Department of Mechanical Engineering

In cloud manufacturing systems, the multi-granularity of service resource and design task models leads to the complexity of cloud service matching. In order to satisfy the preference of resource requesters for large-granularity service resources, we propose a multistage cloud-service matching strategy to solve the problem of matching tasks and resources with different granularity sizes. First, a multistage cloud-service matching framework is proposed, and the basic strategy of matching tasks with cloud services is planned. Then, the context-aware task-ontology modeling method is studied, and a context-related task-ontology model is established. Thirdly, a process-decomposition method of design tasks is studied, and the product development process with small granularity tasks is established. Fourthly, a matching strategy of ontology tasks and cloud services is studied, and the preliminary matching is accomplished. Finally, intelligent optimization is carried out, and the optimal cloud service composition is found with the optimal design period as the objective function. With the help of the preceding method, the service matching of maximizing the task granularity is realized on the premise of ensuring the matching success rate, which meets the preference of resource requesters for large-granularity service resources.

System current state estimation (or condition monitoring) and future state prediction (or failure prognostics) constitute the core elements of condition-based maintenance programs. For complex systems whose internal state variables are either inaccessible to sensors or hard to measure under normal operational conditions, inference has to be made from indirect measurements using approaches such as Bayesian learning. In recent years, the auxiliary particle filter (APF) has gained popularity in Bayesian state estimation; the APF technique, however, has some potential limitations in real-world applications. For example, the diversity of the particles may deteriorate when the process noise is small, and the variance of the importance weights could become extremely large when the likelihood varies dramatically over the prior. To tackle these problems, a regularized auxiliary particle filter (RAPF) is developed in this paper for system state estimation and forecasting. This RAPF aims to improve the performance of the APF through two innovative steps: (1)regularize the approximating empirical density and redraw samples from a continuous distribution so as to diversify the particles; and (2)smooth out the rather diffused proposals by a rejection/resampling approach so as to improve the robustness of particle filtering. The effectiveness of the proposed RAPF technique is evaluated through simulations of a nonlinear/non-Gaussian benchmark model for state estimation. It is also implemented for a real application in the remaining useful life (RUL) prediction of lithium-ion batteries. © 2011 IOP Publishing Ltd.

The realization of long-range ferromagnetic order in two-dimensional van der Waals crystals, combined with their rich electronic and optical properties, could lead to new magnetic, magnetoelectric and magneto-optic applications. In two-dimensional systems, the long-range magnetic order is strongly suppressed by thermal fluctuations, according to the Mermin-Wagner theorem; however, these thermal fluctuations can be counteracted by magnetic anisotropy. Previous efforts, based on defect and composition engineering, or the proximity effect, introduced magnetic responses only locally or extrinsically. Here we report intrinsic long-range ferromagnetic order in pristine Cr₂Ge₂Te₆ atomic layers, as revealed by scanning magneto-optic Kerr microscopy. In this magnetically soft, two-dimensional van der Waals ferromagnet, we achieve unprecedented control of the transition temperature (between ferromagnetic and paramagnetic states) using very small fields (smaller than 0.3 tesla). This result is in contrast to the insensitivity of the transition temperature to magnetic fields in the three-dimensional regime. We found that the small applied field leads to an effective anisotropy that is much greater than the near-zero magnetocrystalline anisotropy, opening up a large spin-wave excitation gap. We explain the observed phenomenon using renormalized spin-wave theory and conclude that the unusual field dependence of the transition temperature is a hallmark of soft, two-dimensional ferromagnetic van der Waals crystals. Cr₂Ge₂Te₆ is a nearly ideal two-dimensional Heisenberg ferromagnet and so will be useful for studying fundamental spin behaviours, opening the door to exploring new applications such as ultra-compact spintronics.