基于CPA-OSELM的热轧带钢厚度在线预测

为解决自动厚度控制（automatic gauge control，AGC）系统反馈滞后、耦合强、厚度偏差大等问题，本文提出了一种基于食肉植物算法（carnivorous plant algorithm, CPA）的在线顺序极限学习机（online sequential extreme learning machine, OSELM）预测算法。首先，基于从现场采集的相关数据，建立了OSELM在线厚度预测模型。然后为了提高模型的准确性及稳定性，采用CPA方法优化OSELM的权重和偏置。在此基础上，运用自学习方法进一步提高模型的预测精度。最后，通过实验验证基于CPA-OSELM预测模型的有效性。实验结果表明：基于CPA-OSELM的方法能够对不同规格带钢的出口厚度进行高精度在线预测，预测结果可用于提升AGC模型的控制精度，为提升带钢产品质量奠定基础。

Online prediction of hot-rolled strip thickness based on CPA-OSELM

An online sequential extreme learning machine (OSELM) algorithm with self-learning capability based on carnivorous plant algorithm (CPA) was proposed to solve the problem of feedback lag, strong coupling, and large thickness deviation in thickness prediction in automatic gauge control (AGC) system. Firstly, an online thickness prediction model of OSELM was established based on the data collected from hot rolling site. Next, in order to improve the model accuracy and stability, a CPA method was used to optimize the weights and the biases of OSELM. On this basis, a self-learning method was applied to further improve the prediction accuracy. Finally, the effectiveness of the prediction model based on CPA-OSELM was verified by experiments. The experimental results show that the method based on CPA-OSELM can predict the delivery thickness of different specifications strip online with high accuracy. The prediction results can be used to improve the control accuracy of the AGC system, and provide a firm basis for improving the quality of rolled strip products.