过程参数优化的建模

针对化工厂参数优化问题，提出了一种多项式逼近(即多项式回归)、人工神经网络逼近等其他建模方法——插值建模.这种方法依据大量的原始数据，使用插值法得到一个初步的离散系统模型，然后对上述模型进行滤波，最终得到可用于优化的数值模型.并以一维自变量和二维自变量建模方法为例，阐述了插值建模的算法，给出了仿真结果，初步证明了这种方法的可行性和有效性，从而拓宽了参数优化建模的思路.但是这种方法建模需要大量的计算，所以现阶段仅用于离线优化问题.

Modeling in process parameter optimization

A new method named the interpolation modeling is presented,which is different from the other schemes in process parameter optimization such as polynomial approach,ANN and so on.The main idea is that a discrete system model can be obtained with a mess of original data,and then the final model will be obtained after filtering the discrete system model.The interpolation algorithm is described and illustrated in the terms of one-and two-dimensional variables respectively.Finally it is proved by simulation that the(proposed) method is feasible and effective.And thus,the modeling method can be extended in parameter(optimization.) Because of calculation cost during the modeling,this method can only be used in offline(parameter) optimization.