یکپارچه سازی شبکه های عصبی فازی مبتنی بر بهینه سازی ازدحام ذرات و شبکه های عصبی مصنوعی برای انتخاب تامین کنندگان

This study is intended to develop an intelligent supplier decision support system which is able to consider both the quantitative and qualitative factors. It is composed of (1) the collection of quantitative data such as profit and productivity, (2) a particle swarm optimization (PSO)-based fuzzy neural network (FNN) to derive the rules for qualitative data, and (3) a decision integration model for integrating both the quantitative data and fuzzy knowledge decision to achieve the optimal decision. The results show that the decision support system developed in this study make more precise and favorable judgments in selecting suppliers after taking into account both qualitative and quantitative factors.

An ever-increasing trend in today’s industrial firms is to exploit outsourcing for those products and activities deemed to be outside the company’s core business. This is because that under the changing environment, industrial firms have to focus more on their core components and employ suppliers to outsource. Thus, selecting suitable suppliers becomes a very important issue in supply chain. However, managers always give too much self-concern making an accurate decision. Recently, Artificial Intelligent (AI)-based models have very promising results in a number of areas, like forecasting. Therefore, this study intends to present an AI-based decision support system so that managers can choose the suitable suppliers to get more business benefit. In addition, the other objective is to consider the qualitative as well as the quantitative factors, since both of them are critical for decision making. The proposed system is composed of (1) collection for quantitative factors, (2) a fuzzy neural network (FNN) model for handling qualitative data, and (3) decision integration model. The quantitative factors include quality, finance, location, price, delivery deadline and productivity. The fuzzy IF-THEN rules are summarized from the questionnaire filled by the experts and generated by a proposed fuzzy neural network (FNN) with initial weights generated by particle swam optimization (PSO) algorithm. Finally, the results from the above two parts are integrated through a feed-forward artificial neural network (ANN) with error back-propagation (EBP) learning algorithm. The simulation results show that PSO-based FNN is able to learn the fuzzy relationship between fuzzy inputs and outputs. In addition, a real world problem from a well-known Laptop computer company indicates that the proposed decision support system can provide better result compared with regression model. The rest of this paper is organized as follows. Section 2 presents brief review of the backgrounds, while the proposed method is illustrated in Section 3. The real world problem results are shown in Section 4. Finally, the concluding remarks are made in Section 5.

The quality of suppliers definitely makes great influence on a company. This study has demonstrated that using ANNs for supplier selection to learn both the qualitative and quantitative data is really can provide more promising results than considering either one. We also have shown that the using evolutionary approaches, PSO algorithm, with the structure of FNN are able to learn the fuzzy relationship between fuzzy factors and their corresponding impacts. The company can more easily formalize their influence on the supplier selection. Thus, the proposed system can help managers or decision makers make a more precise judgment. This is the merit that the traditional methods cannot have. For the practical applications, once the system is developed, the purchasing managers only have to collect all suppliers’ corresponding data, the proposed system can provide the ranking for decision support. Thus, it is very practical for industrial purpose. In the future, the proposed decision support system will be adopted for green supplier selection, since WEEE and RoHS have become two important issues for supplier selection. Besides, it also can be applied to forecasting systems, cataloging systems, etc.