面向变维多阶段间歇过程的闭环故障诊断与容错控制策略研究

As one of the industrial production mode, the batch process has always received extensive attention in the problem of production safety. Batch processes are involved with the characteristics of batch reproducibility, 2D nature, multi-phase and other complicated features, and the dimension may be different in different phases. Combined with the wide application in closed-loop control, these factors all make the research on fault diagnosis and fault-tolerant control for batch processes become a difficult and challenging problem to be solved urgently. Intending to focus emphatically on some phenomena, such as the multi-phase feature existing in batches, variable-dimension switching between the adjacent phases, etc., using the strategy of 2D variable-dimension hybrid system, the closed-loop fault diagnosis and fault-tolerant control for batch processes will be studied. The main problem to be solved in this project are: The model establishment of variable-dimension multi-phase closed-loop system based on real industrial background, from which the quantitative analysis about the impact caused by closed-loop control on fault diagnosis performance index will be obtained; The design of closed-loop fault diagnosis unit under the condition of variable-dimension and multi-phase, from which the strategy of closed-loop fault diagnosis satisfying actual demands will be given; The collaborative design between the closed-loop fault diagnosis and fault-tolerant control strategy under the condition of variable-dimension and multi-phase, from which the fault-tolerant control strategy closely matching with the results of fault diagnosis will be figured out. The theoretical achievements achieved in this project will be verified and optimized on a typical batch process, namely the injection molding, through which the powerful theoretical basis and technical support will be provided for effective solution on the safe operation of batch processes in our country.

作为工业生产方式之一的间歇过程，其安全生产问题一直备受关注。间歇过程具有批次重复性、二维性、多阶段性等特点，不同阶段系统的维数也可能不同，加之闭环控制的广泛应用，这些都使得间歇过程故障诊断和容错控制的研究成为亟待解决的挑战性难题。本项目拟重点围绕间歇过程存在的批次过程多阶段、相邻阶段间的变维切换等现象，采用2D变维混杂系统策略，对其闭环故障诊断及容错控制进行研究，主要解决：实际工业背景下变维多阶段闭环系统模型的建立问题，由此得出闭环控制对故障诊断性能指标影响的定量分析结论；变维多阶段条件下闭环故障诊断单元的设计问题，由此给出满足实际需求的闭环故障诊断策略；变维多阶段条件下闭环故障诊断与容错控制律的协同设计问题，由此得出与故障诊断结果高度匹配的容错控制律。项目的理论成果将在典型的间歇过程——注塑过程中进行验证和优化，为我国间歇生产过程安全运行问题的有效解决提供有力的理论依据和技术支持。

间歇过程是一类复杂的批次生产过程，生产中是由多个阶段相互影响生产某一产品，不仅某一阶段需要控制，多个阶段也要一起考虑实现切换控制。故障发生，极有可能影响系统稳定运行甚至造成系统崩溃、财产损失或人员伤亡。针对上述情况及在间歇过程中的人们一直关注高精控制及算法优化等问题，本项目做了一系列工作，取得了较好的创新性成果。内容为：.一：针对带有不确定性的间歇过程，提出了一种鲁棒迭代学习预测跟踪控制策略，给出控制律优化设计。考虑输入时滞，依然利用上述方法，解决了系统具有时滞、输入输出约束和扰动等问题。针对带有不确定性的多阶段间歇过程，提出了一种鲁棒迭代学习控制方法，利用切换系统理论，结合相关切换条件并利用平均驻留时间方法，给出不同部分稳定运行时子系统的最小运行时间和控制律增益。接着又考虑了系统模型中具有时变时滞和外部扰动的情况，实现了多阶段间歇过程成本控制的研究及控制律优化设计。.二： 针对具有执行器故障的间歇过程，研究结果从线性系统出发延伸至非线性系统等控制及优化。针对线性系统具有的执行器故障，一种解决方案是将其视为干扰，设计故障允许范围内的鲁棒控制律，实现系统稳定运行，同时给出其故障范围。另一种方案是针对故障情况设计控制律，只需设计切换策略即可，即什么情况设计什么情况控制律，然后按照某一策略切换控制器即可。对那种故障少有发生的设备，这种办法极好，可实现节能减耗。当然，针对故障频发系统而然，设计可靠控制律即可。针对非线性系统，提出了一种二维模糊约束模型容错预测控制器的设计方法。设计了约束条件下的迭代学习容错控制律，实现了非线性系统的模糊控制。对于多阶段间歇过程，类似于上述正常情况下那样，所设计的容错控制律同样实现了故障情况下系统平稳运行及控制优化。.最后，以工业生产典型代表注塑成型过程、搅拌釜或三容水箱等设备为例，实现了这些设备中某些参数的控制。案例分析验证了所提方法的有效性和实用价值。