基于双电源二次激励法的湿型砂组分快速预测用网络模型的研究

The network for predicting active clay content and moisture of green sand based on double power source exciting method

双电源二次激励法是本文提出的一种利用温型粘土砂导电特性参数，快速预测其组分的新方法。利用人工神经网络（ＡＮＮｓ）研究型砂有效粘土含量和含水量与型砂交流电导率、直流电导率、直流电导率变化率之间的复杂关系，通过ＢＰ算法建立了信息参数与预测参数之间的非线性映照关系。实验结果表明，双电源二次激励法与人工神经网络相结合，可以实现型砂有效粘土含量和含水量的快速在线预测。

Double Power Source Exciting Method(DPSEM) is a new method for predicting thecomposition of green sand by testing its conductivity parameters. In this paper, the relationshipsamong active clay content, moisture, AC conductivity, DC conductivity, DC conductivity changeratio and compactness of green sand is investigated by means of Artificial Neural Networks(ANNs) and the nonlinear mapping function between information parameters and predictingones has been established. It is proved by experiments that DPSEM together with ANNsmakes the fast on line prediction of active clay content and moisture of green sand feasible.