Atomic　clock　difference　prediction　is　the　key　step　of　atomic　clock　time　scale　calculation　and　atomic　clock　control.　Good　prediction　of　the　clock　difference　can　significantly　improve　the　precision　of　the　atomic　clock　time　scale　and　atomic　clock　control.　In　order　to　further　improve　the　prediction　precision　of　the　clock　difference　of　the　hydrogen　atomic　clock,　this　paper　presents　an　improved　BP　neural　network　algorithm　to　predict　the　atomic　clock　difference,　which　is　verified　by　the　actual　hydrogen　atomic　clock　data　of　the　time-keeping　laboratory　in　National　Institute　of　Metrology,　China.　The　verification　results　show　that　compared　with　the　clock　difference　prediction　algorithms　based　on　linear　regression　and　SVM　used　at　present,　the　improved　BP　neural　network　clock　difference　prediction　algorithm　significantly　improves　the　prediction　precision　of　hydrogen　atomic　clock,　and　has　a　good　promoting　effect　on　improving　the　precision　of　atomic　clock　time　scale　calculation　and　atomic　clock　control.　©　2016,　Science　Press.　All　right　reserved.