Artificial neural networks approach on solar parabolic dish cooker

This paper presents heat transfer analysis of solar parabolic dish cooker using Artificial Neural Network (ANN). The objective of this study to envisage thermal performance parameters such as receiver plate and pot water temperatures of the solar parabolic dish cooker by using the ANN for experimental data. An experiment is conducted under two cases (1) cooker with plain receiver and (2) cooker with porous receiver. The Back Propagation (BP) algorithm is used to train and test networks and ANN predictions are compared with experimental results. Different network configurations are studied by the aid of searching a relatively better network for prediction. The results showed a good regression analysis with the correlation coefficients in the range of 0.9968–0.9992 and mean relative errors (MREs) in the range of 1.2586–4.0346% for the test data set. Thus ANN model can successfully be used for the prediction of the thermal performance parameters of parabolic dish cooker with reasonable degree of accuracy.     

Heat transfer analysis for shell-and-tube heat exchangers with experimental data by artificial neural networks approach

This work applied Artificial Neural Network (ANN) for heat transfer analysis of shell-and-tube heat exchangers with segmental baffles or continuous helical baffles. Three heat exchangers were experimentally investigated. Limited experimental data was obtained for training and testing neural network configurations. The commonly used Back Propagation (BP) algorithm was used to train and test networks. Prediction of the outlet temperature differences in each side and overall heat transfer rates were performed. Different network configurations were also studied by the aid of searching a relatively better network for prediction. The maximum deviation between the predicted results and experimental data was less than 2%. Comparison with correlation for prediction shows superiority of ANN. It is recommended that ANN can be used to predict the performances of thermal systems in engineering applications, such as modeling heat exchangers for heat transfer analysis.     

An enhanced radial basis function network for short-term electricity price forecasting

This paper proposed a price forecasting system for electric market participants to reduce the risk of price volatility. Combining the Radial Basis Function Network (RBFN) and Orthogonal Experimental Design (OED), an Enhanced Radial Basis Function Network (ERBFN) has been proposed for the solving process. The Locational Marginal Price (LMP), system load, transmission flow and temperature of the PJM system were collected and the data clusters were embedded in the Excel Database according to the year, season, workday and weekend. With the OED applied to learning rates in the ERBFN, the forecasting error can be reduced during the training process to improve both accuracy and reliability. This would mean that even the “spikes” could be tracked closely. The Back-propagation Neural Network (BPN), Probability Neural Network (PNN), other algorithms, and the proposed ERBFN were all developed and compared to check the performance. Simulation results demonstrated the effectiveness of the proposed ERBFN to provide quality information in a price volatile environment.     

Natural cooling of horizontal cylinder using Artificial Neural Network (ANN)

Artificial Neural Network (ANN) is used to determine natural convection heat transfer and fluid flow around a cooled horizontal circular cylinder having constant surface temperature. Governing equations of natural convection were solved using finite volume technique by writing a FORTRAN code to generate the database for ANN scheme and Rayleigh number is changed from Ra = 10⁶ to 10⁸. Results obtained from numerical solutions were used for training and testing the ANN approach. A comparison was performed among the soft programming (ANN), experimental observation and Computational Fluid Dynamic (CFD) code. It is observed that ANN soft programming code can be used more efficiently to determine cold plume and thermal field generated around a cold cylinder. Based on the results a new correlation is developed for natural convection of cooled horizontal cylinders.     

A comparative assessment of predicting CH4 adsorption on different activated carbons using generalized regression neural network (GRNN), and adaptive network-based fuzzy inference system (ANFIS)

The aim of this work is to predict the adsorption of methane on various activated carbon using to intelligent models including Generalized Regression Neural Network (GRNN), and adaptive network-based fuzzy inference system (ANFIS). Methane is the major component of natural gas, coal bed gas, and some exhaust gases of petrochemical or chemical units. Therefore, a fundamental study on the adsorption was encouraged by engineering concerns. In this regards, the precise prediction of CH₄ adsorption is of great interest and importance. The model is developed using a comprehensive database obtained from the literature. The outcomes of the model were compared with the experimental data. The values of the statistical parameters R², RMSE, and AARD% reveal that the ANFIS model is more accurate. Results showed that the developed model accurately predicts CH₄ adsorption on activated carbons with an overall R² and AARD% values of 0.921% and 0.657%, respectively.     

Comparative assessment of response surface methodology quadratic models and artificial neural network method for dry reforming of natural gas

The performance of 0.5% wt Rh/γ-AL₂O₃catalyst for the dry reforming of natural gas using carbon dioxide has been considered. In the present work a comparative study has been performed using Radial Basis Function Neural Network (RBFNN) and Response Surface Methodology (RSM) quadratic models to investigate their predictive ability for the effect of two different operating parameters, naSSSmely the hourly space velocity and the reaction temperature on the conversion of the different components comprising commercial natural gas. The predictive capabilities of the two methodologies were compared employing statistical error functions. The results indicated the superiority of RBF in the prediction capability; for example, the F-ratio for the CH₄ reactant is 86 and 1088185 employing RSM and ANN methods, respectively. Also for the various components involved in the reaction system R² ranges from 0.74 to 0.96 in case of RSM while it is 1.0 for all components employing ANN. This is due to ANN ability to approximate the non-linearity between the input and output variables.     

An artificial intelligence approach to inverse heat transfer modeling of an irradiative dryer

In this work, a new solution approach was developed for heat estimation class of inverse heat transfer problems where radiation provides the dominant mode thermal energy transport. An Artificial Neural Network (ANN) was designed, trained and employed to estimate the heat emitted to irradiative batch drying process.     

基于神经网络的内燃机活塞-缸套磨损的故障诊断

分析了BP算法。在MATLAB环境下以改进的BP网络为识别模型对内燃机活塞-缸套磨损的几种故障进行分类训练,并应用待识别的故障样本识别仿真。结果表明,该方法在活塞-缸套磨损诊断中是行之有效的。     

Aerodynamic Performance Improvement of a Highly Loaded Compressor Airfoil with Coanda Jet Flap

Coanda jet flap is an effective flow control technique,which offers pressurized high streamwise velocity to eliminate the boundary layer flow separation and increase the aerodynamic loading of compressor blades.Traditionally,there is only single-jet flap on the blade suction side.A novel Coanda double-jet flap configuration combining the front-jet slot near the blade leading edge and the rear-jet slot near the blade trailing edge is proposed and investigated in this paper.The reference highly loaded compressor profile is the Zierke&Deutsch double-circular-arc airfoil with the diffusion factor of 0.66.Firstly,three types of Coanda jet flap configurations including front-jet,rear-jet and the novel double-jet flaps are designed based on the 2D flow fields in the highly loaded compressor blade passage.The Back Propagation Neural Network(BPNN)combined with the genetic algorithm(GA)is adopted to obtain the optimal geometry for each type of Coanda jet flap configuration.Numerical simulations are then performed to understand the effects of the three optimal Coanda jet flaps on the compressor airfoil performance.Results indicate all the three types of Coanda jet flaps effectively improve the aerodynamic performance of the highly loaded airfoil,and the Coanda double-jet flap behaves best in controlling the boundary layer flow separation.At the inlet flow condition with incidence angle of 5°,the total pressure loss coefficient is reduced by 52.5%and the static pressure rise coefficient is increased by 25.7%with Coanda double-jet flap when the normalized jet mass flow ratio of the front jet and the rear jet is equal to 1.5%and 0.5%,respectively.The impacts of geometric parameters and jet mass flow ratios on the airfoil aerodynamic performance are further analyzed.It is observed that the geometric design parameters of Coanda double-jet flap determine airfoil thickness and jet slot position,which plays the key role in supressing flow separation on the airfoil suction side.Furthermore,there exists an optimal combination of front-jet and rear-jet mass flow ratios to achieve the minimum flow loss at each incidence angle of incoming flow.These results indicate that Coanda double-jet flap combining the adjust of jet mass flow rate varying with the incidence angle of incoming flow would be a promising adaptive flow control technique.     

Heat transfer prediction for radiant floor heating/cooling systems using artificial neural network (ANN)

Radiant floor cooling and heating systems (RHC) are gaining popularity as compared with conventional space conditioning systems. An understanding of the heat transfer capacity of the radiant system is desirable to design a space conditioning system using RHC technology. In the present work, a simplified heat flux model for RHC is developed for both cooling and heating modes of operation. The Artificial Neural Network (ANN) technique is used for the development of the simplified model. Experimental data from literature covering a wide operating range of the RHC is considered for model development and validation. Operating parameters such as mass flow rate (m_f), heat resistance (R_s), mean temperature of water flowing through the pipe (T_m), and operative temperature (T_op) are considered independent variables influencing the heat flux (q_t). The neural network consists of four input layers, one output layer, and one hidden layer with a feed-forward-back-propagation algorithm. A study on the selection of the optimum number of neurons in the range of 1–9 for the hidden layer is also performed. On the basis of the performance parameters, namely, average-absolute-relative-deviation (AARD = 0.11283) percentage, mean-square-error (MSE = 0.00055), and the coefficient of determination (R² = 0.9984), a hidden layer is modeled with five neurons.     

Artificial neural networks used for the performance prediction of a thermosiphon solar water heater

Artificial Neural Networks (ANN) are widely accepted as a technology offering an alternative way to tackle complex and ill-defined problems. They can learn from examples, are fault tolerant, are able to deal with non-linear problems, and once trained can perform prediction at high speed. ANNs have been used in diverse applications and they have shown to be particularly effective in system modelling as well as for system identification. The objective of this work is to train an artificial neural network (ANN) to learn to predict the performance of a thermosiphon solar domestic water heating system. This performance is measured in terms of the useful energy extracted and of the stored water temperature rise. An ANN has been trained using performance data for four types of systems, all employing the same collector panel under varying weather conditions. In this way the network was trained to accept and handle a number of unusual cases. The data presented as input were, the storage tank heat loss coefficient (U-value), the type of system (open or closed), the storage volume, and a total of fifty-four readings from real experiments of total daily solar radiation, total daily diffuse radiation, ambient air temperature, and the water temperature in storage tank at the beginning of the day. The network output is the useful energy extracted from the system and the water temperature rise. The statistical coefficient of multiple determination (R²-value) obtained for the training data set was equal to 0.9914 and 0.9808 for the two output parameters respectively. Both values are satisfactory because the closer R²-value is to unity the better is the mapping. Unknown data for all four systems were subsequently used to investigate the accuracy of prediction. These include performance data for the systems considered for the training of the network at different weather conditions. Predictions with maximum deviations of 1 MJ and 2.2°C were obtained respectively. Random data were also used both with the performance equations obtained from the experimental measurements and with the artificial neural network to predict the above two parameters. The predicted values thus obtained were very comparable. These results indicate that the proposed method can successfully be used for the estimation of the performance of the particular thermosiphon system at any of the different types of configuration used here. The greatest advantage of the present model is the capacity of the network to learn from examples and thus gradually improve its performance. This is done by embedding experimental knowledge in the network.     

Mapping daily global solar irradiation over Spain: A comparative study of selected approaches

Three methods to estimate the daily global solar irradiation are compared: the Bristow-Campbell (BC), Artificial Neural Network (ANN) and Kernel Ridge Regression (KRR). BC is an empirical approach based on air maximum and minimum temperature. ANN and KRR are non-linear approaches that use temperature and precipitation data (which have been selected as the best combination of input data from a gamma test). The experimental dataset includes 4 years (2005-2008) of daily irradiation collected at 40 stations and temperature and precipitation data collected at 400 stations over Spain. Results show that the ANN method produces the best global solar irradiation estimates, with a mean absolute error 2.33 MJ m⁻² day⁻¹. Daily maps of solar irradiation over Spain at 1-km spatial resolution are produced by applying the ANN method to temperature and precipitation maps generated from ordinary kriging.     

基于BP神经网络的温度控制系统

文中介绍了基于BP(Back Pmpagation)的神经网络气化炉温度控制系统。对BP神经网络控制算法作了详细的介绍，运用模糊逻辑控制概念赋予隐层含义，并决定其节点数，同时用高斯核函数作为节点激励函数，并做了仿真研究，叙述了系统的硬件与软件构成，试验表明所设计的系统操作方便、安全可靠，所选择的控制算法适应性强，控制效果良好。     

Prediction of the physicochemical properties of woody biomass using linear prediction and artificial neural networks

This article aims at using Artificial Neural Networks (ANNs) and linear prediction to predict the physicochemical properties of woody biomass, including gross calorific value, carbon content, and oxygen content. By analyzing 43 data groups, it was found that Multilayer Feedforward Neural Network (MLFN) with 11 nodes is the best model for predicting the gross calorific value, with a root mean square (RMS) error of 0.85; General Regression Neural Network (GRNN) is the best model for predicting the carbon content, with an RMS error of 1.66; and linear prediction is the best model for predicting the oxygen content, with an RMS error of 2.11.     

A new methodology for the production of furfural as a renewable energy source from bagasse in acidic aqueous media

Sugarcane is one of the most promising agricultural sources of biomass energy. Sugarcane produces mainly two types of biomass, cane trash and bagasse. Furfural is synthesized from bagasse hydrolysis in an acidic environment and as a result of pentose dehydration. This study has been focused on the production of furfural by using sulfuric acid plus an inorganic salt (NaCl + H₂SO₄) as catalyst in a pilot plant. The obtained experimental data show that sulfuric acid plus NaCl can be more effective in the production of furfural. Furthermore, in order to predict the outlet furfural percentage from reactors, a three-layer Feed-Forward Neural Network using temperature and pressure of reactor, time of reaction, sulfuric acid percentage, and bagasse humidity has been considered. In fact, using H₂SO₄ + NaCl as the catalyst, significant improvement is observed in the furfural production process and energy consumption.     

Application of machine learning and Box-Behnken design in optimizing engine characteristics operated with a dual-fuel mode of algal biodiesel and waste-derived biogas

Waste-derived biogas and third-generation algal biodiesel are attractive alternative fuels to substitute fossil diesel in a diesel engine. However, using biodiesel as a pilot liquid fuel and biogas as the main fuel in a diesel engine is a complicated and highly non-linear process. The current study seeks to predict and optimize the combustion and exhaust emission characteristics of a variable compression dual-fuel combustion engine. Data from experiments were obtained at a variety of engine loads, compression ratios, pilot fuel injection pressures, and timings. A multi-layer perceptron network was employed to develop an Artificial Neural Network (ANN) based prognostic model using the experimental data. The developed prognostic model was used to estimate brake thermal efficiency, biogas flow rates, peak in-cylinder pressure, carbon dioxide, unburned hydrocarbons, oxides of nitrogen, and carbon monoxide. The predictive model's robustness is demonstrated by statistical metrics such as R (0.9723–0.988) and R² (0.9453–0.9761), Nash-Sutcliffe model efficiency (94–97%), and mean absolute percentage error (0.013–0.128%), Kling-Gupta efficiency (0.9548–0.9836), and Theil's U2 model uncertainty (0.162–0.368). To optimize the parameters of dual-fuel combustion, the Multi-Output Response Surface Methodology (RSM) was employed. The trade-off assessment between emission and efficiency using the desirability approach revealed that 84% engine load, 244 bar of fuel injection pressure, 28 °BTDC of injection timing, and 17.5 compression ratio are the best-operating conditions for the test engine. An experimental investigation was used to corroborate the RSM research findings, and errors were less than 9%. It was revealed that ANN-linked RSM is a good hybrid technique for modeling, prediction, and optimization of the performance of a dual-fuel engine.     

Forecasting of global and direct solar irradiance using stochastic learning methods, ground experiments and the NWS database

We develop and validate a medium-term solar irradiance forecasting model by adopting predicted meteorological variables from the US National Weather Service’s (NWS) forecasting database as inputs to an Artificial Neural Network (ANN) model. Since the inputs involved are the same as the ones available from a recently validated forecasting model, we include mean bias error (MBE), root mean square error (RMSE), and correlation coefficient (R²) comparisons between the more established forecasting model and the proposed ones. An important component of our study is the development of a set of criteria for selecting relevant inputs. The input variables are selected using a version of the Gamma test combined with a genetic algorithm. The solar geotemporal variables are found to be critically important, while the most relevant meteorological variables include sky cover, probability of precipitation, and maximum and minimum temperatures. Using the relevant input sets identified by the Gamma test, the developed forecasting models improve RMSEs for GHI by 10-15% over the reference model. Prediction intervals based on regression of the squared residuals on the input variables are also derived.     

Transient Schemes for Capturing Interfaces of Free-Surface Flows

This article presents a new methodology for the development of Transient Interpolation for Capturing of Surfaces schemes suitable for the simulation of free-surface flows, which is given the acronym TICS. The newly developed approach is based on a switching strategy that combines a bounded high-order transient scheme with a bounded compressive transient scheme. Bounded high-order and compressive transient schemes are constructed by discretizing the transient term in the volume-of-fluid (r) equation over a temporal control-volume in a way similar to the discretization of the convection term over a spatial control-volume, allowing advances in building convective schemes to be exploited in the development of bounded high-order and compressive transient schemes. Following that approach, a bounded version of the second-order upwind Euler scheme is constructed (B-SOUE). The B-SOUE is used to develop a family of temporal compressive schemes that is denoted by the B-CE^m family, where “m” refers to the slope of the scheme on a temporal normalized variable diagram. The TICS methodology is then applied to the B-SOUE scheme and the B-CE^m family of schemes to create a new family of transient interface-capturing schemes that is designated by TICS^m. The virtues of the TICS^m family, in producing a steep interface for the volume-of-fluid (r) field that defines the volume fraction occupied by the different fluids in a computational domain, are demonstrated through results generated using two schemes of the family (TICS^1.75 and TICS^2.5). The accuracy of the new transient TICS schemes is compared to the first-order Euler scheme, the Crank-Nicolson scheme, and the B-SOUE scheme by solving four pure advection test problems (advection of hollow shapes in an oblique flow field and advection of a solid body in a rotational flow field) and one flow problem (the break of a dam) using both the SMART and the STACS convective schemes. Results, displayed in the form of interface contours, demonstrate that predictions obtained with TICS^1.75 and TICS^2.5 are far more accurate and less diffusive, preserving interface sharpness and boundedness at all Courant number values considered.     

High Efficiency Large Area Back Contact Concentrator Solar Cells with a Multilevel Interconnection

Very high efficiencies have been demonstrated under concentration with silicon solar cells having interdigitated contacts on the backside. However, only laboratory cells of small dimension have reached very high efficiencies. The need for developing a multilevel metallization technology for back contact concentrator solar cells of large area is demonstrated. The particular features required for such a multilevel interconnection are studied and a process using anodic oxidation of aluminum is presented. Back contact silicon solar cells of 0.64 cm² have been processed in this technology resulting in 26.2% efficiencies at 10W/cm² (100 suns AM1.5, 25.5 ^°C). the highest efficiency reported to date for a solar cell of this area. The one-sun efficiency of this cell is 21.7% (AMI.5, 25.2^°C). We propose also a new design for the metallization of back contact cells which allows an increase in the size of the cell without increasing the series resistance.     

Prediction of flow fields and temperature distributions due to natural convection in a triangular enclosure using Adaptive-Network-Based Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN)

Artificial Neural Network (ANN) and Adaptive-Network-Based Fuzzy Inference System (ANFIS) were used to predict the natural convection thermal and flow variables in a triangular enclosure which is heated from below and cooled from sloping wall while vertical wall is maintained adiabatic. Governing equations of natural convection were solved using finite difference technique by writing a FORTRAN code to generate database for ANN and ANFIS in the range of Rayleigh number from Ra = 10⁴ to Ra = 10⁶ and aspect ratio of triangle AR = 0.5 and AR = 1. Thus, the results obtained from numerical solutions were used for training and testing the ANN and ANFIS. A comparison was performed among the soft programming and Computational Fluid Dynamic (CFD) codes. It is observed that although both ANN and ANFIS soft programming codes can be used to predict natural convection flow field in a triangular enclosure, ANFIS method gives more significant value to actual value than ANN.