Study on Supplier Selection and Evaluation in Serviced-oriented Manufacturing Network Based on NN

The nature of service-oriented manufacturing network is modular service network for capacity need based. The difference in service capability increases the manufacturing networks' uncertainty, making the choice of supplier more complex. A strategy for supplier selection applying FWNN (fuzzy wavelet neural network) was presented in this paper. The first stage makes use of fuzzy system, with strong competency in knowledge representation, to convert fuzzy input parameters into crisp parameters. The artificial neural network was applied to overcome adverse effects resulting from supplier selection of non-liner and dynamic property in the second stage. The simulation was made to check the performance of the algorithm. The result indicates that this method can overcome the insufficiencies of subjectivity, arbitrariness, complicated algorithm and fuzzy input parameters in vendor selection, which realizes real-time dynamic evaluation and rapid selection under the service-oriented manufacturing environment, and its maximum relative error is less than 0.4841%.