Network Response Surface Research Articles

Mixing efficiency optimization of Tesla-type flow channel for total temperature simulation device

The space vehicle total temperature simulation device introduces the actual gas total temperature signal into the vehicle development process. It enables the simulation of flight environments, reduces development time, and decreases overall costs. The uniform and stable temperature signal is paramount in accurately simulate the vehicle's real flight speed and altitude. However, the existing ground test facilities for space vehicles, characterized by their large scale, face significant challenges including insufficient uniformity in gas mixing and inadequate simulation accuracy. The Tesla-type flow channel (TFC) is widely applied for its excellent mixing capabilities in gas and liquid mixing across various domains. In this paper, from the working principle of the total temperature simulation device of space vehicle, according to the characteristics of TFC, a high mixing efficiency optimization design method of TFC is proposed by using RBF neural network response surface and NSGA-II algorithm. The optimized TFC is implemented in the mixing chamber of the total temperature simulation device to enhance the mixing efficiency. This improvement ultimately leads to enhanced accuracy in simulating the flight environment of space vehicles during semi-physical simulations. By utilizing the Pareto optimal solution, the optimal pressure drop is 985.50 Pa, while the standard deviation of temperature is 39.37 K. The results demonstrate a significant improvement in mixing efficiency within the total temperature simulation device due to the introduction of the TFC. This study serves as a valuable reference for enhancing the mixing performance of the total temperature simulation device for space vehicles, while also addressing the need for total temperature simulation in smaller laboratory environments.

RSM and ANN Comparative Modelling with a Granulation Treatment in Mixed Waters

AbstractA Box‐Behnken design was used for the analysis using a gray wolf optimizer (GWO)‐coupled artificial neural network (ANN) model and response surface methodology (RSM) to analyze the effect of three operating parameters (volumetric exchange ratio [VER], aeration rate [AR], and cycle time [CT]) manipulated during an aerobic granular sludge process (AGS) sequencing batch reactor on modeling the removal of chemical oxygen demand (COD) in mixed wastewater. The most efficient architecture for COD showed the highest efficiency for modeling the AGS. The RSM model and plot results indicate that the CT and AR were the most influential on COD removal efficiency. When compared with models with statistical indices, GWO‐ANN demonstrated higher performance compared to RSM.

Complex modal optimization of the disk brake based on thermal–structural coupling

The brake disk and friction material parameters were taken as design variables, with the negative damping ratio of the unstable mode as the optimization target. The central composite design method was employed to obtain suitable data sample points. A neural network response surface model was constructed to predict the real part values of the unstable mode. Structural optimization design was then conducted using a screening method to derive optimized parameters for the brake disk structure and physical parameters such as density, elastic modulus, and Poisson’s ratio of the friction material. The optimized coefficient of the negative damping ratio of the brake system’s unstable mode was enhanced, thereby reducing the likelihood of brake squeal. Harmonic response analysis showed a significant optimization effect on the curve of brake disk amplitude fluctuations with frequency. These findings offer valuable insights for improving the stability of disk brake systems.

A combination of the particle swarm optimization-artificial neurons network algorithm and response surface method to optimize energy consumption and cost during milling of the 2017A alloy

This research aims to predict the cost and energy consumption associated with pocket and groove machining using the hybrid particle swarm optimization-artificial neurons network (PSO-ANN) algorithm and the response surface method (RSM). A parametric study was conducted to determine the best predictions by adjusting the swarm population size (pop) and the number of neurons (n) in the hidden layer. The results showed that machining strategies and sequences can have a significant impact on energy consumption, reaching a difference of 99.25% between the minimum and maximum values. The cost ( Ctot) and energy consumption ( Etot) values with the PSO-ANN algorithm increased significantly by 99.99% and 92.41%, respectively, compared to the RSM model. The minimum mean square error values for Etot and Ctot with the PSO-ANN models are 3.0499 × 10−5 and 4.6296 × 10−10, respectively. This study highlights the potential of the hybrid PSO-ANN algorithm for multi-criteria prediction and highlights the potential for improved machining of 2017A alloy.

Removal of aqueous methylene blue dye over Vallisneria Natans biosorbent using artificial neural network and statistical response surface methodology analysis

Azo-dye compounds prevalent in textile wastewater pose environmental risks. This research focuses on examining the kinetics and removal percentage of methylene blue (MB) using the biosorbent Vallisneria Natans, also known as pata dam (PD). Eight isotherm adsorption models were explored to comprehend the adsorption behaviours. Here, the data was better fit by Toth isotherm than others, indicating the nature of the adsorbent surface. The MB adsorption data were well fit with pseudo-second-order kinetic model, indicating a constant rate evaluation. The artificial neural network (ANN) and response surface methodology (RSM) effectively determined the conditions that resulted in an observed 88.34 % removal of methylene blue (MB). These optimal conditions comprised a 0.85 g/L adsorbent dose, a pH level of 7, a temperature of 42 °C, and a dye concentration of 40 mg/L. The best-performing ANN architecture was a 4–8–1 topology. ANN demonstrate superior modelling capabilities with an R2 = 0.99405, compared to RSM, with an R2 = 0.9837 when predicting the removal percentage of MB. Fourier transform infrared spectroscopy (FTIR) plays a crucial role in comprehending the interactions occurring during the adsorption of MB on the surface of the PD biosorbent. These findings showed that PD could effectively remove MB contaminated wastewater.

Using halloysite nanotubes modified by tetraethylenepentamine for advanced carbon capture: Experimental and modeling via RSM and ANNs

In this research, halloysite nanotube was studied as a natural adsorbent for CO2 capture. The sorbents were prepared by the impregnation of several amounts of tetraethylenepentamine (TEPA)(10, 20, 30, and 40 wt%) on IMSiNTs support. Carbon dioxide adsorption was studied with the help of artificial neural network modeling and response surface methodology, and the effects of parameters were determined. Using the surface response method (RSM) simulation technique, the model obtained a suitable orientation to optimize the effective parameters in the process to select the best combination of operating conditions, and the adsorption capacity was measured in the temperature range of 20∼50 °C and at pressures of 1∼9 bar via tetraethylenepentamine (TEPA)-functionalized pristine nanotubes of halloysite (HNTs). Artificial neural network modeling was used to predict the adsorption process. The multi-layer perceptron (MLP) and radial-based function (RBF) models using different algorithms were investigated. According to the optimization process of the neurons in the MLP structure, the number of neurons in the first hidden layer was 25, and in the second hidden layer was 5. In addition, in RBF, there was a single hidden layer in which the number of neurons was 30. Response surfaces were created using the central composite design and a correlation coefficient of 0.992 based on a quadratic model was used. The CO2 capture experimental and simulation results were found to agree competently. A value of 9.3041 mmol/g at 9 bar and the temperature of 20 °C resulted in the maximum amount of CO2 adsorption capacity.

Geometrically compatible integrated design method for conformal rotor and nacelle of distributed propulsion tilt-wing UAV

The inner rotors of distributed propulsion tilt-wing Unmanned Aerial Vehicles (UAVs) are often folded in the cruising state and deployed in vertical take-off and landing to cope with the huge difference in thrust requirements. However, the blades of the conventional rotor have poor conformality with the nacelle profile, which will greatly increase the drag of the UAV after folding. This paper proposes an integrated method for the design of rotor and nacelle considering geometric compatibility to reduce the drag of the folded rotor and nacelle, so as to further improve the aerodynamic efficiency in cruise while ensuring the rotor efficiency in the vertical flight mode. A geometric mapping model based on nacelle design parameters and rotor design parameters is established, and a parametric model and aerodynamic optimization model of the outer arc airfoil family are developed. In addition, a rotor performance analysis model and a neural network response surface model for nacelle drag prediction that meet the requirements of confidence level are established. Based on the oblique inflow blade element momentum theory method, numerical simulation method, and genetic algorithm, an integrated optimization framework of the design of the conformal rotor and nacelle is built. Then, a geometrically compatible integrated optimization for the rotor and nacelle is carried out with the objective of maximizing energy efficiency in the full mission profile. Finally, a conformal rotor and nacelle design solution is obtained, which satisfies geometric compatibility and thrust constraints while providing high thrust efficiency and low cruising drag. A comparison of the results of the integrated design and the conventional rotor optimization design shows that the drag of the conventional rotor is 3.45 times that of the conformal integrated design in the cruising state, which proves the effectiveness and necessity of the proposed method.

Predicting Optimized Dissolution of Selected African Copperbelt Copper-cobalt-bearing Ores by Means of Neural Network Prediction and Response Surface Methodology Modeling

While the uncertainty brought about by a varying feed mineralogy was taken into consideration, the paper investigated the modeling and prediction of the leaching behavior of complex copper-cobalt bearing ores, using an artificial neural network (ANN) with a backforward algorithm. The process optimization is further conducted using the response surface methodology (RSM) employing the Box-Behnken design (BBD). Seven (7) parameters were considered in a multiple linear regression according to the L12 screening plan (27) of Plackett–Burman. From the seven parameters, four including solid percentage (15, 27.5, 40%), time (45, 90, 135 min), particle size passing (53, 75, 105 µm), and Fe2+ ion concentration (2, 4, 6 g/L) are modeled with L27(34) BBD. With a composite desirability of 0.94, leaching yields of 93.46% Cu and 89.43% Co were obtained. The neural network algorithm used is the BFGS (Broyden, Fletcher, Goldfarb and Shanno) algorithm multilayer perceptron with the hyperbolic tangent activation function for the hidden layer and a linear activation function for the neural output. The Multilayer perceptron {4–7-1} structure was chosen as a suitable arrangement for Cu leaching. Comparing the predicted values and those obtained experimentally resulted with a correlation coefficient of 0.9552 for the data trained in the artificial neural network and 0.8742 for the data obtained with the response surface methodology. The synergy of these 2 techniques shows that the prediction can be achieved by means of the ANN giving the values of the root mean square errors (RMSE) of 0.0115, 0.00624, 0.0229, respectively, for the training, testing and validation sets for copper recovery while the correlational study between variables could be done through the RSM. The above includes only the 95% confidence interval while the remaining 5% would be uncertain. The above results and conclusion are accompanied by the relative uncertainty as the ore mineralogy varies. The combination of the synergistic use of ANN and RSM with the sensitivity analysis has approached the process to the physics of the Multi-criteria decision-making.Graphical

Using shock tube species time-histories in Bayesian parameter estimation: Effective independent-data number and target selection

Species time-histories in shock tube experiments provide rich kinetic information for parameter estimation, but there are two problems in using these data in Bayesian approaches. First, the effective independent-data number is not equal to the number of data points in a curve, so brute multiplication of all data points in likelihood function can weaken the constraints from prior information. Second, taking all points of a curve as targets can lead to results different from that of taking several representative points in the curve. In this paper, we employed maximum a posteriori estimation combined with a neural network response surface to optimize a propane mechanism against multispecies time-histories of propane pyrolysis in a shock tube. Three methods of calculating the likelihood function are used: multiplying all points in a curve (C-160), taking the averaged likelihood in each point (C-1), and taking the likelihood of last points (LastP). The influence of effective independent-data number was studied by comparing C-1 and C-160. It was found that C-160 performed slightly better in fitting experimental data, but brute multiplication overtuned the rate constants beyond a reasonable range. The larger the effective independent-data number, the more severe the overtuning, leading to only a slight improvement of model predictions. The influence of target selection is investigated by comparing LastP and C-1. LastP outperformed C-1 slightly, which can be attributed to the fact that larger discrepancies observed between experimental data and model predictions of the last point can increase the weights of likelihood functions. This further implies that several critical points can represent the entire line for point estimation. This paper can provide a reference both for modelers about reasonable utilization of species time-histories, and for experimentalists about the importance of a detailed probability distribution of measurement error, as well as experiment design with emphasis on critical points.

Convolutional neural network modeling and response surface analysis of compressible flow at sonic and supersonic Mach numbers

Ribs as passive control in suddenly expanded flow is exciting to enhance the base pressure around nozzle exit. Furthermore, predictions of compressible flow characteristics in duct expansion using machine learning are very rare. In this work, the evolution of compressible flow through a nozzle regulated by semi-circular rib passive control is analyzed experimentally using the data acquisition technique; to control the base pressure for sonic and four supersonic Mach numbers. Results shows nozzles flowing under favorable pressure becomes effective and significantly increases the base pressure. Artificial neural networks (ANN), deep ANN (DNN), convolutional NN (CNN), and deep CNN (DCNN) were used for the modeling of compressible flow data that are highly sensitive and non-linear. For the first time, CNN used to model high-speed data predicted accurately the flow characteristics. The base pressure continuously decreases, irrespective of Mach number and Nozzle Pressure Ratio, even when the flow is under-expanded due to a higher area ratio. DNN was found to be most effective, with R-squared above 0.9 for the base pressure and pressure loss, while for wall pressure, it is above 0.8. The accuracy of base pressure predictions is between 84% and 99% during the training and testing of all NN models.

Experimental Investigation and Modeling of Bubble Departure Frequency for Pool‐Boiling Heat Transfer

AbstractThe artificial neural network (ANN) method and response surface methodology (RSM) were used to predict the bubble departure frequency for the pool‐boiling heat transfer of pure liquids, by using experimental data. The effects of vapor‐liquid density difference, vapor‐liquid viscosity difference, surface tension, thermal conductivity, and heat flux on the departure frequency of vapor bubbles were investigated by RSM and ANN. The results showed that the outputs of the ANN and RSM had a suitable overlap with the experimental data, but the RSM model was more accurate than the ANN model.

Noise optimization of multi-stage orifice plates based on RBF neural network response surface and adaptive NSGA-II

In the multi-stage orifice plates (MSOP) design of the pipeline system of the nuclear power plant, since the multi-parameter coupling and matching technology has not been broken through, the high noise cannot be effectively suppressed under large flow rate and high pressure drop conditions. This paper proposes an optimization method for the low noise design of the MSOP based on the RBF neural network response surface and the adaptive NSGA-II algorithm. Cavitation and flow velocity are proposed as the characteristic parameter for evaluating the noise. The relationship between coupling parameters such as diameter ratio, plates thickness, and plates spacing is clarified by correlation analysis. The method increased the minimum pressure by 28.57% and reduced the maximum velocity by 11.49% under the constraints of the high pressure drop, structure size and flow rate. The results prove that this method can effectively realize the design of the MSOP with low noise.

Artificial neural network modeling and optimization of thermophysical behavior of MXene Ionanofluids for hybrid solar photovoltaic and thermal systems

• Thermophysical properties of MXene Ionanofluids are presented for PV/T system. • LMBPNN approach was used to predict the thermophysical behavior of MXene Ionanofluids. • LMBPNN model performance in predicting the thermophysical property behavior. • Optimization and Parametric analysis of thermophysical properties was performed using RSM. Newly developed MXene materials are excellent contender for improving thermal systems' high energy and power density. MXene Ionanofluids are novel materials; their optimum thermophysical behavior at various synthesis conditions has not been addressed yet. The aim of this study is to investigate the effect of synthesis conditions (temperature 303–343 K and nanofluids concentration 0.1–0.4 wt%) on the thermophysical properties (thermal conductivity, specific heat capacity, thermal stability, and viscosity) of MXene Ionanofluids. Levenberg Marquardt based Artificial Neural Network (ANN) model and Response Surface Methodology (RSM) based optimization techniques have been adopted for systematic parametric analysis of MXene Ionanofluids thermophysical properties using experimental data. ANN and RSM have predicted the thermophysical behavior of MXene ionanofluids at optimized conditions. The experimental data were used to train, test, and validate the ANN model. The neural network could correctly predict the outcomes for the four properties based on the numerical performance with R 2 values close to 1, and a prediction error is 2%. The performance of the proposed LM-based back-propagation algorithm demonstrates that the error involved has been minimal and acceptable. RSM has developed correction among input parameters and thermophysical properties of MXene Ionanofluids. The comparison between experimental results and the proposed correlations revealed excellent practical compatibility. Optimized thermophysical properties of MXene Ionanofluids thermal conductivity of 0.776 W/m.K, specific heat capacity of 2.5 J/g.K, thermal stability of 0.33931 wt loss %, and viscosity of 11.696 mPa.s were obtained at a temperature of 343 K and nanofluids concentration of 0.3 wt%. MXene Ionanofluids with optimal thermophysical properties could be used for the greatest performance of hybrid solar photovoltaic and thermal system applications.

Optimization of Wood Particleboard Drilling Operating Parameters by Means of the Artificial Neural Network Modeling Technique and Response Surface Methodology

Drilling is one of the oldest and most important methods of processing wood and wood-based materials. Knowing the optimum value of factors that affect the drilling process could lead both to high-quality furniture and low-energy consumption during the manufacturing process. In this work, the artificial neural network modeling technique and response surface methodology were employed to reveal the optimum value of selected factors, namely, drill tip angle, tooth bite, and drill type of the delamination factor at the inlet and outlet, thrust force, and drilling torque. The data set that was used in this work to develop and validate the ANN models was collected from the literature. The results showed that the developed ANN models could reasonably predict the analyzed responses. By using these models and the response surface methodology, the optimum values of analyzed factors were revealed. Moreover, the influences of selected factors on the drilling process of wood particleboards were analyzed.

Modeling of the creep behavior of epoxy/yerba-mate residue composites

This work presents the short-term creep behavior of novel epoxy composites reinforced by post-consumed yerba-mate (YM). Particulate composites were manufactured using 5, 10, and 20 wt.% of YM. The composite morphologies were related to the dynamic mechanical and creep behavior at the glassy state (∼30°C). Creep tests were performed using three different stress loads (1.5, 3.0, and 6.0 MPa). Weibull model, Artificial Neural Network (ANN) approach and Response Surface Methodology (RSM) were used to fit the experimental curves and to predict results. A better fit was obtained using the ANN approach than the Weibull model due to the capability of ANN to learn from own data and in fitting complex nonlinear data. The RSM approach proved to be an intelligent and reliable technique to access a higher range of results, reducing experimental time and cost and keeping statistical significance. Also, the present methodology can be extended to model and predict other properties and/or optimize parameters.

BPNN-based optimal strategy for dynamic energy optimization with providing proper thermal comfort under the different outdoor air temperatures

An efficient method is used to optimize indoor air distribution in heating systems of stratum ventilation to be beneficial to develop a potential of energy-saving with providing proper thermal comfort. To save much energy with providing proper thermal comfort, asearch for good optimization methods or strategies is required to optimize indoor air distribution affected the parameters (i.e., supply air temperature, supply vane angle, clothing insulation and outdoor air temperature) in stratum ventilation systems. In this study, the air distribution of stratum ventilation for heating is optimized by using the strategy 1 (the optimal strategy of the predicted mean vote (PMV) between −0.7 and 0.7 based on a technique of order preference by similarity to ideal solution (TOPSIS)) and the strategy 2 (the optimal strategy of PMV between −0.7 and 0.7, and draft rate (DR) less than 20). Firstly, the Computational Fluid Dynamics software is used to perform ventilation operations on the parameters for stratum ventilation systems. Secondly, according to the simulation results, the parameters are set as the input variables to construct the predictive model of ventilation performance (i.e., Energy consumption, PMV and DR) by using the Back Propagation Neural Network (BPNN) method and response surface model (RSM) method, respectively. According to the accuracy analysis of models, the BPNN method is chosen to build the predictive model. Finally, the ventilation performance is output through the optimization of the proposed BPNN-based strategy 1 and strategy 2. The optimized result shows that BPNN can build prediction models of great precision for ventilation performance. The proposed strategy 2 can save more energy by providing thermal comfort than the strategy 1. And the strategies can be beneficial to save energy as the outdoor air temperature varies. Moreover, the clothing insulation instead of supply air temperature indirectly saves energy, while providing proper thermal comfort.

Creep parameter inversion for high CFRDs based on improved BP neural network response surface method

The creep parameters of rockfill materials obtained from engineering analogy method or indoor tests often cannot accurately reflect the long-term deformation of high Concrete-Faced Rockfill Dams (CFRDs). This paper introduces an optimized inversion method based on multi-population genetic algorithm-improved BP neural network and response surface method (MPGA-BPNN RSM). The parameters used for inversion are determined by parameter sensitivity analysis based on the statistical orthogonal test method. MPGA-BPNN RSM, validated by root-mean-square error, mean absolute percentage error, squared correlation coefficient (R2), etc., completely reflects the response between the creep parameters and the settlement calculation values obtained by finite element method (FEM). MPGA optimized the objective function to obtain the optimal creep parameters. The results show that the settlement values of Xujixia CFRD calculated by FEM using the inversion parameters has great consistency with the monitored values both in size and in distribution, suggesting that the model parameters obtained by the introduced creep parameter inversion method are feasible and effective. The introduced method can improve the inversion efficiency and the prediction accuracy in FEM applications.

Influence of Material Randomness on Welding Residual Stress in Dissimilar Metal Welded Joints of Nuclear Power Plants

The welding residual stress is one of the main factors that lead to Stress Corrosion Cracking (SCC) in dissimilar metal welded (DMW) joints of the safe-end in nuclear power plants. Based on ABAQUS software, a thermal-elastic plastic finite element method is developed to simulate residual stress for DMW joints of the safe-end. Considering the randomness of material parameters of the welding metal, the neural network response surface method is applied to calculate the change of welding residual stress distribution. Meanwhile, to improve the efficiency of numerical analysis, MATLAB is employed in the secondary development for ABAQUS. With the help of existing experimental data, the effect of random parameters of the welding metal on the residual stress in DMW joints is simulated and analyzed in this study. The results show that the residual stress distribution of the DMW joints is significantly affected by the random parameters. Among the parameters, the randomness of Yield strength and Module of Elasticity has the most significant influence on the uncertainty of the distribution of welding residual stress.

Inspecting the bioenergy potential of noxious Vachellia nilotica weed via pyrolysis: Thermo-kinetic study, neural network modeling and response surface optimization

In the present study, Vachellia nilotica (VN) was valorized via thermal pyrolysis. A thermogravimetric analyzer studied the thermal degradation of VN under non-isothermal conditions. Kinetic parameters were determined using seven isoconversional methods. The Eavg was ranged from 142.15 to 166.21 kJ/mol, with the advanced isoconversional approach showing significantly better performance than general methods. Thermodynamic analysis of VN pyrolysis suggested the possibility of conversion. Neural network modeling was significantly employed to predict the value of dα/dt. The effect of temperature (350–550 °C), heating rate (10–50 °C/min), and particle size (0.1–1 mm) on process performance in terms of products yield were systematically studied and analyzed. The process variables of temperature, heating rate, and particle size were also optimized using response surface methodology (RSM). The ideal operating parameters of temperature at 500 °C, and a heating rate of 25 °C/min with 0.4 mm particle size were determined and experimentally confirmed. The characteristics of the resulting product were determined using ultimate analysis, FTIR, and GC-MS, and it was discovered that the principal constituents in the VN pyrolysis process include phenolics, furfural, and dehydroacetic acid. VN has a high thermochemical conversion potential for bioenergy, as evidenced by their physicochemical properties and thermo-kinetic findings.

Process intensification of hydrogen production by catalytic steam methane reforming: Performance analysis of multilayer perceptron-artificial neural networks and nonlinear response surface techniques

Uncertainty about how process factors affect output might lead to waste of resources in laboratory experiments. To address this constraint, a data-driven method might be used to describe the non-linear connection between process parameters and desired output. A Multi-Layer Perceptron-Artificial Neural Network (MLP-ANN) and non-linear response surface method are used to predict hydrogen generation by catalytic steam methane reforming. The impact of training methods (scaled conjugate and gradient descent), hidden layer variation, artificial neuron variation, and activation functions were studied in 80 MLP-ANN combinations (hyperbolic tangent function and sigmoid function). The performance of MLP-ANN models was affected by the training techniques, activation functions, layer count, and number of artificial neurons. The model with the sigmoid function and 3 input layers, 17 artificial neurons in the first layer, 15 artificial neurons in the second layer, and 2 output nodes had the greatest performance among the 40 configurations of scaled conjugate trained ANNs. It projected an 89.55% maximal hydrogen yield with a coefficient of determination (R2) of 0.997 and reduced errors with Mean absolute percentage error (MAPE) and mean squared error (MSE) of 0.199 and 0.121, respectively. Similarly, the gradient descent ANN model with hyperbolic tangent activation function had the greatest performance among the 40 gradient descent trained-ANN configurations. The 3–15–7–2 gradient descent trained ANN model projected a maximum hydrogen output of 89.73% compared to the experimental results of 89.51%. The MLP-ANN models outperformed nonlinear response surface methods, with R2, MAPE, and MSE of 0.231, 0.191, and 0.988, respectively. The updated Garson algorithm indicated that the input parameters impacted the hydrogen production in the sequence reaction temperature>methane partial pressure>steam partial pressure. The sensitivity analysis might assist identify how resources should be spent.