降维神经网络在印染废水混凝试验中的应用

Application of dimension-reduced artificial neural network in mixed coagulation experiments of dyeing and printing wastewaters

运用降维人工神经网络,揭示了在用复合絮凝剂处理印染废水混凝试验中,多维工艺操作参数与经处理后水的透光率之间的关系.在二维Z平面上目标函数的等值线图,可全景式地展现出样本数据集操作空间的面貌和特征,直观地显示出最优操作点和最优操作区域,也可通过逆映射算法将最优操作点还原到多维空间,实现混凝试验优化设计.

Dimension-reduced artificial neural network was used to reveal the relations between the operation parameters of multi-dimension processes and the transparency of the treated water in the mixed coagulation experiments of dyeing and printing wastewaters with composed coagulants. In the target function isograms on the 2-dimension plane Z, the figures and features of the operation spaces of the sample data were presented in a full-scenic way, the optimal points and zones of operation were shown directly, and the optimal points could be regressed to the multi-dimension spaces through the reverse mapping arithmetic, thus realizing the optimized design of mixed coagulaton experiments.