In　order　to　improve　the　performance　of　high　dimensional　time　series　classification,　a　High　Dimensional　Time　Series　Classification　Model　(HDTSCM)　based　on　multi-layer　perceptron　and　moving　average　model　is　proposed.　By　constructing　a　multi-layer　perceptron　neural　network　model　and　applying　moving　average　model　in　the　backward　propagation　of　the　network　model,　it　realizes　the　train　of　the　model　and　classification　of　high　dimensional　time　series.　Experimental　results　on　8　UCRArchive　datasets　show　that　the　classification　error　rates　of　HDTSCM　are　significantly　lower　than　the　classification　methods　of　Euclidean　distance　and　dynamic　time　warping,　relatively　reduced　by　49.76%　at　most.