l. naderloo

The use of corn oil in diets is due to its positive effects on cardiovascular and immune systems. Corn oil is composed of 99% triacylglycerol, with 59% unsaturated fatty acids and 13% saturated fatty acids. Of the unsaturated fatty acids, 24% contain a double bond. Because of this composition, corn oil can be a good alternative to other oils high in saturated fatty acids, as it reduces blood cholesterol levels.This study employed an electrical nasal system to detect the amount of palm oil present in corn oil. The properties extracted from the signals obtained by the device were processed using principal component analysis, artificial neural networks, infusion, and response surface methods. The results were then compared to find the best method for detecting palm oil levels in corn oil.

The required palm oil was obtained from the Nazgol Oil Agro-industrial Plant, while the corn oil was obtained from natural lubrication centers. To prepare samples with different percentages of palm oil, 75 grams of palm oil and corn oil with the specified percentages were mixed and stored in special containers.In the electrical nose system, ten metal oxide semiconductor sensors (MOS) were used to collect output data. Pre-processing operations were performed on this data using RSM, ANFIS, PCA, and ANN methods to estimate the percentage of palm oil in corn oil. The Unscrambler V.9 software, Design Expert 8.07.1, and MATLAB R2013a were used to analyze the results.

Based on the Score plot, PC-1 and PC-2 explain 53% and 25%, respectively, describing the variance between samples for a total of 78 data points. The analysis indicates that sensors 7 and 8 have minimal impact on the detection process and can be removed from the sensor array. When reducing the cost of the olfactory system's sensor array, sensor 6 plays a more significant role than other sensors in detecting corn oil with palm composition.According to the loading diagram of palm percentage in corn oil, the MQ6 sensor had the least effect in classifying different percentages of palm in corn oil and pattern identification. Out of all functional parameters (accuracy, sensitivity, and specificity), the RSM method is deemed more appropriate for determining the percentage of palm in corn oil.Regarding the separation of corn oil and palm oil by ANFIS, RSM, and ANN, the results in Table 3-1 indicate that the RSM method is better suited for classifying corn and palm oil.

In this study, we used an electronic multi-sensor system based on metal oxide sensors to analyze various aromatic compounds in different oil and palm samples and to detect the presence of palm. The system provided comparable information for classifying different samples of palm oils. Using PCA, ANN, ANFIS, and RSM methods, we evaluated the system's performance in differentiating and classifying various oil and palm samples.The results obtained from the loading diagrams for the detection of palm in corn oil indicated that the MQ6 sensor had the least impact on the detection process. Therefore, this sensor can be removed from the sensor array.Additionally, our analysis showed that using the RSM method is more effective in detecting different percentages of palm in corn oil. Overall, our study demonstrates the efficacy of the electronic multi-sensor system in analyzing different oil and palm samples and detecting the presence of palm.

The aim of this study was to determine the authenticity of honey by processing microscopic images and obtaining an algorithm for classifying various honey frauds. In this study, sucrose, fructose, and fructose-glucose solution at a ratio of 0.9 were used to make honey adulteration. The level of adulterated honey was based on the weight percentages of 2.5, 5, 7.5, 10, 20, 30, 40, 50, 60, 70, 80, 90 and 100 by stirring. Different samples were imaged under a microscope. Each image was processed in 33 monochrome color spaces and 15 parameters were extracted from it. The three main and effective parameters of various color spaces were selected using sensitivity analysis for modeling honey fraud by adaptive Fuzzy Neural Inference System (ANFIS), Artificial Neural Network (ANN), and response surface methodology. Various criteria were used to evaluate the performance of the models such as coefficient of determination, mean square error, sum of squared estimate of errors, and mean absolute errors. The results showed that the determination coefficient and the mean square error of the artificial neural network model was 0.974 and 0.0024, respectively. Finally, using the desirability function, the artificial neural network model was selected as the best model due to less prediction error values and desirability of 0.948.