Optimization of ANFIS with GA and PSO estimating α ratio in driven piles

This study aimed to optimize Adaptive Neuro-Fuzzy Inferences System (ANFIS) with two optimization algorithms, namely, Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) for the calculation friction capacity ratio (α) in driven shafts. Various studies are shown that both ANFIS are valuable methods for prediction of engineering problems. However, optimizing ANFIS with GA and PSO has not been used in the area of pile engineering. The training data set was collected from available full-scale results of the driven piles. The input parameters used in this study were pile diameter (m), pile length (m), relative density (I_d), embedment ratio (L/D), both of the pile end resistance (q_c) and base resistance at relatively 10% base settlement (q_b0.1) from CPT result, whereas the output was α. A learning fuzzy-based algorithm was used to train the ANFIS model in the MATLAB software. The system was optimized by changing the number of clusters in the FIS and then the output was used for the GA and PSO optimization algorithm. The prediction was compared with the real-monitoring field data. As a result, good agreement was attained representing reliability of all proposed models. The estimated results for the collected database were assessed based on several statistical indices such as R², RMSE, and VAF. According to R², RMSE, and VAF, values of (0.9439, 0.0123 and 99.91), (0.9872, 0.0117 and 99.99), and (0.9605, 0.0119 and 99.97) were obtained for testing data sets of the optimized ANFIS, GA–ANFIS, and PSO–ANFIS predictive models, respectively. This indicates higher reliability of the optimized GA–ANFIS model in estimating α ratio in driven shafts.

     

Neuro-genetic,neuro-imperialism and genetic programing models in predicting ultimate bearing capacity of pile

Prediction of pile-bearing capacity developing artificial intelligence models has been done over the last decade. Such predictive tools can assist geotechnical engineers to easily determine the ultimate pile bearing capacity instead of conducting any difficult field tests. The main aim of this study is to predict the bearing capacity of pile developing several smart models, i.e., neuro-genetic, neuro-imperialism, genetic programing (GP) and artificial neural network (ANN). For this purpose, a number of concrete pile characteristics and its dynamic load test specifications were investigated to select pile cross-sectional area, pile length, pile set, hammer weight and drop height as five input variables which have the most impacts on pile bearing capacity as the single output variable. It should be noted that all the aforementioned parameters were measured by conducting a series of pile driving analyzer tests on precast concrete piles located in Pekanbaru, Indonesia. The recorded data were used to establish a database of 50 test cases. With regard to data modelling, many smart models of neuro-genetic, neuro-imperialism, GP and ANN were developed and then evaluated based on the three most common statistical indices, i.e., root mean squared error (RMSE), coefficient determination (R²) and variance account for (VAF). Based on the simulation results and the computed indices’ values, it is observed that the proposed GP model with training and test RMSE values of 0.041 and 0.040, respectively, performs noticeably better than the proposed neuro-genetic model with RMSE values of 0.042 and 0.040, neuro-imperialism model with RMSE values of 0.045 and 0.059, and ANN model with RMSE values of 0.116 and 0.108 for training and test sets, respectively. Therefore, this GP-based model can provide a new applicable equation to effectively predict the ultimate pile bearing capacity.

     

Optimum model for bearing capacity of concrete-steel columns with AI technology via incorporating the algorithms of IWO and ABC

In a composite column, the performance of both the concrete and steel has a considerable effect on the structural behaviour under different loading conditions. This study applies several artificial intelligence (AI) techniques to optimise the bearing capacity of concrete-filled steel tube (CFST) columns. First, the bearing capacity values of the CFST columns are estimated by an artificial neural network (ANN) technique. Using 303 datasets, the outer diameter, concrete compressive strength, tensile yield stress of the steel column, thickness of the steel cover, and length of the applied samples are considered as the model inputs. Following a series of analyses, several ANN models are developed. The ANN model with 8 neurons and 250 iterations is determined as the best model to predict the bearing capacity of the CFST columns. Subsequently, the invasive weed optimisation (IWO) technique, which is considered the most current optimisation algorithm, is developed to maximise the results of the bearing capacity by considering the selected ANN model. To highlight the ability of IWO, the artificial bee colony (ABC) algorithm is also applied. Consequently, it is found that both optimisation algorithms can design input parameters such that the maximum value of the bearing capacity can be obtained. The bearing capacity of the CFST columns from the ABC and IWO techniques indicates that IWO has a better capability of maximising the bearing. Thus, IWO can optimise similar problems with a high rate of performance.

     

The prediction of ultimate pure bending moment of concrete-filled steel tubes by adaptive neuro-fuzzy inference system (ANFIS)

In this study, different modelling techniques such as multiple regression and adaptive neuro-fuzzy inference system (ANFIS) are used for predicting the ultimate pure bending of concrete-filled steel tubes (CFTs). The behaviour of CFT under pure bending is complex and highly nonlinear; therefore, forward modelling techniques can have considerable limitations in practical situations where fast and reliable solutions are required. Linear multiple regression (LMR), nonlinear multiple regression (NLMR) and ANFIS models were trained and checked using a large database that was constructed and populated from the literature. The database comprises 72 pure bending tests conducted on fabricated and cold-formed tubes filled with concrete. Out of 72 tests, 48 tests were conducted by the second author. Input variables for the models are the same with those used by existing codes and practices such as the tube thickness, tube outside diameter, steel yield strength, strength of concrete and shear span. A practical application example, showing the translation of constructed ANFIS model into design equations suitable for hand calculations, was provided. A sensitivity analysis was conducted on ANFIS and multiple regression models. It was found that the ANFIS model is more sensitive to change in input variables than LMR and NLMR models. Predictions from ANFIS models were compared with those obtained from LMR, NLMR, existing theory and a number of available codes and standards. The results indicate that the ANFIS model is capable of predicting the ultimate pure bending of CFT with a high degree of accuracy and outperforms other common methods.

     

Settlement prediction of the rock-socketed piles through a new technique based on gene expression programming

The settlement design of bored piles socketed into rock has received considerable attention. Although many design methods of pile settlement are recommended in the literature, proposing new/practical technique(s) with higher performance prediction is of advantage. A new model based on gene expression programming (GEP) is presented in this paper for predicting the settlement of the rock-socketed pile. To do this, 96 piles socketed in different types of rock (mostly granite) as part of the Klang Valley Mass Rapid Transit project, Malaysia, were studied. In order to propose a predictive model with higher performance prediction, a series of GEP analyses were conducted using the most important factors on pile settlement, i.e. ratio of length in soil layer to length in rock layer, ratio of total length to diameter, uniaxial compressive strength, standard penetration test and ultimate bearing capacity. For comparison purpose, using the same dataset, linear multiple regression (LMR) technique was also performed. After developing the equations, their prediction performances were checked through several performance indices. The results demonstrated the feasibility of GEP-based predictive model of settlement. Coefficients of determination (CoD) values of 0.872 and 0.861 for training and testing datasets of GEP equation, respectively, show superiority of this model in predicting pile settlement while these values were obtained as 0.835 and 0.751 for the LMR model. Moreover, root mean square error (RMSE) values of (1.293 and 1.656 for training and testing) and (1.737 and 1.767 for training and testing) were achieved for the developed GEP and LMR models, respectively.

     

Several non-linear models in estimating air-overpressure resulting from mine blasting

This research presents several non-linear models including empirical, artificial neural network (ANN), fuzzy system and adaptive neuro-fuzzy inference system (ANFIS) to estimate air-overpressure (AOp) resulting from mine blasting. For this purpose, Miduk copper mine, Iran was investigated and results of 77 blasting works were recorded to be utilized for AOp prediction. In the modeling procedure of this study, results of distance from the blast-face and maximum charge per delay were considered as predictors. After constructing the non-linear models, several performance prediction indices, i.e. root mean squared error (RMSE), variance account for (VAF), and coefficient of determination (R ²) and total ranking method are examined to choose the best predictive models and evaluation of the obtained results. It is obtained that the ANFIS model is superior to other utilized techniques in terms of R ², RMSE, VAF and ranking herein. As an example, RMSE values of 5.628, 3.937, 3.619 and 2.329 were obtained for testing datasets of empirical, ANN, fuzzy and ANFIS models, respectively, which indicate higher performance capacity of the ANFIS technique to estimate AOp compared to other implemented methods.     

Performance evaluation of hybrid Wavelet-ANN and Wavelet-ANFIS models for estimating evapotranspiration in arid regions of India

This paper evaluates the ability of wavelet transform in improving the accuracy of artificial neural network (ANN) and adaptive neuro-fuzzy interface systems (ANFIS) models. In this study, the performance of hybrid Wavelet-ANN and Wavelet-ANFIS models for estimating daily evapotranspiration in arid regions was evaluated. Prior to the development of models, gamma test was used to identify the best input combinations that could be used under limited data scenario. Performance of the proposed hybrid models was compared to ANN, ANFIS, and conventionally used Hargreaves equation. The results revealed that use of wavelet transform as data preprocessing technique enhanced the efficiency of ANN and ANFIS models. Wavelet-ANN and Wavelet-ANFIS performed reasonably better than other models. Better handling of wavelet-decomposed input variables enabled Wavelet-ANN models to perform slightly better than the Wavelet-ANFIS models. W-ANN2 (RMSE = 0.632 mm/day and R = 0.96) was found to be the best model for estimating daily evapotranspiration in arid regions. The proposed W-ANN2 model used second-level db3 wavelet-decomposed subseries of temperature and previous day evapotranspiration values as inputs. The study concludes that hybrid Wavelet-ANN and Wavelet-ANFIS models can be effectively used for modeling evapotranspiration.

     

Response model based on weighted bagging GMDH

Identifying customers who are more likely to respond to a mail or call is an important issue in direct marketing, and constructing response model is an effective approach to solve this problem. Logistic regression and neural network are two of the most commonly used models in direct marketing. However, both models have their own limitations. Based on the group method of data handling (GMDH) theory, in this study we propose a weighted bagging GMDH model (WB-GMDH) as a response model. First, multiple GMDH models are trained with bootstrap replicas of the original data set. Then, the final classification results are produced by aggregating the outputs of the base GMDH models based on the weights, which are computed according to the classification accuracy of each base model in the validation data set. The model is applied to two real-world data sets and its performance is measured by accuracy, ROC curve and lift chart. The results show that the WB-GMDH model has better performance than other models. Finally, analyses are carried out according to the explicit expression obtained from this model, which will help marketing analysts to develop new business plans better.     

Adaptive neuro‐fuzzy inference modelling and sensitivity analysis for capacity estimation of fiber reinforced polymer‐strengthened circular reinforced concrete columns

In this study, a fuzzy logic‐based model for predicting the ultimate strength of FRP‐confined circular reinforced concrete (RC) columns is presented. The adaptive neuro‐fuzzy inference system (ANFIS) model was generated using valid experimental data with seven input variables. Input parameters were considered in such a way that all the parameters affecting the compressive strength of the column were simultaneously involved. Different models for compressive strength of fiber reinforced polymer (FRP)‐confined concrete including the model in American Concrete Institute (ACI), to calculate the maximum stress endured by the column under axial load, were presented and compared with the results of the ANFIS model. Also, for similarity to other models, the ACI equation for calculating the maximum compressive strength tolerated by a column was considered without reducing coefficients as ACI‐N and was compared with other models. The results obtained from the ANFIS model were compared with results from other models. ANFIS model showed the highest accuracy among all models in predicting the experimental results.     

Performance evaluation of hybrid FFA-ANFIS and GA-ANFIS models to predict particle size distribution of a muck-pile after blasting

Accurately predicting the particle size distribution of a muck-pile after blasting is always an important subject for mining industry. Adaptive neuro-fuzzy inference system (ANFIS) has emerged as a synergic intelligent system. The main contribution of this paper is to optimize the premise and consequent parameters of ANFIS by firefly algorithm (FFA) and genetic algorithm (GA). To the best of our knowledge, no research has been published that assesses FFA and GA with ANFIS for fragmentation prediction and no research has tested the efficiency of these models to predict the fragmentation in different time scales as of yet. To show the effectiveness of the proposed ANFIS-FFA and ANFIS-GA models, their modelling accuracy has been compared with ANFIS, support vector regression (SVR) and artificial neural network (ANN). Intelligence predictions of fragmentation by ANFIS-FFA, ANFIS-GA, ANFIS, SVR and ANN are compared with observed values of fragmentation available in 88 blasting event of two quarry mines, Iran. According to the results, both ANFIS-FFA and ANFIS-GA prediction models performed satisfactorily; however, the lowest root mean square error (RMSE) and the highest correlation of determination (R²) values were obtained from ANFIS-GA model. The values of R² and RMSE obtained from ANFIS-GA, ANFIS-FFA, ANFIS, SVR and ANN models were equal to (0.989, 0.974), (0.981, 1.249), (0.956, 1.591), (0.924, 2.016) and (0.948, 2.554), respectively. Consequently, the proposed ANFIS-GA model has the potential to be used for predicting aims on other fields.

     

Design and implementation of a new tuned hybrid intelligent model to predict the uniaxial compressive strength of the rock using SFS-ANFIS

This study proposes a novel design to systematically optimize the parameters for the adaptive neuro-fuzzy inference system (ANFIS) model using stochastic fractal search (SFS) algorithm. To affirm the efficiency of the proposed SFS-ANFIS model, the predicting results were compared with ANFIS and three hybrid methodologies based on ANFIS combined with genetic algorithm (GA), differential evolution (DE), and particle swarm optimization (PSO). Accurate prediction of uniaxial compressive strength (UCS) is of great significance for all geotechnical projects such as tunnels and dams. Hence, this study proposes the use of SFS-ANFIS, GA-ANFIS, DE-ANFIS, PSO-ANFIS, and ANFIS models to predict UCS. In this regard, the fresh water tunnel of Pahang–Selangor located in Malaysia was considered and the requirement data samples were collected. Different statistical metrics such as coefficient of determination (R²) and mean absolute error were used to evaluate the models. Referring to the efficiency results of SFS-ANFIS, it can be found that the SFS-ANFIS (with the R² of 0.981) has higher ability than PSO-ANFIS, DE-ANFIS, GA-ANFIS, and ANFIS models in predicting the UCS.

     

Predicting discharge capacity of triangular labyrinth side weir located on a straight channel by using an adaptive neuro-fuzzy technique

Side weirs are widely used for flow diversion in irrigation, land drainage, urban sewage systems and also in intake structures. It is essential to correctly predict the discharge coefficient for hydraulic engineers involved in the technical and economical design of side weirs. In this study, the discharge capacity of triangular labyrinth side weirs is estimated by using adaptive neuro-fuzzy inference system (ANFIS). Two thousand five hundred laboratory test results are used for determining discharge coefficient of triangular labyrinth side weirs. The performance of the ANFIS model is compared with multi nonlinear regression models. Root mean square errors (RMSE), mean absolute errors (MAE) and correlation coefficient (R) statistics are used as comparing criteria for the evaluation of the models’ performances. Based on the comparisons, it was found that the ANFIS technique could be employed successfully in modeling discharge coefficient from the available experimental data. There are good agreements between the measured values and the values obtained using the ANFIS model. It is found that the ANFIS model with RMSE of 0.0699 in validation stage is superior in estimation of discharge coefficient than the multiple nonlinear and linear regression models with RMSE of 0.1019 and 0.1507, respectively.     

Design of hybrid differential evolution and group method of data handling networks for modeling and prediction

This paper proposes a hybrid modeling approach based on two familiar non-linear methods of mathematical modeling; the group method of data handling (GMDH) and differential evolution (DE) population-based algorithm. The proposed method constructs a GMDH self-organizing network model of a population of promising DE solutions. The new hybrid implementation is then applied to modeling tool wear in milling operations and also applied to two representative time series prediction problems of exchange rates of three international currencies and the well-studied Box-Jenkins gas furnace process data. The results of the proposed DE–GMDH approach are compared with the results obtained by the standard GMDH algorithm and its variants. Results presented show that the proposed DE–GMDH algorithm appears to perform better than the standard GMDH algorithm and the polynomial neural network (PNN) model for the tool wear problem. For the exchange rate problem, the results of the proposed DE–GMDH algorithm are competitive with all other approaches except in one case. For the Box-Jenkins gas furnace data, the experimental results clearly demonstrates that the proposed DE–GMDH-type network outperforms the existing models both in terms of better approximation capabilities as well as generalization abilities. Consequently, this self-organizing modeling approach may be useful in modeling advanced manufacturing systems where it is necessary to model tool wear during machining operations, and in time series applications such as in prediction of time series exchange rate and industrial gas furnace problems.     

ANN- and ANFIS-based multi-staged decision algorithm for the detection and diagnosis of bearing faults

Condition monitoring of roller element bearing defects has been one of the biggest problems in predictive maintenance since bearing failures may give catastrophic results on the machining operations and may cause downtime. Two of the well-established and widely used methods for bearing fault detection and diagnosis are the artificial neural network (ANN) and adaptive neuro-fuzzy inference systems (ANFIS). For this aim, a multi-staged decision algorithm was developed in this study based on ANN and ANFIS models. Both time and frequency domain parameters extracted from the vibration and current signals were used to train the ANN and ANFIS models, which were then used to detect and diagnose the severity of the bearing fault. Experimental data collected from a shaft-bearing mechanism were used to test the performances of the two schemes. The system was operated under four different speeds, for a brand-new bearing and bearings with artificially introduced local defects with various sizes. The experimental results showed that the proposed scheme is an effective means for detecting and diagnosing bearing faults. Furthermore, the results revealed that ANFIS-based scheme was superior to the ANN-based one especially in diagnosing fault severity.

     

Connectionist approaches for solubility prediction of n-alkanes in supercritical carbon dioxide

Carbon dioxide injection is a known promising and economical technology for improving oil recovery. Despite its immense effect on oil recovery, the application of this technique in modern recovery industry has been limited due to poor solubility of n-alkanes in supercritical CO₂. Therefore, it is very consequential to investigate the solubility of different n-alkanes in supercritical CO₂. Since experimental methods for measuring the solubility of n-alkanes in supercritical CO₂ at different temperatures and pressures are not economical and usually take a long time, feasibility of applying intelligent tools in the solubility prediction of different n-alkanes in supercritical CO₂ at pressures up to 45.9 MPa was conducted in this study. For this purpose, two models including an artificial neural network and an adaptive neuro-fuzzy interference system (ANFIS) both trained with particle swarm optimization (PSO) algorithm were used for simulating this process. Calculated mole fractions of n-alkanes in supercritical CO₂ from ANFIS–PSO model were excellently consistent with actual measured values. Moreover, comparison between these models and Chrastil semiempirical correlation show superiority and accuracy of the proposed ANFIS–PSO approach. Results of this study indicate that ANFIS–PSO method is a powerful technique for predicting solubility of n-alkanes in supercritical CO₂.

     

A new hybrid ANFIS–PSO model for prediction of peak particle velocity due to bench blasting

In this paper, a novel hybrid approach is proposed for predicting peak particle velocity (PPV) due to bench blasting in open pit mines. The proposed approach is based on the combination of adaptive neuro-fuzzy inference system (ANFIS) and particle swarm optimization (PSO). In this approach, the PSO is used to improve the performance of ANFIS. Furthermore, a model is developed based on support vector regression (SVR) approach. The models are trained and tested based on actual data compiled from 120 blast rounds in Sarcheshmeh copper mine. To determine the accuracy and efficiency of ANFIS–PSO and SVR models, a statistical model (USBM equation) is applied. According to the obtained results, both techniques can be used to predict the PPV, but the comparison of models shows that the ANFIS–PSO model provides better results. Root mean square error (RMSE), variance account for (VAF), and coefficient of determination (R ²) indices were obtained as 1.83, 93.37 and 0.957 for ANFIS–PSO model, respectively.     

The effect of ICA and PSO on ANN results in approximating elasticity modulus of rock material

Reliable determination/evaluation of the rock deformation can be useful prior any structural design application. Young’s modulus (E) affords great insight into the characteristics of the rock. However, its direct determination in the laboratory is costly and time-consuming. Therefore, rock deformation prediction through indirect techniques is greatly suggested. This paper describes hybrid particle swarm optimization (PSO)–artificial neural network (ANN) and imperialism competitive algorithm (ICA)–ANN to solve shortcomings of ANN itself. In fact, the influence of PSO and ICA on ANN results in predicting E was studied in this research. By investigating the related studies, the most important parameters of PSO and ICA were identified and a series of parametric studies for their determination were conducted. All models were built using three inputs (Schmidt hammer rebound number, point load index and p-wave velocity) and one output which is E. To have a fair comparison and to show the capability of the hybrid models, a pre-developed ANN model was also constructed to estimate E. Evaluation of the obtained results demonstrated that a higher ability of E prediction is received developing a hybrid ICA–ANN model. Coefficient of determination (R²) values of (0.952, 0.943 and 0.753) and (0.955, 0.949 and 0.712) were obtained for training and testing of ICA–ANN, PSO–ANN and ANN models, respectively. In addition, VAF values near to 100 (95.182 and 95.143 for train and test) were achieved for a developed ICA–ANN hybrid model. The results indicated that the proposed ICA–ANN model can be implemented better in improving performance capacity of ANN model compared to another implemented hybrid model.

     

Rock brittleness prediction through two optimization algorithms namely particle swarm optimization and imperialism competitive algorithm

Brittleness index (BI) is a significant rock parameter when dealing with projects performed in rocks. The main goal of this research work is to propose the novel practical models to predict the BI through particle swarm optimization (PSO) and imperialism competitive algorithm (ICA). For this aim, two forms of equations, i.e., linear and power are considered and the weights of these equations are optimized by PSO and ICA. In the other words, four predictive models, namely ICA linear, ICA power, PSO linear, and PSO power models are developed to predict BI in this study. In the modeling of the predictive models, 79 datasets are used, so that Schmidt hammer rebound number, wave velocity, density, and Point Load Index (Is₅₀) are selected as the independent (input) parameters and the BI values are considered as the dependent (output) parameter. Then, the performances of the proposed predicting models are checked using two error indices, namely coefficient correlation (R²) and root mean squared error (RMSE). The results showed that the PSO power model has superior fitting specification for the prediction of the BI compared to the other prediction models and is quite practical for use. As a result, linear and power models of PSO received higher performance prediction compared to ICA. PSO power (with R² train = 0.937, R² test = 0.959, RMSE train = 0.377 and RMES test = 0.289) showed the most powerful technique to predict BI of the granite samples.

     

Developing GEP tree-based,neuro-swarm,and whale optimization models for evaluation of bearing capacity of concrete-filled steel tube columns

The type of materials used in designing and constructing structures significantly affects the way the structures behave. The performance of concrete and steel, which are used as a composite in columns, has a considerable effect upon the structure behavior under different loading conditions. In this paper, several advanced methods were applied and developed to predict the bearing capacity of the concrete-filled steel tube (CFST) columns in two phases of prediction and optimization. In the prediction phase, bearing capacity values of CFST columns were estimated through developing gene expression programming (GEP)-based tree equation; then, the results were compared with the results obtained from a hybrid model of artificial neural network (ANN) and particle swarm optimization (PSO). In the modeling process, the outer diameter, concrete compressive strength, tensile yield stress of the steel column, thickness of steel cover, and the length of the samples were considered as the model inputs. After a series of analyses, the best predictive models were selected based on the coefficient of determination (R²) results. R² values of 0.928 and 0.939 for training and testing datasets of the selected GEP-based tree equation, respectively, demonstrated that GEP was able to provide higher performance capacity compared to PSO–ANN model with R² values of 0.910 and 0.904 and ANN with R² values of 0.895 and 0.881. In the optimization phase, whale optimization algorithm (WOA), which has not yet been applied in structural engineering, was selected and developed to maximize the results of the bearing capacity. Based on the obtained results, WOA, by increasing bearing capacity to 23436.63 kN, was able to maximize significantly the bearing capacity of CFST columns.

     

Application of adaptive neuro-fuzzy technique and regression models to predict the compressive strength of geopolymer composites

This article introduces an adaptive network-based fuzzy inference system (ANFIS) model and two linear and nonlinear regression models to predict the compressive strength of geopolymer composites. Geopolymers are highly complex materials which involve many variables which make modeling its properties very difficult. There is no systematic approach in the mix design for geopolymers. The amounts of silica modulus, Na₂O content, w/b ratios, and curing time have a great influence on the compressive strength. In this study, by developing and comparing parametric linear and nonlinear regressions and ANFIS models, we dealt with predicting the compressive strength of geopolymer composites for possible use in mix-design framework considering the mentioned complexities. ANFIS model developed by generalized bell-shaped membership function was recognized the best approach, and the prediction results of linear and nonlinear regression models as empirical methods showed the weakness of these models comparing ANFIS model.