Neural Networks And Fuzzy Control for Bulk Terminal Operating Management
From Proceeding of Summer Computer Simulation Conference,SCSC95, Ottawa, July 24-26 1994
Pietro Giribone, Agostino G.Bruzzone
CONCLUSIONS
The authors are currently trying to apply methods to the procedure applied in this study to reduce the model
experimentation time. Obviously, since this involves applications for real problems it is impossible to generalise; however,
the experience acquired with similar industrial situations has shown that it is possible to obtain a certain level of
standardisation at least relative to very specific applications.
In any case, the decision process selected can still be understood by the manager, who thus can logically validate the
suggestions provided by the DSS and thus obtain additional useful information.
The results are rather promising for what concerns the use of the neural network as a tool to study the time series relative
to chaotic and stochastic phenomena. However, such results still cannot completely solve the relative operating problems
given that the time frame is still limited for which such results are considered reliable.
However, the development potential in this sector is quite high. This part was developed in the study since it was necessary
to verify the trigging cause of the high level of predictive uncertainty and to identify a possible additional support for the
decision process.
In the specific case, the physical system is designed in such a way that it must operate under particularly critical conditions
to be economically profitable. In fact, this research team carried out a parallel study which verified that the chaotic
threshold of this plant is much less than the traffic operating conditions imposed to guarantee profitability.
Though it is necessary to operate under these conditions, the supports proposed in this study are still very helpful and, if
properly integrated with the available information, may simplify planning problems and thus ensure considerable
economic savings.
