基于 ABC-LSSVM 的铝热连轧板凸度软测量建模

摘要

The plate crown is an important indicator of the aluminum strip in Aluminum hot strip mill flatness control. In order to accurately predict the aluminum hot strip mill plate crown,a soft sensor model was established on the base of artificial bee colony(ABC)and least squares support vector machine(LSSVM). ABC was applied to the process of the model parameters optimization because prediction accuracy and generalization ability of LSSVM model depends on the choice of the parameters. The prediction performance of the model was tested by sample data collected at the scene of a factory 1 + 4 aluminum hot rolling. The model was compared with Marquardt and the LSSVM model whose parameters were optimized by Genetic Algorithms( GA). The simulation result shows that the ABC-LSSVM soft sensor model parameters optimization fast,simple structure and high precision .%在铝热连轧板形控制中，板凸度是铝板带的重要指标之一。为了准确预测铝热连轧板凸度，提出了一种基于人工蜂群（ABC）和最小二乘支持向量机（LSSVM）的板凸度软测量模型。由于 LSSVM 的精度和泛化能力取决于模型参数的选择，故引入 ABC 进行参数优化。根据某厂1+4铝热轧现场采集的数据验证软测量模型的预测性能，并与 GA-LSSVM 模型和 Marquardt 模型做比较，仿真结果表明：建立的 ABC-LSSVM 板凸度软测量模型参数优化速度快、结构简单，并且具有较高精度。