An　accurate　electro-hydraulic　power　steering　system　(EHPSS)　model　is　essential　to　analyze　dynamic　steering　performance　and　advanced　nonlinear　control.　The　main　obstacle　to　establish　an　accurate　model　is　the　complex　structure,　including　steering　mechanism,　valve　controlled　dual　hydraulic-actuator,　and　heavy　duty　tires.　This　paper　constructs　a　suitable　model　incorporating　these　parts,　based　on　a　Lagrange　equation　describing　the　steering　trapezoid　mechanism　and　dual　cylinder　hydraulic　dynamics,　regarding　steering　resisting　moment　as　an　external　load.　A　simplified　tire　model　is　used　to　represent　the　tire　basic　steering　load　characteristics.　Due　to　the　complexity　of　the　kinematic　relationship　in　the　model,　several　expressions　are　fitted　using　back　propagation　neural　networks　to　significantly　reduce　calculation　difficulty.　Experimental　measurements　and　simulation　using　Matlab/Simulink　and　experiment　are　realized　for　the　case　of　in　situ　steering,　and　the　results　validate　the　proposed　model　accuracy.　Thus,　the　proposed　model　is　suitable　to　analyze　the　system　and　design　advanced　controllers.