Recent research has shown that the Macroscopic Fundamental Diagram (MFD) is an effective way to manage traffic flow and mitigate congestion, which regulates bidirectional transfer flows at regional boundaries via perimeter control. In practice, perimeter control may fail to achieve desired control targets due to adverse conditions such as inclement weather or accidents, which may result in loss of actuator effectiveness faults. To address this challenge, this paper proposes two adaptive fault-tolerant perimeter control schemes to handle both time-invariant and time-varying actuator faults. To cope with time-invariant actuator faults, a data-driven fault-tolerant perimeter control approach is presented, which employs adaptive dynamic programming to approximate the solution of the Hamilton-Jacobi-Bellman equation. For dealing with time-varying faults, an adaptive fault-tolerant perimeter controller is designed by linearizing the MFD function and ensuring tracking of a given reference model. Theoretical analysis proves that the tracking error converges to zero asymptotically. Finally, simulation studies validate the effectiveness of the proposed schemes in improving traffic management under actuator fault conditions.

As an emerging topic in industrial digital transformation, digital business development in the context of platformization has received widespread attention. A large number of industrial companies have established new platform-based systems for digital business development by integrating their original information systems. The unified platform development mode promotes the integration of previously decentralized knowledge. However, the massive expansion of the knowledge system under platformization causes it to be no easier for developers to master or understand the core knowledge (context, concepts, and elements) of the business to be developed. According to the above dilemmas we have observed in the industry, in this article, a domain knowledge network modeling method for the knowledge system under platformization and a GP-based rule generation method for recognizing core business knowledge in the domain knowledge network are proposed for the first time. Our experiment and practical case study verify that our method can recognize a set of core business knowledge from a large knowledge network efficiently, which could help developers understand the business to be developed with a lower cognitive load. We hope the idea of platform-based business development and core business knowledge recognition can provide a reference for those companies that need efficient digital business development.