Exponential Gaussian Process Regression Research Articles

AI-driven battery ageing prediction with distributed renewable community and E-mobility energy sharing

Electrochemical battery storages play multiple functions in district energy systems, like peak shaving, load coverage, renewable penetration, load coverage, frequency regulation and short-term fast response. Battery cycling ageing will significantly affect the battery performance and the ignorance or inaccurate estimation on battery cycling ageing will lead to performance overestimation. However, current literature provides limited progress on these topics. In this study, a coastal district energy community is developed with district buildings, transportations, solar-wind systems, electrochemical battery storages, power grid and multi-directional power interactions. The ultimate target is to predict the dynamic state of health of various batteries with high accuracy and efficiency. In respect to frequent charging from intermittent renewables and discharging from stochastic demands and vehicle driving s, a series of cycling ageing models are developed, tested and compared, using mathematical fitting and machine learning approaches. Considering modelling complexity, time-consuming and product-dependent characteristic of mathematical models, machine learning-based regression learners are applied to firstly learn the underlying ageing mechanism, and then make accurate predictions. A top-down approach, following statistical analysis in training, cross-validation and prediction processes, is proposed for algorithm selection to accelerate the cycling ageing prediction. Results indicate that, during training and cross-validation processes, the final algorithm is identified as the exponential Gaussian Process Regression. Demonstration on battery cycling ageing verifies the feasibility and effectiveness in different battery categories and energy paradigms, so as to avoid techno-economic performance overestimation.

Modelling and optimizing microgrid systems with the utilization of real-time residential data: a case study for Palapye, Botswana

Microgrids are becoming a realistic choice for residential buildings due to the increasing need for affordable and sustainable energy solutions in developing nations. Through modeling and simulation, the main goal is to evaluate the viability and performance of a solar microgrid system. Residential load modeling is used, which is vital to developing an effective Energy Management System (EMS) for the microgrid. A residential household’s load metering data is examined using statistical methods, including time series and regression analysis. For the residential community load in this research, Proportional-Integral-Derivative (PID) controllers and Fuzzy Logic Controllers (FLC) are used to generate the necessary Direct Current (DC) microgrid voltage. The simulation research shows that FLC have benefits over PID controllers. The FLC technique performs better at reducing total harmonic distortion, which improves the microgrid system’s overall power quality. The Seasonal Autoregressive Integrated Moving Average (SARIMA) model was found to be the most appropriate and reliable model for the dataset after the performance of the models was evaluated using the metrics. The optimization results also showed that FLC optimization improves the microgrid system’s stability. The exponential Gaussian process regression (GPR) produced the highest R-squared measure of 0.49 and RSME measure of 7.9646, making it the best goodness fit for modeling the total daily energy usage and the peak daily usage.

Machine Learning Modeling of Wheel and Non-Wheel Path Longitudinal Cracking

Roads degrade over time due to various factors such as traffic loads, environmental conditions, and the quality of materials used. Significant investments have been poured into road construction globally, necessitating regular evaluations and the implementation of maintenance and rehabilitation (M&R) strategies to keep the infrastructure performing at a satisfactory level. The development and refinement of performance prediction models are essential for forecasting the condition of pavements, especially to address longitudinal cracking distress, a major issue in thick asphalt pavements. This research leverages multiple machine learning methods to create models predicting non-wheel path (NWP) and wheel path (WP) longitudinal cracking using data from the Long-Term Pavement Performance (LTPP) program. This study highlights the marked differences in distress conditions between WP and NWP, underscoring the importance of precise models that cater to their unique features. Aging trends for both types of cracking were identified through correlation analysis, showing an increase in WP cracking with age and a higher initial International Roughness Index (IRI) linked to NWP cracking. Factors such as material characteristics, kinematic viscosity, pavement thickness, air voids, particle size distribution, temperature, KESAL, and asphalt properties were found to significantly influence both WP and NWP cracking. The Exponential Gaussian Process Regression (GPR) emerged as the best model for NWP cracking, showcasing exceptional accuracy with the lowest RMSE of 89.11, MSE of 7940.72, and an impressive R-Squared of 0.63. For WP cracking, the Squared Exponential GPR model was most effective, with the lowest RMSE of 12.00, MSE of 143.93, and a high R-Squared of 0.62. The GPR models, with specific kernels for each cracking type, proved their adaptability and efficiency in various pavement scenarios. A comparative analysis highlighted the superiority of our new machine learning model, which achieved an R2 of 0.767, outperforming previous empirical models, demonstrating the strength and precision of our machine learning approach in predicting longitudinal cracking.

Combination of Machine Learning and RGB Sensors to Quantify and Classify Water Turbidity

Turbidity is one of the crucial parameters of water quality. Even though many commercial devices, low-cost sensors, and remote sensing data can efficiently quantify turbidity, they are not valid tools for the classification it. In this paper, we design, calibrate, and test a novel optical low-cost sensor for turbidity quantification and classification. The sensor is based on an RGB light source and a light detector. The analyzed samples are characterized by turbidity values from 0.02 to 60 NTUs, and have four different sources. These samples were generated to represent natural turbidity sources and leaves in the marine areas close to agricultural lands. The data are gathered using 64 different combinations of light, generating complex matrix data. Machine learning models are compared to analyze this data, including training, validation, and test datasets. Moreover, different alternatives for data preprocessing and feature selection are assessed. Concerning the quantification of turbidity, the best results were obtained using averaged data and principal components analyses in conjunction with exponential gaussian process regression, achieving an R2 of 0.979. Regarding the classification of the turbidity, an accuracy of 91.23% is obtained with the fine K-Nearest-Neighbor classifier. The cases in which data were misclassified are characterized by turbidity values lower than 5 NTUs. The obtained results represent an improvement over the current solutions in terms of turbidity quantification and a completely novel approach to turbidity classification.

Application of machine learning technique for dynamic analysis of confined geomaterial subjected to vibratory load

Computer programming-based numerical programs are firmly established in geotechnical engineering, with rapid growth of finite element modeling and machine learning techniques gaining much attention both in practice and academia. This study is intended to expedite the dissemination of advanced computer applications in terms of finite element simulation and machine learning models by investigating the dynamic response of geomaterials subjected to vibratory loads. Several trial models were built to perform the experimental investigations with a vibratory shaker, signal generator, several accelerometers, a data collection system, and other ancillary devices. The implicit integration techniques in commercialized software were adopted for numerical simulations. After data collection from numerical simulation, models were chosen, trained, and assessed to produce predictions that were then used in this study. Several technologies, including the ensemble boosted tree, squared exponential Gaussian Process Regression (GPR), Matern 5/2 GPR, exponential GPR, and decision tree architectures (fine and medium), were used to forecast the displacement of confined geomaterial. The displacement-depth ratio was found rising to 80% in the frequency range of 5 to 25 Hz, suggesting a considerable change in the behavior of the geomaterial. The Matern 5/2 GPR model showed better accuracy with an R2 value of 0.99, indicating an outstanding predictive ability. The Matern 5/2 GPR and boosted tree models could help better understand the links between displacement and its distribution along the direction of load application. The outcomes of this study based on computer-aided finite element programs can be effectively implemented in machine learning to develop computer programs. In conclusion, the computational machine learning models adopted in this study offer a new insight for uncovering hidden intrinsic laws and creating new knowledge for geotechnical researchers and practitioners.

Optimizing Solar Energy Harvesting: Supervised Machine Learning-Driven Peak Power Point Tracking for Diverse Weather Conditions

Solar Power is one of the significant prevalent forms of clean energy due to its perceived to be pollution-free and easily accessible. The market for renewable energy was established by the rapid development in electrical energy consumption and the diminution of conventional energy resources (CER). Under varying weather condition extracted energy from solar system is not constant and maximum. This study suggests the applicability of machine learning algorithm (MLA) in Peak power point tracking (P3T) methods to maximize power of a PV arrangement under varying weather conditions. Machine learning methods optimize peak power point tracking in solar photovoltaic systems by bringing agility, data-driven decision-making, and increased accuracy. MLAs improve the overall efficiency, stability, and dependability of these systems by handling the unpredictability of solar energy production under varying weather circumstances and PSCs Because MLAs are able to learn and adjust to non-linear relationships between solar intensity and PVS output. In this study, the squared multiple squared exponential Gaussian process regression method SGPRA tested in three rapidly varying ecological conditions. The performance of ML-P3T methods is validated using Matlab/Simulink, and the simulation outcome are compared with one of the most used algorithms, the variable step size incremental conductance algorithm (VINA). The Matlab/Simulink findings show that SGPRA operates significantly better under varying weather circumstances, harnessing more peak power efficiency 90%, shorter tracking time 0.13 sec, a mean error of 0.042, and superior stability.

The informative power of heart rate along with machine learning regression models to predict maximal oxygen consumption and maximal workload capacity

Prediction of maximal oxygen consumption (VO2max) and maximal workload capacity (MWC) through submaximal exercise tests is an important topic for sports sciences. Numerous studies highlighted the predictive power of submaximal heart rate (HR) and oxygen consumption (VO2) in predicting VO2max and MWC. The challenge is achieving the best possible precision and accuracy by identifying the best predictors and regression models. This project assessed the performance of different indexes along with machine learning regression models to estimate VO2max and MWC. Predictors consisted of biodata (age, weight, and height) along with different combinations of change-scores of HR and VO2 between 0–50 Watts, 50–65 Watts, and 65–80 Watts (Δ0–50, Δ50–65, and Δ65–80, respectively). The use of biodata + HR Δ50–65 + HR Δ65-80 via a Squared Exponential Gaussian Process Regression model resulted in the best performance in predicting VO2max, while the use of biodata + HR Δ0–50 via a Robust Linear Regression model resulted in the best performance in predicting MWC. These results suggest that information provided by HR only during submaximal exercise offers the best predictive mean for estimating VO2max and MWC, while the use of VO2 changes or its addition along with HR changes does not improve predictions. Moreover, different predictors need to be selected for the best estimation of VO2max and MWC. Change-scores refer to absolute value changes, providing information to develop athlete assessment protocols through standardized workloads. These results show practical applicability for sports assessments to be performed indirectly, rapidly, sub-maximally, and through the simple measurement of HR.

Enhancing sediment transport predictions through machine learning-based multi-scenario regression models

Machine learning is one effective way of increasing the accuracy of sediment transport models. Machine learning captures patterns in the sequence of both structured and unstructured data and uses it for forecasting. In this research, the different regression models were train to forecast sediment data using 8 years of measured sediment data collected in Sg. Linggui suspended sediment station. Data from different scenarios were used where each scenario indicates the number of lags. Seven regression models, namely, Linear Regression, Regression Trees, Support Vector Machines, Gaussian Process Regression, Kernel Approximation, Ensemble of Trees, and Neural Network were trained using the data and compared. The trained models were evaluated using Root Mean Square Error (RMSE) and Coefficient of Determination (R2). The best-performing models from two different types of regression models were chosen and they were tested using the test data to find the Relative Percentage Error (RPE) of the predicted data. The Exponential Gaussian Process Regression model performs much better than the other models in terms of RMSE and R2 values. When the exponential models from all 3 scenarios are compared, scenario 3 seems to have a better-performing model but only by a very small margin, after using testing data, the result shows scenario 3 has less RPE compared to other models. Hence, it can be deduced that the exponential gaussian process regression model from scenario 3 is the best-performing model overall in terms of RSME, R2, and RPE.

Effect of Nozzle Parameters on the Cooling Conditions in the Secondary Cooling Zone of a Slab Caster

The following contribution focuses on the secondary cooling zone of a slab caster, analyzing the effects of nozzle parameters on cooling conditions. The research group employs experimental measurements, including the Nozzle Measuring Stand (NMS), to determine local heat transfer coefficients (HTC) and water distribution. These measurements are crucial for accurately setting the boundary conditions in the simulation. The implementation of these boundary conditions is realized by an Exponential Gaussian Process Regression (EGPR) to predict HTC values. The results highlight the importance of an accurate HTC assessment in regions with overlapping sprays. The developed methodology aids in enhancing the precision of simulations for continuous casting processes, contributing to a better product quality and process efficiency using the non-commercial software platform “m2CAST”.

Modeling spinel oxide based-photocatalytic degradation of organic pollutants from industrial wastewater

Rapid population growth has resulted in rapid growth in industrialization to meet various human needs. As a result of this, huge volume of effluent is being generated from the industrial processes and released into the water bodies. These anthropogenic activities are often detrimental to human and aquatic lives. In this study, a modeling approach to evaluate the photocatalytic degradation of organic pollutants from industrial wastewater using spinel oxide is investigated. Four machine learning algorithms namely, linear regression, decision tree ensemble, medium Gaussian support vector machine, and exponential Gaussian process regression were employed. The parametric analysis of the predictors (particle size of the spinel oxides, the initial dye concentration, the amount of photocatalysts, the band gap, and the irradiation time) and the targeted output of the photocatalytic degradation efficiency shows that a non-linear relationship exists between the predictors and the targeted output. This was further confirmed by the linear regression model with R2 of 0.220. Besides, the decision tree ensemble and medium Gaussian support vector machine regression offer poor performances in predicting the photocatalytic degradation efficiency as indicated by R2 of 0.420 and 0.490, respectively. A superior performance in predicting the photocatalytic degradation efficiency was displayed by the exponential Gaussian process regression with R2 of 0.991.

Quantitative structure-retention relationship by databases of illegal additives

The importance of small molecule retention time in detecting illegal additives cannot be overstated. In order to provide a basis for identification in small-scale databases lacking corresponding chemical standards, a quantitative structure-retention relationship (QSRR) model was developed using two liquid chromatography databases under the same column but with different mobile phases. The QSRR model workflow includes feature extraction, feature selection, model construction and evaluation. Molecular descriptors were calculated by molecules optimized using Consistent force field (CFF) and Chemistry at HARvard Macromolecular Mechanicets (CHARMm) force field during the feature extraction stage. Two selection algorithms, minimum redundancy-maximum relevance (MRMR) and F-test, were compared during the feature selection stage. Five categories of machine learning algorithms, Regression Trees (Reg-T), Support Vector Machines (SVM), Gaussian Process Regression Models (GPR), Ensembles of Trees, and Kernel approximation models, were compared during the model construction phase resulting in 14 models being obtained. The best-performing model was found to be the Exponential Gaussian Process Regression Model (E-GPR). This was tested on Database 1 with coefficient of determination (R2)= 0.84 ± 0.02 and on Database 2 with R2 = 0.83 ± 0.03. The results indicate that this QSRR model can accurately predict retention times for small molecules within small-scale databases.

Regression Model-Based Short-Term Load Forecasting for Load Despatch Centre

Forecasting load is an integral part of the planning, operation, and control of power systems. This paper is part of a research effort aimed at developing better energy demand forecasting models for load dispatch centers (LDCs) in Indian states as part of an ambitious project utilizing artificial intelligence-based load forecasting models. In this paper, we present a half hourly load forecasting method for the energy management system of the project that will be used at 33 /11 kV and 0.415 kV substations with good accuracy. The paper uses the half-hourly load consumption dataset collected from MSEDCL for Maharashtra from July 1, 2020 through August 31, 2022. This paper evaluates 24 regression model-based half hourly based load forecasting algorithms for ALE PHATA load based on the load consumption dataset and the collected meteorological dataset. The 24 models in MATLAB Regression belong to five types of regression models: Linear Regression, Regression Trees, Support Vector Machines (SVM), Gaussian Process Regression (GPR), Ensemble of Trees, and Neural Networks. As a consequence of their nonparametric kernel-based probabilistic nature, the GPR family of models demonstrates the best load forecasting performance. Least squares estimation was used to determine the regression coefficients. There is a direct correlation between load in an electrical power system and temperature, due point, and seasons, as well as a correlation between load and previous load consumption. Therefore, the input variables are Wet Bulb Temperature at 2 Meters (C), Dew/Frost Point at 2 Meters (C), Temperature at 2 Meters (C), Relative Humidity at 2 Meters (%), Specific Humidity at 2 Meters (g/kg) and Wind Speed at 10 Meters (m/s). The mean absolute percentage error and the R squared are used to validate or verify the accuracy of the model, which is shown in the results section.  Based on this study, two GPR models are recommended for load forecasting, the Rational Quadratic GPR and the Exponential GPR and Exponential GPR as final model.

Proxy quality control of biomass particles using thermogravimetric analysis and Gaussian process regression models

AbstractThe temperature experienced by reactants during preparation in a reactor is a key component in determining the yield and homogeneity of usable chemical products such as biomass particles. Thermocouples with sensors can be used to monitor spatial temperature gradients within reactors but these sensors are often too expensive and/or invasive. The present work proposes a strategy to identify optimal machine learning models to infer the maximum effective temperature experienced by particles during oxidative biomass torrefaction using key thermochemical combustion parameters. The maximum rate of weight loss, the corresponding temperature, and fixed carbon content on a dry‐ash‐free basis are used as literature‐based predictor variables obtained from thermogravimetric analysis. The evaluation of 24 machine‐learning models using the standard tenfold cross‐validation method suggests that the exponential Gaussian process regression (GPR) model is the most effective, followed by other GPR models. These high‐performing GPR models were also utilized to predict the effective preparation temperature distribution of reactor‐produced biomass particles under eight conditions of varying residence time and air‐to‐biomass ratio. The effective preparation temperature and residence time of individual biomass particles were then encoded into the torrefaction severity factor and used to estimate the energy yield of the reactor output as a novel quality control method. © 2023 The Authors. Biofuels, Bioproducts and Biorefining published by Society of Industrial Chemistry and John Wiley & Sons Ltd.

Comparison of the Performance of the Regression Models in GPS-Total Electron Content Prediction

The ionosphere is an important layer that provides radio communication in the upper atmosphere. The ionosphere is located between 50 km and 1000 km above the atmosphere. Electron density, which is the most important parameter of the ionosphere, changes depending on location, time, seasons, altitude, solar, geomagnetic and seismic activity. A significant measurable amount of electron density is Total Electron Content (TEC), which is used to probe the structure of the ionosphere and upper atmosphere. The Global Positioning System (GPS), which has a low cost and widespread receiver network is prominent used in TEC estimation. The IONOLAB-TEC data estimated from GPS is used in this study. Prediction of TEC is important phenomenon to operate and to plan the Earth-space and satellite-to-satellite communication systems, to generate the earthquake precursor signals using TEC and to detect of anomalies in the ionosphere. In this study, IONOLAB-TEC data obtained from GPS is estimated using regression models. Among the tested algorithms, it is observed that the Exponential Gaussian Process Regression and Interactions Linear Regression algorithms are very successful and high-performance models for TEC estimation.

Feasibility analysis for predicting the compressive and tensile strength of concrete using machine learning algorithms

Concrete is the most utilized material (i.e., average production of 2 billion tons per year) for the construction of buildings, bridges, roads, dams, and several other important infrastructures. The strength and durability of these structures largely depend on the compressive strength of the concrete. The compressive strength of concrete depends on the proportionality of the key constituents (i.e., fine aggregate, coarse aggregate, cement, and water). However, the optimization of the constituent proportions (i.e., matrix design) to achieve high-strength concrete is a challenging task. Furthermore, it is essential to reduce the carbon footprint of the cementitious composites through the optimization of the matrix. In this research, machine learning algorithms including regression models, tree regression models, support vector regression (SVR), ensemble regression (ER), and gaussian process regression (GPR) were utilized to predict the compressive and tensile concrete strength. Also, the model performance was characterized based on the number of input variables utilized. The dataset used in this research was compiled from journal publications. The results showed that the exponential GPR had the highest performance and accuracy. The model had an impressive performance during the training phase, with a R2 of 0.98, RMSE of 2.412 MPa, and MAE of 1.6249 MPa when using 8 input variables to predict the compressive strength of concrete. In the testing phase, the model maintained its accuracy with a R2 of 0.99, RMSE of 0.0025134 MPa, and MAE of 0.0016367 MPa. In the training and testing phases, the exponential GPR also demonstrated high accuracy in predicting the tensile strength with an R2, RMSE, and MAE of 0.99, 0.00049247 MPa, and 0.00036929 MPa, respectively. Furthermore, in the prediction of tensile strength the number of variables utilized had an insignificant effect on the performance of the models. However, in predicting the compressive strength, an increase in the number of input variables lead to an enhancement in the performance metrics. The results of this research can allow for the quick and accurate prediction of the strength of a given concrete mixture design.

Blood glucose estimation based on ECG signal.

Successful self-management of diabetes requires Continuous Glucose Monitors (CGMs). These CGMs have several limitations such as being invasive, expensive and limited in terms of use. Many techniques, in vain, have been proposed to overcome these limitations. Nowadays, with the help of the Internet of Medical Things (IoMT) technologies, researchers are working to find alternative solutions. They succeed to predict hypoglycemia and hyperglycemia peaks using Electrocardiogram (ECG) signals. However, they failed to use it to estimate the Blood Glucose Concentration (BGC) directly and in real time. Three patients with 08days of measurements from the D1namo dataset contributed to the study. A new technique has been proposed to estimate the BGC curves based on ECG signals. We used a convolutional neural network to segment the different regions of ECG signals as well as we extracted ECG features that were required for the next step. Then, five regression models have been employed to estimate BGC using as input sixth ECG parameters. We were able to segment the ECG signals with an accuracy of 94% using the convolutional neural network algorithm. The best performance among all simulated models was provided by Exponential Gaussian Process Regression (GPR) with Root Mean Squared Error (RMSE) values of 0.32, 0.41, 0.67 and R-squared (R2) values of 98%, 80%, and 70% for patients 01, 02 and 03 respectively. The method indicates the potential use of ECG wearable devices as non-invasive for continuous blood glucose monitoring, which is affordable and durable.

Prediction of monthly dry days with machine learning algorithms: a case study in Northern Bangladesh

Dry days at varied scale are an important topic in climate discussions. Prolonged dry days define a dry period. Dry days with a specific rainfall threshold may visualize a climate scenario of a locality. The variation of monthly dry days from station to station could be correlated with several climatic factors. This study suggests a novel approach for predicting monthly dry days (MDD) of six target stations using different machine learning (ML) algorithms in Bangladesh. Several rainfall thresholds were used to prepare the datasets of monthly dry days (MDD) and monthly wet days (MWD). A group of ML algorithms, like Bagged Trees (BT), Exponential Gaussian Process Regression (EGPR), Matern Gaussian Process Regression (MGPR), Linear Support Vector Machine (LSVM), Fine Trees (FT) and Linear Regression (LR) were evaluated on building a competitive prediction model of MDD. In validation of the study, EGPR-based models were able to better capture the monthly dry days (MDD) over Bangladesh compared to those by MGPR, LSVM, BT, LR and FT-based models. When MDD were the predictors for all six target stations, EGPR produced highest mean R2 of 0.91 (min. 0.89 and max. 0.92) with a least mean RMSE of 2.14 (min. 1.78 and max. 2.69) compared to other models. An explicit evaluation of the ML algorithms using one-year lead time approach demonstrated that BT and EGPR were the most result-oriented algorithms (R2 = 0.78 for both models). However, having a least RMSE, EGPR was chosen as the best model in one year lead time. The dataset of monthly dry–wet days was the best predictor in the lead-time approach. In addition, sensitivity analysis demonstrated sensitivity of each station on the prediction of MDD of target stations. Monte Carlo simulation was introduced to assess the robustness of the developed models. EGPR model declared its robustness up to certain limit of randomness on the testing data. The output of this study can be referred to the agricultural sector to mitigate the impacts of dry spells on agriculture.

Analysis of the exhaust hydrogen characteristics of high-compression ratio, ultra-lean, hydrogen spark-ignition engine using advanced regression algorithms

• Exhaust H 2 and NO x were investigated using heavy-duty, hydrogen spark-ignition engine. • Lower excess air ratio exhibited lower H 2 emission, but greater heat transfer loss. • The regression model achieved a high accuracy under various operating conditions. • The proposed H 2 emission model is applicable to various HICEs. Hydrogen is a leading alternative fuel for eliminating the carbon emissions of internal combustion engines. Moreover, the use of hydrogen improves the combustion owing to its high flame speed. The hydrogen selective catalytic reduction is a promising solution for the reduction of nitric oxides (NOx), which is the sole regulated gas species in hydrogen-fueled internal combustion engines. In the present study, hydrogen and NOx emissions, the crucial species for the catalyst performance, were investigated using a heavy-duty, hydrogen spark-ignition engine. The ratio of H 2 to NOx was above 100 at the excess air ratio (lambda) of 2.5 or higher. Three-dimensional, numerical simulation showed that the in-cylinder hydrogen distribution was dependent on the engine load and the lambda value. A regression analysis showed that the exhaust hydrogen quantity had a strong correlation with the seven parameters, namely engine speed, the gross indicated effective pressure (IMEPg), lambda, spark timing, total combustion duration, peak in-cylinder temperature, and peak pressure rise rate. Among these parameters, the IMEPg and the lambda value exhibited the highest weights in a neighborhood component analysis. The regression model with the seven features exhibited the highest R 2 value of 0.94 with the squared exponential Gaussian process regression. The proposed prediction model can contribute to not only reducing NOx emission with aftertreatment, but also maximizing the thermal efficiency of hydrogen-fueled internal combustion engines.

Performance analysis and modeling of bio-hydrogen recovery from agro-industrial wastewater

Significant volumes of wastewater are routinely generated during agro-industry processing, amounting to millions of tonnes annually. In line with the circular economy concept, there could be a possibility of simultaneously treating the wastewater and recovering bio-energy resources such as bio-hydrogen. This study aimed to model the effect of different process parameters that could influence wastewater treatment and bio-energy recovery from agro-industrial wastewaters. Three agro-industrial wastewaters from dairy, chicken processing, and palm oil mills were investigated. Eight data-driven machine learning algorithms namely linear support vector machine (LSVM), quadratic support vector machine (QSVM), cubic support vector machine (CSVM), fine Gaussian support vector machine (FGSVM), binary neural network (BNN), rotation quadratic Gaussian process regression (RQGPR), exponential quadratic Gaussian process regression (EQGPR) and exponential Gaussian process regression (EGPR) were employed for the modeling process. The datasets obtained from the three agro-industrial processes were employed to train and test the models. The LSVM, QSVM, and CSVM did not show an impressive performance as indicated by the coefficient of determination (R2) < 0.7 for the prediction of hydrogen produced from wastewaters using the three agro-industrial processes. The LSVM, QSVM, and CSVM models were also characterized by high prediction errors. Superior performance was displayed by FGSVM, BNN, RQGPR, EQGPR, and EQGPR models as indicated by the high R2 > 0.9, an indication of better predictability with minimized prediction errors as indicated by the low root mean square error (RMSE), mean square error (MSE), and mean absolute error (MAE).

Estimation of carotid-femoral pulse wave velocity from finger photoplethysmography signal

Objective. This project compared a new method to estimate the carotid-femoral pulse wave velocity (cf-PWV) to the gold-standard cf-PWV technique. Approach. The cf-PWV was estimated from the pulse transit time (FPS-PTT) calculated by processing the finger photoplethysmographic signal of Finapres (FPS) and subject’s height only (brief mode) as well as along with other variables (age, heart rate, arterial pressure, weight; complete mode). Doppler ultrasound cf-PWVs and FPS-PTTs were measured in 90 participants equally divided into 3 groups (18–30; 31–59; 60–79 years). Predictions were performed using multiple linear regressions (MLR) and with the best regression model identified by using MATLAB Regression Learner App. A validation set approach (60 training datasets, 30 testing datasets; VSA) and leave-one-out cross-validation (LOOCV) were used. Main results. With MLR, the discrepancies were: 0.01 ± 1.21 m s−1 (VSA) and 0.001 ± 1.11 m s−1 (LOOCV) in brief mode; −0.02 ± 0.83 m s−1 (VSA) and 0.001 ± 0.84 m s−1 (LOOCV) in complete mode. Using a linear support vector machine model (SVM) in brief mode, the discrepancies were: 0.01 ± 1.19 m s−1 (VSA) and −0.01 ± 1.06 m s−1 (LOOCV). Using an Exponential Gaussian process regression model (GPR) in complete mode, the discrepancies were: −0.03 ± 0.79 m s−1 (VSA) and 0.01 ± 0.75 m s−1 (LOOCV). Significance. The cf-PWV can be estimated by processing the FPS-PTT and subjects’ height only, but the inclusion of other variables improves the prediction performance. Predictions through MLR qualify as acceptable in both brief and complete modes. Predictions via linear SVM in brief mode improve but still qualify as acceptable. Interestingly, predictions through Exponential GPR in complete mode improve and qualify as excellent.