Artificial Neural Network Output Research Articles

Training improvement methods of ANN trajectory predictors in power systems

AbstractThis paper proposes training improvement methods of artificial neural networks (ANN) trajectory predictors. First, a dynamic power system time‐series trajectory is split into several different segments to simplify the original ANN training problem. Moreover, the time‐derivative of the trajectory is included to obtain an augmented loss function. Compared to previous studies which mainly focused on increasing the prediction accuracy, the aim of these novel techniques is to reduce the computational burden where the ANN output performance is still acceptable. The effectiveness of the developed methods is validated based on the WSCC three‐machine nine‐bus and IEEE 39‐bus system models. The mean absolute error (MAE) and trajectory prediction results are analysed, in which the numbers of neurons, hidden layers, and training epochs are constrained during the ANN training process. Rotor‐angle difference between generators and the system frequency are investigated as the dynamic trajectories of the power system models. The approaches are revealed to be effective when the ANN architecture and epochs are constrained. The MAE results can be reduced by up to 65% in the power system models depending on the ANN hyperparameters and training epochs. The ANN training results can better reflect the original trajectory as well.

An Improved Vehicle Path-Tracking Model Based on Adaptive Nonlinear Model Predictive Control via Online Big Bang—Big Crunch Algorithm and Artificial Neural Network

<div>In this article, a novel tuning approach is proposed to obtain the best weights of the discrete-time adaptive nonlinear model predictive controller (AN-MPC) with consideration of improved path-following performance of a vehicle at different speeds in the NATO double lane change (DLC) maneuvers. The proposed approach combines artificial neural network (ANN) and Big Bang–Big Crunch (BB–BC) algorithm in two stages. Initially, ANN is used to tune all AN-MPC weights online. Vehicle speed, lateral position, and yaw angle outputs from many simulations, performed with different AN-MPC weights, are used to train the ANN structure. In addition, set-point signals are used as inputs to the ANN. Later, the BB–BC algorithm is implemented to enhance the path-tracking performance. ANN outputs are selected as the initial center of mass in the first iteration of the BB–BC algorithm. To prevent control signal fluctuations, control and prediction horizons are kept constant during the simulations. The results showed that all AN-MPC weights are successfully tuned online and updated during the maneuvers, and the path-following performance of the ego vehicle is improved at different NATO DLC speeds using the proposed ANN + BB–BC, compared to the method where ANN is used only.</div>

Performance of Machine Learning, Artificial Neural Network (ANN), and stacked ensemble models in predicting Water Quality Index (WQI) from surface water quality parameters, climatic and land use data

Assessing water quality is essential for managing freshwater resources, safeguarding ecosystems, and guaranteeing public health. Traditional water quality assessment methods suffer from seasonal sampling, multi-parameter requirements, and labor-intensive sampling processes, which are major constraints for the frequent monitoring of vast river basins. To overcome this issue, the study modeled the remote sensing-based climatic and land use parameters with Principal Component Analysis (PCA) to leverage Artificial Neural Networks (ANN) and machine learning (ML) algorithms to predict the Water Quality Index (WQI). The Weighted Arithmetic Water Quality Index (WAWQI) method was used to calculate the WQI of the Godavari River Basin for the available 19 stream water quality parameters (SWQPs). Further, PCA was applied to reduce the dimensionality of the parameters from 19 to 6. These results led to the development of two modeling methods to predict the WQI. In the first method, the correlation-based model was developed to predict WQI by evaluating six SWQPs. The second method, the causal-effect model, uses land use and meteorological factors to determine WQI using causality. Using advanced AutoML techniques, the initial pool of 40 ML models was meticulously evaluated and refined, culminating in the selection of the top three exemplary models such as Extreme Gradient Boosting (XGB), Extra Trees (ET), and Random Forest (RF). In both methods, XGB models show better prediction, with the coefficient of determination (R2) value of 0.95 during training and 0.83 during testing in method one. Whereas in the second method, R2 of 0.93 in training and 0.80 in testing are obtained. Further, XGB, ET, and ANN outputs were stacked with each model to enhance these results in both methods. Among these three stacked models, the stacked ANN_ML model performed better compared to stacked XGB_ML and stacked ET_ML. In the first method, the stacked ANN_ML model predicts R2 values of 0.95 and 0.91 for training and testing. In the second method, 0.95 and 0.90 for training and testing are obtained using stacked ANN_ML model. These findings emphasize the stacked model prediction ability to capture nonlinear relationships in the parameters and the novel approach of land use and climate parameters based WQI prediction, which replace the laborious, time-consuming SWQP measurements.

Using Artificial Neural Networks to Predict Operational Parameters of a Drinking Water Treatment Plant (DWTP)

The scope of the present study is the estimation of key operational parameters of a drinking water treatment plant (DWTP), particularly the dosages of treatment chemicals, using artificial neural networks (ANNs) based on measurable in situ data. The case study consists of the Aposelemis DWTP, where the plant operator had an estimation of the ANN output parameters for the required dosages of water treatment chemicals based on observed water quality and other operational parameters at the time. The estimated DWTP main operational parameters included residual ozone (O3) and dosages of the chemicals used: anionic polyelectrolyte (ANPE), poly-aluminum chloride hydroxide sulfate (PACl), and chlorine gas (Cl2(g)). Daily measurable results of water sample analysis and recordings from the DWTP Supervisory Control and Data Acquisition System (SCADA), covering a period of 38 months, were used as input parameters for the artificial neural network (1188 values for each of the 14 measurable parameters). These input parameters included: raw water supply (Q), raw water turbidity (T1), treated water turbidity (T2), treated water residual free chlorine (Cl2), treated water concentration of residual aluminum (Al), filtration bed inlet water turbidity (T3), daily difference in water height in reservoir (∆H), raw water pH (pH1), treated water pH (pH2), and daily consumption of DWTP electricity (El). Output/target parameters were: residual O3 after ozonation (O3), anionic polyelectrolyte (ANPE), poly-aluminum chloride hydroxide sulfate (PACl), and chlorine gas supply (Cl2(g)). A total of 304 different ANN models were tested, based on the best test performance (tperf) indicator. The one with the optimum performance indicator was selected. The scenario finally chosen was the one with 100 neural networks, 100 nodes, 42 hidden nodes, 10 inputs, and 4 outputs. This ANN model achieved excellent simulation results based on the best testing performance indicator, which suggests that ANNs are potentially useful tools for the prediction of a DWTP’s main operational parameters. Further research could explore the prediction of water chemicals used in a DWTP by using ANNs with a smaller number of operational parameters to ensure greater flexibility, without prohibitively reducing the reliability of the prediction model. This could prove useful in cases with a much higher sample size, given the data-demanding nature of ANNs.

Comparison of experimental thermal conductivity of water-based Al2O3–Cu hybrid nanofluid with theoretical models and artificial neural network output

Comparison of experimental thermal conductivity of water-based Al2O3–Cu hybrid nanofluid with theoretical models and artificial neural network output

Use of Tungsten Inert Gas Welding on Mild Steel Weldment to Optimize Welding Process Variables on Electrode Melting Rate

One of the most crucial factors taken into account when evaluating the performance of Tungsten Inert Gas (TIG) welding is the Electrode Melting Rate, which shows how much of the heat deposited by the welding operation is used to generate melting. A dense weld pool forms in the field of welding when there is a good melting rate. Therefore, this paper investigates the use of tungsten inert gas welding on mild steel weldment to optimize welding process variables on electrode melting rate employing Artificial Neural Networks and Response Surface Methodology in the analysis. The ideal electrode melting rate of 4.6539 mm/s can be achieved by combining the following parameters: wire diameter of 2.55 mm, welding speed of 3.02 mm/s, and welding current of 208.22A, according to the RSM results. R values of 0.9119 were displayed by the RSM for the electrode melting rate. A regression plot of the ANN results indicates that the total R value is 0.93807. Because the Artificial Neural Network's output fits the experimental data more closely than the Response Surface Methodology's, it is chosen as the superior predictive model.

Two-dimensional stiffness identification by inversion of local wavenumbers based on artificial neural network

Impact and fatigue damages in carbon fiber reinforced polymers (CFRP) are unavoidable in severe operating conditions, which will lead to the stiffness properties degrading and redistributing in space. The local wavenumbers of guided waves at a specific point depend on the stiffness at that spatial location. Therefore, the two-dimensional stiffness properties can be determined by inversion of local wavenumbers. In this paper, the relationship between the local wavenumbers and the stiffness coefficients was established by the artificial neural network (ANN) method. For each spatial point, based on the semi-analytical finite element (SAFE) method, the wavenumbers (inputs of ANN) with different frequencies can be obtained at given stiffness coefficients (outputs of ANN). Furthermore, the accuracy of the ANN method was validated by the finite element model. Results show that the predicted two-dimensional stiffness coefficients of the damaged region are close to the true ones, thus demonstrating the accuracy of the two-dimensional stiffness identification method.

Modeling the Drying Process of Onion Slices Using Artificial Neural Networks

One of the food preservation technologies is the drying process, which requires heat and is significantly energy-intensive, resulting in high costs. This caused the search for new design solutions for dryers, which requires continuous experimental research and the creation of new decision-supporting models for the optimization of drying processes. In this work, four models of the kinetics of convective onion drying were developed using Artificial Neural Networks (ANNs), taking into account pre-treatment before drying and the different temperatures of the drying agent. The moisture content in the dried material at a specific moment in time was taken as the dependent variable (ANN output). The following were accepted as independent variables (ANN inputs): drying temperature, initial sample thickness, initial moisture content, initial mass of the sample, time of drying, and material pre-treatment (no pre-treatment—blanching–osmotic dehydration). Four semantic models were formulated, the general Ann1 model taking into account all input variables and three detailed Ann2 models for individual types of pre-treatment. For the best Ann1, the MAPE values were 5.88–7.02% (for different data: Training, Test, Validation). For the detailed Ann2 models, the error values were more than twice lower. The MAPE values ranged from 1.14% to 3.12%.

Performance prediction for network-level flexible pavements using an optimised ANN approach

ABSTRACT Considering that the road network shares traffic flow as a system, this study aimed to predict the network-level flexible pavement performance, using an optimised artificial neural network (ANN) approach to predict the pavement performance of each individual road, and the Ford-Fulkerson algorithm to determine the weight coefficients of roads. ANNs were developed to predict distress conditions, functional conditions and structural performance using the inputs of the pavement age and structural, traffic and climatic conditions. ANNs were trained with Long-Term Pavement Performance (LTPP) program data, and almost all the coefficient of determination values between ANN outputs and measured results are larger than 0.9. This ANN approach was optimised by establishing grey models to provide predictive short-term performance data for training ANNs, and employing Kalman Filter to modify the long-term performance prediction. Applications in typical LTPP sections validated the effectiveness of the optimisation method. This study used the reduction in road network capacity due to the assumed unavailability of a certain road to quantify its role (weight coefficient), which was calculated using the Ford-Fulkerson algorithm. This network-level performance prediction approach was applied in a 25-expressways network in Shanghai, which validated the good generalisation capabilities of ANNs and feasibility of the Ford-Fulkerson algorithm.

Breast Cancer Subtype Prediction Model Employing Artificial Neural Network and 18F-Fluorodeoxyglucose Positron Emission Tomography/ Computed Tomography.

Although positron emission tomography/computed tomography (PET/CT) is a common tool for measuring breast cancer (BC), subtypes are not automatically classified by it. Therefore, the purpose of this research is to use an artificial neural network (ANN) to evaluate the clinical subtypes of BC based on the value of the tumor marker. In our nuclear medical facility, 122 BC patients (training and testing) had 18F-fluoro-D-glucose (18F-FDG) PET/CT to identify the various subtypes of the disease. 18F-FDG-18 injections were administered to the patients before the scanning process. We carried out the scan according to protocol. Based on the tumor marker value, the ANN's output layer uses the Softmax function with cross-entropy loss to detect different subtypes of BC. With an accuracy of 95.77%, the result illustrates the ANN model for K-fold cross-validation. The mean values of specificity and sensitivity were 0.955 and 0.958, respectively. The area under the curve on average was 0.985. Subtypes of BC may be categorized using the suggested approach. The PET/CT may be updated to diagnose BC subtypes using the appropriate tumor maker value when the suggested model is clinically implemented.

From Detection to Action: Implementing Deep Learning Inference in PLC Systems via Docker Services

This paper focuses on the implementation of recognition and detection algorithms for programmable logic controllers (PLCs) based on artificial intelligence (AI). AI methods based on deep neural networks (DNN) are characterized by a large number of hidden layers and neurons in hidden layers, which increases the computational complexity of computing the outputs of artificial neural networks (ANNs). Popular convolutional neural networks (CNN) are among such ResNet, AlexNet, GoogLeNet and others. A remote cloud solution is proposed in this paper to compute CNN outputs. The CNN implementation was tested on object recognition and detection applications. Training and testing of the CNN and YOLO detector was carried out in the Python environment.

Rule extraction based on PROMETHEE-assisted multi-objective genetic algorithm for generating interpretable neural networks

Rule extraction from artificial neural networks (ANNs) refers to the process of identifying the underlying decision-making logic used by the network in its predictions. This involves converting the complex, non-linear relationships between inputs and outputs in the network into a set of interpretable rules. The main objective of rule extraction is to enhance the interpretability of decision-making in artificial neural networks (ANNs) and make it more understandable for human experts, especially in critical domains like healthcare, where clear decision-making is essential. Despite the ability of machine learning methods to interpret ANNs, interpreting complex neural networks remains limited, as most studies tend to sacrifice one criterion, related to rule quality, in favor of another. Rule extraction methods applied directly to complex neural networks may produce overly specific rules that are tailored to the training data. Such rules can suffer from overfitting and may not generalize well to unseen data. Also, The use of multiobjective genetic algorithms for rule extraction can result in a large number of rules, making it challenging for decision-makers to interpret them. This study proposes an innovative approach to address challenges in rule extraction from complex artificial neural networks (ANNs) by integrating a multiobjective genetic algorithm with the PROMETHEE method. By integrating these techniques, the study aims to identify meaningful relationships between inputs and outputs of ANNs, generating concise and reliable rules that are ranked based on support, confidence, and lift values. The use of PROMETHEE's rule ranking mechanism enabled the consideration of rule priority during candidate selection, guiding the genetic algorithm towards discovering high-quality rule sets and efficiently exploring the search space. This approach not only enhances the quality of extracted rules but also facilitates efficient decision-making by striking a balance between rule accuracy, fidelity, and comprehensibility, thus contributing to advancing the understanding of complex neural networks. By equally weighting the criteria (support, confidence, and lift) during the rule ranking mechanism, our approach achieved high coverage rates ranging from 94.88 % to 100 % and generated a manageable number of rules that could be easily interpreted by human experts, ranging from 5 to 10.5. These findings demonstrate the potential of our approach to significantly improve decision-making accuracy and interpretability across various real-world applications, making it a promising tool for applications in healthcare, finance, and other fields.

Using Fundamental Parameters and Artificial Neural Network for Controlling the Laser Welding of Metals

The present work aims to design an artificial neural network (ANN) for controlling the geometry of the fusion zone in different welding conditions. AA5754 aluminium plates with 6 mm thickness were welded in butt configuration by using an Yb-doped fiber laser in continuous wave regime. Laser power, travel speed, beam diameter and shielding gas were considered as process-related factors, while the weld geometry was evaluated in terms of penetration depth and bead width. The accuracy of the model was endorsed by using untrained test data. Results showed a good agreement between the ANN output and the experimental values.

Modeling and optimization of the corrosion resistance of Cr-free and Cr-based chemical conversion coatings on nickel foil by artificial neural network and response surface method

Chemical conversion treatment is one of the most powerful methods to prevent the corrosion of metals. However, determining the optimal addition amount of various active ingredients for the enhancement of corrosion resistance requires a large number of repeated experiments, which is very time-consuming and is also not helpful for figuring out the interactions between active ingredients. In the present study, Cr-free conversion coatings (CFCC) and Cr-based conversion coatings (CCC) were prepared on nickel foil and characterized by scanning electron microscopy and X-ray photoelectron spectroscopy to determine their morphology and composition of them. The corrosion resistance of conversion coatings was measured by salt spray test and further investigated using the response surface method (RSM) and artificial neural network (ANN) approach. The constructed models displayed high predictive accuracy, and the coefficient of correlation value of the RSM-CFCC, ANN-CFCC, RSM-CCC, and ANN-CCC models is 0.9441, 0.9513, 0.9535 and 0.9564, respectively. This comparative study recommends the ANN model best fits the experimental data for both CFCC and CCC. By using the ANN outputs, satisfactory results can be estimated with superior accuracy and efficiency while the RSM model can describe the interactions between factors.

Combining Weather Station Data and Short-Term LiDAR Deployment to Estimate Wind Energy Potential with Machine Learning: A Case Study from the Swiss Alps

Wind energy potential in complex terrain is still poorly understood and difficult to quantify. With Switzerland’s current efforts to shift to renewable energy resources, it is now becoming even more crucial to investigate the hidden potential of wind energy. However, the country’s topography makes the assessment very challenging. We present two measurement campaigns at Lukmanier and Les Diablerets, as representative areas of the complex terrain of the Swiss Alps. A general understanding of local wind flow characteristics is achieved by comparing wind speed measurements from a near-surface ultra-sonic anemometer and from light detection and ranging (LiDAR) measurements further aloft. The measurements show how the terrain modifies synoptic wind for example through katabatic flows and effects of local topography. We use an artificial neural network (ANN) to combine the data from the measurement campaign with wind speed measured by weather stations in the surrounding area of the study sites. The ANN approach is validated against a set of LiDAR measurements which were not used for model calibration and also against wind speed measurements from a 25-meter mast, previously installed at Lukmanier. The statistics of the ANN output obtained from multi-year time series of nearby weather stations match accurately the ones of the mast data. However, for the rather short validation periods from the LiDAR, the ANN has difficulties in predicting lowest wind speeds at both sites, and highest wind speeds at Les Diablerets.

GDP responses to supply chain disruptions in a post-pandemic era: Combination of DL and ANN outputs based on Google Trends

With the recent Russian-Ukraine conflict, the frequency and intensity of disruptive shocks on major supply chains have risen, causing increasing food and energy security concerns for regulators. That is, the combination of newly available sophisticated deep learning tools with real-time series data may represent a fruitful policy direction because machines can identify patterns without being pre-conditioned calibration thanks to experimental data training. This paper employs Deep Learning (DL) and Artificial Neural Network (ANN) algorithms and aimed predicts GDP responses to supply chain disruptions, energy prices, economic policy uncertainty, and google trend in the US. Sampled data from 2008 to 2022 are monthly wrangled and embed different recession episodes connected to the subprime crisis of 2008, the COVID-19 pandemic, the recent invasion of Ukraine by Russia, and the current economic recession in the US. Both DL and ANN outputs empirically (and unanimously) demonstrated how sensitive monthly GDP variations are to dynamic changes in supply chain performances. Findings identify the substantial role of google trends in delivering a consistent fit to predicted GDP values, which has implications While a comparative discussion over the larger forecasting performance of DL compared to ANN experiments is offered, implications for global policy, decision-makers and firm managers are finally provided.

New neural network-based algorithm for predicting fatigue life of aluminum alloys in terms of machining parameters

Various classification of aluminum alloys is one of the most common applied materials in transportation industries due to the specific mechanical and material properties. Body of vehicles, including cars, ships, and airplanes, are constantly exposed to different cyclic loads that lead to surface damage and finally, fatigue failure occurs by crack growth. Therefore, the quality of the machined surface has a direct effect on the fatigue life of the products. In this regard, the most important source of surface roughness is the choice of machining parameters. To understand it, turning operations were performed on 2xxx and 7xxx series aluminum alloys considering different values of process parameters. For any series of aluminums, 189 high-cycle fatigue testing specimens were prepared, and experiments were performed by axial tension–compression fatigue test machine in 7 levels of stress (all tests were repeated three times and the mean failure cycles were reported as the fatigue life). Next, an Artificial Neural Network (ANN) based on the Back Propagation (BP) error algorithm was developed to predict fatigue life of Al alloys which are machined with different conditions. To this end, the parameters considered as input variables to the neural network structure include the Yield Strength (YS) and Ultimate Tensile Strength (UTS) to identify the aluminum series, as well as the process parameters such as cutting depth (d), rotational speed (R), and feed speed (V). In addition, the applied cyclic stress (S) to specimen was considered as input. Furthermore, surface roughness (Z) and number of cycles to failure (N) were considered as ANN output. The comparison of the obtained results from ANN predicting with the experimental values indicates that this approach was tuned finely. Eventually, the presented model can be a suitable alternative to perform fatigue tests with high costs and time-consuming.‌‌ Also, the most effective machining parameter on the surface roughness and fatigue limit was reported.

Artificial Intelligence Can Guide Antibiotic Choice in Recurrent UTIs and Become an Important Aid to Improve Antimicrobial Stewardship.

A correct approach to recurrent urinary tract infections (rUTIs) is an important pillar of antimicrobial stewardship. We aim to define an Artificial Neural Network (ANN) for predicting the clinical efficacy of the empiric antimicrobial treatment in women with rUTIs. We extracted clinical and microbiological data from 1043 women. We trained an ANN on 725 patients and validated it on 318. The ANN showed a sensitivity of 87.8% and specificity of 97.3% in predicting the clinical efficacy of empirical therapy. The previous use of fluoroquinolones (HR = 4.23; p = 0.008) and cephalosporins (HR = 2.81; p = 0.003) as well as the presence of Escherichia coli with resistance against cotrimoxazole (HR = 3.54; p = 0.001) have been identified as the most important variables affecting the ANN output decision predicting the fluoroquinolones-based therapy failure. A previous isolation of Escherichia coli with resistance against fosfomycin (HR = 2.67; p = 0.001) and amoxicillin-clavulanic acid (HR = 1.94; p = 0.001) seems to be the most influential variable affecting the output decision predicting the cephalosporins- and cotrimoxazole-based therapy failure. The previously mentioned Escherichia coli with resistance against cotrimoxazole (HR = 2.35; p < 0.001) and amoxicillin-clavulanic acid (HR = 3.41; p = 0.007) seems to be the most influential variable affecting the output decision predicting the fosfomycin-based therapy failure. ANNs seem to be an interesting tool to guide the antimicrobial choice in the management of rUTIs at the point of care.

Minerals resource rent responses to economic performance, greener energy, and environmental policy in China: Combination of ML and ANN outputs

The relationship of mineral resources rent with economic performance, greener energy, and environmental policy is crucial because it could impact how welfare and eco-friendly are treated in a nation. This paper employed machine learning (ML) and Artificial Neural Networks (ANN) to predict minerals resourcesthrougheconomic growth, the producer price index, market prices, the environmental policy stringency index (EPSI), and greener energy in China. The ML and ANN outputs showed how responsive quarterly minerals resource rent variations are too dynamic shifts in economic performance and renewable energy use. The more critical prediction accuracy of ML experiments compared to ANN experiments is highlighted by data from mean absolute percentage, mean square, root mean square, and root mean absolute errors, as well as the coefficient of determination. Though China is abundant in natural resources, there is a need to employ these resources efficiently to achieve long-term eco-friendly performance. In addition to comparing the results of ML and ANN, it also provided causality conclusions and policy implications.

Application of artificial neural networks in performance prediction of cement mortars with various mineral additives

The machine learning technique for prediction and optimization of building material performances became an essential feature in the contemporary civil engineering. The Artificial Neural Network (ANN) prognosis of mortar behavior was conducted in this study. The model appraised the design and characteristics of seventeen either building or high-temperature mortars. Seven different cement types were employed. Seventeen mineral additives of primary and secondary origin were embedded in the mortar mixtures. Cluster Analysis and Principal Component Analysis designated groups of similar mortars assigning them a specific purpose based on monitored characteristics. ANN foresaw the quality of designed mortars. The impact of implemented raw materials on the mortar quality was assessed and evaluated. ANN outputs highlighted the high suitability level of anticipation, i.e., 0.999 during the training period, which is regarded appropriate enough to correctly predict the observed outputs in a wide range of processing parameters. Due to the high predictive accuracy, ANN can replace or be used in combination with standard destructive tests thereby saving the construction industry time, resources, and capital. Good performances of altered cement mortars are positive sign for widening of economical mineral additives application in building materials and making progress towards achieved carbon neutrality by reducing its emission.