Expert Systems: Forecasting Petrol Orders for PETRONAS Petrol Station

This system aims to develop an expert system (ES) prototype to assist PETRONAS
petrol station's manager to predict the weekly petrol sales. Accurate result is very
important to the manager because this will determine the amount of petrol left in the
petrol station storage tank. An empty tank will interrupt the petrol station daily
operation and the manager needs to bear the cost as well. Currently the manager does
not have any assisting material (training only provided before the petrol station start
operating) to support them during planning and most of them based solely on their
experienced. There is no guideline to guide inexperience manager particularly the new
petrol station manager. The scope of study is to understand the process during making
decision to order petrol from the depot, the underlying factors that involved in
forecasting, external parties that influenced the forecast and the method used to develop
the ES. The system will develop using rule-based ES that introduced the if-then rules.
The system will allow more flexibility to the users to customize their own rules
according to their needs. Two interfaces will be used to distinguish different user roles
that involved in the system; administrator and operator. The system is depending on the
system knowledge base and the inference engine to formalize and analyze the rules
from the users inserts data. Later, the system will come out with the forecasted order
and other details to guide the petrol station manager to make decision. The significant
of this project will improve the performance, efficiency of the petrol station and also
increase PETRONAS petrol selling throughout the nation.