Hu Yefa

Correlation-aware resource service composition and optimal-selection in manufacturing grid

For a multi-resource service request task (MRSRTask) in manufacturing grid (MGrid) system, in addition to the search for all qualified resource services according to each subtask, the system selects one candidate resource service for each subtask. Then the system generates a new composite resource service (CRS) and selects the optimal resource service composite path from all possible paths to execute the task with the given multi-objective (e.g., time minimization, cost minimization and reliability maximization) and multi-constraints. The above problem is defined as multi-objective MGrid resource service composition and optimal-selection (MO-MRSCOS) problem. The formulation is presented for an MO-MRSCOS problem. The correlations among resource services are taken into account during MGrid resource service composition, and a QoS description mode supporting resource service correlation is presented. The basic resource service composite modes (RSCM) for CRS are described, and the principles for translating a complicated RSCM into a simple sequence RSCM are presented for simplifying the resolving process and complexity of MO-MRSCOS problem. A new method based on the principles of particle swarm optimization (PSO), is proposed for solving MO-MRSCOS problem. Unlike previous works: (a) the proposed PSO algorithms combine the non-dominated sorting technique to achieve the selection of global best position and private best position; (b) the parameters of particle updating formulation in PSO are dynamical generated in order to make a compromise between the global exploration and local exploitation abilities of PSO; (c) permutation-based and objective-based population trimming operators are applied in PSO to maintain diversity of solutions in population. The experimental results and performance comparison show that the proposed method is both effective and efficient.

Resources publication and discovery in manufacturing grid

In the manufacturing grid’s architecture, Resources Management System (RMS) is the central component responsible for disseminating resource information across the grid, accepting requests for resources, discovering and scheduling the suitable resources that match the requests for the global grid resource, and executing the requests on scheduled resources. In order to resolve the problem of resources publication and discovery in Manufacturing Grid (MGrid), the classification of manufacturing resources is first researched after which the resources encapsulation class modes are put forward. Then, a scalable two-level resource management architecture is constructed on the model, which includes root nodes, domain nodes and leaf nodes. And then an RMS is proposed, and the resources publication and discovery mechanism are detailedly described. At last, an application prototype is developed to show the validity and the practicability of the proved theory and method.

A Redundant-Sensor-Based Fault Reasoning Technique for Multi-Sensors

In magnetic bearing systems or other relative devices with multi-sensors which states are not easy to be monitored and tested, multi-valued logic algebra based on sequential variables is suggested to be utilized in the fault diagnosis of the multi-sensors. The states of sensors and devices are both discretized to be three logic values: normal, abnormal and a transition state between them. On the basis of proving several relative theorems that can reveal the relation between the states of the sensors and that of the devices, the experiment arrangement is discussed in detail when the state of a device is anyone of the three states respectively. All the completed experiments should be necessary and be without redundant ones. A novel redundant-sensor-based fault reasoning technique is presented to be used for fault diagnosis of multi-sensors according to the experiment results. The reasoning result shows that it is not in all the occasions that the state of every sensor can be determined. Fault diagnosis should be carried out early when there are not too many sensors with fault.

A WWW-based customer-driven product design and manufacturing solution for mass customization

Consumers needs are too diversified to be fit individually, mass customization become more and more necessary driven by the customer-oriented market. In this paper, the concept of customer-driven product design and manufacturing of mass customization are discussed. The Web-based solution which enables the interactive design and collaborative development of customizing products is put forward. The architecture and main modules of prototyping implementation platform are introduced. These provide the integrated platform for mass customization. Customers can have direct access to the product design project and incorporate new product development with designer by using a Web-browser or other application systems based on the platform. An instance is given in the last part of this article.

Magnetic Bearings Control Using Fuzzy Logic and a Single Neuron

Linear controllers cannot get satisfactory results for active magnetic bearings (AMB) due to the highly nonlinear and open-loop unstable electromechanical dynamics. This paper proposes a combined controller which integrates a fuzzy logic controller (FLC) with a single neuron controller for AMB. The FLC operates with a small single neuron control component when the system error is big; the FLC and the single neuron controller operate simultaneously when the system error is small. The proposed controller overcomes the defect of the FLC which has a steady state error, and the defect of a single neuron controller which has unsatisfied dynamic performance. Simulation results demonstrate that the proposed controller has improved the dynamic performance and the disturbance rejection ability of AMB.

Inner Mechanism-Based Real Time Data Fusion of the Multisensors of Magnetic Bearings

On the basis of introducing the basic principle of magnetic bearings the data model of the multisensors of magnetic bearings is established. The data are considered to obey normal distribution. Variance analysis method is proposed in the data fusion. The condition is derived to determine whether or not the sensors are wrong. It is pointed out that an industrial computer or two embedded single microprocessors can realize the real time control and data process. When two microprocessors are used one is used to control the rotor keeping at the balance place, and the others task is to finish data fusion. The system will provide fault prompt when there exists fault according to the result of data fusion. In order to positioning the fault inner mechanism should be considered in fault analysis. Inner mechanism is mainly decided by the position arrangement of the sensors and vibration of the rotor. Some relative experimental results are given.