Abstract
California bearing ratio (CBR) and resilient modulus are critical factors for designing pavements. However, the measurement of CBR and resilient modulus of crushed waste rocks that are widely used for the construction of mine haul roads can be costly and time-consuming which is often prohibitive, especially since service lifetime of haul roads is relatively short. The recent development of machine learning techniques makes it possible to develop more efficient models, but many algorithms exist and it is not always clear which one is better for predicting CBR and resilient modulus. The main objective of this study was therefore to evaluate and compare the performance of multiple models, such as multiple linear regression (MLR), k-nearest neighbors (KNN), decision tree (DT), random forest (RF), multilayer perceptron (MLP), and neuroevolution of augmenting topologies (NEAT) for predicting CBR and resilient modulus of crushed waste rocks. Thirty-nine and 2320 datasets, obtained from a series of CBR and repeated load triaxial tests, were applied to develop CBR and resilient modulus models, respectively. The study of input features and hyperparameters was conducted to determine the optimal architecture of the machine learning models. A comparison study showed that RF models provided better results with coefficient of determination R2 greater than 0.9. NEAT models showed good generalizability and simple structure although their performance was lower than RF models.
Similar content being viewed by others
References
AASHTO T307-99. (2017) Standard method of test for determining the resilient modulus of soils and aggregate materials. Washington, DC.
Adeli H (2001) Neural networks in civil engineering: 1989–2000. Comput-Aided Civil Infrastruc Eng 16(2):126–142
Ai X, Yi J, Zhao H, Chen S, Luan H, Zhang L, et al (2020) An empirical predictive model for the dynamic resilient modulus based on the static resilient modulus and california bearing ratio of cement-and lime-stabilised subgrade soils. Road Mater Pavement Des 1–20
Al-Qahtani FH, Crone SF (2013) Multivariate k-nearest neighbour regression for time series data—a novel algorithm for forecasting uk electricity demand. Paper presented at the The 2013 international joint conference on neural networks (IJCNN)
Alavi AH, Gandomi AH, Lary DJ (2016) Progress of machine learning in geosciences: Preface. Geosci Front 7(1):1–2
ASTM D1883-16. (2016) Standard test method for california bearing ratio (CBR) of laboratory-compacted soils. ASTM International, West Conshohocken, PA
ASTM D2487-17e1. (2017) Standard practice for classification of soils for engineering purposes (unified soil classification system). ASTM International. West Conshohocken, PA
Breiman L (1999) 1 random forests--random features
Breiman L, Friedman J, Stone CJ, Olshen RA (1984) Classification and regression trees. CRC Press, Boca Raton
Bussière B (2007) Colloquium 2004: hydrogeotechnical properties of hard rock tailings from metal mines and emerging geoenvironmental disposal approaches. Can Geotech J 44(9):1019–1052
Ceylan H, Bayrak MB, Gopalakrishnan K (2014) Neural networks applications in pavement engineering: a recent survey. Int J Pavement Res Technol 7(6)
Chen L (2010) A multiple linear regression prediction of concrete compressive strength based on physical properties of electric arc furnace oxidizing slag. Int J Appl Sci Eng 7(2):153–158
Chollet F (2018) Deep learning mit python und keras: Das praxis-handbuch vom entwickler der keras-bibliothek. MITP-Verlags GmbH & Co KG, Wachtendonk
Coronado O, Caicedo B, Taibi S, Gomes Correia A, Souli H, Fleureau J-M (2016) Effect of water content on the resilient behavior of non standard unbound granular materials. Transp Geotech 7:29–39
Curram SP, Mingers J (1994) Neural networks, decision tree induction and discriminant analysis: an empirical comparison. J Oper Res Soc 45(4):440–450
de Souza WM, Ribeiro AJA, da Silva CAU (2020) Use of ann and visual-manual classification for prediction of soil properties for paving purposes. Int J Pavement Eng 1–9
Díaz E, Tomás R (2021) Upgrading the prediction of jet grouting column diameter using deep learning with an emphasis on high energies. Acta Geotech 16(5):1627–1633
EN 13286-7. (2004) Unbound and hydraulically bound mixtures–cyclic load triaxial test for unbound mixtures. British Standard Institute
Erzin Y, Turkoz D (2016) Use of neural networks for the prediction of the cbr value of some aegean sands. Neural Comput Appl 27(5):1415–1426
Ferentinou M, Fakir M (2018) Integrating rock engineering systems device and artificial neural networks to predict stability conditions in an open pit. Eng Geol 246:293–309
Ferreira A (2017) Application of ann in pavement engineering: State-of-art
Ghorbani A, Hasanzadehshooiili H (2018) Prediction of ucs and cbr of microsilica-lime stabilized sulfate silty sand using ann and epr models; application to the deep soil mixing. Soils Found 58(1):34–49
Ghorbani B, Arulrajah A, Narsilio G, Horpibulsuk S (2020) Experimental and ann analysis of temperature effects on the permanent deformation properties of demolition wastes. Transp Geotech 24:100365
Ghorbani B, Arulrajah A, Narsilio G, Horpibulsuk S (2020) Experimental investigation and modelling the deformation properties of demolition wastes subjected to freezethaw cycles using ann and svr. Constr Build Mater 258:119688
Ghorbani B, Arulrajah A, Narsilio G, Horpibulsuk S, Bo MW (2020) Development of genetic-based models for predicting the resilient modulus of cohesive pavement subgrade soils. Soils Found 60(2):398–412
Ghorbani B, Arulrajah A, Narsilio G, Horpibulsuk S, Bo MW (2020) Hybrid formulation of resilient modulus for cohesive subgrade soils utilizing cpt test parameters. J Mater Civil Eng 32(9):06020011
Golder (2019) Environmental impact assessment and environmental management programme report for the proposed metsimaholo underground coal mine
Gu C, Ye X, Wang J, Cai Y, Cao Z, Zhang T (2020) Resilient behavior of coarse granular materials in three-dimensional stress state. Can Geotech J 57(9):1280–1293
Gu C, Zhan Y, Wang J, Cai Y, Cao Z, Zhang Q (2020) Resilient and permanent deformation of unsaturated unbound granular materials under cyclic loading by the large-scale triaxial tests. Acta Geotech 15(12):3343–3356
Hanandeh S, Ardah A, Abu-Farsakh M (2020) Using artificial neural network and genetics algorithm to estimate the resilient modulus for stabilized subgrade and propose new empirical formula. Transp Geotech 24:100358
Hao S, Pabst T (2021) Estimation of resilient behavior of crushed waste rocks using repeated load cbr tests. Transp Geotech 28:100525
Hausknecht M, Lehman J, Miikkulainen R, Stone P (2014) A neuroevolution approach to general atari game playing. IEEE Trans Comput Intell AI Games 6(4):355–366
Hecht-Nielsen R (1987) Kolmogorov’s mapping neural network existence theorem. Paper presented at the Proceedings of the international conference on Neural Networks
Jain A, Nandakumar K, Ross A (2005) Score normalization in multimodal biometric systems. Pattern Recogn 38(12):2270–2285
James M, Aubertin M, Bussière B (2013) On the use of waste rock inclusions to improve the performance of tailings impoundments. Paper presented at the Proceedings of the 18th International Conference Soil Mechanics and Geotechnical Engineering, Paris, France
Kaloop MR, Gabr AR, El-Badawy SM, Arisha A, Shwally S, Hu JW (2019) Predicting resilient modulus of recycled concrete and clay masonry blends for pavement applications using soft computing techniques. Front Struct Civ Eng 13(6):1379–1392
Karlaftis MG, Vlahogianni EI (2011) Statistical methods versus neural networks in transportation research: Differences, similarities and some insights. Transp Res Part C Emerg Technol 19(3):387–399
Khedr S (1985) Deformation characteristics of granular base course in flexible pavements. Transp Res Rec 1043:131–138
Kim S-H, Yang J, Jeong J-H (2014) Prediction of subgrade resilient modulus using artificial neural network. KSCE J Civ Eng 18(5):1372–1379
Kirby HR, Watson SM, Dougherty MS (1997) Should we use neural networks or statistical models for short-term motorway traffic forecasting? Int J Forecast 13(1):43–50
Laverdière A (2019) Effet de la granulométrie sur le comportement géotechnique de roches stériles concassées utilisées comme surface de roulement sur des routes minières. École Polytechnique de Montréal
LeCun Y, Bengio Y, Hinton G (2015) Deep learning. Nature 521(7553):436–444
Lee D-H, Cheon E, Lim H-H, Choi S-K, Kim Y-T, Lee S-R (2021) An artificial neural network model to predict debris-flow volumes caused by extreme rainfall in the central region of south korea. Eng Geol 281:105979
Lee Rodgers J, Nicewander WA (1988) Thirteen ways to look at the correlation coefficient. Am Stat 42(1):59–66
Lekarp F, Isacsson U, Dawson A (2000) State of the art. I: resilient response of unbound aggregates. J Transp Eng 126(1):66–75
Lin P, Ni P, Guo C, Mei G (2020) Mapping soil nail loads using federal highway administration (FHWA) simplified models and artificial neural network technique. Can Geotech J 57(10):1453–1471
Liu X, Zhang X, Wang H, Jiang B (2019) Laboratory testing and analysis of dynamic and static resilient modulus of subgrade soil under various influencing factors. Constr Build Mater 195:178–186
Mahmoodzadeh A, Mohammadi M, Hashim Ibrahim H, NarimanAbdulhamid S, Ghafoor Salim S, Farid Hama Ali H et al (2020) Artificial intelligence forecasting models of uniaxial compressive strength. Transp Geotech 27:100499
Mata A, Corchado JM (2009) Forecasting the probability of finding oil slicks using a cbr system. Expert Syst Appl 36(4):8239–8246
Mozumder RA, Laskar AI (2015) Prediction of unconfined compressive strength of geopolymer stabilized clayey soil using artificial neural network. Comput Geotech 69:291–300
Mukherjee A, Jain DK, Goswami P, Xin Q, Yang L, Rodrigues JJ (2020) Back propagation neural network based cluster head identification in mimo sensor networks for intelligent transportation systems. IEEE Access 8:28524–28532
Nazzal MD, Tatari O (2013) Evaluating the use of neural networks and genetic algorithms for prediction of subgrade resilient modulus. Int J Pavement Eng 14(4):364–373
NCHRP (2004) Guide for mechanistic-empirical design of new and rehabilitated pavement structures (Final Rep. No. 1-37A)
Nguyen-Thanh VM, Anitescu C, Alajlan N, Rabczuk T, Zhuang X (2021) Parametric deep energy approach for elasticity accounting for strain gradient effects. Comput Methods Appl Mech Eng 386:114096
Oskooei PR, Mohammadinia A, Arulrajah A, Horpibulsuk S (2020) Application of artificial neural network models for predicting the resilient modulus of recycled aggregates. Int J Pavement Eng, 1–13
Pan L, Feng X, Sang F, Li L, Leng M, Chen X (2019) An improved back propagation neural network based on complexity decomposition technology and modified flower pollination optimization for short-term load forecasting. Neural Comput Appl 31(7):2679–2697
Park H, Kweon G, Lee SR (2009) Prediction of resilient modulus of granular subgrade soils and subbase materials using artificial neural network. Road Mater Pavement Des 10(3):647–665
Pekel E (2020) Estimation of soil moisture using decision tree regression. Theoret Appl Climatol 139(3):1111–1119
Poulsen AP, Thorhauge M, Funch MH, Risi S (2017) Dlne: A hybridization of deep learning and neuroevolution for visual control. Paper presented at the 2017 IEEE Conference on Computational Intelligence and Games (CIG)
Qi S, Cui Y, Dupla JC, Chen R, Wang H, Su Y et al (2020) Investigation of the parallel gradation method based on the response of track-bed materials under cyclic loadings. Transp Geotech 24:100360
Ren J, Vanapalli SK, Han Z, Omenogor KO, Bai Y (2019) The resilient moduli of five canadian soils under wetting and freeze-thaw conditions and their estimation by using an artificial neural network model. Cold Reg Sci Technol 168:102894
Sadrossadat E, Heidaripanah A, Osouli S (2016) Prediction of the resilient modulus of flexible pavement subgrade soils using adaptive neuro-fuzzy inference systems. Constr Build Mater 123:235–247
Saha S, Gu F, Luo X, Lytton RL (2018) Use of an artificial neural network approach for the prediction of resilient modulus for unbound granular material. Transp Res Rec 2672(52):23–33
Samaniego E, Anitescu C, Goswami S, Nguyen-Thanh VM, Guo H, Hamdia K et al (2020) An energy approach to the solution of partial differential equations in computational mechanics via machine learning: Concepts, implementation and applications. Comput Methods Appl Mech Eng 362:112790
Seif El Dine B, Dupla JC, Frank R, Canou J, Kazan Y (2010) Mechanical characterization of matrix coarse-grained soils with a large-sized triaxial device. Can Geotech J 47(4):425–438
Shahin MA (2016) State-of-the-art review of some artificial intelligence applications in pile foundations. Geosci Front 7(1):33–44
Shen S-L, AtanganaNjock PG, Zhou A, Lyu H-M (2021) Dynamic prediction of jet grouted column diameter in soft soil using bi-lstm deep learning. Acta Geotech 16(1):303–315
Singh A, Yadav A, Rana A (2013) K-means with three different distance metrics. Int J Comput Appl 67(10)
Sinha S, Singh T, Singh V, Verma A (2010) Epoch determination for neural network by self-organized map (som). Comput Geosci 14(1):199–206
Sobol’ IYM (1990) On sensitivity estimation for nonlinear mathematical models. Matematicheskoe Modelirovanie 2(1):112–118
Sohangir S, Rahimi S, Gupta B (2013) Optimized feature selection using neuroevolution of augmenting topologies (neat). Paper presented at the 2013 Joint IFSA World Congress and NAFIPS Annual Meeting (IFSA/NAFIPS)
Sola J, Sevilla J (1997) Importance of input data normalization for the application of neural networks to complex industrial problems. IEEE Trans Nucl Sci 44(3):1464–1468
Stanley KO, Bryant BD, Miikkulainen R (2005) Real-time neuroevolution in the nero video game. IEEE Trans Evol Comput 9(6):653–668
Stanley KO, Miikkulainen R (2002) Evolving neural networks through augmenting topologies. Evol Comput 10(2):99–127
Tannant D, Regensburg B (2001) Guidelines for mine haul road design
Taskiran T (2010) Prediction of california bearing ratio (cbr) of fine grained soils by ai methods. Adv Eng Softw 41(6):886–892
Tenpe AR, Patel A (2020) Application of genetic expression programming and artificial neural network for prediction of cbr. Road Mater Pavement Des 21(5):1183–1200
Thompson R, Peroni R, Visser AT (2019) Mining haul roads: Theory and practice. CRC Press, Boca Raton
Thompson RJ (2011) Mine haul road design, construction and maintenance management. Curtin University, Bentley
Thompson RJ, Visser AT (2003) Mine haul road maintenance management systems. J S Afr Inst Min Metall 103(5):303–312
Tremblay G, Hogan C (2001) Mend manual: Volume 1, summary
Tso GKF, Yau KKW (2007) Predicting electricity energy consumption: A comparison of regression analysis, decision tree and neural networks. Energy 32(9):1761–1768
Varghese VK, Babu SS, Bijukumar R, Cyrus S, Abraham BM (2013) Artificial neural networks: A solution to the ambiguity in prediction of engineering properties of fine-grained soils. Geotech Geol Eng 31(4):1187–1205
Vidal A, Kristjanpoller W (2020) Gold volatility prediction using a CNN-LSTM approach. Expert Syst Appl 157:113481
Wang L, Wu C, Tang L, Zhang W, Lacasse S, Liu H et al (2020) Efficient reliability analysis of earth dam slope stability using extreme gradient boosting method. Acta Geotech 15(11):3135–3150
Wen R, Guo Z, Zhao T, Ma X, Wang Q, Wu Z (2017) Neuroevolution of augmenting topologies based musculor-skeletal arm neurocontroller. Paper presented at the 2017 IEEE International Instrumentation and Measurement Technology Conference (I2MTC)
Xu M, Watanachaturaporn P, Varshney PK, Arora MK (2005) Decision tree regression for soft classification of remote sensing data. Remote Sens Environ 97(3):322–336
Yilmazkaya E, Dagdelenler G, Ozcelik Y, Sonmez H (2018) Prediction of mono-wire cutting machine performance parameters using artificial neural network and regression models. Eng Geol 239:96–108
Yin Z-Y, Hicher P-Y, Dano C, Jin Y-F (2017) Modeling mechanical behavior of very coarse granular materials. J Eng Mech 143(1):C4016006
Yuksel ME (2018) Agent-based evacuation modeling with multiple exits using neuroevolution of augmenting topologies. Adv Eng Inform 35:30–55
Zhang W, Goh ATC (2013) Multivariate adaptive regression splines for analysis of geotechnical engineering systems. Comput Geotech 48:82–95
Acknowledgements
This work was carried out with the financial support from FRQNT and the industrial partners of Research Institute on Mines and the Environment (http://irme.ca/). The repeated load triaxial and CBR test equipment used in this study was acquired with a CFI grant.
Funding
Fonds Québécois de la Recherche sur la Nature et les Technologies, 2017-MI-202860, Thomas Pabst
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
System configuration and parameters in the NEAT models
Startup parameters | Settings | Definition |
---|---|---|
Number of generations | 30,000 | The number of generations to perform for a run |
Population size | 150 | The number of individuals in each generation |
Activation function | ReLU | The default activation function attribute assigned to hidden neurons |
Activation mutation rate | 0.0 | The probability that mutation will replace the neuron’s activation function |
Bias initial mean | 0.0 | The mean of the normal/Gaussian distribution |
Bias initial standard deviation | 1.0 | The standard deviation of the normal/Gaussian distribution |
Bias max. value | 30.0 | The maximum allowed bias value |
Bias min. value | − 30.0 | The minimum allowed bias value |
Bias mutation power | 0.5 | The standard deviation of the zero-centered normal/Gaussian distribution from which a bias value mutation is drawn |
Bias mutation rate | 0.7 | The probability that mutation will change the bias of a neuron by adding a random value |
Bias replace rate | 0.1 | The probability that mutation will replace the bias of a neuron with a newly chosen random value |
Compatibility disjoint coefficient | 1.0 | The coefficient for the disjoint and excess gene counts’ contribution to the network distance |
Compatibility weight coefficient | 0.5 | The coefficient for each weight, bias, or response multiplier difference’s contribution to the network distance (for homologous neurons or connections) |
Connection add probability | 0.5 | The probability that mutation will add a connection between existing neurons |
Connection delete probability | 0.2 | The probability that mutation will delete an existing connection |
Enabled default | True | The default enabled attribute of newly created connections |
Enabled mutation rate | 0.03 | The probability that mutation will replace the enabled status of a connection |
Feed forward | True | Generated networks will not be allowed to have recurrent connections |
Initial connection | Full | Each input neurons are connected to all output neurons |
Node add probability | 0.3 | The probability that mutation will add a new neuron |
Node delete probability | 0.25 | The probability that mutation will delete an existing neuron |
Number of hidden neurons | 0 | The number of hidden neurons to add to each genome in the initial population |
Number of input neurons | 8 and 3 | The number of input neurons, through which the network receives inputs |
Number of output neurons | 1 | The number of output neurons, to which the network delivers outputs |
Weight initial mean | 0.0 | The mean of the normal/Gaussian distribution used to select weight attribute values for new connections |
Weight initial standard deviation | 1.0 | The standard deviation of the normal/Gaussian distribution used to select weight values for new connections |
Weight max. value | 30.0 | The maximum allowed weight value |
Weight min. value | − 30.0 | The minimum allowed weight value |
Weight mutation power | 0.5 | The standard deviation of the zero-centered normal/Gaussian distribution from which a weight value mutation is drawn |
Weight mutation rate | 0.8 | The probability that mutation will change the weight of a connection by adding a random value |
Weight replace rate | 0.1 | The probability that mutation will replace the weight of a connection with a newly chosen random value |
Compatibility threshold | 3.0 | Individuals whose network distance is less than this threshold are considered to be in the same species |
Max. stagnation | 20 | Species that have not shown improvement in more than this number of generations will be considered stagnant and removed |
Species elitism | 2 | The number of species that will be protected from stagnation; mainly intended to prevent total extinctions caused by all species becoming stagnant before new species arise |
Elitism | 2 | The number of fittest individuals in each species that will be preserved as is from one generation to the next |
Survival threshold | 0.2 | The fraction for each species allowed to reproduce each generation |
Rights and permissions
About this article
Cite this article
Hao, S., Pabst, T. Prediction of CBR and resilient modulus of crushed waste rocks using machine learning models. Acta Geotech. 17, 1383–1402 (2022). https://doi.org/10.1007/s11440-022-01472-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11440-022-01472-1