论文部分内容阅读
针对产品的性能要求制定合理的热轧工艺,提出将组织性能预测与控制技术应用于热轧工艺的优化设计。基于大量生产数据,建立了包含10个BP神经网络的模型组以描述化学成分、工艺和力学性能的对应关系,屈服强度、抗拉强度和伸长率的预测精度分别达到了±6%、±6%和±4%。结合多目标粒子群优化算法,针对客户对性能的需求,在化学成分和工艺约束已知的条件下,对热轧工艺进行了优化计算。工艺优化计算结果与现场生产数据吻合良好,验证了工艺优化设计的有效性,从而为热轧最优工艺设计提供指导。
According to the performance requirements of the product, a reasonable hot rolling process is established, and the optimal design of the technology for predicting and controlling the structure properties is proposed for the hot rolling process. Based on mass production data, a model group containing 10 BP neural networks was established to describe the correspondence between chemical composition, process and mechanical properties. The prediction accuracy of yield strength, tensile strength and elongation reached ± 6% and ± 6 respectively % And ± 4%. Combined with multi-objective particle swarm optimization algorithm, according to the customer’s demand for performance, the hot rolling process is optimized and calculated under the condition of chemical composition and process constraints. The result of process optimization is in good agreement with the field data, which verifies the effectiveness of the process optimization design and provides guidance for the optimal process design of hot rolling.