【作者】 何纯玉；

【导师】 吴迪； 赵宪明；

【作者基本信息】 东北大学 ， 材料加工工程， 2009， 博士

【摘要】 本文针对中厚板生产过程中由于不对称因素所引起的侧弯问题,提出了侧弯预测与控制模型,研究了温度预测的在线修正方法、变形抗力自学习方法和影响函数方法,并基于机器视觉技术,开发出集侧弯检测、侧弯预测与侧弯控制的应用软件,在生产现场进行了在线应用,取得良好的效果。研究内容和取得的主要进展如下：(1)实现了中厚板生产过程中温度的高精度预测。基于隐式差分方法和有限元方法建立了中厚板轧制过程温度演变模型,确定了在不同传热条件下的换热系数；根据中厚板的生产特点,提出了根据模型预测温度和实测温度的互相检验,获得接近于真实的轧件温度的测量方法,该方法在测量轧件温度的同时,不断的对模型参数进行学习,保证了测量的准确性。(2)开发出基于遗传算法的变形抗力自学习系统。中厚板生产过程中,变形抗力模型参数的取值决定了轧制力模型的预测精度,通过对轧制力计算误差的影响因素分析,将遗传算法引入到变形抗力自学习方法中,充分利用遗传算法的空间搜索和优化能力,在实际生产数据中寻找变形抗力参数的最优值。在变形抗力自学习系统的开发过程中,实现了决策变量的选择,编码与解码、适应度评价及终止条件的处理。计算结果表明,这种方法的优化速度和精度能够满足生产要求。(3)提出中厚板轧制过程的侧弯预测方法。根据轧制过程中发生侧弯时轧件的运动特点,推导出预测轧件侧弯曲率的理论公式。利用修正后的适应轧件宽度变化的影响函数方法,结合温度模型和轧制力模型,计算轧件断面厚度分布。将影响函数方法与侧弯理论公式相耦合组成侧弯预测模型,可以分析不对称因素对轧件侧弯的影响。(4)提出一种轧件塑性系数的在线计算方法。针对塑性系数计算不准确或不稳定的问题,基于轧制理论模型,利用二次曲线拟合生产过程中的塑性曲线,在拟合曲线上求解压下量点处的切线斜率就得到了实际的塑性系数。这种处理方法使塑性系数的计算不依赖于过程计算机,可直接嵌入基础自动化中进行计算,根据轧制力的变化和辊缝的变化,不断的对塑性系数进行修正,提高系统的厚度补偿精度。(5)开发出过程控制支撑平台,并实现了在线应用。通过对过程控制系统结构的研究,在分析系统功能需求的基础上,开发出基于通讯进程和模型进程相分离的过程控制支撑平台,该平台封装了过程控制软件中的通讯环节,建立了一个基于事件调度、适合模型的编程、维护和调试的系统架构。该平台已经在多个中厚板的生产线实现了成功应用,极大的提高了过程控制系统的开发效率。(6)提出中厚板轧件平面尺寸测量方法。根据中厚板轧件平面尺寸的测量特点,提出测量方案,基于机器视觉技术,以灰度直方图均衡、快速中值滤波、Sobel算子边缘检测、Hough直线检测、采用空间矩方法进行亚像素定位以及8点标定方法为基础,建立了轧件侧弯测量系统,测量结果表明,测量精度可以满足工程要求。(7)提出等效轧件的概念,并应用于轧件的侧弯控制。在轧制过程中,轧件本身产生的不对称影响因素可以近似用一个位于轧制中心,带有一定楔形量的等效轧件来代替,对等效轧件实施侧弯控制与对实际轧件的侧弯控制是等效的,这种处理方法统一了不对称影响因素,便于进行计算和控制。(8)结合侧弯模型、轧件断面形状计算模型和轧件平面尺寸检测模型,开发出侧弯检测和控制应用软件。实现了轧件侧弯控制的在线应用,将生产过程中的轧件侧弯曲率限制在一定范围内,提高了产品的成材率。本文结合我国的中厚板生产过程,对中厚板轧制中的轧件侧弯现象进行了深入研究,开发出具有自主知识产权的侧弯检测与控制软件,对提高我国中厚板生产水平具有重要的意义。更多还原

【Abstract】 To the camber of rolled pieces problems by asymmetric influencing factors existed in plate production process, this paper puts forward camber prediction and control model, the on line correction method of temperature prediction, the self-learning method of deformation resistance and influence function method were analyzed. The Application software included camber measurement, camber prediction and camber control were developed based on the machine vision technology, and has obtained very good results in the on-line application of production practice.(1) The high accuracy prediction of temperature was realized in plate production process. The temperature evolution model of plate production have been built based on implicit difference method and finite element method, and determine the heat transfer coefficients of different conditions. According to the characteristics of plate production, the approach to the real temperature measuring method of rolled pieces were proposed on the basis of the verification each other between prediction and measured temperature. The method constantly study to the model parameters in measuring temperature of rolled pieces, and ensures the measured accuracy.(2) The self-learning system of deformation resistance was developed based on genetic algorithm. The model parameters value of deformation resistance determines the prediction accuracy of rolling force model. According to the influencing factors analysis of rolling force calculation error, the genetic algorithm were introduced the self-learning method of deformation resistance, and search the optimal value of deformation resistance on the basic of space exploration and optimization ability of genetic algorithm. The decision variable selection, the coding and decoding, the fitness evaluation and the terminal conditions process were implemented during development process of self-learning system. The results show that the optimization speed and accuracy can meet production requirement.(3) A camber predictive method of plate rolling was put forward. The theoretical formulas that predict the camber curvature of rolled piece were derived on the basis of the movement characteristics of rolled piece. In combination of temperature model and rolling force model, the section thickness distributions of rolled piece were calculated by using of improved influence function method for adaption to width variations. The camber predictive model for the improved influence function method coupling with camber theoretical formulas can analyze the effect of asymmetry factors on camber of rolled piece.(4) An on-line calculation method of plasticity coefficient was put forward. Based on the rolling theory model, a plasticity coefficient curve was fitted by using conic according to the inaccurate or unstable problems of plasticity coefficient calculation. The plasticity coefficient can be obtained by the tangent slope at reduction amount of conic. The process method does not depend on process control computer, and can be embedded into basic automation system immediately. According to the change of rolling force and roll gap, the plasticity coefficient can be adjusted constantly, and improves the compensation precision of AGC system.(5) A supporting platform of process control was developed and was used on-line. According to the research of process control system, a supporting platform of process control, based on the separated structure between communication process and model process, was developed on the basic of the analysis of system functional requirement. The platform encapsulates the communication procedure, and establishes system architecture based on event scheduling, suitable to programming and debugging, it has been applied to plate production line successfully, and greatly promote development efficiency of process control system.(6) A measured method of plane dimension was presented for plate rolled piece. A measured scheme was presented on the basic of the measuring characteristic of plane dimension for plate rolled piece. The camber measuring system, based on machine vision technology including gray histogram equilibrium, fast median filter, Sobel operator edge detection, Hough line detection, sub-pixel edge location using spatial moment and eight point’s calibration method, is established. The measuring results show that the measuring accuracy can meet the project requirements.(7) The concept of equivalent rolled piece was given, and applied to the camber control of rolled piece. The asymmetry factor can be replaced by using an equivalent rolled piece, which is located in rolling centerline, and has a certain amount wedge. The camber control way between actual and equivalent rolled piece is the same. The process way unifies asymmetry factor, and carries out calculation and control conveniently.(8) The Application software of camber detection and control was developed on the basic of camber model, section shape calculation model and plane dimension measuring model. The software implements on-line application of camber control, and can restricts the camber curvature of rolled piece in a certain scope, and improves finishing product rate.Under the background of the development of plate rolling in our country, this dissertation carried the analysis of camber phenomenon of plate rolling. The camber detection and control software for plate mill was developed, and it improves the plate production level of our country.更多还原