Bayesian Regularized Artificial Neural Network Research Articles

Data-driven prediction of tool wear using Bayesian regularized artificial neural networks

The prediction of wear in cutting tools is pivotal for boosting productivity and reducing manufacturing costs. Although current data-driven models in machine learning and deep learning have advanced predictive capabilities, they often lack generality and demand substantial data training. This paper presents a novel approach using Bayesian Regularized Artificial Neural Networks (BRANNs) to precisely forecast wear in milling tools. Unlike conventional machine learning models, BRANNs merge the strengths of artificial neural networks (ANNs) and Bayesian regularization, yielding a more robust and generalized predictive model. We utilized three open-access datasets from the literature alongside an in-house dataset generated by our milling setup. Initially, we assessed the model’s predictive ability by training and testing it against individual open-access datasets. We investigated the impact of input features, training data size, hidden units, training algorithms, and transfer functions on the model’s predictive capability. Subsequently, we trained the model using three open-access datasets and tested it against our in-house data. Our findings demonstrate that the developed model surpasses existing state-of-the-art models in accuracy and transferability

Robust prediction of workability properties for 3D printing with steel slag aggregate using bayesian regularization and evolution algorithm

This study addresses the growing interest in utilizing steel slag as a sustainable alternative to river sand in additive manufacturing of concrete, driven by the increasing scarcity of natural resources. The rheological properties of fresh material significantly impact the quality of 3D-printed filament, necessitating suitable workability for printability. The research focuses on evaluating the influence of steel slag aggregate and key admixtures, such as silica fume and superplasticizer, on the workability properties of fresh concrete. Through an extensive series of 90 slump tests, optimal combinations ensuring the desired workability for 3D printing applications were identified. To enhance the manufacturing of fresh concrete and develop user-friendly tools, a novel soft computing approach is introduced—Adaptive Elitist Differential Evolution coupled with Bayesian Regularization Artificial Neural Network (aeDE-BRANN). This advanced model considers five critical input parameters: silica fume to Portland cement ratio, steel slag to cement ratio, water to cement ratio, cement content, and superplasticizer dosage. The model outputs crucial workability metrics, including slump flow and slump. In a comprehensive comparative study, the aeDE-BRANN framework demonstrates superior performance in terms of both simplicity and efficiency when compared to other forecasting models. Feature analysis techniques, including Shapley values from game theory and partial dependence plots, provide valuable insights into the intricate relationships between input variables and workability properties. The findings underscore the water to cement ratio as the most influential factor on workability, followed by silica fume to Portland cement ratio and steel slag to cement ratio. This study contributes a reliable tool for predicting workability properties, aiding in the selection of suitable steel slag aggregate and admixtures. Ultimately, it facilitates the sustainable and innovative evolution of 3D concrete printing practices.

Tree Biomass Modeling Based on the Exploration of Regression and Artificial Neural Networks Approaches

Understanding the dynamics of tree biomass is a significant factor in forest ecosystems, and accurate quantitative knowledge of its development provides support for the optimization of forest management. This work aimed to employ innovative practices in tree biomass modeling, artificial neural network approaches along with the least-squares regression methodology, in order to construct reliable and accurate estimation and prediction models that contribute to solving the emerging problems in the field of sustainable forest management. Based on this aim, different modeling strategies were developed and explored. The nonlinear seemingly unrelated regression (NSUR) methodology, the generalized regression (GRNN), the resilient propagation (RPNN) and the Bayesian regularization (BRNN) artificial neural network algorithms were utilized for the construction of reliable biomass models to attain the most accurate and reliable tree biomass components and total tree biomass estimations. The work showed that GRNN models provided a significantly better performance compared with the other modeling methodologies tested. Considering the non-parametric nature of the GRNN neural network algorithm, the fact that it was designed for nonlinear regression-type problems capable of dealing with small datasets, this modeling approach warrants consideration as an effective alternative to nonlinear regression or to other neural network approaches to the field of tree biomass modeling.

The Predictive Power of Macroeconomic Variables on the Indian Stock Market Utilizing an Ann Model Approach: An Empirical Investigation Based on BSE Sensex

Abstract Research background The paper focuses on the use of Artificial Neural Networks (ANNs) for forecasting time series data of the stock market since ANNs are dynamic and are more capable of handling complex data sets in comparison to conventional forecasting techniques such as regression, Logistic regression, and have massive potential for the prediction of stock market prices. Purpose Artificial neural networks are an effective method for forecasting time series. Therefore, this study aims to forecast the closing price of the BSE Sensex using artificial neural networks (ANNs). Research methodology The study uses nine input variables, including macroeconomic and global stock market factors, to estimate the BSE Sensex using scaled conjugate gradient algorithm artificial neural networks (SCGANNs) and Bayesian regularized artificial neural networks (BRANN). Results As per the empirical results of the study, the ANN model can forecast the closing values of the BSE Sensex with a Bayesian Regularization (BR) method with an accuracy of over 99 percent, thus leading to significant implications for domestic institutional investors (DIIs), foreign institutional investors (FIIs), investment houses, and so on. This study adds more value to the existing literature by proving that the BRANN models outperform SCGANN in stock market forecasting. Novelty This is the first study to employ macroeconomic variables as input variables for predicting the Indian stock market using ANN. The study highlights the ANN model’s forecasting potential, giving investors robust and accurate stock value prediction capabilities.

Genome-enabled prediction of indicator traits of resistance to gastrointestinal nematodes in sheep using parametric models and artificial neural networks

This study aimed to assess the predictive ability of parametric models and artificial neural network method for genomic prediction of the following indicator traits of resistance to gastrointestinal nematodes in Santa Inês sheep: packed cell volume (PCV), fecal egg count (FEC), and Famacha© method (FAM). After quality control, the number of genotyped animals was 551 (PCV), 548 (FEC), and 565 (FAM), and 41,676 SNP. The average prediction accuracy (ACC) calculated by Pearson correlation between observed and predicted values and mean squared errors (MSE) were obtained using genomic best unbiased linear predictor (GBLUP), BayesA, BayesB, Bayesian least absolute shrinkage and selection operator (BLASSO), and Bayesian regularized artificial neural network (three and four hidden neurons, BRANN_3 and BRANN_4, respectively) in a 5-fold cross-validation technique. The average ACC varied from moderate to high according to the trait and models, ranging between 0.418 and 0.546 (PCV), between 0.646 and 0.793 (FEC), and between 0.414 and 0.519 (FAM). Parametric models presented nearly the same ACC and MSE for the studied traits and provided better accuracies than BRANN. The GBLUP, BayesA, BayesB and BLASSO models provided better accuracies than the BRANN_3 method, increasing by around 23% for PCV, and 18.5% for FEC. In conclusion, parametric models are suitable for genome-enabled prediction of indicator traits of resistance to gastrointestinal nematodes in sheep. Due to the small differences in accuracy found between them, the use of the GBLUP model is recommended due to its lower computational costs.

Artificial intelligence application in adsorption of uremic toxins: Towards the eco-friendly design of highly efficient with potential applications as hemodialysis membranes

Six Functionalized Activated Carbon Cloths (FACCs) were designed to obtain fundamental information for training a Bayesian Regularized Artificial Neural Network (BRANN) capable of predicting adsorption capacity of the FACCs to synthesize tailor-made materials with potential application as dialysis membranes. Characterization studies showed that FACCs have a high surface area (1354–2073 m2 g−1) associated with increased microporosity (W0, average: 0.57 cm3 g−1). Materials are carbonaceous, with a carbon content between 69 and 92%. Chemical treatments modify the pHpzc of materials between 4.1 and 7.8 due to incorporating functional groups on the surface (C=O, –COOH, –OH, –NH, –NH2). Uremic toxins tests showed a high elimination rate of p-cresol (73 mg g−1) and creatinine (90 mg g−1) which is not affected by the matrix (aqueous solution and simulated serum). However, in the case of uric acid, adsorption capacity decreased from 143 mg g−1 to 71 mg g−1, respectively. When comparing the kinetic constants of the adsorption studies in simulated serum versus the studies in aqueous solution, it can be seen that this does not undergo significant changes (0.02 min−1), evidencing the versatility of the material to work in different matrices. The previous studies, in combination with characterization of the materials, allowed to establish the adsorption mechanism. Thus, it permitted to train the BRANN to obtain mathematical models capable to predict the kinetic adsorption of the toxins studied. It is concluded that the predominant adsorption mechanism is due to π−π interactions between the adsorbate unsaturations with the material's pseudo-graphitic planes. Results show that FACCs are promising materials for hemodialysis membranes. Finally, taking into consideration the adsorption capacities and rates, as well as the semiquantitative analysis of the environmental impact associated with the preparation of the adsorbents, the best adsorbent (CC, Eco-Scale = 91.5) was selected. The studies presented show that the material is eco-friendly and highly efficient in the elimination of uremic toxins.

Determining the Proper Times and Sufficient Actions for the Collision Avoidance of Navigator-Centered Ships in the Open Sea Using Artificial Neural Networks

Ship collisions are a major maritime accident; various systems have been proposed to prevent them. Through investigating and analyzing the causes of maritime accidents, it has been established that ship collisions can either caused by delaying actions or not taking the sufficient actions to avoid them. Recognizing the limitations in providing quantitative numerical values for avoiding ship collisions, this study aimed to use Bayesian regularized artificial neural networks (BRANNs) to suggest the proper time and sufficient actions required for ship collision avoidance consistent with the Convention on the International Regulations for Preventing Collisions at Sea. We prepared the data by calculating the proper times and sufficient actions based on precedent research and used them to train, validate, and assess the BRANNs. Subsequently, an artificial neural network controller was designed and proposed. The data of the proposed neural network controller were verified via simulation, validating the controller. This study is limited in cases such as overtaking a ship in front. However, it is expected that this controller can be improved by establishing the criteria for an appropriate overtaking distance after further examining the closest point of approach (CPA) and time to the CPA (TCPA) for overtaking a ship in front and using the method presented herein.

A spatiotemporal monitoring model of TSM and TDS in arid region lakes utilizing Sentinel-2 imagery

The assessment of lake water quality is crucial for protecting and managing natural resources. Total suspended matter (TSM) and total dissolved solids (TDS) are significant parameters in evaluating lake water quality, and their concentration levels provide useful information about the health of the ecosystem. However, measuring TSM and TDS concentrations in remote arid areas can be challenging, necessitating the development of a simple, efficient, and cost-effective monitoring system. In this study, the aim was to retrieve TDS and TSM concentrations in the arid region using an empirical and physics-based method simultaneously. Measured data and remote sensing images for Chah-Nimeh Reservoirs (CNRs), Iran, were used to establish the TSM and TDS concentration model. Twenty-seven Sentinel-2 multispectral instrument (MSI) data acquired from 2018 to 2021 were employed to determine the spatiotemporal distribution of TDS and TSM concentrations. The results indicate that Sentinel-2 MSI data is a productive tool for retrieving TSM concentration using the Case-2 Regional/Coastal Color (C2RCC) processor, with a high level of performance in the dry ecosystem of CNRs (R2 = 0.92). Furthermore, the Bayesian regularization artificial neural network (BRANN) algorithm was used to evaluate TDS concentration, and the results demonstrated that the BRANN algorithm outperforms other models, with a high level of performance (R2 = 0.89). It is noteworthy that evaluating TDS concentration is generally more challenging than TSM. The study provides a valuable model for monitoring TSM and TDS concentrations in remote arid areas using Sentinel-2 MSI data. The proposed approach can facilitate effective spatiotemporal monitoring of TSM and TDS concentrations and aid in the evaluation of lake water quality, specifically in regions where in-situ measurements are not feasible. The findings have significant implications for the management and conservation of natural resources in arid and remote regions.

Multi-peak emission of In2O3 induced by oxygen vacancy aggregation

Oxygen vacancy is crucial to the optical properties in In2O3, however, the single oxygen vacancy model fails to explain the observed multi-peak emission in the experiment. Herein, we have theoretically investigated the diversity of oxygen vacancy distribution, revealing the relationship between the defect configurations and the optical properties. Combining the first-principles calculations and bayesian regularized artificial neural networks, we demonstrate that the structural stability can be remarkably enhanced by multi-oxygen vacancy aggregation, which will evolve with the defect concentration and temperature. Notably, our results indicate that the single oxygen vacancy will induce the emission peaks centered at 1.35 eV, while multi-peak emission near 2.35 eV will be attributed to the distribution of aggregated double oxygen vacancies. Our findings provide a comprehensive understanding of multi-peak emission observed in In2O3, and the rules of the vacancy distribution may be extended for other metal oxides to modulate the optical properties in practice.

Simultaneous retrieval of sugarcane variables from Sentinel-2 data using Bayesian regularized neural network

Quantifying biophysical and biochemical vegetation variables is of great importance in precision agriculture. Here, the ability of artificial neural networks (ANNs) to generate multiple outputs is exploited to simultaneously retrieve Leaf area index (LAI), leaf sheath moisture (LSM), leaf chlorophyll content (LCC), and leaf nitrogen concentration (LNC) of sugarcane from Sentinel-2 spectra. We apply a type of ANNs, Bayesian Regularized ANN (BRANN), which incorporates the Bayes' theorem into a regularization scheme to tackle the overfitting problem of ANN and improve its generalizability. Quantitatively assessing the result accuracy indicated RMSE values of 0.48 (m2/m2) for LAI, 2.36 (% wb) for LSM, 5.85 (μg/cm2) for LCC, and 0.23 (%) for LNC, applying simultaneous retrieval. It was demonstrated that simultaneous retrievals of the variables outperformed the individual retrievals. The superiority of the proposed BRANN over a conventional ANN trained with the Levenberg-Marquardt algorithm was confirmed through statistical comparison of their results. The model was applied over the entire Sentinel-2 images to map the considered variables. The maps were probed to qualitatively evaluate the model performance. The results indicated that the retrievals reasonably represent spatial and temporal variations of the variables. Generally, this study demonstrated that the BRANN simultaneous retrieval model can provide faster and more accurate retrievals than those obtained from conventional ANNs and individual retrievals.

RETRIEVAL OF SUGARCANE LEAF AREA INDEX FROM PRISMA HYPERSPECTRAL DATA

Abstract. The PRecursore IperSpettrale della Missione Applicativa (PRISMA) satellite of the Italian Space Agency, lunched in 2019, has provided a new generation source of hyperspectral data showing to have high potential in vegetation variable retrieval. In this study, the newly available PRISMA spectra were exploited to retrieve Leaf Area Index (LAI) of sugarcane using a new kind of Artificial Neural Networks (ANN) so-called Bayesian Regularized Artificial Neural Network (BRANN). The suggested BRANN retrieval model was implemented over a dataset collected during a field campaign in Amir Kabir Sugarcane Agro-Industrial zone, Khuzestan, Iran, in 2020. Principle Component Analysis (PCA) was utilized to reduce the dimensionality of PRISMA data cube. An accuracy assessment based on the bootstrapping procedure indicated RMSE of 0.67 m2/m2 for the LAI retrieval by applying the BRANN model. This study is a confirmation of the high performance of the BRANN method and high potential of PRISMA images to retrieve sugarcane LAI.

Evaluation of Predictive Ability of Bayesian Regularized Neural Network Using Cholesky Factorization of Genetic Relationship Matrices for Additive and Non-additive Genetic Effects

This study aimed to explore the effects of additive and non-additive genetic effects on the prediction of complex traits using Bayesian regularized artificial neural network (BRANN). The data sets were simulated for two hypothetical pedigrees with five different fractions of total genetic variance accounted by additive, additive x additive, and additive x additive x additive genetic effects. A feed forward artificial neural network (ANN) with Bayesian regularization (BR) was used to assess the performance of different nonlinear ANNs and compare their predictive ability with those from linear models under different genetic architectures of phenotypic traits. Effective number of parameters and sum of squares error (SSE) in test data sets were used to evaluate the performance of ANNs. Distribution of weights and correlation between observed and predicted values in the test data set were used to evaluate the predictive ability. There were clear and significant improvements in terms of the predictive ability of linear (equivalent Bayesian ridge regression) and nonlinear models when the proportion of additive genetic variance in total genetic variance ( ) increased. On the other hand, nonlinear models outperformed the linear models across different genetic architectures. The weights for the linear models were larger and more variable than for the nonlinear network, and presented leptokurtic distributions, indicating strong shrinkage towards 0. In conclusion, our results showed that: a) inclusion of non-additive effects did not improve the prediction ability compared to purely additive models, b) The predictive ability of BRANN architectures with nonlinear activation function were substantially larger than the linear models for the scenarios considered.

Probability Prediction Approach of Fatigue Failure for the Subsea Wellhead Using Bayesian Regularization Artificial Neural Network

The subsea wellhead (SW) system is a crucial connection between blowout preventors (BOPs) and subsea oil and gas wells. Excited by cyclical fatigue dynamic loadings, the SW is prone to fatigue failure, which would lead to the loss of well integrity and catastrophic accidents. Based on the Bayesian Regularization Artificial Neuron Network (BRANN), this paper proposes an efficient probability approach to predict the fatigue failure probability of SW during its entire life. In the proposed method, the BRANN fatigue damage (BRANN-FD) model reflecting the non-linear relationship between the input and output data was developed by the limited fatigue damage analysis data, which was utilized to generate thousands of non-numerical fatigue damage data of SW rapidly. Combining parametric and non-parametric estimation methods, the probability density function (PDF) of SW fatigue damage was determined to calculate the accumulation fatigue damage during service life. Using the logistic regression, the fatigue failure probability of SW was predicted. The application of the proposed approach was demonstrated by a case study. The results illustrated that the fatigue damage of SW would be viewed as obeying the Lognormal distribution, which could be used to obtain the accumulation fatigue damage in operation conveniently. Furthermore, the fatigue failure probability of SW nonlinearly increased with the increment in the accumulation fatigue damage of SW, which could be helpful to ensure the operation safety of SW in deepwater oil and gas development, especially for aged wellhead.

Service level modelling of motorized three-wheelers at un-signalized intersections under heterogeneous traffic conditions

In developing nations like India, due to non-homogeneity in traffic streams, the flow of motorized three-wheelers gets clogged. While studying the literature, researchers do not find any significant Level of Service (LOS) models for forecasting motorized three-wheelers' service quality at uncontrolled un-signalized intersections under heterogeneous traffic conditions. This study brings to an AI-based Three-Wheeler Level of Service (3WhLOS) model to evaluate the service quality offered by un-signalized intersections operating under mixed traffic conditions. Data are collected from 21 uncontrolled intersections located at 7 different cities of India. Spearman's correlation analysis is performed to fathom the influence of service parameters towards perceived 3WhLOS score. Bayesian Regularized Artificial Neural Network (BRANN) is adopted for the prediction of 3WhLOS scores. Sensitivity Analysis is also executed to determine the relative importance of each parameter and help the transport authorities to identify the issues and improvise them for the betterment of the users.

Prediction of axial load bearing capacity of PHC nodular pile using Bayesian regularization artificial neural network

Pre-stressed precast high strength concrete (PHC) nodular piles with hyper-MEGA construction method are favorably used in medium to high-rise building foundations. In this study, a feed-forward neural network (FFNN) was adopted to investigate the ultimate axial load bearing capacity of the PHC nodular pile. The network receives the composite pile and geotechnical conditions with eight input neurons and outputs the nodular pile's ultimate axial load bearing capacity. Among numerous possible FFNN network architectures, the most accurate one is determined by optimizing the hidden layer. Network training is conducted with Bayesian regularization backpropagation (BRB); the training datasets consist of static pile load test and standard penetration test index of soil profile collected from various projects in Vietnam. The significance of each input parameter is quantified with importance-based sensitivity analysis. An explicit function has been constructed from weights and bias values at each neuron in the FFNN to estimate the axial load bearing capacity. The excellent agreement of all output values by the proposed FFNN with the measured values proved the model’s robustness and reliability. The predictive capacity of the proposed FFNN model has significantly outperformed all current empirical formulas. The outcome of this study can be directly put into engineering practice to furnish an economically optimal design of the composite nodular pile.

Prediction of Casing Collapse Strength Based on Bayesian Neural Network

With the application of complex fracturing and other complex technologies, external extrusion has become the main cause of casing damage, which makes non-API high-extrusion-resistant casing continuously used in unconventional oil and gas resources exploitation. Due to the strong sensitivity of string ovality, uneven wall thickness, residual stress, and other factors to high anti-collapse casing, the API formula has a big error in predicting the anti-collapse strength of high anti-collapse casing. Therefore, Bayesian regularization artificial neural network (BRANN) is used to predict the external collapse strength of high anti-collapse casing. By collecting full-scale physical data, including initial defect data, geometric size, mechanical parameters, etc., after data preprocessing, the casing collapse strength data set is established for model training and blind measurement. Under the classical three-layer neural network, the Bayesian regularization algorithm is used for training. Through empirical formula and trial and error method, it is determined that when the number of hidden neurons is 12, the model is the best prediction model for high collapse resistance casing. The prediction results of the blind test data imported by the model show that the coincidence rate of BRANN casing collapse strength prediction can reach 96.67%. Through error analysis with API formula prediction results and KT formula prediction results improved by least square fitting, the BRANN-based casing collapse strength prediction has higher accuracy and stability. Compared with the traditional prediction method, this model can be used to predict casing strength under more complicated working conditions, and it has a certain guiding significance.

Crack identification system on MOH cold rolled grain oriented sheets: Application of K-fold cross validated BRANN

To increase efficiency of transformers, iron (hysteresis and Fuko losses), copper and abnormal losses must be prevented. Hysteresis loss may be significantly reduced by adding silicon to iron, and Fuko loss may be significantly reduced by using cold-rolled grain-oriented steel (CRGO) in the form of thin sheets with high resistance. To reduce abnormal losses, it is necessary not to use CRGO with ferromagnetic features including cracks in the core sequence. In line with this, currently, the magnetic flux leakage (MFL) method is commonly used to research cracks in ferromagnetic materials. Generally, literature studies researched the variation characteristics of MFL signals collected with magnetic sensors and size of amplitude values according to the physical properties (crack width, crack depth, etc.) of the crack in the ferromagnetic material. Different to the literature, our primary aim in this study is to determine this variation according to the type and geometric features of an artificial full crack with known physical properties created in MOH CRGO steel. The secondary aim of our study is to develop an artificial neural network predicting the type and geometric features of a full crack with unknown geometric features using the data obtained. In line with this, our study first produced an MFL measurement system. Then, artificial full crack samples were prepared with different types and geometric features using MOH CRGO steel. These artificial full crack samples were magnetized by being placed in the core of an electromagnet with 50 kHz AC signal and the surface of the material was scanned in a single dimension with a location-controlled fluxgate sensor. Harmonics of signals obtained from the fluxgate sensor were investigated with DSP lock-in amplifier and the amplitude values for the harmonics with greatest variation were uploaded to the computer. Finally, 3 different Bayesian regularized artificial neural networks (BRANN) were developed and trained using the harmonic amplitude values for full crack samples with known geometric features and used to predict the type and geometric features of full cracks with unknown geometric features. The first of these BRANN was used to identify the fixed or variable width of the full crack type, while the others were used to find the geometric features according to the full crack type. For the BRANNs trained with the K-fold cross-validation method, accuracy degrees of R = 0.99934, R = 0.99999 and R = 0.91654 were obtained, respectively. The trained BRANNs provided results compatible with reality for artificial full crack models with unknown crack type and geometric properties.

Yield loss assessment of grapes using composite drought index derived from landsat OLI and TIRS datasets

Drought is one of the most challenging natural phenomena and occurs due to inadequate and erratic precipitation (snowfall and rainfall) as well as the effects of high temperature on agronomical and horticultural crops. Among horticultural crops, grapes suffer extensively from inadequate berry formation due to drought. Therefore, developing a comprehensive drought monitoring tool with a drought assessment system that incorporates multiple agrometeorological variables into a single drought indicator is critical for supporting growers. In this context, the objective of this research was to assess drought severity using the composite drought index (CDI) and determine the variations in grape yields in drought-affected vineyards. In this study, the CDI was developed for the years 2016–2020 by comprising five indices: the vegetation condition index (VCI), temperature condition index (TCI), deviation of NDVI from the long-term mean (NDVI DEV), normalized difference moisture index (NDMI) and precipitation condition index (PCI). Moreover, the quantitative principal component analysis (PCA) approach was used to assign a weight for each input parameter, and then the weights of all the indices were combined into one composite drought index. Finally, Bayesian regularized artificial neural networks (BRANNs) were used to evaluate the yield variation in each affected vineyard. The results were validated with the standard precipitation index (SPI) from 2016 to 2020. Moderate to severe drought was detected with the CDI across Kabul Province during 2016 and 2018. The correlation coefficient between the CDI and SPI-1 in the active growth stages (April to October) was relatively high, at 0.64. A validation was performed with the correlating yield losses, and losses of 3.4 ton/ha and 4.7 ton/ha were found in 2016 and 2018, respectively, under severe drought conditions. The findings suggested that the CDI can be used for vineyard drought detections and agricultural organization support to more precisely aid farmers during severe drought conditions in grape-growing regions based on satellite remote sensing datasets.

A Hybrid Physics-Based and Stochastic Neural Network Model Structure for Diesel Engine Combustion Events

Estimation of combustion phasing and power production is essential to ensuring proper combustion and load control. However, archetypal control-oriented physics-based combustion models can become computationally expensive if highly accurate predictive capabilities are achieved. Artificial neural network (ANN) models, on the other hand, may provide superior predictive and computational capabilities. However, using classical ANNs for model-based prediction and control can be challenging, since their heuristic and deterministic black-box nature may make them intractable or create instabilities. In this paper, a hybridized modeling framework that leverages the advantages of both physics-based and stochastic neural network modeling approaches is utilized to capture CA50 (the timing when 50% of the fuel energy has been released) along with indicated mean effective pressure (IMEP). The performance of the hybridized framework is compared to a classical ANN and a physics-based-only framework in a stochastic environment. To ensure high robustness and low computational burden in the hybrid framework, the CA50 input parameters along with IMEP are captured with a Bayesian regularized ANN (BRANN) and then integrated into an overall physics-based 0D Wiebe model. The outputs of the hybridized CA50 and IMEP models are then successively fine-tuned with BRANN transfer learning models (TLMs). The study shows that in the presence of a Gaussian-distributed model uncertainty, the proposed hybridized model framework can achieve an RMSE of 1.3 × 10−5 CAD and 4.37 kPa with a 45.4 and 3.6 s total model runtime for CA50 and IMEP, respectively, for over 200 steady-state engine operating conditions. As such, this model framework may be a useful tool for real-time combustion control where in-cylinder feedback is limited.

Predicting heat release properties of flammable fiber-polymer laminates using artificial neural networks

Heat release rate is an important fire reaction property used to quantify the flammability of composite materials in fire. In this study, an artificial neural network (ANN) model was developed to predict the heat release properties of composites. The ANN model was trained using 10,419 data points for heat release rate extracted from the results of cone calorimetry tests performed on 14 sets of composite laminates. Two machine learning algorithms of Multiple Linear Regression (MLR) and Bayesian regularized artificial neural network with Gaussian prior (BRANNGP) are compared. The composites used to demonstrate the predictive accuracy of the ANN model were phenolic-based laminates containing different types and amounts of flame retardant additives. The BRANNGP model is capable of predicting the heat release rate-time curve, peak heat release value and total heat release of the composites. In addition, the BRANNGP model with outlier-elimination strategy can estimate with good accuracy the complex non-linear relationship between heat release rate and heat flux exposure time without considering the mechanistic interactions between the input and output parameters.