基于经济性能指标的MPC监控

发布时间：2018-05-23 06:46

本文选题：模型预测控制 + 经济效益　；参考：《浙江大学》2016年硕士论文

【摘要】：随着控制理论的不断发展,以MPC(模型预测控制)为代表的先进控制器为工业过程取得了良好的经济效益,但同时人们也意识到控制器的运行初期能够满足各种性能要求,但控制器没经过定期维护其性能会随着时间推移而退化进而影响经济效益。因此要保证生产过程安全、高效、优质、低耗的运行,就需要对其进行控制性能和经济效益的监控与评估。本文研究基于经济效益指标的MPC监控问题,并提出了一套监控方案。具体工作如下：1.基于MV(过程输入)统计指标的MPC控制性能监控。针对MPC性能下降问题,提出了一种MPC控制性能监控方法,通过分析MV的统计指标,区分模型失配和不可测干扰的影响。通过IMC(内模控制)对MV的均值和方差进行理论分析,说明了利用MV的统计指标进行MPC控制性能监控的可行性。与利用CV统计指标的监控方法进行对比,表明利用MV的统计指标进行监控对于模型失配和不可测干扰更加敏感,更容易区分模型失配和不测可干扰的影响。最后通过仿真验证了方法的正确性。2.MPC经济效益监控。建立了一种基于历史数据的MPC经济效益监方案,定义了经济效益公式和历史数据基准。在此方法中将待监控变量转化成以货币单位衡量的经济价值,可以更加直观地衡量全厂的经济效益。同时提出了一种基于系统辨识的优化变量监控方案,此方法可以监控工业过程中的优化变量是否处于最优状态。并以火电厂为实例,对其热力发电单元、燃烧单元、负荷跟踪特性、脱硝单元的经济效益进行了分析,并提出了多种基准下的经济效益监控方案。3.基于经济效益的诊断系统及监控软件实现。此诊断方案主要用于分析控制系统经济效益下降的原因并且利用相关变量分析方法定位引起经济效益下降的控制变量。本文最后将提出的基于经济效益的MPC监控方案进行了完整的软件实现,并利用电厂实际数据进行了仿真。
[Abstract]:With the development of control theory, the advanced controller, represented by MPC (Model Predictive Control), has achieved good economic benefits for the industrial process, but at the same time, people also realize that the controller can meet various performance requirements in the early stage of operation. However, the performance of the controller will degrade with time without regular maintenance, and then affect the economic benefits. Therefore, in order to ensure the safe, efficient, high quality and low consumption operation of the production process, it is necessary to monitor and evaluate its control performance and economic benefits. In this paper, the problem of MPC monitoring based on economic benefit index is studied, and a set of monitoring scheme is put forward. The work is as follows: 1. MPC control performance monitoring based on MVstatistical index. Aiming at the problem of MPC performance degradation, a MPC control performance monitoring method is proposed. By analyzing the statistical indexes of MV, the effects of model mismatch and undetectable interference are distinguished. Through the theoretical analysis of the mean and variance of MV (Internal Model Control), the feasibility of using the statistical index of MV to monitor the control performance of MPC is demonstrated. Compared with the monitoring method using CV statistical index, it shows that the MV statistical index is more sensitive to model mismatch and undetectable interference, and it is easier to distinguish the influence of model mismatch and unexpected disturbance. Finally, the correctness of the method is verified by simulation. 2. MPC economic benefit monitoring. A MPC economic benefit monitoring scheme based on historical data is established, and the economic benefit formula and historical datum are defined. In this method, the variables to be monitored can be transformed into the economic value measured by monetary units, which can be used to measure the economic benefits of the whole plant more intuitively. At the same time, an optimal variable monitoring scheme based on system identification is proposed, which can monitor whether the optimal variables are in the optimal state in the industrial process. Taking thermal power plant as an example, the economic benefits of thermal power generation unit, combustion unit, load tracking unit and denitration unit are analyzed. The realization of diagnosis system and monitoring software based on economic benefit. This diagnostic scheme is mainly used to analyze the reasons for the decline of economic benefits of the control system and to locate the control variables that cause the decline of economic benefits by using the method of correlation variable analysis. At the end of this paper, the MPC monitoring scheme based on economic benefit is implemented with complete software, and the actual data of power plant are simulated.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F426.61;TP13

【参考文献】