【作者】 张月蓉；

【导师】 王勇；

【作者基本信息】 山东大学 ， 机械制造及其自动化， 2012， 博士

【摘要】 调节阀是流体输送系统中重要的控制部件,在流体管道输送中,合理采用调节阀对调节介质的流量、压力等参数进行有效控制,进而控制流体脉动,对降低能耗,提高能源利用率具有十分重要的意义。调节阀广泛应用于化工、石油、冶金、电力、轻工、水利、船舶以及航空等领域,随着工业发展的需要,流体管道输送向着更大流量、更高压力的方向发展,因此大流量调节阀的应用更为广泛。钢厂利用混合煤气进行燃气-蒸汽联合循环发电的工程中,气体供给量大,对混合煤气的流量要求高达54000Nm3/h以上,用于混合煤气压力控制的调节阀公称通径超过200mm。大流量混合煤气的稳压控制是钢厂循环发电工程的关键技术问题。通常采用预防性定期检修的手段保证大流量调节阀工作的安全性和可靠性,但是预防性检修存在着“维修过剩”的问题,致使维修费用过高,导致可靠性降低,故障率上升；另一方面国内工业领域对大流量调节阀的需求量非常大,但是我国相关的调节阀故障诊断工作的深入研究不足,气体压力调节阀的故障诊断研究报道少见,尤其大流量气体压力调节阀故障诊断的报道就更为鲜见,高性能大流量调节阀的最先进技术都被国外公司垄断,而且许多关键的调节阀价格高达数十万,甚至数百万人民币。因此大流量调节阀的故障诊断方法从理论研究和实际应用都有大量的工作要做。由于大流量混合煤气调节阀应用场合的特殊性,在正常运行工况下很难提取故障发生时的事故特征,而且由于流场脉动,直接信号检测分析方法难以甄别异常的信号产生的原因；当开发调节阀新产品时,用于故障诊断的知识获取成为实现故障诊断的瓶颈。而基于模型的故障诊断方法将系统的模型和实际系统冗余运行,通过对比产生残差信号,剔除控制信号对系统的影响因素,基于模型的方法适合于调节阀的故障诊断。对于基于模型的故障诊断方法,建立数学模型是关键问题。调节阀中的介质流动受物理守恒定律的支配,因此建立调节阀系统的数学模型采用“灰箱理论”的方法,即将理论建模方法和系统辨识建模方法结合起来：机理已知的部分采用理论建模方法,机理未知的部分采用系统辨识建模的方法。输入输出数据是系统辨识的基础,由于大流量调节阀的工作环境的特殊性,采取做样机试验的方法在试验室环境下获取数据难以实现。随着计算机技术以及计算流体动力学分析技术的发展,对于传统的现场测试数据耗时长、成本高以及试验条件下模拟实际工况的局限性等问题,利用数值模拟手段分析流体流动问题的优越性和可靠性越来越明显。从计算流体动力学的角度对物理守恒定律的数学描述可以得到调节阀系统在时间和空间上定量描述流场的数值解,因此应用计算流体动力学的方法研究调节阀流场,可以克服试验室条件的局限性,在场变量近似值的基础上获取输入输出数据,然后按照最小二乘法辨识方法,建立调节阀系统的数学模型。通过混合煤气的组分相关数据的分析计算,给出了在调节阀流场中混合煤气特性的判断方法和依据；研究不同开度情况下调节阀流道CAD模型的快速建模方法和分块划分网格方法,可以满足不同情况的数值试验要求、提高建模效率,并得到网格质量较好的网格模型,从而保证计算准确性；确立调节阀流场的控制方程、离散化方法以及数值计算方法,得到调节阀流场CFD模型的建模和求解的思路,解决调节阀流场数值试验中关于试验方案、试验过程以及结果后处理等问题,并分别完成了大流量调节阀正常工作状态、泄漏故障状态以及堵塞故障状态的数值试验。根据数值试验的结果分析输入输出变量之间的变化关系,确定系统辨识模型的模型类型。根据参数辨识方法建立准则函数,对调节阀流场的基于压力和开度的瞬态模型进行参数估计；利用数值试验结果分析压力损失和开度变化对流场滞后现象的影响规律,确立采样时间的函数类型,通过参数估计建立调节阀出口压力的采样时间控制模型；根据正常状态和故障状态的数值试验结果,考察瞬态模型预测的调节阀出口压力变化规律,采用实际出口压力相对于预测值的变化趋势,即出口压力相对于预测值偏差的大小和趋势作为故障征兆,为实现泄漏故障和堵塞故障的诊断提供依据。根据调节阀流场的数学模型研究结果,给出调节阀故障诊断中残差的设计原则以及阈值和故障分析残差的定义,提出基于统计方法的残差算法,完成调节阀故障指示器的设计；根据确定的故障监测与诊断的详细步骤,分析硬件和软件的功能要求,确定所需硬件,选择软件开发平台,结合开发平台特点确定调节阀故障监测与诊断过程的数据流向,为基于瞬态模型的调节阀故障监测与诊断的相关软件开发与应用提供指导方案。选择试验平台中的测试设备和试验管路,确定相关的结构尺寸和元件安装尺寸,在稳压试验平台进行调节阀流场的测试试验；同时根据测试平台的试验管路结构尺寸,利用快速建模方法分别建立不同开度下试验管路中小流量调节阀的三维模型并分块划分网格,进行小流量调节阀的数值试验,并根据测试元件的安装尺寸处理试验结果；根据试验结果,建立小流量调节阀流场的瞬态模型和采样时间控制模型,对测试试验结果和小流量调节阀瞬态模型的预测结果进行比对。比对结果证明：论文提出通过对残差信号的分析实现调节阀故障监测与诊断的方法是可行的,即瞬态模型对于调节阀流场状态的预测结果具有有效性,同时证明采用CFD方法进行调节阀流场数值试验具有可靠性。综上所述,论文以大流量混合煤气的压力调节阀为研究对象,应用计算流体动力学的有关理论,结合CFD软件,对大流量调节阀的瞬态压力场规律进行研究；应用系统辨识的有关理论得到了调节阀的压力场参数,建立了调节阀瞬态流场的数学模型；应用故障诊断理论提出大流量调节阀故障监测与诊断的方法,从而既保证调节阀在工作过程中的可靠性,又避免调节阀的“维修过剩”问题。论文研究结果对推动钢铁厂的燃气-蒸汽联合循环发电工程的进一步发展具有重要的理论意义和工程实践价值,也为高性能大流量调节阀应用的国产化提供科学的依据和技术支撑。本课题得到国家高技术研究发展计划项目“钢厂循环发电工程中大流量煤气的智能优化控制及应用技术(2008AA042130)”和高等学校博士学科点专项科研基金资助项目“大流量高频响核电控制阀的多场耦合振动与泄漏自感知机理研究”(20110131110042)的支持。更多还原

【Abstract】 In the process of piping fluid, the control valve regulates the flow and the pressure of a medium to control fluid pulse. Along with the development of the industry, new demands have been presented on the performance of the control valves, such as greater flow, higher pressure, etc. Usually, the security and the reliability of control valve are ensured by preventive maintenance, which can result in a "maintenance-surplus" problem. The maintenance surplus could result in some adverse effects, such as larger maintenance costs, reduced reliability and increased failure rate.On the other hand, the control valves are in great demand in the domestic industry. But the fault diagnosis research of the control valve is still insufficient in China. Correspondingly, there are few studies of gas pressure valve, especially fault diagnosis of the reports about large valve. The most advanced technology of high-performance control valve is monopolized by overseas manufactory, and many control valves used in the key portion are of high price, as much as thousands of, or even millions of Chinese yuan. Therefore, much work has to be done on fault diagnosis method of the large control valve, which includes theoretical research and practical application.The aim of this paper is to study the control valve, which regulates combination gas pressure, and to determine its fault detection and diagnosis method, and then to ensure the reliability of the control valve. In this way, the maintenance-surplus problem could be resolved, and scientific basis and technological support are provided in order to realize the localization of the application for the large control valve with high performance.Fault diagnosis method is mainly divided into:analytic model-based method, signal processing method, and knowledge-based method. Combined with the characteristics of the three methods, the system of control valve is analyzed. First, due to the particularity of the application situation, it is difficult to extract failure characteristics when the accident occurs in normal operation condition. In addition, the method using direct signal detection cannot discriminate against the reasons of abnormal signals for the flow pulsation. Second, the acquisitions of knowledge used for fault diagnosis become the bottleneck of realizing fault diagnosis when developing new products. Third, medium flow in control valves follows the conservation laws of physics, so the mathematical description of the system, by which the math models of control valve are established, can be determined in the view of fluid dynamics.Therefore, it is feasible to realize fault detection and diagnosis of the control valves using the residual signals. The residual results from the redundancy, which is generated from the mathematic mold and real system. The paper works on the research of system theory and experiment for model based fault diagnosis method of the control valve, which regulates mixed gas flow. The work provids the important theoretical basis and technical support for the application of fault detection and diagnosis method.As to the model-based method, the mathematical model is the key problem. The method of "Grey box" theory is used to establish a mathematic model of the control valves system. Using "Grey box" theory means to combine theoretical modeling method with identify modeling method. In other words, one part of the mechanism which is known uses the theoretical modeling method, and the other part of the mechanism which is unknown uses the identify modeling method. According to system identification standards, the mathematic model should be gained. Input and output data are the foundation of system identification.Due to the particularity of working environment for large control valve, it is difficult to achieve data by the method of prototype test in the laboratory environment. It is effective to overcome the limitations of laboratory conditions to research flow field of the control valve by the method of computational fluid dynamics. According to the present approximation for variable input/output data, it is viable to establish mathematic models of the control valve system by method of the least square identification.Through analyzing the component’s data of the blended gas related data, the rules and method are given in the paper, by which the characteristics of the combined gas in the flow field of the control valve could be discriminated. It is proven effective in practice to use the method of automatic rapid3-D modeling for Flow Field to improve modeling efficiency. The method of block grid is useful to get a better quality of grid mesh. The control equations of flow field about the control valve are established, and then the methods of discretization and numerical calculation are determined. The tests, including normal condition, leakage and block fault states have been done based on the above work.According to the. results of the tests, the relationship between input and output variables has been gained by a numerical analysis in order to establish the type of system identification model. The transient model is based on pressure, and the opening is given by the method of system identification in the paper. In additional, the control model of sampling time for outlet pressure is gained by analyzing the pressure loss and opening change similarly. Comparing the results of normal and faults state, it has been found that there are the differences between detect value and test value for outlet pressure in different states. The differences present a trend, by which the fault detects and’ diagnosis for control valves can be realized.The tests in the laboratory have been done in order to prove the validity of the transient model and the reliability of CFD method. The main contents are choosing the equipment and pipes in experimental platform, and ensuring the size of structure and installation. While according to the size, it is necessary to establish the3D models and mesh models for the control valves in different openings. Through the results of the CFD test for the control valve in small size, the mathematic model for small-size valve could be gained. The theory and method of fault detect and diagnoses are submitted, which is through the model-based method for the large control valve. The emphases are the principle of residual design and design of a fault indicator. According to detailed steps of the fault detection and diagnosis, it is needed to analyze the function of hardware and software in order to select the development platform of software and ensure the data flow when a fault detection and diagnosis is going on. The guidelines of development and application are given therewith; of which are the fault detection and diagnosis of control valve with the large flow based on a transient model.This research is based upon work supported by National High-tech Research Development Program of China under Grant No.2008AA04Z130and by Specialized Research Fund for the Doctoral Program of Higher Education under Grant No.20110131110042.更多还原