Jin Yi-hui

A MES design based multi-Agent system

As a center of enterprise integration, the manufacturing execution system (MES) integrated with the organisation layer and production layer business software forms a total solution of enterprise scheme. However, MES still can not meet the expectations of application completely. This paper presents a solution based on the multi-agent system, by which MES can be implemented better for its many virtues.

Modeling supply chain based on multi-agent system

Now supply chain management and optimization are becoming the focus for enterprises to improve their comprehensive competitive ability. However, as a typical dynamic complex system, the supply chain imposes great difficulty on traditional modeling theories and approaches. The autonomy of an agent and the coordination mechanism based on message-passing make a multi-agent system (MAS) a promising tool to model complex systems. We first introduce the concept of an agent, and then describe the structure, dependencies and coordination mechanism of a general MAS in an analytical form. Finally, the structure, dependencies and control in a supply chain are discussed based on its constituent elements, and a framework and approach of modeling a supply chain based on a MAS are proposed.

The research on knowledge management and knowledge portal framework based on agent mechanism

With the fast development of information techniques, the problem of processing and transferring information with low efficiency is improved. But another problem is arising, in which the scarcity of knowledge and the creating ability for knowledge become the bottlenecks of the social and economic development. The concept of knowledge management (KM) is presented to improve the organization and utilization of information and knowledge. The concept of knowledge management and the difference between KM and information management (IM) are studied in the paper. After analyzing and extracting the key problem in KM, the concept of a knowledge portal is presented. A knowledge portal framework based on an agent mechanism is presented with further analysis on the function demand of the knowledge portal. The primary analysis and design of the function and structure of the framework are introduced and the general structure of agent is proposed.

A continuous process model for real-time scheduling based on hybrid Petri net

The structure and methodological features of a new optimal planning and dynamic scheduling system for a total plant continuous manufacturing process is introduced. A hybrid Petri net, a combination of a timed continuous Petri net and extended Petri net, is developed to model the process together with static equipment models and expert knowledge. Definitions of the system components and model construction steps are discussed, while a simplified example is also presented.

Coordination of multi-factory production planning in the process industry based on internal-prices

Abstract A large enterprise may consist of many factories with decision-making authority. The traditional centralized planning method is usually inefficient for this kind of enterprise. A distributed coordination algorithm based on internal-prices is presented in this paper. Within this algorithm, only the factory production planning model needs to be set up, which can be solved by its local information. Through updating the internal-prices in the coordination center, a near optimal solution of the multi-factory production planning is achieved. Numerical testing results show that the internal-prices converge.

Bottleneck analysis of the generalized networks

Network simplex algorithm is an effective algorithm for the generalized network, which gains negative cost graphs by constructing augmented forest structure, and then augments flows on the negative cost graphs. Bottleneck structure is presented based on the augmented forest structure in this paper, then negative cost augmented networks are constructed, the arcs that block the negative cost augmented network are the elements of the bottleneck. The bottleneck analysis of the whole network is completed by decomposing the network. An example is presented in the end.

The optimal control of flatness in four-roll CVC mill

This paper describes the optimal control and mathematical model of the four-roll CVC mill. From the practical applications, it is suitable for flatness control of the similar mills.

The flatness preset on cold rolling and its adaptation and inheritance algorithm

Describes the presetting of flatness control for cold rolling and its adaptation and inheritance algorithm.

Multivariable Optimal Control of the Cold Mill Flatness

Abstract In this paper, the two kinds of techniques for flatness control are analyzed and investigated. The first one is based on flatness defect identification, and the other is based on multivariable optimal search technique. After the mathematical analysis and the comparison with the practical process data, both the advantages and disadvantages of the two approaches are evaluated.

A Modelling Method Based on the Wavelet Transform and Artificial Neural Networks

Abstract A new modelling method called Discrete Wavelet Transform Model Identification (DWTMI) algorithm is presented. First, the basic concept of the wavelet transfonn is introduced. It is adopted to process and reconstruct the input and output data of a low sample rate system. An Artificial Neural Network (ANN) model of this kind of system is established from the processed data, which can give out high sample rate data. The process control quality can be improved with this model. The model is also useful for unit operating optimization. Extensive simulation results show that the algorithm is effective and feasible. The method is also successfully applied to a practical example