Application of Neural Networks to Assess the Resource Value of Oil-Contaminated Waste Storage Facilities

Abstract

The article presents a methodology for evaluating the efficiency of oil industry waste recycling systems using multi-layer artificial neural networks. As an indicator of the efficiency of the recycling system, the indicator of the resource value of oil-contaminated waste (OCW) is used. For training neural networks, the data sets are formed using the resource value assessment algorithm based on the Data Envelopment Analysis (DEA) method of multi-factor evaluation of the efficiency of production systems. The development and training of neural networks are performed using the free software Neuroph Studio. A comparative analysis of the quality of the assessment of the OCW resource value depending on the size and number of layers in a multi-layer neural network is carried out. The obtained results demonstrate the prospects of the proposed approach. Recommendations for improving the accuracy of resource value assessment by an artificial neural network are given.