Division of Information Technology, Engineering and the Environment
School of Advanced Manufacturing and Mechanical Engineering
School of Advanced Manufacturing and Mechanical Engineering
Thesis Abstract
Supplier selection is perhaps one of the most important functions of the purchasing department in a supply chain. The selection of the “best” supplier has huge impact on both short and long term performance of the organisation and the supply chain as a whole. The selection process is a complicated one since it requires collecting accurate information as an input to the decision making model, selecting a suitable model to process those information and then making the final decision of selecting the best supplier or a list of the best possible suppliers.
Adding to the complexity of the problem, in a real life situation, data available for the decision makers regarding supplier selection is usually uncertain, sometimes vague and incomplete. The task of establishing a robust, practical and accurate model to represent/model the problem and provide the most accurate solution is still far from accomplished.
In addition, the supplier selection criteria are usually qualitative and subjective. They depend heavily on human judgment, and in most of the time, the available data is uncertain. Hence, Fuzzy Logic (FL) appears here as the most suitable and powerful technique for representing such information. In addition there is a need to take into consideration other factors such as the company learning in regards to previous supplier selection decisions. Thus, the neural networks approach appears here as a potential methodology for establishing this stage of the decision system.
The aim of this research is to build a new decision model that addresses all these complexities and better represent the current developments in the supplier selection problem and SCM and helps close the gap in this field.
The research utilises the capabilities of both MATLAB & Simulink softwares and their tool boxes particularly Fuzzy Logic and Neutral Nets. In addition, SimEvents tool box will be used to model and simulate the Supplier Evaluation & Selection Decision problem. The simulation approach is used to build a more generic model that is not bound by specific or limited survey data or consultations with experts. As a result, any number of suppliers can be considered with any number of selection criteria (decision attributes) of any data type. This will be demonstrated via simulated scenarios during which 'actual' supplier performances or profiles are compared with an 'ideal' or minimum required profile and the best supplier is selected based on minimum profile equivalent cost. The Model will then be enhanced by designing a neural network that is trained to find the best supplier and then can be used for future application.
