Adaptive Neuro-fuzzy Interface System Model Research Articles

Development of an intelligent system based on ANFIS model for predicting soil erosion

This study explores capability of adaptive neuro-fuzzy interface system (ANFIS) model for soil erosion assessment of which various parameters are involved and the parameters interaction is highly nonlinear. Soft computing technique has been applied widely in various fields, but its application in soil science is limited. A number of models have been established in predicting soil erosion such as a revised universal soil loss equation (RUSLE) model whereby parameters of the RUSLE model are employed in testing the ANFIS model in this study. Besides, a new parameter of biomass microbe (Mf) is introduced and its effectiveness in controlling erosion is demonstrated. A hilly area near the Guthrie Corridor Expressway, Malaysia, is chosen as a study area whereby five experimental plots size of 8 × 8 m and 5 × 5 m are developed. These five plots are characterized by either natural or planted vegetation or either with or without microbe bio-fertilizer. Input parameters of RUSLE, i.e. rainfall erosivity, soil erodibility, slope length and steepness, vegetation and support practice, are measured, and observed erosion is monitored for 1 year. The microbe factor presented as Mf is examined by pairing the plot with and without microbe while keeping other similar characteristic for those plots. Result shows a satisfactory achievement between observed and predicted model by ANFIS with R2, RMSE and MAE being 0.8275, 1.4276 and − 0.0165, respectively. In terms of a relationship between the Mf factor and soil erosion, it is noticed that the soil erosion is inversely proportional to the Mf when the entire erosion conservation practices are unchanged for a given period. Therefore, the application of microbes could be an alternative or additional erosion conservation technique in soil conservation works.

Application of ANN and ANFIS for Predicting the Ultimate Bearing Capacity of Eccentrically Loaded Rectangular Foundations

Extensive laboratory model tests were conducted on a rectangular embedded foundation resting over homogeneous sand bed and subjected to an eccentric load to determine the ultimate bearing capacity. Tests were conducted for foundations with width-to-length ratios (B/L) of zero (strip case), 0.333, 0.5, and 1. The depth of embedment varies from 0 to 1B with an increment of 0.5B; where B is the width of foundation and the eccentricity ratio (e/B) varies from 0 to 0.15 with an increment of 0.05. Based on the laboratory model test results, two different approaches are proposed to determine the ultimate bearing capacity. Firstly, a neural network model is developed to estimate the reduction factor. The reduction factor can be used to estimate the ultimate bearing capacity of an eccentrically loaded foundation from the ultimate bearing capacity of a centrally loaded foundation. A thorough sensitivity analysis was carried out to determine the important parameters affecting the reduction factor. Importance was given to the construction of neural interpretation diagram. Based on this diagram, whether direct or inverse relationships exist between the input and output parameters were determined. Secondly, an adaptive neuro-fuzzy interface system (ANFIS) is used to predict the ultimate bearing capacity. The neuro-fuzzy models combine the transparent, linguistic representation of a fuzzy system with learning ability of artificial neural networks (ANNs). The results from the ANN and ANFIS were compared with the laboratory model test results. It is clearly seen that the performance of the ANFIS model in our study is better than that of the ANN model.

Application of adaptive neuro fuzzy interface system optimized with evolutionary algorithms for modeling CO2-crude oil minimum miscibility pressure

CO2 injection is known as one of the most reliable enhanced oil recovery techniques. The success of every gas injection process depends highly on the minimum miscibility pressure (MMP) needed for the injected gas and the oil to reach miscibility. Therefore, determination of the MMP between the two fluids is of great importance. In this study, a novel intelligent model based on adaptive neuro fuzzy interface system (ANFIS) was developed for predicting MMP values between pure/impure CO2 and reservoir oil at different reservoir conditions based on 270 experimental data points for pure and impure CO2, dead and live oils, reservoir temperature ranges from 293.72K to 388.73K, and experimental MMP values range from 6.54MPa to 31.30MPa. The ANFIS model was optimized by five different approaches; including Back Propagation (BP), Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO) and Differential Evolution (DE). A large number of data points from various sources of literature were gathered for model development and verification. In addition, various statistical and graphical error analyses were employed to evaluate the performance of the developed models as well as to compare them with major literature models for MMP prediction. The results showed that the ANFIS model optimized by PSO has the highest accuracy among all the models developed in this study as well as literature models, with an average absolute percent relative error of only 7.53%. In addition, all models developed in this study are more accurate than the existing models and their accuracy are as follows: ANFIS-PSO>ANFIS-GA>ANFIS-ACO>ANFIS-BP>ANFIS-DE. Lastly, the ANFIS models developed here can be inserted in any simulator to increase the accuracy of predicting pure and impure CO2-crude oil MMP.

An adaptive neuro-fuzzy interface system model for traffic classification and noise prediction

In present study, two adaptive neuro-fuzzy models have been developed for traffic classification and noise prediction, respectively. The traffic classification model (ANFIS-TC) classifies extracted sound features of different categories of vehicles based on their acoustic signatures. The model also compute total number of vehicles passes through a particular sampling point. The results have been used for the estimation of the equivalent traffic flow ( $$Q_\mathrm{E})$$ . The noise prediction model (ANFIS-TNP) has three inputs, namely equivalent traffic flow ( $$Q_\mathrm{E})$$ , equivalent vehicle speed ( $$S_\mathrm{E})$$ and honking. The equivalent traffic flow ( $$Q_\mathrm{E})$$ is the output of ANFIS-TC model, while equivalent vehicle speed ( $$S_\mathrm{E})$$ and honking are computed from observed averaged speed of different categories of vehicles and number of recorded horns blow per minute. The model assumes that the distance between sound level meter and road centerline is fixed for particular sampling point. The performance of both the models has been validated by field observations. The results show that traffic classification is 100% accurate, while correlation coefficients between observed and predicted traffic noise range from 0.75 to 0.96. Both the models are validated with random samples of data, and it is observed that both the models are generalized and could be employed for traffic classification and traffic noise prediction in small urban heterogeneous traffic environment for noise pollution assessment and control.

Adaptive neuro-fuzzy interface system for gap acceptance behavior of right-turning vehicles at partially controlled T-intersections

Gap acceptance theory is broadly used for evaluating unsignalized intersections in developed countries. Intersections with no specific priority to any movement, known as uncontrolled intersections, are common in India. Limited priority is observed at a few intersections, where priorities are perceived by drivers based on geometry, traffic volume, and speed on the approaches of intersection. Analyzing such intersections is complex because the overall traffic behavior is the result of drivers, vehicles, and traffic flow characteristics. Fuzzy theory has been widely used to analyze similar situations. This paper describes the application of adaptive neuro-fuzzy interface system (ANFIS) to the modeling of gap acceptance behavior of right-turning vehicles at limited priority T-intersections (in India, vehicles are driven on the left side of a road). Field data are collected using video cameras at four T-intersections having limited priority. The data extracted include gap/lag, subject vehicle type, conflicting vehicle type, and driver’s decision (accepted/rejected). ANFIS models are developed by using 80 % of the extracted data (total data observations for major road right-turning vehicles are 722 and 1,066 for minor road right-turning vehicles) and remaining are used for model validation. Four different combinations of input variables are considered for major and minor road right turnings separately. Correct prediction by ANFIS models ranges from 75.17 % to 82.16 % for major road right turning and 87.20 % to 88.62 % for minor road right turning. The models developed in this paper can be used in the dynamic estimation of gap acceptance in traffic simulation models.

Impact of multi-resolution analysis of artificial intelligence models inputs on multi-step ahead river flow forecasting

Summary In this paper an attempt is made to show that the performance of daily river flow forecasting is improved when data-preprocessing techniques are used with computational intelligence methods. Especially for forecasting longer lead-times, one of the inherent problems in all forecasting methods is that the reliability of forecasting decreases with increasing the lead-time. Therefore wavelet multi-resolution analysis is coupled with artificial neural networks (ANN) and adaptive neuro-fuzzy interface system (ANFIS) which are two promising methods for forecasting nonlinear and non-stationary time series. Different models with a combination of the different input data sets are developed for 1, 2, 3, 4 and 5 days ahead forecasting in Harvey River, Western Australia. Daubechies and Symlet wavelets are used to decompose river flow time series to different levels. Hybrid wavelet neural networks (WNN) models are trained using Levenberg–Marquart (LM) algorithm and the wavelet neuro-fuzzy (WNF) models with subtractive clustering method to find the optimum number of fuzzy rules. Comparing the results with those of the original ANN and ANFIS models indicates that the hybrid models produce significantly better results, especially for the peak values and longer lead-times.

Forecasting the Level of Reservoirs Using Multiple Input Fuzzification in ANFIS

Estimation the Level of water is one of the crucial subjects in reservoir management influencing on reservoir operation and decision making. One of the most accurate artificial intelligence model used broadly in water resource aspects is adaptive neuro-fuzzy interface system (ANFIS) taking in to account the membership functions (MF) on the basis of the smoothness characteristics and mathematical components each for set of input data. All researches in hydrological estimation used ANFIS, merely a type of MF has been noticed for all sets of inputs without considering the response of each of them. This study is applying a specified certain MFs for each type of input to improve the accuracy of ANFIS model in forecasting the water level in Klang Gates Dam in Malaysia. On the basis of the previous studies, two most popular MFs, Generalized Bell Shape MF and, Gaussian MF, are employed for examine the new pattern in two inputs ANFIS architecture resulted less stress in error performance, and higher accuracy in estimation, compare to the traditional ANFIS model. The aim is achieved by evaluating the performance in and fitness of the model in daily reservoir estimation.

Numerical modeling and optimization of wastewater treatment using porous polymeric membranes

AbstractA new modeling approach was developed for prediction of ammonia removal from water by means of porous membranes. The model was based on adaptive neuro‐fuzzy interface system (ANFIS) to simulate ammonia stripping from water by means of hollow‐fiber membrane contactors. The predictions aimed to obtain optimum conditions for ammonia stripping using the Taguchi method. The initial concentrations of ammonia, pH of the ammonia solution, velocity of the feed, and the presence of excess ions in the ammonia feed solution were considered as the input properties. On the other hand, mass transfer coefficient was considered as output. The prediction results revealed that the pH of the ammonia feed has a significant effect on the separation of ammonia from water. The results also showed that the prediction of ANFIS model and experimental data match well and that the model can be used for prediction of porous membranes. Furthermore, simulated annealing was also used to determine controllable conditions to find the highest mass transfer coefficient. POLYM. ENG. SCI., 2013. © 2012 Society of Plastics Engineers

Prediction of landslides using ASTER imagery and data mining models

The aim of this study was to identify landslide-related factors using only remotely sensed data and to present landslide susceptibility maps using a geographic information system, data-mining models, an artificial neural network (ANN), and an adaptive neuro-fuzzy interface system (ANFIS). Landslide-related factors were identified in Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) satellite imagery. The slope, aspect, and curvature of topographic features were calculated from a digital elevation model that was made using the ASTER imagery. Lineaments, land-cover, and normalized difference vegetative index layers were also extracted from the imagery. Landslide-susceptible areas were analyzed and mapped based on occurrence factors using the ANN and ANFIS. The generalized bell-shaped built-in membership function of the ANFIS was applied to landslide susceptibility mapping. Analytical results were validated using landslide test location data. In the validation results, the ANN model showed 80.42% prediction accuracy and the ANFIS model showed 86.55% prediction accuracy. These results suggest that the ANFIS model has a better performance than does the ANN in predicting landslide susceptibility.

Modeling daily discharge responses of a large karstic aquifer using soft computing methods: Artificial neural network and neuro-fuzzy

This paper compares two methods for modeling karst aquifers, which are heterogeneous, highly non-linear, and hierarchical systems. There is a clear need to model these systems given the crucial role they play in water supply in many countries. In recent years, the main components of soft computing (fuzzy logic (FL), and Artificial Neural Networks, (ANNs)) have come to prevail in the modeling of complex non-linear systems in different scientific and technologic disciplines. In this study, Artificial Neural Networks and Adaptive Neuro-Fuzzy Interface System (ANFIS) methods were used for the prediction of daily discharge of karstic aquifers and their capability was compared. The approach was applied to 7 years of daily data of La Rochefoucauld karst system in south-western France. In order to predict the karst daily discharges, single-input (rainfall, piezometric level) vs. multiple-input (rainfall and piezometric level) series were used. In addition to these inputs, all models used measured or simulated discharges from the previous days with a specified delay. The models were designed in a Matlab™ environment. An automatic procedure was used to select the best calibrated models. Daily discharge predictions were then performed using the calibrated models. Comparing predicted and observed hydrographs indicates that both models (ANN and ANFIS) provide close predictions of the karst daily discharges. The summary statistics of both series (observed and predicted daily discharges) are comparable. The performance of both models is improved when the number of inputs is increased from one to two. The root mean square error between the observed and predicted series reaches a minimum for two-input models. However, the ANFIS model demonstrates a better performance than the ANN model to predict peak flow. The ANFIS approach demonstrates a better generalization capability and slightly higher performance than the ANN, especially for peak discharges.