【作者】 吴庆双；

【导师】 隋振；

【作者基本信息】 吉林大学 ， 控制理论与控制工程， 2008， 硕士

【摘要】 预测控制是20世纪70年代中后期在欧美工业领域出现的,它是在新型计算机控制算法基础上发展起来的,是一种基于模型的先进控制技术。Clarke等人于1984年提出了基于模型的广义预测控制(GPC)。它是以受控自回归积分滑动平均模型(CARIMA)为基础,在优化中引入了多步预测的思想,抗负载扰动、随机噪声、时延变化等能力显著提高,并结合辨识和自校正机制,表现出良好的鲁棒性。因此近年来受到学术和工程界的广泛注意和重视,并已有工业应用成果。本文参考大量国内外文献,针对辗环机系统模型做了仿真研究,实现辗环机加工的不可逆过程控制无超调。首先,对辗环机的工作原理进行阐述,目前国内外采用的控制算法还停留在经验算法的水平上,引出生产工艺的不可逆过程问题。介绍应天津天海精锻厂要求设计的D51-250型辗环机控制系统的设计。其次,对辗环机控制系统进行系统建模,本文改进得到一个较精确,便于实现的系统模型,以便进行仿真研究,和工业实现。然后,就解决上述问题,借鉴国内外先进科研成果,总结一种能够有效抑制超调的算法,并比传统的经验算法速度快,而且继承了广义预测控制优点:1.抗负载扰动、随机噪声、时延变化等能力显著。2.结合辨识和自校正机制,表现出良好的鲁棒性。本文在介绍辗环机的特殊性基础上,将辗环机系统模型与广义预测控制的抑制超调算法综合到一起,对不可逆过程的控制进行了系统的研究。通过引入广义预测控制的抑制超调算法,实现了不可逆过程的精确控制。在理论上给予了充分的证明,而且通过仿真实验进一步。仿真结果表明了本文所提策略的有效性,不仅有效抑制了超调,而且比经验算法响应速度快,满足了用户对辗环机精度和生产效率的需求。为其应用于工业现场控制奠定了基础。仿真实验的后续工作将是在辗环机设备上进行实验。由于时间紧迫、条件有限,未能在辗环机设备上进行实验,是本文工作的一大缺憾。更多还原

【Abstract】 Predictive Control was born in the late 1970s in Europe and the United States in the field of industry, which is developed on the control of a computer algorithm, is a model based on the advanced control technology. With the research and development of adaptive control algorithm, and the development and application of the impulse response to the step response for the non-parametric model predictive control algorithm, Clarke and others raised the generalized model predictive control ( GPC) in 1984 which is based on controlled autoregressive moving average model of integration(CARIMA), and it used the method of multi-step forecast, the capability of anti-load disturbance, random noise, delay variation was significantly increased. And combining literacy and self-correction mechanism, it has a good performance of robustness. So it is widely taken attention by the academic and engineering in recent years, and has applied in industry. In this paper, we read a large number of literature reference about Rolling Machine, and had a simulation about the model, we make an effort to resolve the irreversible process issue and the realization of control without overshoot.The paper introduces D51-250 ring-rolling machine control system designing, and introduces the principle of Rolling Machine, the control algorithm remain used the algorithm of experience on the current domestic and international, even though it can achieve a certain degree of processing requirements, but in terms of advanced degrees, but also it can not reach a higher extent. Because the Rolling Machine is a nonlinear system, delayed, the time-varying characteristics of complex systems, the establishment of model is very complicated, it not only need to do a lot of experiments and calculation, but also it need have a strong theoretical foundation of nonlinear system modeling. The process of Rolling Machine needs non-overshoot, because its processing is a irreversible process. In the production process of irreversible, it is not permitted overshoot in the work-piece machining, because these processes is different from the process of temperature adjusting, the temperature adjusting allows the temperature is above or below the set value, but in certain process, if the actual value is more than the set value then it will be impossible to return, such as Rolling Ring calls for the outer diameter of D, but the processing gets to the D`> D, then the outer diameter of the work-piece can no longer meet the requirements of diameter D, it is only greater than the set value. And this is an irreversible process. Irreversible process widely exists in forging process, such as coiling machine and paper machine process.Secondly, we build a system modeling for the Rolling Machine Control System. Because of the special nature of the Rolling Machine system, it is very difficult to build, but if the accuracy of the algorithm use a kind of model is not very high, and to building a system of the model which contain the basic feature of the model is feasible, thus, some of the factor does not affect the system, we build a Rolling Machine model which neglected the factor. The model is based on adopted analytical method and system identification method combination of methods. Rolling Machine System model, taking into account the whole process of the major factors, can reflect a more complete about the matter of fact; can be used to design Rolling Machine open loop control system, optimal control system, adaptive control system, predictive control systems.Then, to solve the above problems, learn from the advanced research at domestic and abroad, we summed up a algorithm which can effectively inhibit overshoot, and the algorithm has much faster speed than the traditional experience algorithm, and has the GPC advantages:1) Anti-load disturbance, random noise and delayed variation was significantly improved;2) Combining literacy and self-correction mechanism, good performance robustness.The main approach is to use the model derived j step, then output the optimal value of y ( k + j), in the moment of k can get the compensation value for time t of the moment of k +1 by calculation, that is to say, to use the changed trend of incremental control to amend the variable which get from the Optimal control of open-loop control. Then we get the expression u(k+1) in the moment of k, we used the changed trend to amend the control incremental. This will be more precise tracking, and it played a very prominent role in inhibiting the overshoot.The whole system of Rolling Machine is a complex system, it’s nonlinear, time-delay, and has many other characteristics, it needs a large number of experimental and calculated to build an accurate system model, and this system is a highly nonlinear systems, which requires strong nonlinear theory of knowledge. And the characteristics of predictive control is through a multi-step prediction, rolling optimization and feedback correction to achieve control strategy that will improve the object delay and order Robust , in other words, the Rolling Machine System does not require model has high precision, and it can achieve better control performance. So it can reasonably neglect many restrictive conditions for complex systems of Rolling Machine System modeling, and a simple interpretations of the model is established, which is by generalized predictive control to achieve control requirements and the no overshoot. Comparing with the original control method, it greatly improved the speed and accuracy of processing.This paper introduces the special nature of the Rolling Machine system, it integrated Rolling Machine and overshoot suppression algorithm of Generalized System Model predictive control together, and have researched the irreversible process of the control system. By citing overshoot suppression algorithm of GPC, it achieves precise control in irreversible process. This paper gave full proof not only in theory, but also has done a lot of simulations. The simulation results show the effectiveness of this strategy. And its industrial application laid the foundation for control of the scene. The simulation result and the Rolling Machine has been molding can prove that the strategy is feasible.更多还原