概率网络化离散事件系统的诊断与监控及概率ω-自动机的判定性

The supervisory control and fault diagnoses of probabilistic discrete-event systems (probabilistic DES) are two frontier issues in the fields of system control and artificial intelligence. In almost the existing researches on probabilistic DES, it is assumed that the communication environment is perfect, that is, there are no delays and losses in the communication between the supervisors/diagnosers and the plants. However, it is inevitable for the communication delays and losses under the shared network environment. Recently, the supervisory control of (conventional) DES is considered with the assumption that there exist communication delays and losses in the channels of observation and control, that is so called networked DES. The present project focuses on investigating the probabilistic DES in the condition of communication delays and losses (we name it as probabilistic networked DES). Specifically, the following problems are involved: (1) The supervisory control problems, including the existence of supervisors and supervisor synthesis, non-blocking supervisory control, decentralized control etc. (2) The fault diagnosis problems, including centralized diagnoses, decentralized diagnoses, safe diagnoses etc. (3) The minimization of communication and the optimal control. (4) The decidability of the equivalence of probabilistic ω-Automata. We would investigate the probabilistic DES in more generalized framework and plan to solve some key problems. We expect this project can further develop the control and diagnosis theory of DES, and also may bring several new theoretical principles and practical solutions for AI.

概率离散事件系统(概率DES)的控制及错误诊断是系统控制领域和人工智能领域中的前沿研究课题。目前概率DES的研究大都是在理想的通信环境中开展，即假设监控器或诊断器与系统之间的通信无延时且无信息丢失。然而，在网络环境下通信延时和信息丢失是难以避免的。因此，最近人们在通信延时和丢失的条件下研究确定性DES，即网络化DES。本项目将在概率通信延时和丢失的条件下研究概率DES(概率网络化DES): (1) 监督控制问题：监控器的存在性及监控器综合问题、非阻塞的监控问题、分散控制问题等；(2) 错误诊断问题：集中式诊断问题、分散诊断问题、安全诊断问题等；(3) 通信最小化和最优控制问题；(4) 作为模型检测的基础之一，研究概率ω-自动机的等价性判定。本项目的研究可能为人工智能的错误诊断提供新的理论依据和解决方案。

概率离散事件系统(概率DES)的控制及错误诊断是系统控制领域和人工智能领域中的前沿研究课题。目前概率DES的研究大都是在理想的通信环境中开展，即假设监控器或诊断器与系统之间的通信无延时且无信息丢失。然而，在网络环境下通信延时和信息丢失是难以避免的。因此，最近人们在通信延时和丢失的条件下研究确定性DES，即网络化DES。本项目将在概率通信延时和丢失的条件下研究概率DES(概率网络化DES): (1) 监督控制问题：监控器的存在性及监控器综合问题、非阻塞的监控问题、分散控制问题等；(2) 错误诊断问题：集中式诊断问题、分散诊断问题、安全诊断问题等；(3) 通信最小化和最优控制问题；(4) 研究与量子自动机相关的量子DES。本项目的研究可能为人工智能的错误诊断提供新的理论依据和解决方案。