摘要提出了一種綜合改進的BP 神經網絡算法，該算法在訓練時對不同的連接權和閾值采用不同的學習速率，由此建立了乙二醇精制塔塔釜乙二醇濃度的神經網絡軟測量模型。結果表明該算法能有效提高乙二醇濃度BP 神經網絡軟測量模型的收斂精度。
關鍵詞BP 神經網絡算法改進乙二醇濃度軟測量模型
Abstract A comprehensively improved BP algorithm is issued in this paper. According to this algorithm，different learning rates are applied to different
connection weights and threshold values when a BP network is trained. A neural network model is setup to be used on soft sensing of the concentration of
the ethylene glycol of the MEG column. Practice shows that this proposed algorithm is effectively able to improve the convergence precision of the BP
neural network in soft sensing for the concentration of ethylene glycol.
Keywords BP neural network Improved algorithm Concentration of ethylene glycol Soft sensing model