Abstract
Lithology prediction is a fundamental problem because the outcome of lithology prediction is the critical underlying data for some basic geological work, e.g., establishing stratigraphic framework or analyzing distribution of sedimentary facies. As the geological formation generally consists of many different lithologies, the lithology prediction is always viewed as a tough work by geologists. Probabilistic neural network (PNN) shows high efficiency when solving pattern recognition problem since learning data do not need to do any pre-training of learning data and calculation results are universally reliable, and then, this model could be considered as an effective solution. However, there are two factors that seriously limit the PNN’s performance: One is existence of the interference variables of learning samples, and the other is selection of the window length of probability density distribution. In view of adverse impact of those two factors, two techniques, mean impact value (MIV) and particle swarm optimization (PSO), are introduced to improve the PNN’s calculation capability. Thus, a new prediction method referred as MIV–PSO–PNN is proposed in this paper. The proposed method is validated by three well-designed experiments, and the corresponding experiment data are recorded by two cored wells of the LULA oilfield. For the three experiments, prediction accuracies of the results provided by the proposed method are 81.67%, 73.34% and 88.34%, respectively, all of which are higher than those provided by other comparative approaches including backpropagation (BP), PNN, and MIV-PNN. The experiment results strongly demonstrate that the proposed method is capable to predict complex lithology.
Similar content being viewed by others
Abbreviations
- \( p({\mathbf{y}}|{\mathbf{X}}_{i} ) \) :
-
The probability density value of \( {\mathbf{y}} \) for the cluster \( {\mathbf{X}}_{i} \)
- \( n \) :
-
The number of variables, or the number of samples in Eq. (2), or the number of output units in Eq. (6)
- \( {\mathbf{x}}_{ik} \) :
-
kth sample of \( {\mathbf{X}}_{i} \)
- \( l \) :
-
The dimension of sample, or the number of hidden units in Eq. (6) and this number is an integer
- \( \sigma \) :
-
The smooth factor, also named as “spread”
- \( {\mathbf{B}}_{fn} \) :
-
The test data
- \( m \) :
-
The number of learning samples, or the number of input units in Eq. (6)
- \( f \) :
-
The number of test samples
- \( {\mathbf{D}}_{fm} \) :
-
The matrix of the Euclidean distance
- \( {\mathbf{P}}_{fm} \) :
-
The matrix of the probability density
- \( {\mathbf{S}}_{fz} \) :
-
The matrix of the total probability density
- \( {\mathbf{C}}_{fz} \) :
-
The ratio matrix containing the pattern recognition ratio of each test sample for each cluster
- \( \beta_{1}^{1} \) :
-
The enlarged coefficient
- \( \beta_{1}^{2} \) :
-
The reduced coefficient
- \( {\mathbf{H}} \) :
-
The learning data without interference variables
- \( {\mathbf{R}} \) :
-
The vector of MIV results
- \( f_{\text{MIV}} ( \cdot ) \) :
-
The MIV results
- \( {\varvec{\Gamma}} \) :
-
The spread population
- \( {\varvec{\upsigma}}_{i} \) :
-
ith spread seed
- \( {\mathbf{P}}_{fm}^{i} \) :
-
The matrix of the probability density scaled by ith spread seed
- \( {\varvec{\Delta}}_{i}^{j} \) :
-
The iteration result of ith spread seed; its corresponding predicted result has the smallest error in the first j iteration times
- \( {\varvec{\upsigma}}_{i}^{j} \) :
-
The iteration result of ith spread seed; its corresponding predicted result has the smallest error in jth iteration time
- \( {\mathbf{O}}_{if}^{j} \) :
-
The predicted result corresponding to the ith spread seed in jth iteration time
- \( {\mathbf{O}}_{{f\_{\text{std}}}} \) :
-
The standard result
- \( t \) :
-
The iteration time
- \( {\varvec{\Delta}}_{g}^{j} \) :
-
The iteration result of one spread seed; its corresponding predicted result has the smallest error in the first j iteration times
- \( {\varvec{\upsigma}}_{\varGamma }^{j} \) :
-
The iteration result of one spread seed; its corresponding predicted result has the smallest error in jth iteration time
- \( {\mathbf{O}}_{\varGamma f}^{j} \) :
-
The predicted result corresponding to one spread seed and has the smallest error in jth iteration time
- \( {\mathbf{W}}_{i}^{j} \) :
-
The iteration step of ith spread seed in jth iteration time
- \( \omega \) :
-
The inertia weight
- \( c_{1} \) :
-
The acceleration coefficient
- \( c_{2} \) :
-
The acceleration coefficient
- \( r_{1} \) :
-
Random value limited in [0,1]
- \( r_{2} \) :
-
Random value limited in [0,1]
- \( \sigma_{\text{max} \_1} \) :
-
The left limit of spread seed
- \( \sigma_{\text{max} \_2} \) :
-
The right limit of spread seed
- \( W_{\text{max} } \) :
-
The limit of iteration step
- \( \hbox{min} ( \cdot ) \) :
-
The function acquiring the minimum
- \( \left\| \cdot \right\|_{2} \) :
-
The function calculating L2 norm
- \( \omega_{\text{max} } \) :
-
The maximum inertia
- \( \omega_{\text{min} } \) :
-
The minimum inertia
- \( t_{\text{max} } \) :
-
The maximum iteration time
- \( v_{\text{nor}} \) :
-
The normalized logging value
- \( v \) :
-
The logging value
- \( v_{\text{max} } \) :
-
The maximum value of one log
- \( v_{\text{min} } \) :
-
The minimum value of one log
- \( a \) :
-
An integer value and its taking range is [1,10]
- \( {\varvec{\upsigma}}_{\# 1} \) :
-
The spread vector used for prediction of #1 well
- \( {\varvec{\upsigma}}_{\# 2} \) :
-
The spread vector used for prediction of #2 well
References
Ahila R, Sadasivam V, Manimala K (2012) Particle swarm optimization-based feature selection and parameter optimization for power system disturbances classification. Appl Artif Intell 26(9):832–861
Akhmanov GG, Silva IP, Erba E, Cita MB (2003) Sedimentary succession and evolution of the Mediterranean Ridge western sector as derived from lithology of mud breccia clasts. Mar Geol 195(1–4):277–299
Alves TM, Fetter M, Lima C, Cartwright JA, Cosgrove J, Gangá A (2017) An incomplete correlation between pre-salt topography, top reservoir erosion, and salt deformation in deep-water Santos Basin (SE Brazil). Mar Pet Geol 79:300–320
Andrew AM (2001) Backpropagation. Kybernetes 30(9–10):85–104
Andrioni M, Lima JAM, Guerra LA, Ribeiro EO, Nunes LMP, Ceccopieri W (2012) Ocean eddies’ influence on Lula field, Santos Basin, Brazil. In: ASME 2012, international conference on ocean, offshore and arctic engineering
Barrow J (2009) Lithology identification and prediction through amplitude variation with offset analysis in an area of the Malay basin, gulf of Thailand/Jerreh barrow. J Braz Chem Soc 20:57–63
Bogdanov YA, Gorbunova ZN, Serova VV (1998) Lithology and mineralogy of bottom sediments from the Bear island area, Norwegian sea. Oceanology 38(4):542–547
Bosch D, Ledo J, Queralt P (2013) Fuzzy logic determination of lithologies from well log data: application to the KTB project data set (Germany). Surv Geophys 34(4):413–439
Boyd A, Souza A, Carneiro G, Machado V, Trevizan W, Coutinho B, Netto P, Polinski R, Bertolini A (2015) Pre-salt carbonate evaluation for Santos Basin, offshore Brazil. Petrophysics 56(6):577–591
Broomhead DS, Lowe D (1988) Radial basis functions, multi-variable functional interpolation and adaptive networks. Adv Neural Inf Process Syst 4148:728–734
Chen CH, Chu CT (2009) High performance iris recognition based on 1-D circular feature extraction and PSO-PNN classifier. Expert Syst Appl 36(7):10351–10356
Chen N, Sun F, Ding L, Wang H (2009) An adaptive PNN-DS approach to classification using multi-sensor information fusion. Neural Comput Appl 18(5):455–467
Chetelat B, Liu CQ, Wang Q, Zhang G (2013) Assessing the influence of lithology on weathering indices of Changjiang river sediments. Chem Geol 359(6):108–115
Chi MV, Wong PK, Wong KI (2014) Simultaneous-fault detection based on qualitative symptom descriptions for automotive engine diagnosis. Appl Soft Comput 22(5):238–248
Cobbold PR, Meisling KE, Mount VS (2001) Reactivation of an obliquely rifted margin, Campos and Santos Basins, southeastern Brazil. AAPG Bull 85(11):1925–1944
D’Afonseca L, Lecerf D, Souza A, Sanchez F, Prigent H, Vidal T (2013) Multi-azimuth imaging for deep-water pre-salt reservoirs in Santos Basin, Brazil. In: EAGE conference and exhibition incorporating SPE EUROPEC
de Almeida JA (2010) Stochastic simulation methods for characterization of lithoclasses in carbonate reservoirs. Earth Sci Rev 101(3–4):250–270
Dev VA, Eden MR (2019) Formation lithology classification using scalable gradient boosted decision trees. Comput Chem Eng 128:392–404
Djafarov IS, Khafizov SF, Syngaevsky PE (2004) NMR application in reservoirs with complex lithology: a case study. Petrophysics 45(2):119–129
Duarte CSL, Viana AR (2007) Santos drift system: stratigraphic organization and implications for late Cenozoic palaeocirculation in the Santos Basin, SW Atlantic Ocean. Miner Pediatr 276(1):171–198
Eberhart R, Shi Y (2001) Particle swarm optimization: developments, applications and resources. In: Proceedings of 2001 IEEE congress on evolutionary computation, vol 1, pp 81–86
Freire EB, Lykawka R, Gabaglia GPR, Rodrigues EB, Terra GJS (2011) Searching for potential Analogues for the pre-salt Santos Basin, Brazil: high-resolution stratigraphic studies of microbialite-bearing successions from Salta Basin, Argentina. In: AAPG international conference and exhibition
Goldszmidt R, Moises M (1997) Bayesian network classifiers. Machine Learning 29(2–3):131–163
Guerra MCM, Underhill JR (2012) Role of halokinesis in controlling structural styles and sediment dispersal in the Santos Basin, offshore Brazil. Geol Soc Lond Spec Publ 363:175–206
Guo Y, Ma H, Ba J, Yu H, Long C (2015) Impact of data distribution on fluid sensitivity analysis: a quantitative investigation. J Appl Geophys 119:1–15
Huc AY (2004) Petroleum in the south Atlantic. Oil Gas Sci Technol 59(59):243–253
Inaba T, Suzuki N, Hirai A, Sekiguchi K, Watanabe T (2001) Source rock lithology prediction based on oil diacholestane abundance in the siliceous-clastic Akita sedimentary basin, Japan. Organ Geochem 32(7):877–890
Jacobson LA (1988) Macroscopic thermal neutron capture cross section measurements. IEEE Trans Nucl Sci 35(1):817–821
Jacobson LA, Wyatt DF (1993) Elemental yields and complex lithology analysis from the pulsed spectral gamma log. SPWLA Annu Log Symp 37(1):50–61
Jain AK, Dubes RC (1988) Algorithms for clustering data. Technometrics 32(2):227–229
Kakouei A, Masihi M, Sola BS, Biniaz E (2014) Lithological facies identification in Iranian largest gas field: a comparative study of neural network methods. J Geol Soc India 84(3):326–334
Kennedy J, Eberhart R (1995) Particle swarm optimization. In: Proceedings of IEEE international conference on neural networks, vol 4, pp 1942–1948
Kominz MA, Patterson K, Odette D (2011) Lithology dependence of porosity in slope and deep marine sediments. J Sediment Res 81(9–10):730–742
Kou Y, Shi YM, Boren LI, Qin XS (2010) The complex lithology rock-electricity features of volcanic rocks in Kelameili gas field. Acta Petrol Sin 26(1):291–301
Lentini MR, Fraser SI, Sumner HS, Davies RJ (2010) Geodynamics of the central south Atlantic conjugate margins: implications for hydrocarbon potential. Petrol Geosci 16(3):217–229
Ma YZ (2011) Lithofacies clustering using principal component analysis and neural network: applications to wireline logs. Math Geosci 43(4):401–419
Malik RF, Rahman TA, Hashim SZM, Ngah R (2007) New particle swarm optimizer with sigmoid increasing inertia weight. Int J Comput Sci Secur 1(2):43–52
Michelon D, Marques E, Figueiredo J, Ferraz H, Barros P (2013) Santonian-Campanian channelized systems of the Santos Basin, Brazil: stratigraphic framework and reservoir potential. In: GCSSEPM proceedings, pp 381–396
Mirzaei-Paiaman A, Saboorian-Jooybari H (2016) A method based on spontaneous imbibition for characterization of pore structure: application in pre-SCAL sample selection and rock typing. J Nat Gas Sci Eng 35(A):814–825
Mirzaei-Paiaman A, Salavati S (2012) The application of artificial neural networks for the prediction of oil production flow rate. Energy Sources A Recov Util Environ Eff 34(19):1834–1843
Mirzaei-Paiaman A, Saboorian-Jooybari H, Pourafshary P (2015) Improved method to identify hydraulic flow units for reservoir characterization. Energy Technol 3(7):726–733
Mirzaei-Paiaman A, Ostadhassan M, Rezaee R, Saboorian-Jooybari H, Chen Z (2018) A new approach in petrophysical rock typing. J Petrol Sci Eng 166:445–464
Mirzaei-Paiaman A, Sabbagh F, Ostadhassan M, Shafiei A, Rezaee R, Saboorian-Jooybari H (2019a) A further verification of FZI* and PSRTI: newly developed petrophysical rock typing indices. J Petrol Sci Eng 175:693–705
Mirzaei-Paiaman A, Saboorian-Jooybari H, Chen Z, Ostadhassan M (2019b) New technique of true effective mobility (TEM-Function) in dynamic rock typing: reduction of uncertainties in relative permeability data for reservoir simulation. J Petrol Sci Eng 179:210–227
Murata N, Yoshizawa S, Amari SI (1994) Network information criterion-determining the number of hidden units for an artificial neural network model. IEEE Trans Neural Netw 5(6):865–872
Nakano C, Pinto AC, Marcusso J, Minami K (2010) Pre-salt Santos Basin extended well test and production pilot in the Tupi area-the planning phase. J Petrol Technol 62(2):66–68
Paola JD, Schowengerdt RA (1993) A review and analysis of backpropagation neural networks for classification of remotely-sensed multi-spectral imagery. Int J Remote Sens 16(16):3033–3058
Parzen E (1962) On estimation of a probability density function and mode. Ann Math Stat 33(3):1065–1076
Pediwal J, Mahor A, Khatri N (2012) Exponential decreasing inertia weight particle swarm optimization in economic load dispatch. Int J Eng Innov Res 1(5):380–384
Porter CR, Pickett GR, Whitman WW (1969) A method of determining rock characteristics for computation of log data: the litho-porosity cross plot. Log Anal 6:16–24
Qi M, Fu Z, Chen F (2016) Research on a feature selection method based on median impact value for modeling in thermal power plants. Appl Therm Eng 94:472–477
Riedmiller M (1994) Advanced supervised learning in multi-layer perceptrons-from backpropagation to adaptive learning algorithms. Comput Stand Interfaces 16(3):265–278
Roberts HH, Sydow J, Bouma AH (1993) Seismic stratigraphy, sedimentology, and reservoir potential of a late Pleistocene shelf-edge delta. AAPG Bull 77(9):1598–1599
Rosenfeld, J.H., Hood, J.F., 2006. Play potential in the deepwater Santos Basin, Brazil. Offshore, 66(9)
Sahoo S, Jha MK (2016) Pattern recognition in lithology classification: modeling using neural networks, self-organizing maps and genetic algorithms. Hydrogeol J 25(2):311–330
Shaw L, Bagha S (2012) Online EMG signal analysis for diagnosis of neuromuscular diseases by using PCA and PNN. Int J Eng Sci Technol 4(10):4453–4459
Shi Y, Eberhart R (1998) A modified particle swarm optimizer. In: Proceedings of 1998 IEEE international conference on evolutionary computation, pp 69–73
Shi Y, Eberhart R (1999) Empirical study of particle swarm optimization. In: Proceedings of 1999 IEEE congress on evolutionary computation, pp 1945–1950
Smyth GK, Speed T (2003) Normalization of cDNA microarray data. Methods 31(4):265–273
Specht DF (1990) Probabilistic neural networks. Neural Netw 3(1):109–118
Stanton N, Ponte-Neto C, Bijani R, Masini E, Fontes S, Flexor JM (2014) A geophysical view of the southeastern Brazilian margin at Santos Basin: insights into rifting evolution. J S Am Earth Sci 55(2):141–154
Sun J, Li Q, Chen M, Ren L, Huang G, Li C, Zhang Z (2019) Optimization of models for a rapid identification of lithology while drilling: a win-win strategy based on machine learning. J Petrol Sci Eng 176:321–341
Thompson DL, Stilwell JD, Hall M (2015) Lacustrine carbonate reservoirs from early cretaceous rift lakes of western Gondwana: pre-salt coquinas of brazil and west Africa. Gondwana Res 28(1):26–51
Trelea IC (2003) The particle swarm optimization algorithm: convergence analysis and parameter selection. Inf Process Lett 85(6):317–325
Wang X, Wu C, Guo Y, Meng Q, Zhang Y, Tao Y (2013) Accumulation feature of Lula oilfield and its exploratory implication for pre-salt reservoirs in Santos Basin, Brazil. China Pet Explor 18(3):61–69
Wertheim PH (2008) Pre-salt discoveries continue in Brazil. Offshore 68(7)
Wertheim PH (2009) Tupi extended well test sheds light on Brazil’s subsalt bonanza. Offshore 69(7)
Wong MLD, Nandi AK (2004) Automatic digital modulation recognition using artificial neural network and genetic algorithm. Signal Process 84(2):351–365
Wu C (2015) Petroleum geology characteristics and exploration targets of pre-salt formations in Santos Basin, Brazil. Petrol Geol Exp 37(1):1–5
Xie Y, Zhu C, Zhou W, Li Z, Liu X, Tu M (2018) Evaluation of machine learning methods for formation lithology identification: a comparison of tuning processes and model performances. J Petrol Sci Eng 160:182–193
Yong-Yan LU, Wang WG (2011) Variable selection of financial distress prediction-the SVM method based on mean impact value. Syst Eng 29(8):73–78
Zhang S (2008) Application of probabilistic neural network technique to lithology inversion of heterogeneous stratum. Acta Petrol Sin 29(4):549–552
Zhang DX, Chen YT, Meng J (2018) Synthetic well logs generation via recurrent neural networks. Pet Explor Dev 45(4):629–639
Zhu LQ, Zhang C, Zhang CM, Wei Y, Zhou XQ, Cheng Y, Huang YY, Zhang L (2018) Prediction of total organic carbon content in shale reservoir based on a new integrated hybrid neural network and conventional well logging curves. J Geophys Eng 15:1050–1061
Zhu LQ, Zhang C, Zhang CM, Zhang ZS, Zhou XQ, Liu WN, Zhu BY (2020) A new and reliable dual model- and data-driven TOC prediction concept: a TOC logging evaluation method using multiple overlapping methods integrated with semi-supervised deep learning. J Petrol Sci Eng 188:106944
Acknowledgements
This work was supported by grants from the National Key Oil and Gas Project of China (No. 2016ZX05033-002-005) and the Sinopec Funded Project (No. P19020-2).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Appendices
Appendix 1
“Euclidean distance” is determined by the power exponent component of Eq. (2).“max in row” means extracting the maximum value of each row.
Appendix 2
Appendix 3
Rights and permissions
About this article
Cite this article
Gu, Y., Zhang, Z., Zhang, D. et al. Complex lithology prediction using mean impact value, particle swarm optimization, and probabilistic neural network techniques. Acta Geophys. 68, 1727–1752 (2020). https://doi.org/10.1007/s11600-020-00504-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11600-020-00504-2