Abstract: “ 1+4 ” aluminum hot tandem rolling mill is a complicated system extremely. In order to predict crown of aluminum strip of the finisher delivery more accurately, this paper puts forward a method of timelapse loop BP network to forecast the sequencing crown， and the reconfiguration of BP network is trained with LM algorithm. With a large number of test data gathered by onthespot experiment and analysis, simulation experiment indicates the predictive precision of relative crown(%) of strip can reach 0.01%, effectivenes and superiority is proved.

Key words: aluminum strip hot tandem rolling mill, crown of strip, timelapse feedback BP network, LM algorithm, prediction of sequence