An　online　energy　management　strategy　for　PHEV　based　on　predictive　control　is　proposed.　It　utilizes　BPNN　to　construct　a　trip　prediction　model,　and　uses　genetic　/　particle　swarm　hybrid　optimization　algorithm　to　improve　the　vehicle-speed　prediction　accuracy　of　the　trip　prediction　model.　On　this　basis,　a　dynamic　programming-based　predictive　control　strategy　is　designed　to　ensure　the　adaptability　of　the　trip　prediction　model　to　trip　conditions　and　the　real-time　performance　of　the　strategy.　Finally,　a　verification　simulation　is　conducted　on　the　strategy　proposed　based　on　trip　condition　data.　The　results　show　that　the　trip　prediction　model　designed　can　effectively　predict　vehicle-speeds　with　an　accuracy　higher　than　93%,　and　the　fuel　consumption,　emissions　and　real-time　performance　with　the　proposed　strategy　are　improved　compared　with　the　existing　real-time　strategies　and　global　optimization　strategies.　©　2019,　Society　of　Automotive　Engineers　of　China.　All　right　reserved.