CAJ | 학술논문

针对传统BP神经网络在训练过程中存在收敛速度慢的缺陷，将LM（levenberg marquardt）算法引入到BP神经网络的训练过程，建立了LM—BP神经网络模型，并将其应用于连铸过程中的漏钢预报系统。结合某钢厂连铸现场历史数据对系统进行了测试，测试结果以96．15％的预报率及100％的报出率，验证了基于LM算法的BP神经网络连铸漏钢预报方案的可行性和有效性。
침대전통BP신경망락재훈련과정중존재수렴속도만적결함，장LM（levenberg marquardt）산법인입도BP신경망락적훈련과정，건립료LM—BP신경망락모형，병장기응용우련주과정중적루강예보계통。결합모강엄련주현장역사수거대계통진행료측시，측시결과이96．15％적예보솔급100％적보출솔，험증료기우LM산법적BP신경망락련주루강예보방안적가행성화유효성。
LM algorithm was introduced to the training process of a BP neural network and a LM--BP neural network model was established aiming at the defects of slow convergence in the train- ing process of the traditional BP neural network. The LM--BP neural network model was applied to the breakout prediction in the continuous casting processes, and it was tested with the historical data collected from a steel mill. The feasibility and the validity of the model are verified by the results with the accuracy rate of 96.15% and the prediction rate of 100%