Membranes　fouling　in　MBR　process　is　caused　by　many　complex　and　interactional　factors.　A　flux　prediction　model　is　put　forward　based　on　the　PSO-BP　neural　network,　which　adjusts　weights　of　BP　neural　network　using　particle　swarm　optimization　(PSO)　rather　than　the　traditional　gradient　descent　method.　First,　principal　component　analysis　(PCA)　is　used　to　reduce　the　dimensions　and　correlations　of　input　parameters.　Second,　the　PSO-BP　is　used　to　optimize　the　weights　and　thresholds　of　the　neural　networks.　Based　on　the　experimental　data　(0.038　μm　polyethersulfone　membrane　for　printing　and　dyeing　wastewater　treatment),　the　simulation　is　performed　with　MATLAB.　Results　show　that　the　PSO-BP　neural　network　has　a　faster　convergence　speed　and　a　better　agreement　with　the　real　data　than　traditional　BP　neural　network.