Ma Xuesen

Look-ahead control of conveyor-serviced production station under stochastic demand

In this paper we consider a conveyor-serviced production station system (CSPS) under stochastic demand, which is composed of production and sales centres. The production centre is mainly to unload the unprocessed parts and process them, and the sales centre mainly provides the processed parts to the customers. The production centre is affected by the sales centre. We first convert the look-ahead control problem of this system to a semi-Markov decision process (SMDP) model and discuss system and performance matrices of this model. Then, policy iteration (PI) is used to find an optimal control policy under either average- or discounted-cost criteria.

Application of Improved Genetic Algorithm in Virtual Enterprise Partnership Selection

Partner selection is a key problem of organizing a virtual enterprise. A multi-objective optimization model, which analyzes these candidate enterprises quantitatively is proposed and accomplished by an improved genetic algorithm. This algorithm sorts several single object fitness using quick-sorting algorithm and selects based on total fitness, crossovers and mutates using the self-adaptive probability, and finds the global optimal solution at last. Through comparison with standard GA and the improved self-adaptive GA, simulation example testifies the latter efficiency.

A Clustering Algorithm Based on IEEE802.15.4

For the sake of managing the mobile node in the large Wireless Sensor Network, a simple Clustering Algorithm is propounded, which can form the network structure rapidly and make use of less control overhead to adapt to mobility. The algorithm is implemented in combination with IEEE 802.15.4 MAC protocol, which has the characteristics of beacon synchronization and Guaranteed Time Slots. Finally, the simulation experiments illustrate the algorithm is able to managed network effectively.