Estimating hydrogen absorption energy on different metal hydrides using Gaussian process regression approach

Scientific Reports volume 12, Article number: 21902 (2022) Cite this article

1374 Accesses

7 Altmetric

Metrics details

Hydrogen is a promising alternative energy source due to its significantly high energy density. Also, hydrogen can be transformed into electricity in energy systems such as fuel cells. The transition toward hydrogen-consuming applications requires a hydrogen storage method that comes with pack hydrogen with high density. Among diverse methods, absorbing hydrogen on host metal is applicable at room temperature and pressure, which does not provide any safety concerns. In this regard, AB2 metal hydride with potentially high hydrogen density is selected as an appropriate host. Machine learning techniques have been applied to establish a relationship on the effect of the chemical composition of these hosts on hydrogen storage. For this purpose, a data bank of 314 data point pairs was used. In this assessment, the different A-site and B-site elements were used as the input variables, while the hydrogen absorption energy resulted in the output. A robust Gaussian process regression (GPR) approach with four kernel functions is proposed to predict the hydrogen absorption energy based on the inputs. All the GPR models' performance was quite excellent; notably, GPR with Exponential kernel function showed the highest preciseness with R2, MRE, MSE, RMSE, and STD of 0.969, 2.291%, 3.909, 2.501, and 1.878, respectively. Additionally, the sensitivity of analysis indicated that ZR, Ti, and Cr are the most demining elements in this system.

Energy demand has increased exponentially in recent years, reaching over 18 TW. In the subsequent years, this market growth is expected to continue1. Nowadays, fossil fuels account for over 80% of global energy consumption2,3,4,5. However, due to the environmental issues, the transition to renewable energy sources is critical6. In this regard, hydrogen can revolutionize renewable energy systems as a fuel and a clean energy carrier. It could be the basis for establishing carbon-free fuels7,8. Hydrogen energy has been among the most popular energy sources in recent years. This is since it has a higher energy content and causes fewer environmental issues than fossil fuels9. Hydrogen has a far higher energy density of 142 Mj kg−1 than fossil fuels, with a density of 47 Mj kg−110. It is estimated that about 35% of European vehicles will be hydrogen-powered by 20409. In addition, hydrogen energy will provide around 34% of the world's energy demands by 205011. Even though hydrogen is a prevalent element in nature, it is rarely found in pure form. As a result, several chemicals, electrochemical, photoelectrochemical, thermal, and microbiological approaches have been developed for producing it12,13,14. More than 50 million tons of hydrogen are produced annually in the world15.

Hydrogen may be stored in three primary ways, including gas, liquid, and solid-phase storage. Solid-phase storage is one of the most promising storage technologies owing to its ability to operate at room temperature and atmospheric pressure, as well as its excellent safety and low energy loss16,17,18,19,20,21. Metal hydrides have been noticed as a hydrogen storage material in solid-state conditions22,23,24,25,26,27 and are produced by absorption of hydrogen molecules on a metallic/intermetallic host28. The gravimetric density of hydrogen absorbed in these compounds is about 1–3 wt%5,29. Different metal hydrides have been identified and examined so far, including AB, AB2, AB3, AB5, and A2B, in which A and B are two types of metals or a group of metals. The AB2 metal hydride is the most promising type for hydrogen storage due to its easy activation, fast kinetics, and favorable pressure conditions30. In AB2 alloys, element A contains hydride constituent of elements such as Ti, Zr, Ta, and Hf, while element B contains transition metals such as Fe, Co, Ni, Mn, Cr, and V31,32. The C14 and C15 with a hexagonal and face-center-cubic structure, respectively, are the laves phases of the AB2 metal hydrides33.

Thus, the element selection for A-site and B-site of AB2 compounds influences their hydrogen storage performance. In order to investigate the effect of different elements or dopants on the hydrogen storage properties, traditional approaches, such as basic laws, computational modeling, and experimental investigations, are costly, time-consuming, and associated with numerous trials and errors, making them challenging and inefficient. Thus, to save time, energy, and cost, mainly when a complicated nonlinear relationship exists between the parameters and the performance, alternative machine learning (ML) techniques could be effective assessment methods.

The ML has become a prominent field of research and approach in developing and selecting advanced energy materials in recent years34,35,36,37,38,39,40. So far, various machine learning algorithms have employed hydrogen storage by metal hydride systems. For example, Griffin and Darsey estimated entropy, enthalpy, the temperature at 1 atm, pressure at 25 C, and the weight percent of hydrogen stored in metal hydrides using artificial neural networks. For the above parameters, the average correlation coefficient of R2 was 0.8888, 0.9561, 0.9381, 0.9935, and 0.9569, respectively41. To estimate the hydrogen storage capacity in metal hydrides, Rahnama et al. utilized four models: linear regression, neural network, Bayesian linear regression, and boosted decision tree. The R2 of the utilized models were 0.50, 0.60, 0.56, and 0.83, respectively, indicating that the boosted decision tree performed better than the other models42. In another study, Rahnama et al. classified metal hydrides using four classifiers: multiclass logistic regression, multiclass decision forest, multiclass decision jungle, and multiclass neural network. The accuracy of the used models was 0.47, 0.60, 0.62, and 0.80, respectively, indicating that the multiclass neural network classifier performed better than the other classifiers. This classification was based on the properties of metal hydrides, including the weight percentage of hydrogen, heat of formation, and operating temperature and pressure43. Suwarno et al. used their research to use multivariate regression, decision tree, and random forest models. The heat of formation, phase abundance, and hydrogen storage capacity of AB2 metal hydrides were all estimated using these models. The random forest model showed the most outstanding performance among the three models, with an average R2 value of 0.72244. Determining the pressure-composition-temperature (PCT) curve is an important issue in metal hydrides. This issue was considered in the research of Kim et al., where random forest (RF), K-nearest neighbor (KNN), and deep neural network (DNN) models were used. The deep neural network (DNN) model exhibited the greatest performance among the three models, with an average correlation coefficient R2 of 0.930745.

In the present study, for the first time, the Gaussian process regression (GPR) model with four kernel functions was used to estimate the energy of hydrogen absorption (ΔH) on the surface of the hydride alloys. The elements of A and B in AB2 compounds were chosen as input variables to establish a relationship between the chemical composition of AB2 and hydrogen storage properties. For this purpose, a substantial experimental data bank was applied. The developed model was evaluated by several error and statistical parameters. Also, sensitivity analysis was performed to find the most determining elements in the hydrogen storage on metal hydrides.

A set of 314 pairs of AB2 alloys were collected and presented in the Supplementary Information from the literature44. They include the information of constituent elements and ΔH absorption (in KJ/(molH2)).

It is worth mentioning that, in the pressure-composition-temperature diagram, some of the ΔH of these alloy couples are tacitly explained but are not clearly stated in the publications. The van't Hoff Law, as shown in the following equation, was used to calculate the aforementioned ΔH.

In order to determine the equilibrium pressure, the computation was done by choosing a midpoint from the plateau of the pressure-composition graph. The temperature value in the pressure-composition phase diagram is constant because R, the universal gas constant, is used in the calculation. The term S is assumed to have a constant value of − 110 kJ/(mol H2 K).

As depicted in Fig. 1, 22 alloying elements of Si, Mo, Fe, C, Ni, Co, Zr, La, Cu, Gd, Al, Mn, Ti, Ce, W, B, Mg, V, Ho, Cr, Sn, and Nb are the input parameters while the ΔH is the output of the model to see the effect of each parameter on the hydrogen storage conditions. In this work, 70% of data was separated coincidentally as training data to develop the model, and the rest (30% data) was used as testing data for prediction to evaluate the model's accuracy. Several statistical factors were calculated to quantify the established model preciseness, including R2, standard deviation (STD), mean-square error (MSE), mean relative error (MRE), and root-mean-square error (RMSE). Consiering y and x as the predicted and experimental values respectively, these factors are defined as follows:

Steps and analysis parameters through the study.

In comparison to support vector machines and artificial neural networks, Gaussian process regression with its super-parameters, which can be adaptively attained, is easy to perform. Also, the confidence interval (i.e., the uncertainty of the model prediction) can be obtained by this method35,46.

In GPR modeling \(L = \left\{ {x_{L \cdot i} \cdot y_{L \cdot i} } \right\}_{i = 1}^{{n_{1} }}\) and \(T = \left\{ {x_{T \cdot i} \cdot y_{T \cdot i} } \right\}_{i = 1}^{{n_{2} }}\) are arbitrarily selected training and testing data sets with input and output parameters of x and y, respectively. The modeling begins by:

where \(\varepsilon\), σ2noise, and In are the observation noise, the variance of the noise, and the unit array. Similar to the traing data, we have for the test data:

In GPR method, f(x) is a random function which defined by its corresponding covariance k(x, x′) (also called kernel) and mean m(x) functions.

Although m(x) can be obtained by applying explicit basis functions, for simplicity, it is usually supposed to zero47.

From Eqs. (7) and (11) the y is achieved as:

Now, based on the introduced parameters:

By summation of these two equations, the Gaussian expression is derived:

To obtain the yT distribution, the conditioning rule of Gaussians can be used:

With ΣT and μT as the covariance and the mean value, respectively. The core of the GPR is the kernel function which generates a covariance matrix to calculate the "distance" between two data points. Thus, various kernel functions have different calculation approaches, affecting the strength and the robustness of the final GPR model48. In the present study, four kernel functions of Matern, Rational quadratic, Exponential, and Squared exponential are chosen to find the most appropriate one, defined as follows

Matern kernel function:

Rational quadratic kernel function:

Exponential kernel function:

Squared Exponential kernel function:

In these equations, ℓ, σ, σ2, and α > 0 indicate the length scale, the amplitude, the variance, and scale-mixture, respectively. Also, v, Kv, and Γ represent a positive parameter, the modified Bessel function, and the gamma function, respectively.

In the present study, we developed GPR models based on four kernel functions in MATLAB software version 2018 and compared their capabilities to estimate enthalpy of absorptions.

Due to the existing errors in experiments or calculation methods, some of the collected data behave differently from other data points, known as suspected data or outliers. Having these data in the data bank leads to improper anticipation for the established models. Accordingly, the presence of the suspected data in the data bank should be investigated to advance the quality of the collected data bank. For this purpose, the Leverage method is used, which defines the Hat matrix and critical leverage limit as follows:

where U is a matrix with the i*j dimension, and i and j are the number of parameters and the training data, respectively. To assess the quality of the collected data bank, William's plot concept is used, through which standardized residuals are portrayed versus hat values. According to the method, the data from the reliable zone, the confined area between standardized residuals of [− 3, 3] and [0, critical leverage limit], are suspected data. In the present work, as shown in Fig. 2, most data is in the reliable area. In detail, for all the developed GPR models, only 14 or 13 data points out of 314 data (about 4%) are out of the reliable zone, confirming the collected data set is appropriate for training and testing.

Detection of suspected data for GPR model with kernel function of (a) Exponential, (b) Matern, (c) Squared exponential, (d) Rational quadratic.

In order to determine the effect of each element on the absorption enthalpy, an analysis of sensitivity is implemented. The relevancy factor, the metric which implies how much a parameter is effective, is derived from the following expression:

where \(X_{k.i}\) and \(Y_{i}\) represent the 'k' th input and 'i' th output, while the average values of input and outputs are denoted by \({\overline{X} }_{k}\) and \(\overline{Y }\), respectively. The input parameter with a larger r means a greater effect on the outcome. The positive sign indicates the parameter affects the output positively and vice versa for negative signs. According to the sensitivity analysis (Fig. 3), Ti and Zr are the most effective elements in the ΔH absorption of hydrogen, with the relevancy factor of − 38.47% and 38.38%, respectively. The opposite sign of these elements is because of their interchange in A site. In other words, when Ti increases, the amount of Zr automatically decreases and vice versa. This result was expected because, as discussed, element A (here are Ti and Zr) in AB2 structures is the hydride forming element, significantly affecting the hydrogen adsorption energy of the alloy49. Among the rest of the metals placed in the B site of the AB2 structures, Cr and V are the most influential input elements, while C and Co have the minimum effect on the alloys-hydrogen absorption ΔH. It can be related to their abundance in the collected dataset. Indeed, Cr, Mn, and V have been extensively applied in this research area, while other metals were often used as dopants/modifiers.

Sensitivity analysis of the input variables for ΔH absorption of hydrogen on metal hydrides.

The statistical parameters and the graphical comparison figures are presented to evaluate the developed model performance in the hydrogen absorption ΔH prediction. The statistical parameters are calculated and listed in 1 for the train, test, and overall dataset. In the training phase, the R2 values of 0.976, 0.976, 0.95, and 0.966 were obtained for established GPR-Exponential, GPR-Matern, GPR-Squared Exponential, and GPR-Rational Quadratic models, respectively. Considering their low amount of MRE, MSE, RMSE, and STD, especially for the GPR-Exponential model, confirms that all the GPR models were trained with enough preciseness. They were used to predict new (testing) data to examine the robustness of the models. Based on Table 1, all the developed models showed their acceptable capability in the ΔH prediction. The GPR-Exponential is slightly more accurate among all models with R2 = 0.969, MRE = 2.291%, MSE = 3.909, RMSE = 2.501, and STD = 1.878.

The simultaneous comparison between the experimental and anticipated amounts of hydrogen absorption ΔH for all the models is illustrated in Fig. 4. It is clear that all the proposed GPR models are predicted in excellent agreement with the actual values of ΔH through which the prediction lines cover the data points accurately.

Comparison of actual and anticipated data for GPR model containing kernel function of (a) Exponential, (b) Matern, (c) Squared exponential, (d) Rational quadratic.

The cross plots for all the GPR models are depicted in Fig. 5. In these graphs, the bisector line of the first quarter is the accuracy merit; the closer data to this line, the more precise model is developed. As shown in Fig. 5, the data are placed very close to the bisector line, and their respective linear fitting equations are almost the same as the merit line (slope of unit with R2 greater than 0.9). Thus, the established GPR models could anticipate the ΔH very well.

Cross plots for GPR model with kernel function of (a) Exponential), (b) Matern, (c) Squared exponential, (d) Rational quadratic.

For more assessment of the GPR models' results, the relative deviation between the actual hydrogen absorption and the predicted ones is calculated and illustrated in Fig. 6. In each of the developed GPR models, most of the calculated absolute deviation data points are smaller than 10%. Also, the GPR model with Exponential kernel function has the minimum mean relative error of 2.291% compared to GPR-Matern (2.957%), GPR-Square exponential (4.416%), and GPR-Rational quadratic (3.929%).

Comparison of actual and anticipated data for GPR model containing kernel function of (a) Exponential, (b) Matern, (c) Squared exponential, (d) Rational quadratic.

In order to anticipate the hydrogen absorption ΔH on the AB2 alloys, a machine learning approach of Gaussian process regression (GPR) with four different kernel functions (Exponential, Matern, Squared exponential, and Rational quadratic) was assessed. The 22 different alloying elements were used as the input. All the developed GPR models performed very well. Among them is the GPR-Exponential model with a little more excellence than others, with R2, MRE, MSE, RMSE, and STD of 0.969, 2.291%, and 3.909, 2.501, and 1.878, respectively, chosen as the best one. According to the sensitivity analysis, the Ti and Zr elements, along with V and Cr, contribute the most to the change of hydrogen absorption ΔH. The results of the presented work could provide the researchers and scientists with a perspective to choose the appropriate elements for AB2 alloys for hydrogen storage.

All data generated or analysed during this study are included in this published article [and its supplementary information files].

Chen, B. et al. Exploring the possible effect of anthropogenic heat release due to global energy consumption upon global climate: A climate model study. Int. J. Climatol. 36, 4790–4796 (2016).

Article Google Scholar

Dominković, D. F., Bačeković, I., Pedersen, A. S. & Krajačić, G. The future of transportation in sustainable energy systems: Opportunities and barriers in a clean energy transition. Renew. Sustain. Energy Rev. 82, 1823–1838 (2018).

Article Google Scholar

Goldemberg, J. The promise of clean energy. Energy Policy 34, 2185–2190 (2006).

Article Google Scholar

Abe, J. O., Popoola, A. P. I., Ajenifuja, E. & Popoola, O. M. Hydrogen energy, economy and storage: Review and recommendation. Int. J. Hydrogen Energy 44, 15072–15086 (2019).

Article CAS Google Scholar

Puszkiel, J. et al. Designing an AB2-type alloy (TIZr-CrMNMO) for the hybrid hydrogen storage concept. Energies 13, 2751 (2020).

Article CAS Google Scholar

Dincer, I. & Acar, C. A review on clean energy solutions for better sustainability. Int. J. Energy Res. 39, 585–606 (2015).

Article Google Scholar

Brandon, N. P. & Kurban, Z. Clean energy and the hydrogen economy. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 375, 20160400 (2017).

Article ADS Google Scholar

Goltsov, V. A. & Veziroglu, T. N. From hydrogen economy to hydrogen civilization. Int. J. Hydrogen Energy 26, 909–915 (2001).

Article CAS Google Scholar

Tarhan, C. & Çil, M. A. A study on hydrogen, the clean energy of the future: Hydrogen storage methods. J. Energy Storage 40, 102676 (2021).

Article Google Scholar

Schlapbach, L. & Züttel, A. Hydrogen-storage materials for mobile applications. In Materials for Sustainable Energy: A Collection of Peer-Reviewed Research and Review Articles From Nature Publishing Group 265–270 (World Scientific, 2011).

Rohland, B., Nitsch, J. & Wendt, H. Hydrogen and fuel cells—The clean energy system. J. Power Sour. 37, 271–277 (1992).

Article ADS CAS Google Scholar

Dawood, F., Anda, M. & Shafiullah, G. M. Hydrogen production for energy: An overview. Int. J. Hydrogen Energy 45, 3847–3869 (2020).

Article CAS Google Scholar

Abdin, Z. et al. Hydrogen as an energy vector. Renew. Sustain. Energy Rev. 120, 109620 (2020).

Article CAS Google Scholar

Hosseini, S. E. & Wahid, M. A. Hydrogen from solar energy, a clean energy carrier from a sustainable source of energy. Int. J. Energy Res. 44, 4110–4131 (2020).

Article Google Scholar

Zhang, J. Z., Li, J., Li, Y. & Zhao, Y. Hydrogen Generation, Storage and Utilization (John Wiley & Sons, 2014).

Book Google Scholar

Marques, F., Balcerzak, M., Winkelmann, F., Zepon, G. & Felderhoff, M. Review and outlook on high-entropy alloys for hydrogen storage. Energy Environ. Sci. 14, 5191–5227 (2021).

Article CAS Google Scholar

Hirscher, M. et al. Materials for hydrogen-based energy storage–past, recent progress and future outlook. J. Alloys Compd. 827, 153548 (2020).

Article CAS Google Scholar

Barthélémy, H., Weber, M. & Barbier, F. Hydrogen storage: Recent improvements and industrial perspectives. Int. J. Hydrogen Energy 42, 7254–7262 (2017).

Article Google Scholar

Niaz, S., Manzoor, T. & Pandith, A. H. Hydrogen storage: Materials, methods and perspectives. Renew. Sustain. Energy Rev. 50, 457–469 (2015).

Article CAS Google Scholar

Graetz, J. New approaches to hydrogen storage. Chem. Soc. Rev. 38, 73–82 (2009).

Article CAS Google Scholar

Züttel, A. Hydrogen storage methods. Naturwissenschaften 91, 157–172 (2004).

Article ADS Google Scholar

Chen, Z. et al. Perspectives and challenges of hydrogen storage in solid-state hydrides. Chin. J. Chem. Eng. 29, 1–12 (2021).

Article ADS CAS Google Scholar

Tarasov, B. P. et al. Metal hydride hydrogen storage and compression systems for energy storage technologies. Int. J. Hydrogen Energy 46, 13647–13657 (2021).

Article CAS Google Scholar

Davids, M. W. et al. Development of a portable polymer electrolyte membrane fuel cell system using metal hydride as the hydrogen storage medium. ECS Trans. 75, 553 (2016).

Article CAS Google Scholar

Davids, M. W. et al. Metal hydride hydrogen storage tank for light fuel cell vehicle. Int. J. Hydrogen Energy 44, 29263–29272 (2019).

Article CAS Google Scholar

Lototskyy, M. et al. Metal hydride hydrogen storage tank for fuel cell utility vehicles. Int. J. Hydrogen Energy 45, 7958–7967 (2020).

Article CAS Google Scholar

Pickering, L., Lototskyy, M. V., Davids, M. W., Sita, C. & Linkov, V. Induction melted AB2-type metal hydrides for hydrogen storage and compression applications. Mater. Today Proc. 5, 10470–10478 (2018).

Article CAS Google Scholar

Afzal, M., Mane, R. & Sharma, P. Heat transfer techniques in metal hydride hydrogen storage: A review. Int. J. Hydrogen Energy 42, 30661–30682 (2017).

Article CAS Google Scholar

Rusman, N. A. A. & Dahari, M. A review on the current progress of metal hydrides material for solid-state hydrogen storage applications. Int. J. Hydrogen Energy 41, 12108–12126 (2016).

Article CAS Google Scholar

Manickam, K., Grant, D. M. & Walker, G. S. Optimization of AB2 type alloy composition with superior hydrogen storage properties for stationary applications. Int. J. Hydrogen Energy 40, 16288–16296 (2015).

Article CAS Google Scholar

Hariyadi, A. et al. Modeling of the hydrogen sorption kinetics in an AB2 laves type metal hydride alloy. J. Alloys Compd. 893, 162135 (2022).

Article CAS Google Scholar

Ivey, D. G. & Northwood, D. O. Storing hydrogen in AB2 Laves-type compounds. Z. Phys. Chem. 147, 191–209 (1986).

Article CAS Google Scholar

Stein, F. & Leineweber, A. Laves phases: A review of their functional and structural applications and an improved fundamental understanding of stability and properties. J. Mater. Sci. 56, 5321–5427 (2021).

Article ADS CAS Google Scholar

Gheytanzadeh, M., Baghban, A., Habibzadeh, S., Mohaddespour, A. & Abida, O. Insights into the estimation of capacitance for carbon-based supercapacitors. RSC Adv. 11, 5479–5486 (2021).

Article ADS CAS Google Scholar

Gheytanzadeh, M. et al. Towards estimation of CO2 adsorption on highly porous MOF-based adsorbents using Gaussian process regression approach. Sci. Rep. https://doi.org/10.1038/s41598-021-95246-6 (2021).

Article Google Scholar

Baghban, A., Bahadori, M., Lemraski, A. S. & Bahadori, A. Prediction of solubility of ammonia in liquid electrolytes using least square support vector machines. Ain Shams Eng. J. 9, 1303–1312 (2018).

Article Google Scholar

Ahmadi, M. H. et al. An insight into the prediction of TiO2/water nanofluid viscosity through intelligence schemes. J. Therm. Anal. Calorim. 139, 2381–2394 (2020).

Article CAS Google Scholar

Baghban, A. & Khoshkharam, A. Application of LSSVM strategy to estimate asphaltene precipitation during different production processes. Pet. Sci. Technol. 34, 1855–1860 (2016).

Article CAS Google Scholar

Baghban, A., Abbasi, P. & Rostami, P. Modeling of viscosity for mixtures of Athabasca bitumen and heavy n-alkane with LSSVM algorithm. Pet. Sci. Technol. 34, 1698–1704 (2016).

Article CAS Google Scholar

Bahadori, A. et al. Computational intelligent strategies to predict energy conservation benefits in excess air controlled gas-fired systems. Appl. Therm. Eng. 102, 432–446 (2016).

Article CAS Google Scholar

Griffin, W. O. & Darsey, J. A. Artificial neural network prediction indicators of density functional theory metal hydride models. Int. J. Hydrogen Energy 38, 11920–11929 (2013).

Article CAS Google Scholar

Rahnama, A., Zepon, G. & Sridhar, S. Machine learning based prediction of metal hydrides for hydrogen storage, part I: Prediction of hydrogen weight percent. Int. J. Hydrogen Energy 44, 7337–7344 (2019).

Article CAS Google Scholar

Rahnama, A., Zepon, G. & Sridhar, S. Machine learning based prediction of metal hydrides for hydrogen storage, part II: Prediction of material class. Int. J. Hydrogen Energy 44, 7345–7353 (2019).

Article CAS Google Scholar

Suwarno, S. et al. Machine learning analysis of alloying element effects on hydrogen storage properties of AB2 metal hydrides. Int. J. Hydrogen Energy 47, 11938–11947 (2022).

Article CAS Google Scholar

Kim, J. M., Ha, T., Lee, J., Lee, Y.-S. & Shim, J.-H. Prediction of Pressure-Composition-Temperature Curves of AB2-Type Hydrogen Storage Alloys by Machine Learning. Available SSRN 3999210

Rui, J. et al. TOC content prediction based on a combined Gaussian process regression model. Mar. Pet. Geol. 118, 104429 (2020).

Article CAS Google Scholar

Fu, Q. et al. Prediction of the diet nutrients digestibility of dairy cows using Gaussian process regression. Inf. Process. Agric. 6, 396–406 (2019).

Google Scholar

Minh, H. Q., Niyogi, P. & Yao, Y. Mercer's theorem, feature maps, and smoothing. In Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Vol. 4005 LNAI, 154–168 (2006).

Handbook of Hydrogen Storage. Handbook of Hydrogen Storage (2010). https://doi.org/10.1002/9783527629800

Download references

Surface Reaction and Clean Energy Materials Laboratory, Chemical Engineering Department, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran

Majedeh Gheytanzadeh & Sajjad Habibzadeh

Chemical Engineering Department, Fouman Faculty of Engineering, University of Tehran, Fouman, Iran

Fatemeh Rajabhasani

Chemical Engineering Department, Amirkabir University of Technology (Tehran Polytechnic), Mahshahr Campus, Mahshahr, Iran

Alireza Baghban

College of Engineering and Technology, American University of the Middle East, 54200, Egaila, Kuwait

Otman Abida & Muhammad Tajammal Munir

Department of Chemical Engineering, School of Engineering Technology and Industrial Trades, College of the North Atlantic - Qatar, Doha, Qatar

Amin Esmaeili

You can also search for this author in PubMed Google Scholar

All authors collaborated in writting, software developing, conception, data gathering, and resources.

Correspondence to Alireza Baghban or Sajjad Habibzadeh.

The authors declare no competing interests.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

Gheytanzadeh, M., Rajabhasani, F., Baghban, A. et al. Estimating hydrogen absorption energy on different metal hydrides using Gaussian process regression approach. Sci Rep 12, 21902 (2022). https://doi.org/10.1038/s41598-022-26522-2

Download citation

Received: 15 August 2022

Accepted: 15 December 2022

Published: 19 December 2022

DOI: https://doi.org/10.1038/s41598-022-26522-2

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.