By : Dr. Franyi Sarmiento, Ph.D., Inspenet, April 8, 2022
A research team from the University of Cádiz has applied two different artificial intelligence tools to compare their precision in the gasoline quality control process and classify it based on its self-combustion capacity. Both conclusions will allow the industry to have updated, precise and real-time information on the composition of this hydrocarbon.
Specifically, they have worked with two analytical methods based on identifying patterns in massive data and making predictions, known as machine learning. They have been applied separately and together in an ‘electronic nose’, designed by this same team of experts to detect traces of flammable liquids in a fire, and an infrared measurement system.
As reflected in this study, entitled Comparison of different processing approaches by SVM and RF on HS-MS eNose and NIR Spectrometry data for the discrimination of gasoline samples and published in the Microchemical Journal, they have separately analyzed the utility of both methods in two processes different, one based on algorithms, instructions that process data, and another obtaining chemical measurements of gasoline. Likewise, they have also evaluated the effectiveness of their combination to discriminate and classify hydrocarbon samples
While the electronic nose offers data on the volatile profile of the samples, spectroscopic techniques, on the other hand, focus on analyzing non-volatile compounds. The union of the information from both methodologies is used to generate predictive models that allow discriminating and classifying gasoline samples based on their octane number. “This combination represents a real alternative to automate the quality control process of this petroleum derivative, which currently depends on the experience of the analyst who carries out this work”, explains the UCA researcher, Marta Barea, responsible for this work.
By providing concrete information, the applications of this new methodology in the petrochemical industry greatly contribute to optimizing quality processes, as well as in other areas. “With this data, refineries will have fast quality management systems, on the spot and with a very precise level of detail. It is also very useful in the field of forensic chemistry if, for example, a fire breaks out and it is necessary to determine what flammable liquid has caused it and from there follow indications until locating its origin”, according to the responsible author.
The identification models generated in this study can be used to create web applications for computers, tablets and mobiles and facilitate the automation of the quality processes of this petroleum derivative: “these patterns can be established as an alternative to the interpretation methods conventional for analysts to evaluate the analytical results in a faster and, above all, objective approach”.
To obtain these results, they studied the data from a total of 50 95 and 98 octane gasoline samples, which they analyzed using these two techniques. First, they trained the model, giving it the set of all information it can access. In order to check if it was also capable of interpreting new samples, they included other unknown data for the model unknown models. “The objective of this process is to find out which technical machine learning algorithm can correctly predict whether the gasoline is 95 or 98 octane and the precision of each of them,” argues the person in charge of the study.
In conclusion, the experts have obtained good performance in both algorithms, allowing to classify and determine the samples correctly. However, they have observed that the electronic nose provides more accurate information on gasoline according to its octane rating due to the classification of volatile compounds. “This technique, and specifically the aromatic profile of the samples, makes it possible to better discriminate the amount of octane content that each one contains,” says Barea.
This work has been carried out with funds from the University of Cádiz, the Institute for Vitivinicola and Agro-Food Research (IVAGRO), FEDER Funds and the Department of Economic Transformation, Industry, Knowledge and Universities of the Junta de Andalucía.
References : Marta Barea-Sepúlveda, Marta Ferreiro-González, José Luis P. Calle, Gerardo F. Barbero, Jesús Ayuso, Miguel Palma: ‘Comparison of different processing approaches by SVM and RF on HS-MS eNose and NIR Spectrometry data for the discrimination of gasoline samples’. MicrochemicalJournal. January 2022.
Source and photo : https://www.uca.es/noticia/aplican-una-nariz-electronica-e-infrarrojos-para-evaluar-la-calidad-de-la-gasolina/
Loading...
