Control, identification and diagnosis of networked and hybrid systems for fault tolerance

该提案寻求对一项研究方案的支持，该方案旨在为联网和混合配置中的技术系统/流程开发先进的容错技术。在全球对工厂优化和信息集成的需求的推动下，将网络化引入系统自动化和控制成为流程工业的一个突出趋势。因此，我们现在正在经历一场技术系统的戏剧性变化，从传统的同构和集中式配置到混合和分布式配置。在这种情况下，容错问题变得更加严格。事实上，即使在偏远地区，系统一个部分的故障也可能通过其他部分传播并影响其他部分。作为一个重要的目标，我们打算研究网络带来的约束和资源，并提出新的基于网络数据传输的故障诊断设计工具，以取代传统的有线信号传输。将考虑时延、数据丢失和时变不确定性等基本问题。另一个目标是从混杂系统的角度研究容错控制系统，其中不同的系统操作，如多个故障，由离散模式描述，而每个模式下的系统动态行为由连续时间或离散时间状态演化控制。对这类混合系统的系统辨识及其在容错控制中的应用进行了研究。实际上，故障检测问题可以归结为离散模式估计问题，而在这种情况下，可以采用切换控制策略进行控制重构。最后，在这个项目中，我们试图通过将结果应用于真实世界的问题来开发故障诊断和容错控制的实际应用。拟议的研究方案是跨学科的，涉及系统和控制、信号、信息系统和运筹学的各种主题。它不仅在安全关键和高度自治的系统中有广泛的应用，而且在过程控制、制造和能源行业也有广泛的应用，这些行业现在对可靠性和可持续性的要求越来越高。

This proposal seeks support for a research program on developing advanced fault tolerance techniques for technical systems/processes in networked and hybrid configurations. Driven by the global need for plant optimization and information integration, introducing networking into system automation and control becomes a prominent trend in process industries. As a result, we are now experiencing a dramatic change of technical systems from traditional homogeneous and centralized configurations to hybrid and distributed ones. The fault tolerance issue becomes more stringent in this case. Indeed, faults in one part of the system can propagate through and affect the other parts even in remote locations. As one important objective, we intend to study the constraints and resources brought by the networks, and propose new design tools for fault diagnosis based on network data transmissions instead of traditional wired signal transmissions. Fundamental issues, e.g. time-delay, data loss, and time-varying uncertainties will be considered. Another objective is to study fault tolerant control systems from a hybrid system perspective, where different system operations, e.g. multiple faults, are described by discrete modes while the system dynamic behavior under each mode is governed by continuous-time or discrete-time states evolutions. System identification for such a hybrid system and its application to fault tolerant control will be researched. In fact, fault detection can be treated as the discrete mode estimation problem while the switching control strategy can be adopted for control reconfiguration in this case. At last in this program we seek to develop practical applications of fault diagnosis and fault tolerant control by applying the results to real-world problems. The proposed research program is inter-disciplinary that involves various topics in systems and control, signals, information systems and operations research. It can find vast applications not only in safety critical and highly autonomous systems but also in process control, manufacturing and energy industries, which nowadays have an increasing requirement on reliability and sustainability.