Generalization of adaptive neuro-fuzzy inference systems

The adaptive network-based fuzzy inference systems (ANFIS) of Jang (1993) is extended to the generalized ANFIS (GANFIS) by proposing a generalized fuzzy model (GFM) and considering a generalized radial basis function (GRBF) network. The GFM encompasses both the Takagi-Sugeno (TS)-model and the compositional rule of inference (CRI) model. The conditions by which the proposed GFM converts to TS-model or the CRI-model are presented. The basis function in GRBF is a generalized Gaussian function of three parameters. The architecture of the GRBF network is devised to learn the parameters of GFM, where the GRBF network and GFM have been proved to be functionally equivalent. It Is shown that GRBF network can be reduced to either the standard RBF or the Hunt's RBF network. The issue of the normalized versus the non-normalized GRBF networks is investigated in the context of GANFIS. An interesting property of symmetry on the error surface of GRBF network is investigated. The proposed GANFIS is applied to the modeling of a multivariable system like stock market.     

Modelling of evaporation from the reservoir of Yuvacik dam using adaptive neuro-fuzzy inference systems

Adaptive neuro-fuzzy inference system (ANFIS) models are proposed as an alternative approach of evaporation estimation for Yuvacik Dam. This study has three objectives: (1) to develop ANFIS models to estimate daily pan evaporation from measured meteorological data; (2) to compare the ANFIS model to the multiple linear regression (MLR) model; and (3) to evaluate the potential of ANFIS model. Various combinations of daily meteorological data, namely air temperature, relative humidity, solar radiation and wind speed, are used as inputs to the ANFIS so as to evaluate the degree of effect of each of these variables on daily pan evaporation. The results of the ANFIS model are compared with MLR model. Mean square error, average absolute relative error and coefficient of determination statistics are used as comparison criteria for the evaluation of the model performances. The ANFIS technique whose inputs are solar radiation, air temperature, relative humidity and wind speed, gives mean square errors of 0.181 mm, average absolute relative errors of 9.590% mm, and determination coefficient of 0.958 for Yuvacik Dam station, respectively. Based on the comparisons, it was found that the ANFIS technique could be employed successfully in modelling evaporation process from the available climatic data.     

Structure identification of generalized adaptive neuro-fuzzy inference systems

This paper presents a method to identify the structure of generalized adaptive neuro-fuzzy inference systems (GANFISs). The structure of GANFIS consists of a number of generalized radial basis function (GRBF) units. The radial basis functions are irregularly distributed in the form of hyper-patches in the input-output space. The minimum number of GRBF units is selected based on a heuristic using the fuzzy curve. For structure identification, a new criterion called structure identification criterion (SIC) is proposed. SIC deals with a trade off between performance and computational complexity of the GANFIS model. The computational complexity of gradient descent learning is formulated based on simulation study. Three methods of initialization of GANFIS, viz., fuzzy curve, fuzzy C-means in x/spl times/y space and modified mountain clustering have been compared in terms of cluster validity measure, Akaike's information criterion (AIC) and the proposed SIC.     

Glaucoma detection using adaptive neuro-fuzzy inference system

Purpose. To develop an automated classifier based on adaptive neuro-fuzzy inference system (ANFIS) to differentiate between normal and glaucomatous eyes from the quantitative assessment of summary data reports of the Stratus optical coherence tomography (OCT) in Taiwan Chinese population.Methods. This observational non-interventional, cross-sectional, case–control study included one randomly selected eye from each of the 341 study participants (135 patients with glaucoma and 206 healthy controls). Measurements of glaucoma variables (retinal nerve fiber layer thickness and optic nerve head topography) were obtained by Stratus OCT. Decision making was performed in two stages: feature extraction using the orthogonal array and the selected variables were treated as the feeder to adaptive neuro-fuzzy inference system (ANFIS), which was trained with the back-propagation gradient descent method in combination with the least squares method. With the Stratus OCT parameters used as input, receiver operative characteristic (ROC) curves were generated by ANFIS to classify eyes as either glaucomatous or normal.Results. The mean deviation was −0.67 ± 0.62 dB in the normal group and −5.87 ± 6.48 dB in the glaucoma group (P < 0.0001). The inferior quadrant thickness was the best individual parameter for differentiating between normal and glaucomatous eyes (ROC area, 0.887). With ANFIS technique, the ROC area was increased to 0.925.Conclusions. With Stratus OCT parameters used as input, the results from ANFIS showed promise for discriminating between glaucomatous and normal eyes. ANFIS may be preferable since the output concludes the if–then rules and membership functions, which enhances the readability of the output.     

Improved adaptive neuro-fuzzy inference system

This paper introduces a new type of Adaptive Neuro-fuzzy System, denoted as IANFIS (Improved Adaptive Neuro-fuszzy Inference System). The new structure is realized by the insertion of the error of training of ANFIS in the third layer of this system. The recurrence of the error of training will increase the capability of convergence and the robustness of ANFIS. The proposed IANFIS system is applied to make the identification of nonlinear functions, and the obtained results are compared with these obtained by usual ANFIS to verify the effectiveness of the proposed adaptive neuro-fuzzy system.     

Predicting groutability of granular soils using adaptive neuro-fuzzy inference system

Neural Computing and Applications - In this paper, the applicability of adaptive neuro-fuzzy inference system (ANFIS) for the prediction of groutability of granular soils with cement-based grouts...     

Automatic detection of ophthalmic artery stenosis using the adaptive neuro-fuzzy inference system

In this study, a new approach based on an adaptive neuro-fuzzy inference system (ANFIS) was presented for detection of ophthalmic artery stenosis. The ANFIS was used to detect ophthalmic artery stenosis when two features, resistivity and pulsatility indices, defining changes of ophthalmic arterial Doppler waveforms were used as inputs. The ophthalmic arterial Doppler signals were recorded from 115 subjects, of whom 52 suffered from ophthalmic artery stenosis and the rest were healthy. The proposed ANFIS model combined the neural network adaptive capabilities and the fuzzy logic qualitative approach. Some conclusions concerning the impacts of features on the detection of ophthalmic artery stenosis were obtained through analysis of the ANFIS. The performances of the classifiers were evaluated in terms of training performance and classification accuracies and the results confirmed that the proposed ANFIS classifier has potential in detecting the ophthalmic artery stenosis.     

Modelling electrical conductivity of groundwater using an adaptive neuro-fuzzy inference system

Electrical conductivity is an important indicator for water quality assessment. Since the composition of mineral salts affects the electrical conductivity of groundwater, it is important to understand the relationships between mineral salt composition and electrical conductivity. In this present paper, we develop an adaptive neuro-fuzzy inference system (ANFIS) model for groundwater electrical conductivity based on the concentration of positively charged ions in water. It is shown that the ANFIS model outperforms more traditional methods of modelling electrical conductivity based on the total solids dissolved in the water, even though ANFIS uses less information. Additionally, the fuzzy rules in the ANFIS model provide a categorization of ground water samples in a manner that is consistent with the current understanding of geophysical processes.     

Prediction of landslide displacement with controlling factors using extreme learning adaptive neuro-fuzzy inference system (ELANFIS)

Landslide is a major geo-environmental hazard which imparts serious threat to lives and properties. The slope failures are due to adverse inherent geological conditions triggered by an external factor. This paper proposes a new method for the prediction of displacement of step-like landslides, by accounting the controlling factors, using recently proposed extreme learning adaptive neuro-fuzzy inference system (ELANFIS) with empirical mode decomposition (EMD) technique. ELANFIS reduces the computational complexity of conventional ANFIS by incorporating the theoretical idea of extreme learning machines (ELM). The rainfall data and reservoir level elevation data are also integrated into the study. The nonlinear original landslide displacement series, rainfall data, and reservoir level elevation data are first converted into a limited number of intrinsic mode functions (IMF) and one residue. Then decomposed displacement data are predicted by using appropriate ELANFIS model. Final prediction is obtained by the summation of outputs of all ELANFIS sub models. The performance of proposed the technique is tested for the prediction Baishuihe and Shiliushubao landslides. The results show that ELANFIS with EMD model outperforms other methods in terms of generalization performance.     

基于减法聚类和自适应神经模糊推理系统的递阶模糊系统的设计

提出了一种设计递阶模糊系统的简易而有效的方法.在得到一个单级模糊系统的基础上,用灵敏度分析法对每一个输入变量的重要性进行排序,从而确定每一级子系统的输入变量.利用减法聚类和自适应神经 模糊推理系统逐级对子系统进行训练.所得到的递阶模糊系统可进一步得到简化.仿真实例证实了设计方法的有效性.     

Estimating oxygen consumption from heart rate using adaptive neuro-fuzzy inference system and analytical approaches

In new approaches based on adaptive neuro-fuzzy systems (ANFIS) and analytical method, heart rate (HR) measurements were used to estimate oxygen consumption (VO₂). Thirty-five participants performed Meyer and Flenghi's step-test (eight of which performed regeneration release work), during which heart rate and oxygen consumption were measured. Two individualized models and a General ANFIS model that does not require individual calibration were developed. Results indicated the superior precision achieved with individualized ANFIS modelling (RMSE = 1.0 and 2.8 ml/kg min in laboratory and field, respectively). The analytical model outperformed the traditional linear calibration and Flex-HR methods with field data. The General ANFIS model's estimates of VO₂ were not significantly different from actual field VO₂ measurements (RMSE = 3.5 ml/kg min). With its ease of use and low implementation cost, the General ANFIS model shows potential to replace any of the traditional individualized methods for VO₂ estimation from HR data collected in the field.     

Intelligent mobile manipulator navigation using adaptive neuro-fuzzy systems

The work presented in this paper deals with the problem of autonomous and intelligent navigation of mobile manipulator, where the unavailability of a complete mathematical model of robot systems and uncertainties of sensor data make the used of approximate reasoning to the design of autonomous motion control very attractive.A modular fuzzy navigation method in changing and dynamic unstructured environments has been developed. For a manipulator arm, we apply the robust adaptive fuzzy reactive motion planning developed in [J.B. Mbede, X. Huang, M. Wang, Robust neuro-fuzzy sensor-based motion control among dynamic obstacles for robot manipulators, IEEE Transactions on Fuzzy Systems 11 (2) (2003) 249-261]. But for the vehicle platform, we combine the advantages of probabilistic roadmap as global planner and fuzzy reactive based on idea of elastic band. This fuzzy local planner based on a computational efficient processing scheme maintains a permanent flexible path between two nodes in network generated by a probabilistic roadmap approach. In order to consider the compatibility of stabilization, mobilization and manipulation, we add the input of system stability in vehicle fuzzy navigation so that the mobile manipulator can avoid stably unknown and/or dynamic obstacles. The purpose of an integration of robust controller and modified Elman neural network (MENN) is to deal with uncertainties, which can be translated in the output membership functions of fuzzy systems.     

Reckoner for health risk and insurance premium using adaptive neuro-fuzzy inference system

The paper demonstrates an efficient use of intelligent system for solving the classification problem in the sector of health insurance. A model based on adaptive neuro-fuzzy inference system (ANFIS) is proposed to deal with the fuzziness in the real-life environments. This approach enables the interpretation of majority of health factors of an insurance seeker through a set of fuzzy rules to determine the degree of risk to an individual. A fuzzy neural network has been trained with fuzzy inputs like age, occupation, family size, smoking habits, drinking habits, diabetes history, heart disease and other relevant inputs of individual for risk calculation. The model gets importance in health insurance sector because risk determination is fuzzy in nature, and fuzzy calculations are done more accurately by machines rather than human beings especially for the problems which are repetitive in nature and have large number of vague parameters. The proposed model can help the insurance seeker to identify the degree of risk he is having if he is not taking health insurance. This serves a dual purpose of attracting the insurance seeker to acquire the insurance and facilitate generating business to insurance company. Indicative results are presented and discussed in detail in terms of accuracy and solution interpretability.     

Fault detection and diagnosis of an industrial steam turbine using a distributed configuration of adaptive neuro-fuzzy inference systems

The issue of fault detection and diagnosis (FDD) has gained widespread industrial interest in process condition monitoring applications. An innovative data-driven FDD methodology has been presented in this paper on the basis of a distributed configuration of three adaptive neuro-fuzzy inference system (ANFIS) classifiers for an industrial 440 MW power plant steam turbine with once-through Benson type boiler. Each ANFIS classifier has been developed for a dedicated category of four steam turbine faults. A preliminary set of conceptual and experimental studies has been conducted to realize such fault categorization scheme. A proper selection of four measured variables has been configured to feed each ANFIS classifier with the most influential diagnostic information. This consequently leads to a simple distributed FDD system, facilitating the training and testing phases and yet prevents operational deficiency due to possible cross-correlated measured data effects. A diverse set of test scenarios has been carried out to illustrate the successful diagnostic performances of the proposed FDD system against 12 major faults under challenging noise corrupted measurements and data deformation corresponding to a specific fault time history pattern.     

An expert system of price forecasting for used cars using adaptive neuro-fuzzy inference

An expert system for used cars price forecasting using adaptive neuro-fuzzy inference system (ANFIS) is presented in this paper. The proposed system consists of three parts: data acquisition system, price forecasting algorithm and performance analysis. The effective factors in the present system for price forecasting are simply assumed as the mark of the car, manufacturing year and engine style. Further, the equipment of the car is considered to raise the performance of price forecasting. In price forecasting, to verify the effect of the proposed ANFIS, a conventional artificial neural network (ANN) with back-propagation (BP) network is compared with proposed ANFIS for price forecast because of its adaptive learning capability. The ANFIS includes both fuzzy logic qualitative approximation and the adaptive neural network capability. The experimental result pointed out that the proposed expert system using ANFIS has more possibilities in used car price forecasting.     

Automatic detection of electroencephalographic changes using adaptive neuro-fuzzy inference system employing Lyapunov exponents

This paper describes the application of adaptive neuro-fuzzy inference system (ANFIS) model for automatic detection of electroencephalographic changes. Decision making was performed in two stages: feature extraction by computation of Lyapunov exponents and classification by the ANFIS trained with the backpropagation gradient descent method in combination with the least squares method. Five types of electroencephalogram (EEG) signals were classified by five ANFIS classifiers. To improve diagnostic accuracy, the sixth ANFIS classifier (combining ANFIS) was trained using the outputs of the five ANFIS classifiers as input data. The proposed ANFIS model combined the neural network adaptive capabilities and the fuzzy logic qualitative approach. Some conclusions concerning the saliency of features on classification of the EEG signals were obtained through analysis of the ANFIS. The performance of the ANFIS model was evaluated in terms of training performance and classification accuracies and the results confirmed that the proposed ANFIS model has potential in classifying the EEG signals.     

Fault gear identification and classification using discrete wavelet transform and adaptive neuro-fuzzy inference

In this paper, an intelligent diagnosis for fault gear identification and classification based on vibration signal using discrete wavelet transform and adaptive neuro-fuzzy inference system (ANFIS) is presented. The discrete wavelet transform (DWT) technique plays one of the important roles for signal feature extraction in the proposed system. The abnormal transient signals will show in different decomposition levels and can be used to recognize the various faults by the DWT figure. However, many fault conditions are hard to inspect accurately by the naked eye. In the present study, the feature extraction method based on discrete wavelet transform with energy spectrum is proposed. The different order wavelets are considered to identify fault features accurately. The database is established by feature vectors of energy spectrum which are used as input pattern in the training and identification process. Furthermore, the ANFIS is proposed to identify and classify the fault gear positions and the gear fault conditions in the fault diagnosis system. The proposed ANFIS includes both the fuzzy logic qualitative approximation and the adaptive neural network capability. The experimental results verified that the proposed ANFIS has more possibilities in fault gear identification. The ANFIS achieved an accuracy identification rate which was more satisfactory than traditional vision inspection in the proposed system.     

An adaptive neuro-fuzzy inference system for bridge risk assessment

Bridge risks are often evaluated periodically so that the bridges with high risks can be maintained timely. This paper develops an adaptive neuro-fuzzy system (ANFIS) using 506 bridge maintenance projects for bridge risk assessment, which can help Highways Agency to determine the maintenance priority ranking of bridge structures more systematically, more efficiently and more economically in comparison with the existing bridge risk assessment methodologies which require a large number of subjective judgments from bridge experts to build the complicated nonlinear relationships between bridge risk score and risk ratings. The ANFIS proves to be very effective in modelling bridge risks and performs better than artificial neural networks (ANN) and multiple regression analysis (MRA).     

Missing wind data forecasting with adaptive neuro-fuzzy inference system

In any region, to begin generating electricity from wind energy, it is necessary to determine the 1-year distribution characteristics of wind speed. For this aim, a wind observation station must be constructed and 1-year wind speed and direction data must be collected. For determining the distribution characteristics, the collected data must be statistically analyzed. The continuity and reliability of the data are quite important for such studies on the days when possible faults can occur in any part of the observation unit or on days when, the system is on maintenance, it is not possible to record any data. In this study, it is assumed that the station had not worked at some randomly chosen days and that for these days no data could be recorded. The missing data are predicted using the data that were recorded before and after fault or maintenance by an adaptive neuro-fuzzy inference system (ANFIS). It is seen that ANFIS is successful for such a study.     

Self-adaptive neuro-fuzzy inference systems for classification applications

This paper presents a self-adaptive neuro-fuzzy inference system (SANFIS) that is capable of self-adapting and self-organizing its internal structure to acquire a parsimonious rule-base for interpreting the embedded knowledge of a system from the given training data set. A connectionist topology of fuzzy basis functions with their universal approximation capability is served as a fundamental SANFIS architecture that provides an elasticity to be extended to all existing fuzzy models whose consequent could be fuzzy term sets, fuzzy singletons, or functions of linear combination of input variables. Without a priori knowledge of the distribution of the training data set, a novel mapping-constrained agglomerative clustering algorithm is devised to reveal the true cluster configuration in a single pass for an initial SANFIS construction, estimating the location and variance of each cluster. Subsequently, a fast recursive linear/nonlinear least-squares algorithm is performed to further accelerate the learning convergence and improve the system performance. Good generalization capability, fast learning convergence and compact comprehensible knowledge representation summarize the strength of SANFIS. Computer simulations for the Iris, Wisconsin breast cancer, and wine classifications show that SANFIS achieves significant improvements in terms of learning convergence, higher accuracy in recognition, and a parsimonious architecture.