Pre-developed Artificial Neural Network Model Research Articles

Application of hybrid intelligent systems in predicting the unconfined compressive strength of clay material mixed with recycled additive

A reliable prediction of the soil properties mixed with recycled material is considered as an ultimate goal of many geotechnical laboratory works. In this study, after planning and conducting a series of laboratory works, some basic properties of marine clay treated with recycled tiles together with their unconfined compressive strength (UCS) values were obtained. Then, these basic properties were selected as input variables to predict the UCS values through the use of two hybrid intelligent systems i.e., the neuro-swarm and the neuro-imperialism. Actually, in these systems, respectively, the weights and biases of the artificial neural network (ANN) were optimized using the particle swarm optimization (PSO) and imperialism competitive algorithm (ICA) to get a higher accuracy compared to a pre-developed ANN model. The best neuro-swarm and neuro-imperialism models were selected based on several parametric studies on the most important and effective parameters of PSO and ICA. Afterward, these models were evaluated according to several well-known performance indices. It was found that the neuro-swarm predictive model provides a higher level of accuracy in predicting the UCS of clay soil samples treated with recycled tiles. However, both hybrid predictive models can be used in practice to predict the UCS values for initial design of geotechnical structures.

Imperialist competitive algorithm hybridized with multilayer perceptron to predict the load-settlement of square footing on layered soils

To forecast the value of bearing capacity in shallow footings, a total of 2430 finite element modelling (FEM) simulation is performed. In this regard and to optimize the performance of the artificial neural network (ANN), it is combined with the imperialist competitive algorithm (ICA). The new combined technique is called ICA-MLP (multi-layer perceptron). To develop the ICA-MLP model, the input parameters were the soil type (i.e., having particular soil properties for each of the sandy soil types) installed at the top, the soil type installed at the bottom, the first-layer thickness ratio (h/B) and the applied stress on the footing (kPa), while the output was the vertical settlement (mm) under the square footing. The estimations were compared with a predeveloped ANN model to demonstrate the ability of the ICA-MLP hybrid model. The results showed a high ability of ICA metaheuristic ensembles for understanding the non-linear relationship between the influential factors and the selected target. Meanwhile, a comparison between the used models revealed that the best-combined structure is when the ICA algorithm is followed by the swarm size equal to 350. In this sense, the results from the predeveloped ANN model, based on R2 values, were 0.83 and 0.89 for the training and testing data sets, respectively, whereas the R2 and RMSE values for the ICA-MLP model for the training and testing datasets were 0.983, 0.062 and 0.977, 0.070, respectively. Therefore, the ICA-MLP model can be regarded as a new model that is superior to the conventional MLP technique.

Prediction of Lateral Deflection of Small-Scale Piles Using Hybrid PSO–ANN Model

Piles, as a type of geotechnical structures, are widely utilized to resist different lateral loads sources such as inclined loads and earthquakes. Therefore, the behavior of such structures under lateral loads needs to be investigated. Accordingly, this study examines the piles’ lateral deflection (LD) under various conditions. A total of 183 physical modeling tests were conducted in laboratory considering the most influential parameters on LD values in dried sandy soils. Additionally, a new hybrid model of particle swarm optimization (PSO)–artificial neural network (ANN) was proposed to predict LD of the piles. For comparison purposes, a pre-developed ANN model was also designed for estimation of LD values. In order to evaluate the prediction accuracy of the developed models, several performance indices such as root-mean-squared error (RMSE), coefficient of determination (R2), and variance account for were calculated. The proposed PSO–ANN model was found capable of providing a high accuracy level and, at the same time, a low system error in the LD prediction process. The RMSE values of 0.072 and 0.085 were determined, respectively, for training and testing datasets of the developed PSO–ANN model, while these values were 0.121 and 0.103 for the same datasets of the ANN predictive technique, respectively. It can be concluded that the PSO–ANN model can be relied on as a new hybrid model in field of this study, and also it can be used in other related studies with caution.

Developing a new intelligent technique to predict overbreak in tunnels using an artificial bee colony-based ANN

The drilling and blasting technique is among the common techniques for excavating tunnels with different shapes and sizes. Nevertheless, due to the dynamic energy involved, the rock mass around the excavation zone experiences damage and reduction in stiffness and strength. One of the most common and important issues that occurs during the tunneling process is the overbreak which is defined as the surplus drilled section of the tunnel. It seems that prediction of overbreak before blasting operations is necessary to minimize the possible damages. This paper develops a new hybrid model, namely, an artificial bee colony (ABC)–artificial neural network (ANN) to predict overbreak. Considering the most important parameters on overbreak, many ABC–ANN models were constructed based on their effective parameters. A pre-developed ANN model was also developed for comparison. In order to evaluate the obtained results of this study, a new system, i.e., the color intensity rating (CIR), was introduced and established to select the best ABC–ANN and ANN models. As a result, the ABC–ANN receives a high level of accuracy in predicting overbreak induced by drilling and blasting. The coefficients of determination (R2) for the ANN and ABC–ANN are 0.9121 and 0.9428, respectively, for training datasets. This revealed that the ABC–ANN model (as a new model in the field of this study) is the best one among the models developed in this study.

Prediction of rock interlocking by developing two hybrid models based on GA and fuzzy system

Rock shear strength parameters (interlocking and internal friction angel) are considered as significant factors in the designing stage of various geotechnical structures such as tunnels and foundations. Direct determination of these parameters in laboratory is time-consuming and expensive. Additionally, preparation of good quality of core samples is sometimes difficult. The objective of this paper is introducing and evaluating two hybrid artificial neural network (ANN)-based models by considering genetic algorithm (GA) and fuzzy inference system for prediction of interlocking of shale rock samples. Therefore, hybrid GA-ANN and adoptive neuro-fuzzy inference system (ANFIS) were developed and to show the capability of the hybrid models, the predicted results were compared to those of a pre-developed ANN model. In development of these models, the results of rock index tests, i.e., point load index, dry density, p-wave velocity, Brazilian tensile strength and Schmidt hammer were taken into account as the input parameters, whereas the interlocking of the shale samples was set as the output. The results obtained in this study confirmed the high reliability of the developed hybrid models, however, ANFIS predictive model receives slightly higher performance prediction compared to GA-ANN technique. The obtained results of the developed models were (0.865, 0.852), (0.933, 0.929) and (0.957, 0.965) for ANN, GA-ANN and ANFIS models, respectively, based on coefficient of determination (R2). ANFIS can be introduced as an innovative model to the field of rock mechanics.

Rock Fragmentation Prediction through a New Hybrid Model Based on Imperial Competitive Algorithm and Neural Network

Reliable estimation of rock fragmentation is an important issue in the blasting operations in order to predict quality of the production. Since rock fragmentation is affected by various parameters such as blast pattern and rock mass characteristics, it is very difficult to have an appreciate prediction of it. This paper describes a new hybrid imperialism competitive algorithm (ICA)-artificial neural network (ANN) in order to solve shortcomings of ANN itself for prediction of rock fragmentation. In fact, the influence of ICA on ANN results was studied in this research. By investigating the related studies, the most important parameters of ICA were identified and a series of parametric studies for their determination were conducted. All models were built using 8 inputs and one output which is rock fragmentation. To have a fair comparison and show the capability of the new hybrid model, a pre-developed ANN model was also considered and constructed. Evaluation of the obtained results demonstrated that a higher ability of rock fragmentation prediction is received developing a hybrid ICA-ANN model. Coefficient of determination (R2) values of (0.949 and 0.813) and (0.941 and 0.819) were obtained for training and testing of ICA-ANN and ANN models, respectively which indicated that the proposed ICA-ANN model can be implemented better in improving performance capacity of ANN model in estimating rock fragmentation.

Rock Fragmentation Prediction through a New Hybrid Model Based on Imperial Competitive Algorithm and Neural Network

Reliable estimation of rock fragmentation is an important issue in the blasting operations in order to predict quality of the production. Since rock fragmentation is affected by various parameters such as blast pattern and rock mass characteristics, it is very difficult to have an appreciate prediction of it. This paper describes a new hybrid imperialism competitive algorithm (ICA)-artificial neural network (ANN) in order to solve shortcomings of ANN itself for prediction of rock fragmentation. In fact, the influence of ICA on ANN results was studied in this research. By investigating the related studies, the most important parameters of ICA were identified and a series of parametric studies for their determination were conducted. All models were built using 8 inputs and one output which is rock fragmentation. To have a fair comparison and show the capability of the new hybrid model, a pre-developed ANN model was also considered and constructed. Evaluation of the obtained results demonstrated that a higher ability of rock fragmentation prediction is received developing a hybrid ICA-ANN model. Coefficient of determination (R2) values of (0.949 and 0.813) and (0.941 and 0.819) were obtained for training and testing of ICA-ANN and ANN models, respectively which indicated that the proposed ICA-ANN model can be implemented better in improving performance capacity of ANN model in estimating rock fragmentation.

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.

Proposing a new model to approximate the elasticity modulus of granite rock samples based on laboratory tests results

An accurate examination of deformability of rock samples in response to any change in stresses is deeply dependent on the reliable determination of properties of the rock as analysis inputs. Although Young’s modulus (E) can provide valuable characteristics of the rock material deformation, the direct determination of E is considered a time-consuming and complicated analysis. The present study is aimed to introduce a new hybrid intelligent model to predict the E of granitic rock samples. Hence, a series of granitic block samples were collected from the face of a water transfer tunnel excavated in Malaysia and transferred to laboratory to conduct rock index tests for E prediction. Rock index tests including point load, p-wave velocity and Schmidt hammer together with uniaxial compressive strength (UCS) tests were carried out to prepare a database comprised of 62 datasets for the analysis. Results of simple regression analysis showed that there is a need to develop models with multiple inputs. Then, a hybrid genetic algorithm (GA)-artificial neural network (ANN) model was developed considering parameters with the most impact on the GA. In order to have a fair evaluation, a predeveloped ANN model was also performed to predict E of the rock. As a result, a GA-ANN model with a coefficient of determination (R2) of 0.959 and root mean square error (RMSE) of 0.078 for testing datasets was selected and introduced as a new model for engineering practice; the results obtained were 0.766 and 0.098, respectively, for the developed ANN model. Furthermore, based on sensitivity analysis results, p-wave velocity has the most effect on E of the rock samples.

Optimizing an ANN model with ICA for estimating bearing capacity of driven pile in cohesionless soil

The application of models provided by artificial neural network (ANN) in predicting bearing capacity of driven pile is underlined in several investigations. However, weakness of ANN in slow rate of convergence as well as finding reliable testing output is known to be the major drawbacks of implementing ANN-based techniques. The present study aims to introduce and evaluate an optimized ANN with imperialism competitive algorithm (ICA) model based to estimate bearing capacity of driven pile in cohesionless soil. The training data for optimizing the ICA-ANN structure are based on the in situ study. To develop the ICA-ANN model, the input parameters are internal friction angle of soil located in shaft (φ shaft), and tip (φ tip), pile length (L), effective vertical stress at pile toe (σ v), and pile area (A) while the output is the total driven pile bearing capacity in cohesionless soil. The predicted results are compared with a pre-developed ANN model to demonstrate the ability of the hybrid model. As a result, coefficient of determination (R 2) values of (0.885 and 0.894) and (0.964 and 0.974) was obtained for testing and training datasets of ANN and ICA-ANN models, respectively. In addition, values of variance account for (VAF) of (88.212 for training and 89.215 for testing) and (96.369 for training and 97.369 for testing, respectively) were obtained for ANN and ICA-ANN models, respectively. The obtained results declare high reliability of the developed ICA-ANN model. This model can be introduced as a new model in field of deep foundation engineering.

An expert system based on hybrid ICA-ANN technique to estimate macerals contents of Indian coals

Coal, as an initial source of energy, requires a detailed investigation in terms of ultimate analysis, proximate analysis, and its biological constituents (macerals). The rank and calorific value of each type of coal are managed by the mentioned properties. In contrast to ultimate and proximate analyses, determining the macerals in coal requires sophisticated microscopic instrumentation and expertise. This study emphasizes the estimation of the concentration of macerals of Indian coals based on a hybrid imperialism competitive algorithm (ICA)–artificial neural network (ANN). Here, ICA is utilized to adjust the weight and bias of ANNs for enhancing their performance capacity. For comparison purposes, a pre-developed ANN model is also proposed. Checking the performance prediction of the developed models is performed through several performance indices, i.e., coefficient of determination (R 2), root mean square error and variance account for. The obtained results revealed higher accuracy of the proposed hybrid ICA-ANN model in estimating macerals contents of Indian coals compared to the pre-developed ANN technique. Results of the developed ANN model based on R 2 values of training datasets were obtained as 0.961, 0.955, and 0.961 for predicting vitrinite, liptinite, and inertinite, respectively, whereas these values were achieved as 0.948, 0.947, and 0.957, respectively, for testing datasets. Similarly, R 2 values of 0.988, 0.983, and 0.991 for training datasets and 0.989, 0.982, and 0.985 for testing datasets were obtained from developed ICA-ANN model.

An optimized ANN model based on genetic algorithm for predicting ripping production

Due to the environmental constraints and the limitations on blasting, ripping as a ground loosening and breaking method has become more popular in both mining and civil engineering applications. As a result, a more applicable rippability model is required to predict ripping production (Q) before conducting such tests. In this research, a hybrid genetic algorithm (GA) optimized by artificial neural network (ANN) was developed to predict ripping production results obtained from three sites in Johor state, Malaysia. It should be noted that the mentioned hybrid model was first time applied in this field. In this regard, 74 ripping tests were investigated in the studied areas and the relevant parameters were also measured. A series of GA–ANN models were conducted in order to propose a hybrid model with a higher accuracy level. To demonstrate the performance capacity of the hybrid GA–ANN model, a pre-developed ANN model was also proposed and results of predictive models were compared using several performance indices. The results revealed higher accuracy of the proposed hybrid GA–ANN model in estimating Q compared to ANN technique. As an example, root-mean-square error values of 0.092 and 0.131 for testing datasets of GA–ANN and ANN techniques, respectively, express the superiority of the newly developed model in predicting ripping production.

Ground vibration prediction in quarry blasting through an artificial neural network optimized by imperialist competitive algorithm

This paper presents a new hybrid artificial neural network (ANN) optimized by imperialist competitive algorithm (ICA) to predict peak particle velocity (PPV) resulting from quarry blasting. For this purpose, 95 blasting works were precisely monitored in a granite quarry site in Malaysia and PPV values were accurately recorded in each operation. Furthermore, the most influential parameters on PPV were measured and used to train the ICA-ANN model. Considering the measured data from the granite quarry site, a new empirical equation was developed to predict PPV. For comparison, a pre-developed ANN model was developed for PPV prediction. The results demonstrated that the proposed ICA-ANN model is able to predict blasting-induced PPV better than other presented techniques.

River Bedup Catchment Water Level Prediction Using Pre-developed ANN Model of Siniawan Catchment

This study proposes the application of Artificial Neural Network (ANN) in the prediction of hourly water level under tidal influence for Sadong Basin. An ANN is undoubtedly a robust tool for forecasting various non-linear hydrologic processes, including the water level prediction. It is a flexible mathematical structure which is capable to generalize patterns in imprecise or noisy and ambiguous input and output data sets. In this study, ANN models were developed specifically to forecast the hourly water level for River Bedup Station. Distinctive networks were trained, validated and simulated using hourly data obtained from Department of Irrigation and Drainage, Sarawak in Kuching. The performances of ANN were evaluated based on the coefficient of efficiency, E2 and the coefficient of correlation, R. The back propagation algorithm was adopted for this study. Models used in this study is trained, validated and simulated with scaled conjugate gradient algorithm (trainscg) with two hours of antecedent data, learning rate and the number of neurons in the hidden layer of 0.8 and 40 respectively. In this study, the models generated an accuracy of 100% for all training, validating and simulating stages. It has been found that the ANN has the potential to solve the problems of water level prediction.