Organic　Rankine　cycle　(ORC)　synergistic　multi-objective　optimization　is　the　key　to　obtain　the　actual　waste　heat　recovery　potential　under　dynamic　driving　cycle.　The　ORC　operation　shows　uncertainty　and　hysteresis　under　the　disturbance　of　variable　high　temperature　waste　heat　source.　In　this　paper,　a　synergistic　multi-objective　optimization　mixed　nonlinear　dynamic　modeling　approach　is　proposed　by　deeply　integrating　data　selection,　feature　dimensionality　reduction,　integrated　system,　neural　network　modeling　mechanism,　ensemble　learning　mechanism　and　synergistic　multi-objective　optimization.　Compared　with　direct　modeling,　the　nonlinear　dynamic　modeling　approach　can　reduce　the　data　volume　by　27.65　%　on　average.　The　decision　variables　and　construction　time　are　reduced　at　least　by　69.1　%　and　24.26　%　on　average,　respectively.　Generalization　ability　is　improved　by　at　least　28.14　%　on　average.　Taking　thermodynamic　performance,　economic　performance,　thermoeconomic　performance　and　environmental　impact　as　optimization　objectives,　a　synergistic　multi-objective　optimization　of　ORC　comprehensive　performance　under　driving　cycle　is　carried　out.　The　optimal　emissions　of　CO2　equivalent　(ECE)　will　suppress　the　optimal　power　output　of　per　unit　heat　transfer　area　(POPA)　to　a　certain　extent.　A　decrease　in　optimal　ECE　is　accompanied　by　an　increase　in　optimal　electricity　production　cost.　The　approach　proposed　in　this　paper　can　provide　new　ideas　and　solutions　for　ORC　dynamic　modeling　and　synergistic　multi-objective　optimization　under　road　conditions.　©　2023　Elsevier　Ltd