Adaptive Genetic Programming Research Articles

Insights into modelling and evaluation of thermodynamic and transport properties of refrigerants using machine-learning methods

The thermophysical properties of refrigerating systems should be accurately understood for designing low-temperature refrigeration cycles of economic acceptance. The present work has tried to simplify this complicated procedure by proposing reliable and new correlative methods for determining thermodynamic and transport properties of four refrigerating substance classes, namely halocarbon, inorganic, hydrocarbon, and cryogenic fluids. New machine learning methods e.g., particle swarm optimisation adaptive neuro-fuzzy inference system (PSO-ANFIS), genetic programming (GP), and hybrid adaptive neuro-fuzzy inference system (Hybrid ANFIS) algorithms were utilised. The development of a new, simple and comprehensive correlation was for the first time introduced to estimate saturated vapour enthalpy, entropy, velocity of sound, and viscosity of refrigerants without having in-depth knowledge of complicated parameters. The accuracy and validity of the proposed models were assessed using a variety of statistical and graphical demonstrations. The findings were compared, and it was found that Hybrid ANFIS models are more accurate because Absolute Average Relative Errors (%AARD) for enthalpy, entropy, the velocity of sound, and viscosity were estimated as 0.5558, 1.3105, 0.5215, and 1.5727 in respective order. In addition, the proposed models' results were compared to the results of recently previously published models, and it confirms the reliability of our results. The innovation of this research is the design of reliable correlative methods having elevated precisions for thermodynamic and transport specifications of refrigerating substances.

Education Teaching Evaluation Method Aided by Adaptive Genetic Programming and Robust Scheduling

Process mining technology aims to automatically generate process models by analyzing events, thereby assisting the design and redesign of process models. Although many process mining methods have appeared, they all have deficiencies. These methods focus on mining from the behavioral aspects described by the log, while ignoring the structural nature of the process model itself. The complexity of the process describes the simplicity and ease of understanding of the process. Higher process complexity affects the readability of the process. Genetic programming has strong robustness. Its individual representation based on tree structure can describe the special structure of the process. The introduction of process complexity fitness enables it to consider the complexity of the process model itself while mining log behavior, so as to achieve nonlinear mining of complex processes. This paper analyzes the process mining based on genetic programming and proposes process individuals based on tree structure. By realizing the combination of process complexity measurement and process mining technology, noncomplex process mining can be realized. In this paper, the process complexity is combined with the process mining algorithm based on genetic programming, and a measure of process structural complexity is proposed, which is converted into complexity fitness and introduced into the fitness function of genetic programming, to realize the use of genetic programming. The research results show that the improved new adaptive genetic programming robust scheduling algorithm provides a new idea and method for the evaluation of college sports under different sports conditions. This makes the college sports evaluation management system more intelligent and improves the rational allocation of physical education teaching.

Prediction of welding responses using AI approach: adaptive neuro-fuzzy inference system and genetic programming

Laser welding of thin sheets has widespread application in various fields such as battery manufacturing, automobiles, aviation, electronics circuits and medical sciences. Hence, it is very essential to develop a predictive model using artificial intelligence in order to achieve high-quality weldments in an economical manner. In the present study, two advanced artificial intelligence techniques, namely adaptive neuro-fuzzy inference system (ANFIS) and multi-gene genetic programming (MGGP), were implemented to predict the welding responses such as heat-affected zone, surface roughness and welding strength during joining of thin sheets using Nd:YAG laser. The study attempts to develop an appropriate predictive model for the welding process. In the proposed methodology, 70% of the experimental data constitutes the training set whereas remaining 30% data is used as testing set. The results of this study indicated that the root-mean-square error (RMSE) of tested data set ranges between 7 and 16% for MGGP model, while RMSE for testing data set lies 18–35% for ANFIS model. The study indicates that the MGGP predicts the welding responses in a superior manner in laser welding process and can be applied for accurate prediction of performance measures.

Prediction of Local Scour around Bridge Piers Using Hierarchical Clustering and Adaptive Genetic Programming

ABSTRACT The physics of local scour around bridge piers is fairly complex because of multiple forces acting on it. Existing empirical formulas cannot cover all scenarios and soft computing methods require ever greater amounts of data to cover all cases with a single formula or a neural network. The approach proposed in this study brings together observations from over 40 studies, grouping similar observations with hierarchical clustering, and using genetic programming with adaptive operators to evolve formulas specific to each cluster to predict the scour depth. The resulting formulas are made available along with a basic web-based user interface that finds the closest cluster for newly presented data and finds the scour depth using the formula for that cluster. All formulas have R2 scores over 0.8 and have been validated with validation and testing sets to reduce overfitting. When compared to existing empirical formulas, the generated formulas consistently record higher R2 scores.

Predictive modelling of piezometric head and seepage discharge in earth dam using soft computational models.

Predictions of pore pressure and seepage discharge are the most important parameters in the design of earth dams and assessing their safety during the operational period as well. In this research, soft computing models namely multi-layer perceptron neural network (MLPNN), support vector machine (SVM), multivariate adaptive regression splines (MARS), genetic programming (GP), M5 algorithm, and group method of data handling (GMDH) were used to predict the piezometric head in the core and the seepage discharge through the body of earth dam. For this purpose, the data recorded by the absolute instrument during the last 94 months of Shahid Kazemi Bukan Dam were used. The results showed that all of the applied models had a permissible level of accuracy in the prediction of the piezometric heads. The average error indices for the models in the training phase were R2= 0.957 and RMSE= 0.806 and in the testing phase were equal to R2= 0.949 and RMSE= 0.932, respectively. The performances of all models except the M5 and MARS in predicting seepage discharge are nearly identical; however, the best is the MARS, and the weakest is the M5 algorithm.

Predicting elastic modulus degradation of alkali silica reaction affected concrete using soft computing techniques: A comparative study

Alkali silica reaction (ASR) is a harmful distress mechanism which results in expansion and reduction of mechanical properties of concrete. The latter may cause loss of serviceability and load carrying capacity of affected concrete structures. Influences of ASR on concrete are known to be complex in nature, for which the traditional empirical and curve-fitting approaches are insufficient to provide adequate models to capture such complexity. Recent advancement in soft computing (SC) offers a new tool for tackling the complexity of ASR affected concrete. Most of previous experimental studies agreed that as a result of ASR, the elastic modulus suffers a significant reduction compared with other properties such as compressive and tensile strength of the affected concrete. In this study, an investigation has been conducted, utilising different SC models to quantify ASR-induced elastic modulus degradation of unrestrained concrete. Five SC techniques, namely support vector machine (SVM), artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS), M5P model and genetic expression programming (GEP), are investigated comparatively in this research. The models, on basis of SC techniques, are developed and tested using a comprehensive dataset collected from existing publications. In order to demonstrate the superiorities of SC techniques, the proposed approaches are compared to several empirical models developed using same dataset. The comparative results show that the developed SC models outperform empirical models in a wide range of evaluation indices, which indicates promising applications of the proposed approach.

Comparative evaluation of different statistical tools for the prediction of uniaxial compressive strength of rocks

In this study, uniaxial compressive strength (UCS), unit weight (UW), Brazilian tensile strength (BTS), Schmidt hardness (SHH), Shore hardness (SSH), point load index (Is50) and P-wave velocity (Vp) properties were determined. To predict the UCS, simple regression (SRA), multiple regression (MRA), artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS) and genetic expression programming (GEP) have been utilized. The obtained UCS values were compared with the actual UCS values with the help of various graphs. Datasets were modeled using different methods and compared with each other. In the study where the performance indice PIat was used to determine the best performing method, MRA method is the most successful method with a small difference. It is concluded that the mean PIat equal to 2.46 for testing dataset suggests the superiority of the MRA, while these values are 2.44, 2.33, and 2.22 for GEP, ANFIS, and ANN techniques, respectively. The results pointed out that the MRA can be used for predicting UCS of rocks with higher capacity in comparison with others. According to the performance index assessment, the weakest model among the nine model is P7, while the most successful models are P2, P9, and P8, respectively.

Evolving new strategies to estimate reservoir oil formation volume factor: Smart modeling and correlation development

The precise estimation of oilfield PVT properties is of primary importance for improving field evaluation and development strategies. In the present work, adaptive network-based fuzzy inference system (ANFIS) and genetic programming (GP) models were established for the accurate assessment of reservoir oil formation volume factor (OFVF). A large database including nearly 1200 datapoints from all over the globe was collected for modeling. The performance of the new models was compared to correlations in the existing literature via several graphical and statistical paradigms. Variance approach analysis was implemented to determine the sensitivity of the target parameter as the output with respect to each input variable. It is shown that the highest forecasting accuracy is attained using the developed ANFIS model and GP-based empirical correlation with an average absolute relative deviation (AARD%) of 1.8% and 2.1%, respectively, in comparison to the results obtained from other approaches described in the literature. Lastly, the mathematical techniques designed for estimating OFVF can be of high importance for experts working with reservoir simulation and PVT analysis.

Thermal performance of thin film heat gauges of gold, silver and nano-composite

Thin film materials along with nanotechnology can be a useful tool to efficiently estimate the surface heat flux. The advancement of soft computing methods can also be used for automated estimation of thermal performances of various heat gauges. Thus, in this study, thin film gauges (TFG) made from Ag, Au, CNT/Ag nano-composite topped on insulating substrate have been tested experimentally under impulsive heat loads from laser. The fabrication of TFGs was done using paint method and physical vapor deposition. The fabricated TFGs were calibrated first and then surface heat fluxes were calculated analytically from temperature history by utilizing one-dimensional heat conduction theory applied on semi-infinite solids. The advanced soft computing methods viz. Adaptive Neuro Fuzzy Inference System (ANFIS) and Genetic Programming (GP) were used to estimate heat fluxes for these three TFGs. The statistical evaluation of GP with ANFIS models on the heat flux experimental database reveals that the ANFIS models have performed better in predicting the peak heat fluxes whereas the GP models have performed better in capturing the trend of heat fluxes. This combined approach paved a way to automated estimation of heat fluxes for industrial applications.

Rigorous prognostication of permeability of heterogeneous carbonate oil reservoirs: Smart modeling and correlation development

Permeability estimation has a major role in mapping quality of the reservoir, reservoir engineering calculation, reserve estimation, numerical reservoir simulation and planning for the drilling operations. In carbonate formations, it is of great challenge to predict permeability by reason of natural heterogeneity, nonuniformity of rock, complexity and nonlinearity of parameters. Various approaches have been developed for measuring/predicting this parameter, which are associated with high expenditures, time consuming processes and low accuracy. In this study, comprehensive efforts have been made to the development of radial basis function neural network (RBF-ANN), multilayer perceptron neural network (MLP-ANN), least square support vector machine (LSSVM), adaptive neuro-fuzzy inference system (ANFIS), genetic programming (GP), and committee machine intelligent system (CMIS). For this purpose, a widespread databank of 701 core permeability datapoints as a function of well log data was adopted from the open literature for heterogonous formations. Moreover, several optimization techniques like genetic algorithm (GA), particle swarm optimization (PSO), and levenberg marquardt (LM) were employed to enhance the prediction capability of the proposed tools in this study. For assessing the models efficiency, several tools like crossplot and error distribution diagram were applied in association with statistical calculation. As a result, the CMIS model is identified as the most accurate model with the highest determination coefficient (R2 near to unity) and the lowest root mean square error (RMSE near to zero). As a result of GP mathematical strategy, a new user-friendly empirically-derived correlation was developed for rapid and accurate estimation of reservoir permeability. The outcome of outlier detection shows the validity of dataset used for modeling, and the effective porosity is perceived to be the most affecting parameter on the permeability estimation in terms of sensitivity analysis. The main novelty of this modeling study was the proposal of CMIS and GP-based empirically-derived models for the first time in literature. To this end, the outcome of this study can be of great value for reservoir engineers dealing with simulation and characterization of the heterogonous carbonate reservoirs.

Artificial intelligence design charts for predicting friction capacity of driven pile in clay

In this study, five nonlinear prediction tools are used to model and predict the friction capacity of driven piles installed in clay including classical support vector machine (SVM) and two of its variants, namely regularized generalized proximal SVM and twin SVM, adaptive neuro-fuzzy inference system (ANFIS) and genetic programming (GP). The undrained shear strength, effective vertical stress, pile diameter and pile length are taken as the input parameters of the developed models, and the friction capacity is considered as the output. A total of 80 experimental observations are collected and used to train and test several models estimating the friction capacity of a driven pile in clay. The results demonstrate that, noting the root-mean-square error (RMSE) value, for prediction of the friction capacity of driven piles in clay the ANFIS model gives a better convergence to the in situ results, compared with the GP and SVM models. The developed ANFIS model provided a simple and reliable design structure for proper selection of the pile friction capacity of driven piles installed in clay. Furthermore, a simple mathematical formula is presented based on the GP model. The predicted results are compared with in situ data set models to demonstrate the abilities of the AI models. In order to perform a model evaluation, in addition to RMSE, the regression coefficient of determination is obtained through testing and training of the SVM, ANFIS and GP models. The results show high reliability for the developed models. The presented ANFIS and GP models are introduced as new models in the field of geotechnical engineering.

Increasing the accuracy of loop detector counts using adaptive neural fuzzy inference system and genetic programming

ABSTRACTLoop detectors are devices that are most commonly used for obtaining data at intersections. Multiple detectors are usually required to monitor a location, and this reduces the accuracy of detectors for collecting traffic volumes. The purpose of this paper is to increase the accuracy of loop detector counts using Adaptive Neural Fuzzy Inference System (ANFIS) and Genetic Programming (GP) based on detector volume and occupancy. These methods do not need microscopic analysis and are easy to employ. Four approaches for one intersection are used in a case study. Results show that the models can improve intersection detector counts significantly. Results also show that ANFIS produces more accurate counts compared to regression and GP.

Study of effect of nanofluid concentration on response characteristics of machining process for cleaner production

With the ever-increasing concern for reducing environmental pollution and waste minimization, “green manufacturing” has been successful to draw sufficient amount of attention towards it. Minimum Quantity Lubrication (MQL) is one such technique that has revolutionized the manufacturing industry by not only reducing the amount of working fluid dramatically but also enhancing cutting tool life and reducing material costs. Past studies have reported the use of experiments in MQL based manufacturing but limited computational modeling for optimizing the process parameters Based on the past experimental procedure of machining process (micro-drilling), a computational framework such as Adaptive Neuro Fuzzy Inference System (ANFIS) and Genetic Programming (GP) in quantification of three response characteristics (torque, thrust forces and material removal rate (MRR) is proposed. The performance analysis based on the cross-validation, error metrics, curve fitting and hypothesis tests reveals that among the two models, the GP models have performed better. 2-D and 3-D surface analysis were performed to validate the robustness of the models. Among the three response characteristics, It was found that the nanofluid concentration influences torque the most, which is an important aspect for power consumption. At nanofluid concentration values of 1.4 and 4, the minimum values of torque and thrust forces is achieved respectively. When drill diameter is minimum and the spindle speed is maximum, the values of torque, thrust forces and MRR is the lowest. Thus, the feed rate, nanofluid concentration and drill diameter are most critical for obtaining higher MRR and lower values of torque and thrust force, thus enabling cleaner production and environment.

Computational model for predicting the effect of process parameters on surface characteristics of mass finished components

Purpose Mass finishing is a commonly employed surface finishing process for improving surface characteristics of aerospace engineering components. Optimization of surface characteristics of such critical components require an explicit computational model that can describe the surface characteristics of the finished component. This paper aims to develop an explicit computational model that can describe the surface roughness as a function of various process parameters which influence the mass finishing process. Design/methodology/approach In the present work, the authors propose to study the roughness characteristics using a combined evolutionary computing approach based on Multi-Adaptive Regression Splines (MARS) and Genetic Programming (GP) techniques. Findings The authors conducted sensitivity and parametric analysis to capture the dynamics of surface characteristics by unveiling dominant input variables and hidden non-linear relationships. It is found that by regulating the process time and media size, a greatest variation in surface finish reduction can be achieved in mass finishing process. Originality/value To the best of authors knowledge, for the first time a hybrid evolutionary computational technique has been proposed in this work. The authors combined two powerful evolutionary techniques, namely Multi-variate Adaptive Regressive Splines and Genetic Programming approach. The proposed approach was able to capture the dynamics of surface roughness with higher accuracy as comparable to that of the experiments.

Balancing exploration and exploitation in genetic programming using inversion with individualized self-adaptation

In this article we explore and develop a holistic scheme of self adaptive, individualized genetic operators combined with an adaptive tournament size together with a novel implementation of an inversion genetic operator which is suitable for tree based Genetic Programming. We test this scheme on several benchmark Binary Classification problems and find that the proposed techniques deliver superior performance when compared with both a tuned GP configuration and a feedback adaptive Genetic Programming implementation. Our results also demonstrate that an inversion operator may have a useful role to play in exploitation, particularly when used towards the end of evolution to facilitate gradual convergence of the learning system towards a good solution.

Permeability estimation in heterogeneous oil reservoirs by multi-gene genetic programming algorithm

Abstract Permeability estimation has a significant impact on petroleum fields operation and reservoir management. Different methods were proposed to measure this parameter, which some of them are inaccurate, and some others such as core analysis are cost and time consuming. Intelligent techniques are powerful tools to recognize the possible patterns between input and output spaces, which can be applied to predict reservoir parameters. This study proposed a new approach based on multi-gene genetic programming (MGGP) to predict permeability in one of the heterogeneous oil reservoirs in Iran. The MGGP model with artificial neural networks (ANNs), adaptive neuro-fuzzy inference system (ANFIS) and genetic programming (GP) model were used to predict the permeability and obtained results were compared statistically. The comparison of results showed that the MGGP model can be applied effectively in permeability prediction, which gives low computational time. Furthermore, one equation based on the MGGP model using well log and core experimental data was generated to predict permeability in porous media.

Genetic Programming in Groundwater Modeling

AbstractDetermination of water-table elevation corresponding to aquifer recharge or discharge is an important issue in sustainable groundwater management. This approach necessitates the application of numerical simulation models that may require substantial amounts of input data related to aquifer parameters and specifications, even for modeling only part of the aquifer, which makes the calculations expensive. Moreover, comprehensive aquifer modeling is a time-consuming and computationally intensive process. Artificial intelligence tools can replace simulation models and decrease computational efforts by using input and output data sets without considering complex relations of the system to be modeled. This paper employs an adaptive neural fuzzy inference system (ANFIS) and genetic programming (GP) as artificial intelligence tools to extract governing groundwater flow equations in Ghaen and Karaj aquifers in Iran. For both aquifers, several input-output data sets, for both training and testing data sets, ...

Effects of infill walls on RC buildings under time history loading using genetic programming and neuro-fuzzy

In this study, the efficiency of adaptive neuro-fuzzy inference system (ANFIS) and genetic expression programming (GEP) in predicting the effects of infill walls on base reactions and roof drift of reinforced concrete frames were investigated. Current standards generally consider weight and fundamental period of structures in predicting base reactions and roof drift of structures by neglecting numbers of floors, bays, shear walls and infilled bays. Number of stories, number of bays in x and y directions, ratio of shear wall areas to the floor area, ratio of bays with infilled walls to total number bays and existence of open story were selected as parameters in GEP and ANFIS modeling. GEP and ANFIS have been widely used as alternative approaches to model complex systems. The effects of these parameters on base reactions and roof drift of RC frames were studied using 3D finite element method on 216 building models. Results obtained from 3D FEM models were used to in training and testing ANFIS and GEP models. In ANFIS and GEP models, number of floors, number of bays, ratio of shear walls and ratio of infilled bays were selected as input parameters, and base reactions and roof drifts were selected as output parameters. Results showed that the ANFIS and GEP models are capable of accurately predicting the base reactions and roof drifts of RC frames used in the training and testing phase of the study. The GEP model results better prediction compared to ANFIS model.

Prediction and simulation of monthly groundwater levels by genetic programming

Groundwater level is an effective parameter in the determination of accuracy in groundwater modeling. Thus, application of simple tools to predict future groundwater levels and fill-in gaps in data sets are important issues in groundwater hydrology. Prediction and simulation are two approaches that use previous and previous-current data sets to complete time series. Artificial intelligence is a computing method that is capable to predict and simulate different system states without using complex relations. This paper investigates the capability of an adaptive neural fuzzy inference system (ANFIS) and genetic programming (GP) as two artificial intelligence tools to predict and simulate groundwater levels in three observation wells in the Karaj plain of Iran. Precipitation and evaporation from a surface water body and water levels in observation wells penetrating an aquifer system are used to fill-in gaps in data sets and estimate monthly groundwater level series. Results show that GP decreases the average value of root mean squared error (RMSE) as the error criterion for the observation wells in the training and testing data sets 8.35 and 11.33 percent, respectively, compared to the average of RMSE by ANFIS in prediction. Similarly, the average value of RMSE for different observation wells used in simulation improves the accuracy of prediction 9.89 and 8.40 percent in the training and testing data sets, respectively. These results indicate that the proposed prediction and simulation approach, based on GP, is an effective tool in determining groundwater levels.

Modeling of Water Quality Parameters Using Data-Driven Models

Water has a considerable role in all aspects of human life. Thus, evaluation of water characteristics in general and water quality in particular are necessary to enhance the health of humans and ecosystems. Data-driven models are computing methods that are capable of extracting different system states without using complex relationships. Prediction and simulation are two branches of data-driven modeling that use previous and previous-current data sets to fill gaps in time series. This paper investigates the capability of an adaptive neural fuzzy inference system (ANFIS) and genetic programming (GP) as two data-driven models to predict and simulate water quality parameters (e.g., sodium, potassium, magnesium, sulfates, chloride, pH, electrical conductivity, and total dissolved solids) at the Astane station in Sefidrood River, Iran. The writers considered six combinations of data sets, including the previously noted water quality parameters and river discharge in the previous and previous-current months, as input data. Implementation of the ANFIS and GP models in this paper illustrates the flexibility of GP in time series modeling relative to ANFIS, especially in the testing data set. Accordingly, the writers calculated the coefficient of variation of root mean squared error as the error criterion in different ANFIS and GP models (for assigning achievement probability to an appropriate solution) for each quality parameter. The average of the previously noted values for the six combinations of data sets improved (decreased) 80.51 and 80.89%, respectively, in the training and testing data sets with GP relative to ANFIS. These results indicate that the writers’ proposed modeling, based on GP, is an effective tool for determining water quality parameters.