Coupling Artificial Neural Network Research Articles

Coupling artificial neural network and sperm swarm optimization for soil temperature prediction at multiple depths.

Soil temperature (ST) stands as a pivotal parameter in the realm of water resources and irrigation. It serves as a guide for farmers, enabling them to determine optimal planting and fertilization timings. In the backdrop of regions like Iran, where water resources are scarce, a proficient and economical prediction model for ST, particularly at lower depths, becomes imperative. While recent models have demonstrated adeptness in predicting ST, in general, their error decreases with increasing depth, so that they had the lowest error at a depth of 100 cm. Addressing this gap, our study pioneers a novel hybrid model that excels in accurate daily ST prediction as it delves deeper. The models deployed encompass the multilayer perceptron (MLP) and an enhanced version, MLP coupled with the Sperm Swarm Optimization Algorithm (MLP-SSO). These models prognosticate daily ST across varying depths (5-100cm), leveraging meteorological parameters such as air temperature, relative humidity, wind speed, sunshine hours, and precipitation. These parameters are anchored to the Ahvaz and Sabzevar synoptic stations in Iran, spanned over the period from 1997 to 2022. Evaluation of our research outcomes unveils that the root mean square error (RMSE) witnesses its most substantial reduction at a depth of 100cm. For instance, at the Ahvaz station, the MLP-SSO model diminishes the RMSE value from 1.25 to 1.12°C, in contrast to the MLP model. Similarly, at the Sabzevar station, the RMSE value drops from 1.78 to 1.49°C using the coupled MLP-SSO model. These results robustly highlight the considerable enhancement brought about by the utilization of the MLP-SSO model, clearly surpassing the performance of the standalone MLP model. This emphasizes the potential and promise of the MLP-SSO model for future investigations, offering insights that can significantly advance the domain of soil temperature prediction and its implications for agricultural decision-making.

Efficient production of pullulan by Aureobasidium pullulans using a multi-objective optimization strategy with orthogonal experimental design coupling artificial neural network and genetic algorithm

Efficient pullulan production has long been a central research focus. This study used maltodextrin as the carbon source for pullulan production by Aureobasidium pullulans fermentation. A hybrid optimization approach, integrating orthogonal experimental design (OED), backpropagation artificial neural network (BP-ANN), and elite strategy non-dominated sequential genetic algorithm-II (NSGA-II), was developed. Range analysis based on OED revealed that MgSO4·7H2O significantly affects production but less impacts molecular weight, while pH notably influences molecular weight with a lesser effect on production, underscoring the need for multi-objective optimization. The BP-ANN model showed strong predictive capabilities, with goodness-of-fit values of 0.984 and 0.980 for production and molecular weight, respectively. Using this model as the fitness function for the optimization algorithm enhanced efficiency. Taking cost factors into account, the BP-ANN-NSGA-II algorithm identified the optimal fermentation medium conditions, resulting in a 6.89 % increase in production, a 368.97 % increase in molecular weight, and a 42.49 % reduction in cost. The maximum comprehensive optimization efficiency is 63.73 %, and the multi-objective optimization is better than the single objective optimization. This method significantly guides the improvement of pullulan fermentation optimization efficiency.

Economic optimization of exterior wall insulation in Chinese office buildings by coupling artificial neural network and genetic algorithm

Envelope insulation is an essential measure to reduce building energy consumption and is the focus of many building energy codes. In China, the current standard (GB 55015-2021) only specified limit values for the thermal conductivity (U) of the exterior walls, and it ignored variations in building characteristics and internal parameters. Therefore, it is difficult for architects to quickly determine the exterior wall U-value in building design and retrofit. To address this issue, the exterior wall U-values of office buildings in China were economically optimized by TRNSYS, ANN, and GA. To begin with, the effect of enhancing exterior wall insulation on the heating and cooling energy consumption of office buildings was investigated by TRNSYS software, taking into account variations in building characteristics. Moreover, the exterior wall U-values of office buildings were optimized by coupling ANN and GA. The results showed that the design values of the GB 55015-2021 did not achieve the optimal results in China, and the most economical exterior wall U-value was 0.24 W/(m2·K) in Harbin, 0.24 W/(m2·K) in Beijing, 0.42 W/(m2·K) in Changsha, 1.82 W/(m2·K) in Guangzhou, 0.66 W/(m2·K) in Kunming. Compared with the pre-optimization, the annual life cycle cost of building heating and cooling reduced by 0.96 % in Harbin, 2.07 % in Beijing, 1.44 % in Changsha, 1.68 % in Guangzhou, and 7.37 % in Kunming. In addition, variations in building parameters should be considered in the design phase of new buildings or in the renovation of old buildings, such as internal heat gain, which may lead to different optimal results.

Performance Prediction and Heating Parameter Optimization of Organic-Rich Shale In Situ Conversion Based on Numerical Simulation and Artificial Intelligence Algorithms.

In situ conversion technology is a green and effective way to realize the development of organic-rich shale. Supercritical CO2 can be used as a good heating medium for shale in situ conversion. Numerical simulation is an important means to explore the shale in situ conversion process, but it requires a lot of time and computational cost for in situ conversion simulation under different working conditions. Therefore, a computational framework for rapid prediction of shale in situ conversion development performance and heating parameter optimization is proposed by coupling artificial neural network (ANN) and particle swarm optimization (PSO). The results indicated that kerogen pyrolysis and hydrocarbon product release mainly occurred within 2 years of shale in situ conversion. The production curves of pyrolysis hydrocarbon obviously slowed after in situ conversion for 2 years. The database was constructed by a large number of in situ conversion simulations, and Pearson correlation analysis and the random forest method were adopted to obtain seven main controlling factors affecting reservoir temperature and hydrocarbon production. The determination coefficient of the obtained ANN-based prediction models is higher than 97%, and the mean square error (MSE) is lower than 0.3%. The basic reservoir case can choose to inject 350-450 °C supercritical CO2 (Sc-CO2) fluid with a rate of 600 m3/day to obtain a more promising development effect. The heating parameter optimization for three typical reservoir cases using PSO was performed, and reasonable injection temperature and injection rate were obtained. It realized accurate development prediction and rapid heating parameter optimization, which helps the effective application of shale in situ conversion development design.

Optimization of exterior wall insulation in office buildings based on wall orientation: Economic, energy and carbon saving potential in China

The aim of this study is to optimize the exterior wall insulation of different orientations to further reduce the heating and cooling energy consumption of office buildings, and to analyze its economic, energy and carbon saving potential in China. A six-story office building was selected as the case building and its energy consumption was evaluated using TRNSYS, considering variations in latitude, window-to-wall ratio, aspect ratio, and exterior wall U-values of four orientations. In addition, the exterior wall insulation in four directions was optimized by coupling artificial neural network and genetic algorithm. The main results showed that in the 20°N to 40°N region of China, the GA-optimized exterior wall U-values were highest in the southern direction, followed by the eastern and western directions, and finally the northern direction. In addition, overall, the optimal combination of exterior wall U-values showed a trend of high at low latitudes and low at high latitudes, and increased with increasing WWR and decreased with increasing AR. Finally, compared to the exterior wall requirements for energy efficient buildings in GB 55015-2021, the optimized office buildings at low altitude in China showed a 3.26 %, 6.77 % and 6.59 % reduction in ALCC, total energy consumption and total carbon emissions respectively.

Composition design of high-performance copper alloy by coupling artificial neural network and genetic algorithm

To investigate the complex nonlinear relationship between the composition and properties of high-performance copper alloys, a double-layer artificial neural network model based on a genetic algorithm (ANN-GA) was developed using MATLAB tools. The dataset showed that the correlation coefficient of the model was greater than 95%. Quantitative calculation formulas for the alloy composition, ultimate tensile strength (UTS), and conductivity were established using the transfer function, weights, and thresholds of the ANN-GA model. The effects of the main elements on the mechanical and electrical properties of the alloys were analysed. The simulation results were similar to the experimental results obtained by the ANN-GA, confirming the excellent prediction ability of the model. Four advanced copper alloys with high strength and conductivity were successfully simulated to meet the targets of UTS (≥540 MPa) and conductivity (≥80% IACS). Two alloys were selected for melting, heat treatment, and performance testing. The error between the experimental results and predicted results was within 5%, demonstrating the applicability of the model. The proposed framework is useful for designing high-performance Cu alloys.

Artificial intelligence based structural optimization of solid oxide fuel cell with three-dimensional reticulated trapezoidal flow field

Three-dimensional Reticulated trapezoidal flow field (RTFF) is promising in improving the performance and durability of the solid oxide fuel cell (SOFC). However, the structural complexity makes it challenging for the geometry configuration of the splitter and mixer. To this end, an intelligent optimization framework is proposed by coupling artificial neural network (ANN) and non-dominated sorting genetic algorithm-II (NSGA-II), in order to maximize the net power density and oxygen uniformity simultaneously. The ANN prediction model is trained to obtain the computationally efficient surrogate model of the computational fluid dynamics (CFD) numerical simulation. NSGA-II is used for the multi-objective optimization of the RTFF structural parameters. The results illustrate that the prediction model is of high prediction precision and generalization capability. In comparison to SOFC with conventional parallel flow fields (CPFF), the degree of the performance improvement of SOFC with optimized RTFF depends on the working condition, i.e., fuel and air flow rates and operating temperatures. The SOFC with the optimal RTFF achieves a higher molar concentration of oxygen and a more uniform distribution of oxygen and current density than the CPFF SOFC. The proposed optimization framework provides an efficient design method for the development of the next-generation SOFC flow field.

Optimization of Machining Parameters for Product Quality and Productivity in Turning Process of Aluminum

Modern production process is accompanied with new challenges in reducing the environmental impacts related to machining processes. The turning process is a manufacturing process widely used with numerous applications for creating engineering components. Accordingly, many studies have been conducted in order to optimize the machining parameters and facilitate the decision-making process. This work aims to optimize the quality of the machined products (surface finish) and the productivity rate of the turning manufacturing process. To do so, we use Aluminum as the material test to perform the turning process with cutting speed, feed rate, depth of cut, and nose radius of the cutting tool as our design factors. Product quality is quantified using surface roughness (R_a) and the productivity rate based on material removal rate (MRR). We develop a predictive and optimization model by coupling Artificial Neural Networks (ANN) and the Particle Swarm Optimization (PSO) multi-function optimization technique, as an alternative to predict the model response (R_a) first and then search for the optimal value of turning parameters to minimize the surface roughness (R_a) and maximize the material removal rate (MRR). The results obtained by the proposed models indicate good match between the predicted and experimental values proving that the proposed ANN model is capable to predict the surface roughness accurately. The optimization model PSO has provided a Pareto Front for the optimal solution determining the best machining parameters for minimum R_a and maximum MRR. The results from this study offer application in the real industry where the selection of optimal machining parameters helps to manage two conflicting objectives, which eventually facilitate the decision-making process of machined products.

Coupling Artificial Neural Network with EMMS drag for simulation of dense fluidized beds

The previous sub-grid, energy-minimization multi-scale (EMMS) drag models were all established at certain fixed operating conditions and material properties. In this study, we developed a generic EMMS drag for simulating dense fluidized beds by using the Artificial Neural Network (ANN) to cover a wide range of operating conditions and material properties. To this end, the algorithm of the EMMS model was optimized to provide a huge dataset efficiently and the performance of ANN was tested by training with different numbers of data and hidden layer structures. The EMMS-ANN model was determined by balancing the training precision and computational time and then applied to the simulation of five fluidized beds under different operating conditions and material properties. It was found that the simulation with the EMMS-ANN drag enables reasonable prediction and shows good applicability to a wide range of dense fluidization.

Towards a comprehensive optimization of engine efficiency and emissions by coupling artificial neural network (ANN) with genetic algorithm (GA)

In response to the stringent emission regulations, artificial neural network (ANN) coupled with genetic algorithm (GA) is employed to optimize a novel internal combustion engine strategy named direct dual fuel stratification (DDFS). An enhanced ANN model is introduced to improve the accuracy and stability of predictions. Compared to the conventional computational fluid dynamics (CFD)-GA optimization method, the ANN-GA method can identify a better solution to achieve higher fuel efficiency and lower nitrogen oxide (NOx) emissions with the lower computational time. This is attributed to the lower cost of ANN calculation, which allows ANN-GA to introduce larger population to seek optimal solutions. Through combining the required new training data with the previous ones, the original ANN model can be updated to adapt to a wider parameter range. Thus, ANN-GA can readily deal with the optimization problems with variable parameters and objectives. When more re-optimizations are required, ANN-GA can save the computational time over 75% than CFD-GA owing to the data-driven nature of ANN-GA by fully utilizing the available data. Overall, the ANN-GA method shows the superiority in accuracy, efficiency, expansibility, and flexibility for DDFS strategy optimization. It is promising to integrate ANN with optimization algorithm for further improvements of engine performance.

Coupling artificial neural networks with the artificial bee colony algorithm for global calibration of hydrological models

Hydrological models are widely used tools in water resources management. Their successful application requires an efficient calibration of the model parameters. Nowadays, there are very powerful global search methods applied to this end, but they have the disadvantage of presenting a high computational cost, because the numerical model to be calibrated needs to be evaluated a large number of times with different parameter sets. In this context, surrogate models can reduce significantly the run time of hydrological models, easing the total computational burden of global search methods. In the present work, we propose and explore the combination of a swarm intelligence-based optimization method, the artificial bee colony algorithm, with a surrogate model based on artificial neural networks in order to calibrate hydrological models. The proposed approach (ABC-ANN) is applied to the calibration and validation of a nine-parameter continuous hydrological model in two basins located in the northwest of Spain. Several aspects of the algorithm are evaluated, including its capability to calibrate the model parameters and its efficiency in terms of CPU time compared to a standard implementation of the ABC algorithm. Results show that the ABC-ANN algorithm is able to identify the location of suitable parameter sets with an accuracy rate within 89 and 99 %, and a reduction in CPU time of more than three orders of magnitude when compared to a sequential implementation. In addition, the frequency distribution of the parameter sets identified gives valuable information about the sensitivity of model output to the input parameters.

Artificial Neural Network and Grey Wolf Optimizer Based Surrogate Simulation-Optimization Model for Groundwater Remediation

We herein propose a simulation-optimization model for groundwater remediation, using PAT (pump and treat), by coupling artificial neural network (ANN) with the grey wolf optimizer (GWO). The input and output datasets to train and validate the ANN model are generated by repetitively simulating the groundwater flow and solute transport processes using the analytic element method (AEM) and random walk particle tracking (RWPT). The input dataset is the different realization of the pumping strategy and output dataset are hydraulic head and contaminant concentration at predefined locations. The ANN model is used to approximate the flow and transport processes of two unconfined aquifer case studies. The performance evaluation of the ANN model showed that the value of mean squared error (MSE) is close to zero and the value of the correlation coefficient (R) is close to 0.99. These results certainly depict high accuracy of the ANN model in approximating the AEM-RWPT model. Further, the ANN model is coupled with the GWO and it is used for remediation design using PAT. A comparison of the results of the ANN-GWO model with solutions of ANN-PSO (ANN-Particle Swarm Optimization) and ANN-DE (ANN-Differential Evolution) models illustrates the better stability and convergence behaviour of the proposed methodology for groundwater remediation.

A Dynamic Flow Forecast Model for Urban Drainage Using the Coupled Artificial Neural Network

Dynamic flow forecast, which is one of the critical technologies in the field of future Intelligent Drainage, has great potential for mitigating the damages resulting from extreme rainfalls. This study aims to develop a coupled neural network called RBF-NARX Forecast Model (RNFM) to predict urban drainage outflow. RNFM integrates the architecture advantages of the radial basis function neural network (RBFNN) and the nonlinear autoregressive with an exogenous inputs neural network (NARXNN). By calculating the Square Sum of Error (SSE) between RNFM predictions and SWMM simulations, the network parameters are optimized and the optimal coupling site of RBFNN and NARXNN is found. The urban drainage in Tianjin is presented to justify the feasibility of RNFM, and the average SSE in test rainfalls is only 0.273. Based on the Monte Carlo simulations (MCS), the uncertainty analysis is quantified and the SWMM simulations lie within the 95% prediction confidential interval, which proves that RNFM have great potential in predictions and management of urban runoff.

Next generation prediction model for daily solar radiation on horizontal surface using a hybrid neural network and simulated annealing method

Accurate prediction of solar radiation is essential for optimal design of solar systems. This paper presents an innovative artificial intelligence approach for the determination of the daily solar radiation. A new nonlinear model was developed to predict the daily solar radiation on horizontal surface using a hybrid method coupling artificial neural network (ANN) and simulated annealing (SA), called ANN/SA. This method uses SA-based temperature cycling to improve the ANN calibration performance. A calculation procedure was presented to interpret the ANN/SA model and transform it into a practical design equation. The ANN/SA technique formulates the daily solar radiation in terms of several meteorological parameters. Thousands of daily observations during 1995–2014 in a nominal city in Iran were used to develop the solar radiation models. Validity of the model was verified through different phases. Sensitivity analysis was conducted and discussed. The ANN/SA model accurately predicts the daily solar radiation and outperforms the ANN, support vector machines (SVM), and existing regression and machine learning models.

Prediction of pyrite oxidation in a coal washing waste pile using a hybrid method, coupling artificial neural networks and simulated annealing (ANN/SA)

This paper presents a novel hybrid method coupling artificial neural network (ANN) and simulated annealing (SA), called ANN/SA to predict the fraction of pyrite remaining and therefore the pyrite oxidation rate in the wastes at different depths of a coal washing pile in the Alborz Markazi Coalfield, in northeast Iran. Waste depth, oxygen mole fraction and initial pyrite content in the waste particles were used as inputs to the network. The output of the network was the amount of pyrite content remaining. An ANN/SA model with Levenberg-Marquardt algorithm and a 3-4-3-1 arrangement showed a great capability. The network was used to predict the pyrite content remaining at two trenches E and F over the study waste pile once it was trained with the field-measured data. Simulated results obtained by the ANN/SA model were very closer to the experimental data compared to the outputs of simple ANN and multivariable least squares regression methods. The correlation coefficient (R) value, by the ANN/SA model, was 0.999 for training set, and in testing stage it was 0.998 and 0.99957 for trench E and trench F respectively which shows the model prediction was quite satisfactory. The performance of the model on the training and testing data, mean squared error (MSE) and mean absolute percent error (MAPE), indicate that it has both good predictive ability and generalisation performance.

Robustly Fitting and Forecasting Dynamical Data With Electromagnetically Coupled Artificial Neural Network: A Data Compression Method.

In this paper, a dynamical recurrent artificial neural network (ANN) is proposed and studied. Inspired from a recent research in neuroscience, we introduced nonsynaptic coupling to form a dynamical component of the network. We mathematically proved that, with adequate neurons provided, this dynamical ANN model is capable of approximating any continuous dynamic system with an arbitrarily small error in a limited time interval. Its extreme concise Jacobian matrix makes the local stability easy to control. We designed this ANN for fitting and forecasting dynamic data and obtained satisfied results in simulation. The fitting performance is also compared with those of both the classic dynamic ANN and the state-of-the-art models. Sufficient trials and the statistical results indicated that our model is superior to those have been compared. Moreover, we proposed a robust approximation problem, which asking the ANN to approximate a cluster of input-output data pairs in large ranges and to forecast the output of the system under previously unseen input. Our model and learning scheme proposed in this paper have successfully solved this problem, and through this, the approximation becomes much more robust and adaptive to noise, perturbation, and low-order harmonic wave. This approach is actually an efficient method for compressing massive external data of a dynamic system into the weight of the ANN.

Application of artificial neural network and genetic algorithm to modelling the groundwater inflow to an advancing open pit mine

In this study, a hybrid intelligent model has been designed to predict groundwater inflow to a mine pit during its advance. Novel hybrid method coupling artificial neural network (ANN) with genetic algorithm (GA) called ANN-GA, was utilised. Ratios of pit depth to aquifer thickness, pit bottom radius to its top radius, inverse of pit advance time and the hydraulic head (HH) in the observation wells to the distance of observation wells from the centre of pit were used as inputs to the network. An ANN-GA with 4-5-3-1 arrangement was found capable to predict the groundwater inflow to mine pit. The accuracy and reliability of model was verified by field data. Predicted results were very close to the field data. The correlation coefficient (R) value was 0.998 for training set, and in testing stage it was 0.99.

Prediction of peak ground acceleration of Iran's tectonic regions using a hybrid soft computing technique

A new model is derived to predict the peak ground acceleration (PGA) utilizing a hybrid method coupling artificial neural network (ANN) and simulated annealing (SA), called SA-ANN. The proposed model relates PGA to earthquake source to site distance, earthquake magnitude, average shear-wave velocity, faulting mechanisms, and focal depth. A database of strong ground-motion recordings of 36 earthquakes, which happened in Iran's tectonic regions, is used to establish the model. For more validity verification, the SA-ANN model is employed to predict the PGA of a part of the database beyond the training data domain. The proposed SA-ANN model is compared with the simple ANN in addition to 10 well-known models proposed in the literature. The proposed model performance is superior to the single ANN and other existing attenuation models. The SA-ANN model is highly correlated to the actual records (R = 0.835 and ρ = 0.0908) and it is subsequently converted into a tractable design equation.

Inverse design of aircraft cabin environment by coupling artificial neural network and genetic algorithm

An inverse design method is proposed to achieve the pre-set control objectives of the aircraft cabin environment. The method combines the artificial neural network and the genetic algorithm, and the training and testing data of the artificial neural network is obtained by computational fluid dynamics analysis. Both of the thermal comfort and energy consumption are considered in the inverse design. The artificial neural network is used to identify the relationship between the thermal comfort and the air supply parameters (inlet velocity magnitude and angle, inlet air temperature). The genetic algorithm coupled with the well-trained artificial neural network is used to design the aircraft cabin environment. Numerical results show that the Bayesian regularization algorithm is proved to have better generalization capability than the other training algorithms for the artificial neural network. The increase of training data quantity improves the generalization capability of the artificial neural network, while it spends more simulation time. A computational fluid dynamics database with 60 datasets is shown to be suitable to the present inverse design, and the testing error of the artificial neural network is below 8.2%. Several groups of optimal air supply parameters are found with different trade-offs between the thermal comfort and energy consumption. The best solution of thermal comfort, i.e., the percentage of zone with |PMV| >0.5 in all cabin control domains, is less than 7.8%.

Regional optimal allocation for reducing waste loads via artificial neural network and particle swarm optimization: a case study of ammonia nitrogen in Harbin, northeast China.

Cutting external waste loads can improve water quality. Allocation for reducing waste loads should consider changing variables, such as river flows and pollutant emissions. A particle swarm optimization (PSO) method and coupling artificial neural network (ANN) models have been applied to optimize reduction rates of ammonia nitrogen (NH3-N) loads from sewage outlets in Harbin, northeast China. For the planned water quality functional section (WQFS), the NH3-N concentration is related to emitted pollutant loads and can be well predicted by ANN linkage models. Further, NH3-N load reduction rates of all outlets are optimized by PSO with the water quality standard target. The highest NH3-N concentrations occur in January and February, a typical low-flow period in Harbin. The results delivered optimum NH3-N reduction rates for the five outlets, for January and February 2011. All predicted NH3-N concentrations after the reduction meet the water quality standard. The results indicate that the outlet with the highest NH3-N load has the biggest reduction rate in each WQFS, and outlets in the WQFS with higher background NH3-N concentrations need to cut more NH3-N loads. Decision-makers should not only focus on the outlet with the highest NH3-N emission load, but also ensure that the NH3-N concentration of upper WQFS meets the water quality goal.