponents　and　diversity　subcomponents　separately.　Experimental　results　show　that　MOEA-HG　consumes　less　computational　resources　in　identifying　variable　interactions　than　other　decomposition-based　MOEAs.　Furthermore,　MOEA-HG　has　significant　advantages　over　the　five　state-of-the-art　MOEAs　in　terms　of　optimization　performance.　©　2024　held　by　the　owner/author(s).