The　growing　demand　for　energy　efficiency,　environmental　protection　in　the　heavy　transportation　sector,　particularly　in　large-scale　projects,　highlights　the　importance　of　improving　steering　systems　for　vehicles.　A　pump-controlled　electro-hydraulic　steering　system　is　proposed,　offering　significant　advantages　in　energy　efficiency　under　high　power.　However,　it　leading　to　soft　speed-load　characteristics,　reduced　circuit　stiffness,　and　compromised　performance.　To　address　challenges,　an　improved　back-pressure-controllable　BPC-PC-EHSS　is　introduced,　the　dynamic　and　power　flow　models　are　established.　But　it　increases　power　loss,　conflicting　with　the　energy-saving　objectives.　Therefore,　back-pressure　parameter　identification　that　balances　both　high　performance　and　low　energy-consumption　is　crucial.　The　energy-saving　boundary　is　analyzed　using　the　hydraulic　conductivity　factor,　a　parallel-input　multilayer　neural　network　(PIM-NN)　is　designed　for　nonlinear　system　back-pressure　identification.　Experimental　results　show　that　the　proposed　system　significantly　improves　steering　performance　and　energy-efficiency　with　minimal　change　in　pump　peak　pressure　and　reduced　pressure-vibrations.　Specifically,　under　6　tons　load　the　error　is　1°,which　is　improved　by　55.6　%　compared　to　the　non-identification.　Compared　with　valve-controlled　and　pump-valve　systems　under　same-typical-conditions,　significant　energy-saving　advantages　and　steering　economy　are　demonstrated.　Additionally,　the　real-world　driving　hardware　environment　is　reconstructed,　it　is　validated　that　the　total　steering　input　energy　is　reduced　by　76.19　%　on　the　experimental　road.　©　Elsevier　Ltd