6 January 2025 in IEEE Transactions on Control of Network Systems
It has long posed a challenging task to optimally deploy multi-agent systems (MASs) to cooperatively coverage poriferous environments in real cooperative detection applications. In response to this challenge, this paper proposes a hierarchical coverage control (HCC) protocol for MASs to perform sector-based coverage operations. First, a distributed Voronoi partition-based sweep-and-assign protocol, combined with a sectorial partition method, is developed, enabling the segmentation of a whole poriferous region into multiple sectorial sub-subregions. Following this procedure, an extreme search scheme is developed to determine the optimal amount of agents for each subregion. After this allocation, a distributed controller is proposed to deploy the agents, considering non-holonomic constraints, into designated niche positions. Additionally, sufficient conditions are derived to guarantee the asymptotical stability of the present HCC. The analytical challenge of the present study stems from the non-convex characteristics inherent in the loss function induced by porous environments. This non-convexity renders the task of globally optimizing the assignment of MASs intractable. Finally, numerical simulations are conducted to validate the effectiveness of the present coverage control approach.
1 January 2026 in IEEE Transactions on Industrial Informatics
8 January 2024 in IEEE Journal of Emerging and Selected Topics in Industrial Electronics
To extract the household attribute information from the large volume of smart meter data, this study proposes a resident characteristics estimator. Such an estimator enables energy suppliers to provide personalized services whereas to assist customers to reduce energy consumption. By leveraging the potential connections among different characteristics, a deep convolutional neural network-based cross-task transfer learning (CTTL) scheme is designed, which makes full use of the knowledge learned from one characteristic (like retirement status)-based classification to estimate another relevant characteristics (like age). Extensive experiments are conducted on the Irish dataset with 4232 households to substantiate the superiority of the proposed scheme compared with conventional deep convolutional neural networks-based learning methods.
7 January 2025 in IEEE Transactions on Control of Network Systems
In this paper, a distributed algorithm is proposed to identify the node and edge numbers of anonymous leader-follower networks. The present distributed method works merely by exchanging scalar information by local intercommunication. The merit of the algorithm lies in identifying the required network parameters in time-varying topological networks without special initialization. Sufficient conditions are derived for cascading systems to theoretically guarantee the exponential convergence of the individual estimation to the total number of nodes or edges merely by local information exchanging. Finally, numerical simulations are conducted to substantiate the effectiveness of the present identification algorithm.
1 November 2026 in IEEE Transactions on Control Systems Technology
29 December 2025 in IEEE Transactions on Control of Network Systems
In this paper, a distributed algorithm is proposed to identify the sizes of abundant networked systems. Significantly, the present distributed method works merely via exchanging scalar information by pairwise interactions. By generating a sinusoidal signal with a randomized frequency at each node, the network size can be individually calculated by each node merely according to the frequency decomposition of the locally accessible time series. Accordingly, a prescribed-time distributed average tracking (PT-DAT) protocol is developed for information aggregation and Fast Fourier Transform (FFT) is introduced for each node's frequency extraction. The virtue of the present design lies in eliminating the leader-follower structures required by most of the existing literatures. Moreover, it works for dynamic networks with arbitrary initial conditions. Sufficient conditions are derived for the present PT-DAT protocol to theoretically guarantee the prescribed-time convergence of the present algorithm. Finally, numerical simulations are conducted to substantiate the effectiveness of the present algorithm.