ارزیابی کارآیی الگوریتم های آموزش شبکه‌های عصبی مصنوعی برای برآورد پارامترهای کیفی آب دشت قروه-دهگلان

فصلنامه علمی مهندسی منابع آب

مقاله 6، دوره 16، شماره 59، بهمن 1402، صفحه 79-96   

نوع مقاله: مقاله پژوهشی

شناسه دیجیتال (DOI): 10.30495/wej.2024.24285.2236

نویسندگان

سید اشکان سید ابراهیمی¹؛ ابوذر نکوئی¹؛ محمود رضا ملایی نیا^{* 2}

¹فارغ التحصیل کارشناسی ارشد، گروه مهندسی عمران، دانشکده فنی و مهندسی، دانشگاه زابل، زابل، ایران

²دانشیار، گروه مهندسی عمران، دانشکده فنی و مهندسی، دانشگاه زابل، زابل، ایران

تاریخ دریافت: 09 اسفند 1398،  تاریخ بازنگری: 02 خرداد 1400،  تاریخ پذیرش: 10 بهمن 1402 

چکیده

چکیده
مقدمه: شبکه عصبی مصنوعی (ANN) یک ابزار قدرتمند داده محور است که قادراست روابط خطی و غیرخطی حاکم بر سیستم های مختلف را فراگیرد. اما هنوز تعیین الگوریتم دارای بهترین عملکرد ازنظر سرعت و دقت همگرایی برای یک مسئله خاص، چالش مهم پیش روی کاربران شبکه های عصبی مصنوعی است.
روش­: در این پژوهش، توانایی فرآیندهای پرکاربرد طی چند سال اخیر در بحث شبیه سازی و تخمین پارامترهای غیرخطی کیفیت آب بررسی و اثربخش ترین آن ها تعیین گردید. برای این منظور، 42 مدل از ترکیب توپولوژی شبکه عصبی مصنوعی (تک لایه یا چندلایه)  و فرآیندهای آموزش بررسی گردید. پارامترهای کیفی مشاهده شده در محل 107 چاه  در گستره آبخوان دشت قروه-دهگلان از سال 75 الی 92 جهت آموزش و داده های سال های 93 تا 95 جهت آزمون هر مدل استفاده شد.
یافته­ ها: نتایج نشان داد که شبکه های عصبی مصنوعی دارای یک لایه پنهان که از تعداد بهینه نورون بهره می برند، قادرند رفتار آبخوان را با دقت مطلوب و در مدت‌زمان کمتر شبیه سازی نمایند. همچنین، افزایش تعداد لایه های میانی همگام با افزایش دقت پاسخ، نه‌تنها تعداد سلول های بهینه شبکه بلکه مدت‌زمان تحلیل مساله را افزایش می دهد. به هرحال، شبکه های عصبی مصنوعی مبتنی بر روش Broyden-Fletcher-Goldfarb (BFG) از بیشترین کارآیی در شبیه‌سازی رفتار آبخوان برخوردار است، گرچه کارآیی روش Levenberg Marquart (LM) بسیار نزدیک به آن است. کارآیی بیشتر BFG نسبت به LM می‌تواند به دلیل برخورداری از میانگین خطا و انحراف معیار کمتر (به ترتیب برابر 46/3 و 09/3) آن باشد.
نتیجه ­گیری: شبکه های عصبی مصنوعی با یک لایه پنهان و تعداد بهینه نورون می توانند رفتار آبخوان را با دقت مطلوبی شبیه سازی کنند و نسبت به شبکه های عصبی مصنوعی متشکل از چندین لایه پنهان کارایی بیشتری دارند.

کلیدواژه‌ها

شبیه سازی؛ شبکه عصبی مصنوعی؛ الگوریتم های آموزش شبکه؛ تخمین پارامترهای کیفی؛ دشت قروه-دهگلان

عنوان مقاله [English]

The Efficiency Evaluation of Artificial Neural Network Training Algorithms for Estimation of Water Quality Parameters of Qorveh-Dehgolan Plain

نویسندگان [English]

Seyed Ashkan Seyed Ebrahimi¹؛ Abuzar Nekuie¹؛ Mahmoud Reza Mollaeinia²

¹M.Sc. graduate, Civil Engineering Department, Faculty of Technology and Engineering, University of Zabol, Zabol, Iran

²Associate Professor, Civil Engineering Department, Faculty of Technology and Engineering, University of Zabol, Zabol, Iran

چکیده [English]

Abstract
Introduction: An artificial neural network (ANN) is a powerful data-driven tool capable of learning the linear and nonlinear relationships governing different systems. However, determining the best-performing algorithm in terms of convergence speed and accuracy for a particular problem is still a fundamental challenge for users of artificial neural networks.
Methods: We investigated the most effective algorithm among widely used processes to simulate and estimate nonlinear water quality parameters. For this purpose, we constructed 42 models combining artificial neural network topology (single or multilayer) and training processes. The quality parameters’ data acquired at 107 wells throughout the aquifer of Qorveh-Dehgolan plain were used for training (data from 1996 to 2013) and to test (data from 2014 to 2016) each model.
Findings: The results showed that artificial neural networks with a hidden layer that benefits from the optimal number of neurons could simulate the aquifer behavior with high accuracy and in less time. Also, increasing the number of hidden layers while increasing the response accuracy increases the number of optimal network neurons and the duration of the problem analysis. Finally, artificial neural networks based on the Broyden-Fletcher-Goldfarb (BFG) method had the highest efficiency in simulating aquifer behavior, although the performance of the Levenberg Marquart (LM) method is very close to BFG. BFG is more efficient than LM due to its lower Mean Square Error and Standard Deviation (3.46 and 3.09, respectively).

کلیدواژه‌ها [English]

Simulation, Artificial Neural Network, Network Training Algorithms, Qualitative Parameter Estimation, Qorveh-Dehgolan Plain

مراجع

1. Bahmani P, Mafakheri S, Maleki A, Shah-Mohammadi S (1395) Investigating the chemical quality of groundwater resources in Qorveh-Dehgolan using GIS. Paper presented at  6th Iranian National Water Resources Management Conference, Sanandaj, Iran, 20 March 2016, [in Persian]

2. Mirzaei AA, Nazemi A (2010) Prediction of underground water level using artificial neural networks. Paper presented at  The first national conference on water resources management of coastal lands, Sari, Iran, [In Persian]

3. Bashiri J, Bashiri E, Momenpour E, Jahantabi M (2013) Quantitative and qualitative investigation of groundwater resources in the eastern plains of Kurdistan province and explanation of the effective factors. Paper presented at  The first conference on hydrology of semi-arid regions, Sanandaj, Iran, [In Persian]

4. Osati K (2016) Fluctuations of the underground water level of the aquifers of Dasht Qorveh-Dehgolan: Evidence of improper management of water resources in drought conditions. Paper presented at  6th Iranian National Water Resources Management Conference, Sanandaj, Iran, [In Persian]

5. Daryaei M, Agdarnejad A, Bina M, Radmanesh F (2010) Investigating the effect of river water quality factors on EC and TDS with the help of artificial neural networks. Paper presented at  08th International River Engineering Conference, Ahvaz, Iran, Jan. 2010, [In Persian]

6. Dehghani R, pourhaghi A, Kheiraey M (2016) Comparison of Anfis, Artificial neural network, and Gene expression programming to estimate the amount of Water hardness (Case study: Mazandaran Plain). New Findings in Applied Geology 10 (19):51-62

7. Saiyadi-Shahahraki A, Soltani-Mohammadi A, Naseri A, Mokhtaran A (2016) Simulation of groundwater salinity using Artificial Neural Network (ANN) , Particle Swarm Optimization (PSO) and SEAWAT model. (Case study: Debal khazaie sugarcane plantation). Journal Of Water and Soil Conservation 23 (5):307-316. doi:10.22069/jwfst.2017.10157.2459

8. Rahimzadeh-Kivi Z, Kardanmoghadam H, behbahani SMR (2016) Simulation and prediction nitrate in groundwater (Case study: south Khorasan-Birjand aquifer). Journal of Irrigation and Water Engineering 6 (4):114-127

9. Isazadeh M, Biazar SM, Ashrafzadeh A, Khanjani R (2019) Estimation of quality parameters of Gilan plain aquifer using gamma test, support vector machine models, and artificial neural network. Journal of Environmental Science and Technology 21 (2)

10. Eslamian S, Lavaei N (2009) Modelling nitrate pollution of groundwater using artificial neural network and genetic algorithm in an arid zone. International journal of water 5 (2):194-203

11. Ostad-Ali-Askari K, Shayannejad M, Kharazi H (2017) Artificial Neural Network for Modeling Nitrate Pollution of Groundwater in Marginal Area of Zayandeh-rood River, Isfahan, Iran. KSCE Journal of Civil Engineering 21:134-140. doi:10.1007/s12205-016-0572-8

12. Shariati A, Zolfaghari M, Kariminejad M-T (2016) Investigating temporal-spatial changes in the qualitative index of groundwater in Qorveh-Dehgolan Plain using GIS. Paper presented at  The second national conference on hydrology of semi-arid regions, Sanandaj, Iran, 18 October 2016, [In Persian]

13. Mohebbi M, Saeedi R, Montazeri A, Vaghefi K, Labbafi S, Oktaie S, Abtahi M, Mohagheghian A (2013) Assessment of water quality in groundwater resources of Iran using a modified drinking water quality index (DWQI). Ecological Indicators 30:28–34. doi:10.1016/j.ecolind.2013.02.008

14. Jamshidzadeh Z, Barzi M (2018) Groundwater quality assessment using the potability water quality index (PWQI): a case in the Kashan plain, Central Iran. Environmental Earth Sciences 77. doi:10.1007/s12665-018-7237-5

15. Kenda K, Čerin M, Bogataj M, Senožetnik M, Klemen K, Pergar P, Laspidou C, Mladenić D (2018) Groundwater Modeling with Machine Learning Techniques: Ljubljana polje Aquifer. Proceedings 2 (11). doi:10.3390/proceedings2110697

16. Hydrology TATCoAoANNi (2000) Artifical neural networks in hydrology. II: Hydrologic application. Journal of Hydrologic Engineering 5 (2)

17. Khalili-Aghdam N, Mosaedi A, Soltani A, Kamkar B (2012) Evaluation of ability of LARS-WG model for simulating some weather parameters in Sanandaj. Journal of Water and Soil Conservation 19 (4):85-102

18. Rezaei V, Vahdani E, Naseh G (2015) Investigating the effect of fluctuations in rainfall, evaporation, and relative humidity on the groundwater level of Qorouh-Dehgolan Plain. Paper presented at  The second national congress on environmental pollution and sustainable development, Sanandaj, Iran, March, [In Persian]

19. Mohammadi S, Siosemarde M (2016) Application of Artificial Neural Networks in Order to Predict Mahabad River Discharge. Open Journal of Ecology 06 (07):427-434. doi:10.4236/oje.2016.67040

20. Khalil BM, Awadallah AG, Karaman H, El-Sayed A (2012) Application of Artificial Neural Networks for the Prediction of Water Quality Variables in the Nile Delta. Journal of Water Resource and Protection 4:388-394. doi:10.4236/jwarp.2012.46044

21. Maier HR, Dandy GC (2000) Neural networks for the prediction and forecasting of water resources variables: a review of modelling issues and applications. Environmental Modelling & Software 15:101–124

22. Mustafa MR, Isa MH, Rezaur RB (2012) Artificial Neural Networks Modeling in Water Resources Engineering: Infrastructure and Applications. International Journal of Civil and Environmental Engineering 6 (2):128-136

23. Ehteshami M, Farahani N, Tavassoli S (2016) Simulation of nitrate contamination in groundwater using artificial neural networks. Modeling Earth Systems and Environment 2. doi:10.1007/s40808-016-0080-3

24. Husna N-E-A, Bar SH, Hussain MM, Ur-Rahman MT, Rahman M (2016) Ground water level prediction using artificial neural network. International Journal of Hydrology Science and Technology, 6 (4):371-381

25. Hagan MT, Demuth HB, Beale MH, Jesús OD (2014) Neural Network Design. In: Edition n (ed)

26. Ammari A (2016) MATLAB code of artificial neural networks estimation.

27. Li X, Yeh AG-O (2002) Neural-network-based cellular automata for simulating multiple land use changes using GIS. International Journal of Geographical Information Science 16 (4):323-343

28. Yilmaz I, Erik NY, Kaynar O (2010) Different types of learning algorithms of artificial neural network (ANN) models for prediction of gross calorific value (GCV) of coals. Scientific Research and Essays 5 (16):2242-2249

29. Yesilnacar MI, Sahinkaya E (2012) Artificial neural network prediction of sulfate and SAR in an unconfined aquifer in southeastern Turkey. Environmental Earth Sciences 67 (4):1111-1119

30. Eslamian S, Lavaei N (2009) Modelling nitrate pollution of groundwater using artificial neural network and genetic algorithm in an arid zone. International Journal of Water 5. doi:10.1504/IJW.2009.028726

31. Moasheri S, Rezapour O, Beyranvand Z, Poornoori Z (2013) Estimating the spatial distribution of groundwater quality parameters of Kashan plain with integration method of Geostatistics - Artificial Neural Network Optimized by Genetic-Algorithm. International Journal of Agriculture and Crop Science 5 (30):2434-2442.

32. Mafakheri O, Khaledi S, Shamsipour A, Fallahi-Khoshjoo M, Kermani A (2016) Analysis of drought using the NDVI index in Qaroveh and Dehgolan plains. Journal of Applied Research in Geographical Sciences 41:77-94

33. Asadzadeh F, Kaki M, Shakiba S, Raei B (2016) Impact of Drought on Groundwater Quality and Groundwater Level in Qorveh-Chardoli Plain. Iran Water Resources Research

34. Safavi-Gordini M, Mohammadrezapour O, Bahrami E, Mohammadi-Sadegh M, Salarigazi M (2018) Geostatistical Assessment of spatial and temporal variations of ground water quality parameters in Qorveh and Dehgolan South Plain. Iranian of Irrigation and Water Engineering 9 (1):167-182

35. Nadiri A-A, Jabraili N, Gharekhani M (2019) Comparison of different combination methods ability on groundwater vulnerability assessment in Qorveh- Dehgolan palin aquifer. Ecohydrology 6 (3):821-836

36. Ahmadi S, Soodmand-Afshar R (2018) Monitoring of Land Subsidence in Qorveh and Chahardoli Plains of Hamadan and Kurdistan Provinces using PS-InSAR Technique. Journal of environment and Water engineering 6 (3):219-233

37. Isazadeh M, Arabzadeh R, Darbandi S (2016) Performance Evaluation of Geostatistical Methods and Artificial Neural Network in Estimation of Aquifer Quality Parameters (Case Study: Qorveh Dehghan Plain). Journal of Water and Soil Science 20 (77):197-210. doi:10.18869/acadpub.jstnar.20.77.197

38. Coskun N, Yildirim T The effects of training algorithms in MLP network on image classification. In: Proceedings of the International Joint Conference on Neural Networks, 2003., 2003. IEEE, pp 1223-1226

39. Gopalakrishnan K (2010) Effect of training algorithms on neural networks aided pavement diagnosis. International Journal of Engineering, Science and Technology 2 (2):83-92

40. Parsaie F, Mahmoodi MA, Egdernezhad A (2020) Assessment of Groundwater Quality for Drinking and Agriculture in Qorveh Plain. Journal of Wetland Ecobiology 12 (1):65-80

41. Organization KPAJ (2005) design and Plan management, statistics and planning department. Sanandaj, Iran

42. Adamowski J, Karapataki C (2010) Comparison of multivariate regression and artificial neural networks for peak urban water-demand forecasting: evaluation of different ANN learning algorithms. Journal of Hydrologic Engineering 15 (10):729-743

43. Akdagli A, Kayabasi A (2014) An accurate computation method based on artificial neural networks with different learning algorithms for resonant frequency of annular ring microstrip antennas. Journal of Computational Electronics 13 (4):1014-1019

44. Kayri M (2016) Predictive abilities of bayesian regularization and Levenberg–Marquardt algorithms in artificial neural networks: a comparative empirical study on social data. Mathematical and Computational Applications 21 (2):20