The ability of Artificial Neural Network (ANN) for estimating the natural Gas demand load for the next day and month of the populated cities has shown to be a real concern.As the most applicable network, the ANN with multi-layer back propagation perceptron’s is used to approximate functions. Throughout the current work, the daily effective temperature is determined, and then the weather data with the Gas consumption data of the last days are used for network training. It is shown that nearly 93% and 98.9% of the result is in a good agreement with the real data for the daily Gas load forecasting and those of the monthly respectively. These results clearly show the capability of the presented Networks. The method, however, can further be developed for prediction of other required information in various industries.

One of the main challenges in energy resource management is accurately predicting demand to reduce production and storage costs. The time-series behavior of Gas demand is highly complex and nonlinear due to the influence of variables such as weather conditions, prices, and population. In this study, a hybrid model based on Long Short-Term Memory (LSTM) Neural Networks, combined with the Variational Mode Decomposition (VMD) algorithm and the Cosine-Sine Algorithm (SCA), is proposed. The VMD algorithm decomposes time-series data and extracts the main components, while the SCA optimizes the parameters of the LSTM network to improve prediction accuracy. The prediction accuracy of the model is evaluated using metrics such as Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE), and its performance is compared with three benchmark models: the simple LSTM Neural network, the Support Vector Regression (SVR) model, and the Autoregressive Integrated Moving Average (ARIMA) model. The results show that the proposed model provides higher accuracy than other models and can be used as an efficient tool for energy resource planning and management.

Equilibrium ratios for the mixture of different components are very important for many engineering application processes. Different numerical methods were explored and applied to ensure efficient estimation of Gas-liquid equilibrium ratio. In this paper, the Artificial Neural Network (ANN) approach along with data of experiments performed on 25 Gas condensate reservoirs has been utilized to obtain a relationship of Gas-liquid equilibrium ratios in Gas condensate reservoirs. The relationship between the Gas-liquid equilibrium ratio and parameters of components of a mixture (critical temperature, critical pressure, and acentric factor) has been derived. Finally, the results of ANN have been compared to the proposed correlations in the literature and results of the equation of state. This investigation demonstrated that the result of ANN is more precise than the equation of state and existing empirical correlations. Whereas comparison between experimental data of 3 Gas condensate samples by ANN, EOS, and existing empirical correlation show that the average absolute error for ANN was between 7. 82 to 13. 74% and for others was between 29. 99 to 94. 99%.

Gas lubricated bearings are of tremendous use especially in the biomedical and aerospace industries. Analytical treatment of Gas lubrication is tedious due to high nonlinearity of the pressure equation as the consequence of lubricant compressibility. Howevere, in this paper a feed-forward Neural network approach is employed to investigate the performance of circular as well as two, three and four-lobe noncircular Gas lubricated bearings. The performance parameters considered are stability margins, power loss, bearing load capacity and attitude angle for various values of bearing aspect ratio, eccentricity and compressibility numbers. The results of the Neural network analyses are compared with those obtained from the finite element model. It is observed that results are in good agreement. It is believed that the Neural network model can easily compete with the available theoretical model in predicting the solution of lubrication problems in respect to its simplicity,  as well as its capability of producting accurate results with lesser computer time.

This paper presents an Artificial Neural Network (ANN)-based modeling technique for prediction of outlet temperature, pressure and mass flow rate of Gas turbine combustor. Results obtained by present modeling were compared with those obtained by experiment. The results showed the effectiveness and capability of the proposed modeling technique with reasonable accuracies of about 95 percent. This paper describes a nonlinear SVFAC (State Vector Feedback Adaptive Control) controller scheme for Gas turbine combustor. In order to achieve the satisfied control performance, we have to consider the effect of nonlinear factors contained in controller. The controller is adaptively trained to force the plant output and to track an output reference. The proposed Adaptive control system configuration uses two Neural Networks, a controller network and a model network. The control performance of designed controller is compared with inverse control method and results have shown that, the proposed method has good performance for nonlinear plants such as Gas turbine combustor. SVFAC technique is finally generalized for MIMO systems in this paper.

Gas hydrate often occurs in natural Gas pipelines and process equipment at high pressure and low temperature. Methanol as a hydrate inhibitor injects to the potential hydrate systems and then recovers from the Gas phase and re-injects to the system. Since methanol loss imposes an extra cost on the Gas processing plants, designing a process for its reduction is necessary. In this study, an accurate back propagation Neural network (BPNN) is designed for the prediction of methanol loss by the Gas phase as a function of temperature, pressure, and methanol composition in the aqueous phase. Different configurations of BPNN were trained, tested, and a configuration providing the smallest absolute average relative deviation (AARD%) was chosen as an optimum structure. Finally, comparisons made among the accuracy of the developed BPNN model, process simulators, and probabilistic Neural network (PNN). Results confirm that the designed BPNN model is more accurate than the other considered predictive tools. The BPNN provided an AARD=5. 75% for prediction of experimental data, while Aspen-HYSYS, Aspen-Plus, and PNN presented an AARD% of 9. 71, 12. 57, and 13. 27, respectively.

In this paper, a new histogram-based method is introduced to make object detectors resistant to hostile attacks. In the following, this method was applied to two object detector models, YOLOV5 and FRCNN, and in this way, two models resistant to attacks were introduced. In order to verify the performance of the mentioned models, we performed the adversarial training process of these models with three targeted attacks TOG-vanishing, TOG-mislabeling, and TOG-fabrication and one untargeted attack, DAG. We have checked the efficiency of the introduced models on two data sets MSCOCO and PASCAL VOC, which are among the most famous data sets in the field of object recognition. The results show that this method, in addition to improving the adversarial accuracy, also improves the clean accuracy of the object detector models to some extent. The average clean accuracy of the YOLOv5-n model for the PASCAL VOC dataset, if adversarial attacks are applied to it, in the case where no defense method is applied, is 85.5%, and in the case where the histogram method is applied, the average accuracy is equal to with 87.36%. In the YOLOv5-n model, according to the results, the best adversarial accuracy of this model, which has increased compared to other models, is in TOG-vanishing and TOG-fabrication attacks, which are 48% and 52.36%, respectively.