TY - JOUR ID - TI - Optimal Production Decision Making by using Artificial Neural Networks and Fuzzy Linear Programming AU - Khalan J. Rostam PY - 2022 VL - 56 IS - 3 SP - 601 EP - 621 JO - Journal of The Iraqi University مجلة الجامعة العراقية SN - 18134521 26637502 AB - This research aims to build a mathematical model for some of the products of the Company (Beiji Oil Refinery/ Iraq) and solve the model using the fuzzy linear programming methods in addition to that, how to employ these linear programming models in artificial neural networks by depending on the results of the optimal solution that were reached in the five methods of fuzzy linear programming since artificial neural networks are information processing systems that have the capabilities to imitate the human neural system by developing a model structure to map complex non-linear relationships and processes that are inherent among several influencing variables. The application side of the company implemented (Beiji Oil Refinery/ Iraq) in 2021, such that a neural network was trained using a data set of solved linear programming problems. The objective function used in this training had ten (10) variables and twenty-four (24) constraints equations. This trained neural network was used to optimize oil production profit for 10 different kinds of oil; gas oil, fuel oil, diesel oil, naphtha, light puffs, heavy jet, heavy kerosene, liquid gas, gasoline, and white oil, such that the neural network structure consisted of 274 inputs and 11 outputs with a neural structure of 194 hidden neuron layers. The training algorithm used was Levenberg-Marquardt backpropagation. The neural network results when compared with five methods of fuzziness and comparison between the methods of removing fuzzy in a linear programming model and finding the best method to get maximum profit. The maximum projected profit was up to 98% in a bounded and decomposition method increase from 993423791 IQD to 1943043833 IQD in a day. This paper will increase the current rate of crude oil products in Beiji Oil Refinery and increase the profit of production while maintaining the same quantity of raw materials for daily crude oil products. The paper reached several conclusions, the most prominent of which is that there is a difference in determining the optimal quantities of production and the reflection of this matter on revenues the total achieved when using each of the methods of removing fuzzy, as the results showed that the best way in terms of achieving the highest revenues was when using bounded and decomposition method. Therefore; the importance of this research lies in the topic that I dealt with, which is to make the optimal decision for production using the fuzzy linear programming method and artificial neural networks.

ER -