酸轧机组神经网络模型数据筛选算法研究

Data Sort Algorithm of ANN for the Sour Rolling Mill

酸轧机组神经网络训练之前必须要为各自的网络提供足够代表性的样本数据才能够达到优化的目的,并且所提供的数据要严格符合现场生产工艺要求。介绍了酸轧机组神经网络模型的数据筛选算法及筛选优化过程,重点分析了单纯形算法及其在样本数据筛选过程中的应用。仿真结果表明了算法的有效性,为神经网络模型训练提供了更优的样本数据。

To achieve purpose of optimization, The Sour Rolling Mill nerve network needs to provide enough representation sample data for respective network before trained ,and the data needs provide accords to local production craftwork request. It introduces data sort algorithm and process for sorting and optimizing of The Sour Rolling Mill nerve network model, and importantly analyse Simplex and its application on sample data sort. The result of imitation indicates algorithm's validity, and provides optimization sample data for Sour Rolling Mill nerve network training.