【作者】 潘孝礼；

【导师】 王福利；

【作者基本信息】 东北大学 ， 检测技术与自动化装置， 2008， 博士

【摘要】 目前,国内许多钢铁企业为了提高无缝钢管的生产质量,都采用了加装狄赛尔导盘的斜轧穿孔机进行钢坯穿孔,但是在斜轧穿孔机中,与生产密切相关的工艺参数——导盘转速只能停机离线调整,这大大降低了钢坯穿孔生产效率和质量,成为钢管生产的一个瓶颈。如何实现穿孔机导盘转速的在线可调,进而改善穿孔机穿孔生产的毛管质量,提高无缝钢管生产机组的生产效率,已经成为各钢厂一个迫切需要解决的问题。本文以某钢铁公司钢管分公司SWW斜轧穿孔过程为研究背景,建立了导盘转速软测量模型,最优导盘转速确定模型,以及负载力矩软测量模型,并实现了导盘转速的优化控制。另外,通过对穿孔工艺的深入分析,建立了对穿孔生产具有一定指导意义的穿孔效率预报及优化模型。最后,通过建立毛管质量预报模型,实现了前述优化控制效果的检验及毛管质量的实时监测。本文主要工作如下：针对导盘转速难以在线测量的问题,提出了基于改进PCA-ELM的导盘转速软测量方法。利用PCA方法对数据进行压缩去噪,并且针对传统ELM方法在建模数据较多时,预报速度较慢的缺陷,提出了利用改进PCA-ELM方法建立软测量模型,实现导盘转速的在线预估。同时,利用短期校正与长期校正相结合的方法实现了软测量模型的更新校正。为了提高生产效益,需得到在不同工况下,使毛管质量和穿孔效率较高而能耗较低的最优导盘转速。针对管坯穿孔过程的复杂性及建模数据的特殊性,提出利用适用于非高斯分布数据的ICR建模方法建立独立成分和穿孔机最优导盘转速确定模型,并提出利用误差检验方法确定独立成分个数。最优导盘转速确定模型的建立,可为导盘转速自动控制的最优参数设定提供指导和参考。针对导盘转速的自动控制问题,提出了导盘转速的优化控制系统。其中导盘转速软测量模型实现了难于在线测量的关键参数——导盘转速的在线预估,为实现其直接控制奠定了基础；最优导盘转速确定模型根据管坯原材料、生产要求及生产运行工况确定了导盘转速的最优设定参数,为导盘转速的自动控制提供最优设定；负载力矩软测量模型,实现了系统中的主要扰动——负载力矩的实时预估,并在此基础上实现了负载力矩的前馈补偿控制,大大削弱了该主要扰动对导盘转速的影响。针对无缝钢管穿孔生产过程,难以一直保持较高穿孔效率这一问题,提出了基于均值子时段的MPLS穿孔生产过程穿孔效率预报模型,并根据生产工艺约束,对其进行了优化求解,获得了最优穿孔效率所对应的穿孔生产工艺参数。同时,为了检验穿孔效率的优化结果,实现毛管质量的在线监测,提出了基于均值子时段的MPLS毛管质量在线预报方法。基于均值子时段预估方法的创新之处在于：(1)提出了子时段的思想；(2)生产处于不同操作子时段,对预报参数的影响因素不同,分段选择辅助变量；(3)提出了数据均值的思想。最后,本文给出了穿孔过程关键参数软测量与优化系统的总体结构实施框架,详细地介绍了人机交互系统、下位机以及后台预报与优化系统的功能,同时给出了彼此间的相互联系。更多还原

【Abstract】 Nowadays, in order to improve the quality of seamless steel tubes, many domestic steel enterprises use skew rolling piercer with Diescher guide disc to pierce steel billet. But in the skew rolling piercer, the guide disc rev, which has significant effect on piercing production, can only be regulated offline. This has greatly reduced the production efficiency and quality of steel billet piercing, and thus become a bottleneck in steel tubes manufacturing. It has been an urgent problem with respect to how to online regulate the guide disc rev of piercer so as to improve the quality of steel tubes in piercing production and increase the production efficiency in seamless steel tubes.Based on the SWW skew rolling piercing process in certain Tube Branch of Steel Corporation, this dissertation establishes three types of models, i.e., soft sensing of guide disc rev, determination of optimal guide disc rev and soft sensing of load moment, and realizes the optimization control of guide disc rev. In addition, it builds piercing efficiency prediction and optimization model, which provides instructive significance for piercing producing. Finally, it realizes the effect verification of the above mentioned optimization control and online quality prediction of steel tube based on the construction of quality prediction model of steel tube. The main works of this dissertation are listed as follows:To overcome the problem that it is hard to online measure the guide disc rev, an improved PCA-ELM soft sensing method is proposed for the guide disc rev. In this method, PCA method is used to compress and filter process measurements. Moreover, the soft sensing model of guide disc rev is developed based on an improved PCA-ELM method, which overcomes the prediction speed deficiency of traditional ELM method when dealing with much modeling data and realizes the online prediction of guide disc rev. Meanwhile, the soft sensing model is adaptively updated using the combination of short-term and long-term adjustments.In order to improve production benefit, it is necessary to obtain the optimal guide disc rev under different conditions, which can not only upgrade steel tubes quality and piercing efficiency but also reduce energy cost. Focusing on the complexity of tube billet pierceing process and particularity of modeling data, ICR modeling method, which is suitable for modeling non-Gaussian data, is introduced to extract the independent components and build the optimal decision-making model of guide disc rev, in which, the number of independent components is determined by cross-validation. The development of above-mentioned optimal model can provide instructions and references for setting the optimal parameters of automatic control of guide disc rev.Focusing on the automatic control of guide disc rev, an optimization control system is developed. For the key parameter, the guide disc rev, which is difficult to online measure, the soft sensing model of guide disc rev realizes its online estimation, which establishes the foundation of direct control. The model of optimal guide disc rev determines the optimal setting parameters of guide disc rev according to tube billet raw materials, production requirement and operation status, which can be used to set the optimal setpoint of the guide disc rev automatic control. The soft sensing model of load torque realizes the realtime prediction of load torque, which is the primary disturbance in the system, and further realizes its feedforward compensation control, which, thus, greatly weakens the influence of this disturbance on guide disc rev.Considering that it is hard to maintain high pierceing efficiency during seamless steel tubes pierceing process, a pierceing efficiency prediction model is established using MPLS algorithm based on phase-specific average trajectory. Moreover, the optimal problem is solved under the process operation constrain, and the pierceing process parameters are obtained corresponding to the optimal pierceing efficiency. Meanwhile, to verify the optimal result of pierceing efficiency, and realize the online monitoring of tube quality, an MPLS algorithm based on phase-specific average trajectory is proposed for online tube quality prediction. The innovation of the proposed method lies in:(1) it proposes the idea of subphase; (2) it chooses the phase-specific secondary variables according to the different influences on prediction parameters during different process operation phases; and (3) it proposes the idea of average trajectory.Finally, this dissertation develops the overall framework of soft sensing of key pierceing parameters and optimization control system, detailedly introduces the human-computer interaction system, supervisor computer and background prediction and optimization system, and meanwhile presents their interrelationship.更多还原