基于混沌优化支持向量机的轧制力预测

Rolling force prediction based on chaotic optimized support vector machine

针对带钢热连轧轧制力的精确预测问题,提出一种基于最小二乘支持向量机模型的预测算法.在分析最小二乘支持向量机数学预测模型的基础上,提出一种改进的结合遗传算法的变尺度混沌优化方法,以进行最优模型参数的搜索.利用实测在线数据对模型进行训练并进行轧制力预测.仿真结果表明,利用该方法可使轧制力预测精度得到提高,平均误差率从BP神经网络的±10%降到±5%以下,为进一步提高热连轧厚度控制精度提供了一种有效方法.

Aiming at the exact prediction problem of rolling force in hot strip rolling mills, a prediction algorithm based on least square support vector machine is proposed. The prediction model of least square support vector machine is mathematically analyzed. And an improved multi-scale chaotic optimization algorithm combined with the genetic algorithm is proposed to optimize the model parameters. By using on-line data obtained from the factory, the model is trained and used for rolling force prediction. Simulation results show that the prediction accuracy is improved, and the average error rate decreases from ± 10% achieved by the BP neural network to less than ±5% by using the proposed algorithm. This algorithm provides a new method to improve the thickness control of hot strip rolling.