Fuzzy Neural Network

We undertook a study to determine if the automatic detection and counting of vehicle passengers is feasible. An automated passenger counting system would greatly facilitate the operation of freeway lanes reserved for car-pools (HOV lanes). In the present paper we report our findings regarding the appropriate sensor phenomenology and arrangement for the task. We propose a novel system based on fusion of near-infrared imaging signals and we demonstrate its adequacy with theoretical and experimental arguments.

Most of the power systems protection techniques are related to the definition of system states by means of the identification of patterns from waveform of voltage and associated current. This means that the development of an adaptive protection can essentially be treated as a problem about classification/recognition of patterns (Song et al., 1997). Nevertheless, because the main causes of faults and the operation of nonlinear devices under certain conditions of fault, the methods of recognition of conventional patterns are unsatisfactory in some applications, particularly, in the case of high complexity electrical systems. In this sense, neural networks play an important role due to their unique ability of mapping nonlinear relations. Some successful applications of neural networks in the area of electrical engineering , have demonstrated that they can be used like an alternative method to solve certain problems of great complexity where the conventional techniques have experienced difficulties. Nevertheless, when giving a glance to the different applications of neural networks to electric power systems, it is clear that almost all the developments that have been carried out are based on the multi-layers perceptron with retro-propagation learning algorithms (BP). Although, BP can provide very compact distributed representations of complex data sets, it has some disadvantages such as the following: it exhibits slow learning, it requires great sets of training, they easily fall in local minimums, and in general it shows little robustness (Song et al., 1997). Another type of learning is the non-supervised one that surrounds the learning of patterns without a target. A typical non-supervised learning network is the Self-Organized Mapping (SOMs) developed by Teuvo Kohonen. A SOM network has the advantage of fast learning and small sets of training. Nevertheless, due to the absence of an output "truth" layer in the SOM, its use is not recommendable for the classification of patterns. Instead, it is used as an

The review of the existing approaches and algorithms for designing DSS based on fuzzy logic in case of changing the structure of the vector of the input data is considered in the paper. There was presented the own approach, which lies in correction (editing) the rules of fuzzy knowledge bases at different number of input coordinates of the system. Simulation results confirm the effectiveness and appropriateness of use of the algorithms of editing the rules of fuzzy knowledge bases in the process of DSS designing for management decisionmaking, particularly in solving problems of transport logistics.

In this paper, the integration of Type-2 fuzzy set theory and recurrent wavelet neural network(WNN) is proposed to allow managing of nonuniform uncertainties for identifying non-linear dynamic system. The proposed Type-2 fuzzy WNN is inherently a recurrent multilayered network which constructed based on a set of Type-2 fuzzy rules and recurrent connections in the second layer of the FWNN. Each rule comprises a wavelet function in the consequent part. The structure has both advantages of recurrent and wavelet neural network which expand the basic ability of fuzzy neural network to deal with temporal problems. Both antecedent and consequent parameters update rules are derived based on the gradient descent method. The structure is applied in the identification of dynamic plants which is commonly used in the literature. Simulation result from the identification of a second-order non-linear plant confirms the better performance and effectiveness of the proposed structure.

The proper generation of fuzzy membership function is of fundamental importance in fuzzy applications. The effectiveness of the membership functions in pattern classifications can be objectively measured in terms of interpretability and classification accuracy in the conformity of the decision boundaries to the inherent probabilistic decision boundaries of the training data. This paper presents the Supervised Pseudo Self-Evolving Cerebellar (SPSEC) algorithm that is bio-inspired from the two-stage development process of the human nervous system whereby the basic architecture are first laid out without any activitydependent processes and then refined in activity-dependent ways. SPSEC first constructs a cerebellar-like structure in which neurons with high trophic factors evolves to form membership functions that relate intimately to the probability distributions of the data and concomitantly reconcile with defined semantic properties of linguistic variables. The experimental result of using SPSEC to generate fuzzy membership functions is reported and compared with a selection of algorithms using a publicly available UCI Sonar dataset to illustrate its effectiveness.

Electromechanical oscillations in a power system often exhibit poor damping when the power transfer over a corridor is high relative to the transmission strength. Traditional approaches to aid the damping of power system oscillations include the use of Power System Stabilizers (PSS). Power system stabilizers are used to generate supplementary control signals for the excitation system in order to damp out the low frequency power system oscillations. This paper describes the design procedure for a fuzzy logic based PSS (FLPSS). Speed Deviation of a synchronous machine and its derivative are chosen as the input signals to the FLPSS. The inference mechanism of the fuzzy logic controller is represented by 49 ifthen rules. The proposed technique has the features of a simple structure, adaptivity and fast response and is evaluated on a Single machine and Multi machine Power system under different operating conditions to demonstrate its effectiveness and robustness. Keywords— Power system o...

This paper mainly intends to discuss the solution of fully fuzzy linear systems (FFLS) Ax + b = Cx + d, where A and C are fuzzy matrices, b and d are fuzzy vectors. We transform the systems by using fuzzy numbers with a new parametric form for finding a fuzzy vector x that satisfies in the system.

In this paper, a new approach for solving system of fully fuzzy polynomial equations based on nonlinear programing (shown as NLP) with equality constrain is presented. It is easy to apply the proposed method. This method can also lead to improve numerical methods. In this work, an architecture of NLP is also proposed to find a fuzzy positive root of a system of fuzzy polynomial equations (if exists). Finally, we illustrate our approach by numerical examples.

In this paper, we present a numerical method for solving fully fuzzy polynomials. The proposed method is based on approximating fuzzy neural network. This method can also lead to improving numerical methods. In this work, an architecture of fuzzy neural net-works is also proposed to find a fuzzy root of a fuzzy polynomial (if exists) by introducing a learning algorithm. We propose a learning algorithm from the cost function for adjusting fuzzy weights. Finally, we illustrate our approach by numerical examples.

In this paper, a novel hybrid method based on fuzzy neural network for approximate fuzzy coefficients (parameters) of fuzzy linear and nonlinear regression models with fuzzy output and crisp inputs, is presented. Here a neural network is considered as a part of a large field called neural computing or soft computing. Moreover, in order to find the approximate parameters, a simple algorithm from the cost function of the fuzzy neural network is proposed. Finally, we illustrate our approach by some numerical examples.

This study investigated the stack drillability of unidirectional (UD) carbon fiber-reinforced thermoplastic matrix PAEK/CF composite and aluminum (AA7075) plate utilized in aerospace. The effects of cutting parameters on cutting forces during the drilling process of thermoplastic matrix composites and aluminum materials were experimentally analyzed. 3 different cutting speeds (55, 95, and 165 m/min) and feed rates (0.1, 0.2, and 0.3 mm/rev) were used in the experimental study to investigate the drilling performance of 3 different types of drills (Types A, B, and C) with rounded cutting edge and surface treatment of the same diameter. Drilling operations were carried out on a CNC Vertical Machining Center under dry-cutting conditions. For 3 different drill types, 3 different cutting speeds, and feed rate combinations, a total of 810 holes were drilled in a full factorial experimental design with 30 replicates for each combination. The damage on the hole surfaces on the drilled composite was identified with an optical microscope. The numerical data obtained in the composite testing laboratory of Sabancı University were analyzed using Minitab® 21.1 statistical software and transformed into graphs. The most suitable drill type and cutting parameters were determined for the drillability of PAEK/CF composite with thermoplastic matrix and aluminum (AA7075) plates when stacked.

This study investigated the stack drillability of unidirectional (UD) carbon fiber-reinforced thermoplastic matrix PAEK/CF composite and aluminum (AA7075) plate utilized in aerospace. The effects of cutting parameters on cutting forces during the drilling process of thermoplastic matrix composites and aluminum materials were experimentally analyzed. 3 different cutting speeds (55, 95, and 165 m/min) and feed rates (0.1, 0.2, and 0.3 mm/rev) were used in the experimental study to investigate the drilling performance of 3 different types of drills (Types A, B, and C) with rounded cutting edge and surface treatment of the same diameter. Drilling operations were carried out on a CNC Vertical Machining Center under dry-cutting conditions. For 3 different drill types, 3 different cutting speeds, and feed rate combinations, a total of 810 holes were drilled in a full factorial experimental design with 30 replicates for each combination. The damage on the hole surfaces on the drilled composite was identified with an optical microscope. The numerical data obtained in the composite testing laboratory and we were analyzed using Minitab® 21.1 statistical software and transformed into graphs. The most suitable drill type and cutting parameters were determined for the drillability of PAEK/CF composite with thermoplastic matrix and aluminum (AA7075) plates when stacked.

This research discusses the design process of sliding mode controllers used for sensorless field-oriented induction motor control. A sliding mode regulator uses a sliding surface integration for speed regulation to improve controller performance. The adaptive observer implements the MRAS technique which determines rotor speed estimation used in the speed feedback loop. The experimental research verifies the effectiveness of this proposed method.

The FAIR Principles were introduced to address data challenges and improve the Findability, Accessibility, Interoperability, and Reusability of digital resources, following several Semantic Web standards. 'FAIRness' corresponds to a percentage grade indicating how close a digital object is to abiding by those principles. Several tools have been developed to assess the FAIRness of data digital objects in support of enacting the FAIR Principles. This work offers an appraisal of tools that evaluate the FAIRness of such objects, focusing on fully automated solutions. We conduct a literature review about existing tools, extract from it a set of requirements they aim to fulfill, and assess how each one fares considering this ensemble. Our results help researchers, developers, and data stewards with an overview of the tools, including an analysis of tools' fulfillment to the elicited requirements, making explicit existing gaps, and giving directions to improvements needed.

Neural Networks (NN) proved to be a powerful problem solving mechanism with great ability to learn. The success and speed of training is based on the initial parameter settings such as architecture, initial weights, learning rates and others. The most used method of training NN is the back propagation of the gradient. Although this method provides a global optimal solution in a reasonable time, it can converge towards local minimum, in addition to large number of parameters that should be fixed previously. Within the framework of this study, we propose a new hierarchical genetic algorithm to determine the structure and the to train the NN. Such algorithms are known for structures’ and parameters’ optimization. We prove that hierarchical genetic algorithms can improve the result of backpropagation of gradient.

This paper presents a new online identification algorithm to drive an adaptive affine dynamic model for nonlinear and time-varying processes. The new algorithm is devised on the basis of an adaptive neuro-fuzzy modeling approach. Two adaptive neuro-fuzzy models are sequentially identified on the basis of the most recent input-output process data to realize an online affine-type model. A series of simulation test studies has been conducted to demonstrate the efficient capabilities of the proposed algorithm to automatically identify an online affine-type model for two highly nonlinear and time-varying continuous stirred tank reactor (CSTR) benchmark problems having inherent non-affine dynamic model representations. Adequacy assessments of the identified models have been explored using different evaluation measures, including comparison with an adaptive neuro-fuzzy inference system (ANFIS) as the pioneering and the most popular adaptive neuro-fuzzy system with powerful modeling features.

An important part of the interpretation of a decision process lies on the ascertainment of the in uence of the input features, that is, of how much the implemented model relies on a given input feature to perform the desired task. Recently data analysis techniques based on fuzzy logic have gained attention because of their interpretability. Many real-world applications, however, have very high dimensionality and require very complex decision borders. In this case the numbe rof fuzzy rules can proliferate and the easy interpretability of the fuzzy model can progressively disappear. A method is presented that quanti es the discriminative po wer of the input features in a fuzzy model. The proposed quanti cation helps the interpretation of fuzzy models constructed on high dimensional and very fragmented training sets. First, a measure of the information contained in the fuzzy model is de ned on the basis of its fuzzy rules. The classi cation is then performed along one of the input features, that is, the fuzzy rules are split according to that feature's linguistic values. For each linguistic value, a fuzzy sub-model is generated from the original fuzzy model. The average information contained in these fuzzy sub-models is measured and its relative comparison with the information measure of the original fuzzy model quanti es the information gain that derives from the classi cation performed on the selected input feature. This information gain characterizes the discriminative po wer of that input feature. Therefore, the proposed information gain can beused to obtain better insights into the selected fuzzy classi cation strategy, even in very high dimensional cases, and possibly to reduce the input dimension. Several arti cial and real-world data analysis are reported as examples, in order to illustrate the characteristics and potentialities of the proposed algorithm. As real-world examples, the most informative electrocardiographic measures are detected for an arrhythmia classi cation problem and the role of duration, amplitude and pitch variations of syllabic nuclei in American English spoken sentences is investigated for prosodic stress classi cation.

This correspondence presents an investigation into the comparative performance of an active vibration control (AVC) system using a number of intelligent learning algorithms. Recursive least square (RLS), evolutionary genetic algorithms (GAs), general regression neural network (GRNN), and adaptive neuro-fuzzy inference system (ANFIS) algorithms are proposed to develop the mechanisms of an AVC system. The controller is designed on the basis of optimal vibration suppression using a plant model. A simulation platform of a flexible beam system in transverse vibration using a finite difference method is considered to demonstrate the capabilities of the AVC system using RLS, GAs, GRNN, and ANFIS. The simulation model of the AVC system is implemented, tested, and its performance is assessed for the system identification models using the proposed algorithms. Finally, a comparative performance of the algorithms in implementing the model of the AVC system is presented and discussed through a set of experiments.

As a powerful paradigm for knowledge representation and a simulation mechanism applicable to numerous research and application fields, Fuzzy Cognitive Maps (FCMs) have attracted a great deal of attention from various research communities. However, the traditional FCMs do not provide efficient methods to determine the states of the investigated system and to quantify causalities which are the very foundation of the FCM theory. Therefore in many cases, constructing FCMs for complex causal systems greatly depends on expert knowledge. The manually developed models have a substantial shortcoming due to model subjectivity and difficulties with accessing its reliability. In this paper, we propose a fuzzy neural network to enhance the learning ability of FCMs so that the automatic determination of membership functions and quantification of causalities can be incorporated with the inference mechanism of conventional FCMs. In this manner, FCM models of the investigated systems can be automatically constructed from data, and therefore are independent of the experts. Furthermore, we employ mutual subsethood to define and describe the causalities in FCMs. It provides more explicit interpretation for causalities in FCMs and makes the inference process easier to understand. To validate the performance, the proposed approach is tested in predicting chaotic time series. The simulation studies show the effectiveness of the proposed approach.

The limitations of thermal, vibration, or electrical monitoring of electric machines such as false indications, low sensitivity, and difficulty of fault interpretation have recently been exposed. This has led to a shift in the direction in research towards applying new techniques for improving the reliability of condition monitoring. With the changing environment, the purpose of this article is to provide an overview of the recent trend in the industrial demand and research activity in condition monitoring technology. The new developments in insulation testing, electrical testing, flux analysis, transient analysis, and fault prognostics are summarized. The future challenges and recommendations for future work for the new technologies are also stated to support researchers target research/development efforts according to industrial needs.

The automatic diagnosis of breast cancer is an important medical problem. This paper hybridizes metaphors from cells membranes and intercommunication between compartments with clonal selection principle together with fuzzy logic to produce a fuzzy rule system in order to be used in diagnosis. The fuzzy-membrane-immune algorithm suggested were implemented and tested on the Wisconsin breast cancer diagnosis (WBCD) problem. The developed solution scheme is compared with five previous works based on neural networks and genetic algorithms. The algorithm surpasses all of them. There are two motivations for using fuzzy rules with the membrane-immune algorithm in the underline problem. The first is attaining high classification performance. The second is the possibility of attributing a confidence measure (degree of benignity or malignancy) to the output diagnosis, beside the simplicity of the diagnosis system, which means that the system is human interpretable.

Abstract: Fuzzy neural networks (FNNs) provide a new approach for classification of multispectral data and to extract and optimize classification rules. Neural networks deal with issues on a numeric level, whereas fuzzy logic deals with them on a semantic or linguistic level. FNNs ...

In this paper, we define four types of fuzzy neurons and propose the structure of a four-layer feedforward fuzzy neural network (FNN) and its associated learning algorithm. The proposed four-layer FNN performs well when used to recognize shifted and distorted training patterns. When an input pattern is provided, the network first fuzzifies this pattern and then computes the similarities of this pattern to all of the learned patterns. The network then reaches a conclusion by selecting the learned pattern with the highest similarity and gives a nonfuzzy output. The 26 English alphabets and the 10 Arabic numerals, each represented by 16x 16 pixels, were used as original training patterns. In the simulation experiments, the original 36 exemplar patterns were shifted in eight directions by 1 pixel (6.25% to 8.84%) and 2 pixels (12.5'h to 17.68% ). After the FNN has been trained by the 36 exemplar patterns, the FNN can recall all of the learned patterns with recognition rate. It can also recognize patterns shifted by 1 pixel in eight directions with loo%, recognition rate and patterns shifted by 2 pixels in eight directions with an average recognition rate of 92.01%. After the FNN has been trained by the 36 exemplar patterns and 72 shifted patterns, it can recognize patterns shifted by 1 pixel with recognition rate and patterns shifted by 2 pixels with an average recognition rate of 98.61%. We have also tested the FNN with 10 kinds of distorted patterns for each of the 36 exemplars. The FNN can recognize all of the distorted patterns with 100% recognition rate. The proposed FNN can also be adapted for applications in some other pattern recognition problems. A structure which is composed of many processing elements connected to each other through weights [ 11-[3]. Neural networks (NN's) are built after biological neural systems. A NN stores patterns with distributed coding and is a trainable nonlinear dynamic system. A NN has a faster response and a higher performance than those of a sequential digital computer in emulating the capabilities of the human brain. Recently, NN's have been used in pattem recognition problems, especially where input patterns are shifted in position and scale-changed. Fukushima et al. [4], [SI have presented the Neocognitron, which is insensitive to translation and deformation of input patterns, and used it to recognize handprinted characters. However, the Neocognitron is complex Manuscript

A new method for specific object detection in two-dimensional color images is proposed in this paper. The proposed method uses color histograms of an object on the hue and saturation (HS) color space as detection features. To represent color information by histograms as accurately as possible, a nonuniform partition of HS space is proposed. The whole detection process consists of three stages. In the first stage, the input image is repeatedly sub-sampled by a factor, resulting in a pyramid of images. Scanning on all of the scaled images with a pre-defined window size is performed, where histograms of each window are fed as inputs to a fuzzy classifier. The fuzzy classifier used is a self-organizing Takagi-Sugeno (TS)-type fuzzy network with support vector learning (SOTFN-SV). SOTFN-SV is a fuzzy system constituted by TS-type fuzzy if-then rules. It is constructed by the hybridization of fuzzy clustering and support vector machine. Many candidate objects are detected in this stage. In the second stage, a splitting K-means clustering method is proposed and applied to the detections from Stage 1 so that detections with nearby positions are grouped into the same cluster. The number of clusters is generated automatically by the clustering method according to cluster variances. Final object position is determined from the clusters. In the final stage, size of a detected object is determined. To verify performance of the proposed method, experiments on five specific object detections are conducted and comparisons with different types of detectors are made.

This paper proposes an interval type-2 fuzzy-neural network with support-vector regression (IT2FNN-SVR) for noisy regression problems. The antecedent part in each fuzzy rule of an IT2FNN-SVR uses interval type-2 fuzzy sets, and the consequent part is of the Takagi-Sugeno-Kang (TSK) type. The use of interval type-2 fuzzy sets helps improve the network's noise resistance. The network inputs may be numerical values or type-1 fuzzy sets, with the latter being used for further improvements in robustness. IT2FNN-SVR learning consists of both structure learning and parameter learning. The structure-learning algorithm is responsible for online rule generation. The parameters are optimized for structural-risk minimization using a two-phase linear SVR algorithm in order to endow the network with high generalization ability. IT2FNN-SVR performance is verified through comparisons with type-1 and type-2 fuzzy-logic systems and other regression models on noisy regression problems.

This paper proposes a new recurrent model, known as the locally recurrent fuzzy neural network with support vector regression (LRFNN-SVR), that handles problems with temporal properties. Structurally, an LRFNN-SVR is a five-layered recurrent network. The recurrent structure in an LRFNN-SVR comes from locally feeding the firing strength of each fuzzy rule back to itself. The consequent layer in an LRFNN-SVR is a Takagi-Sugeno-Kang (T-S-K)-type consequent, which is a linear function of current states, regardless of system input and output delays. For the structure learning, a one-pass clustering algorithm clusters the input-training data and determines the number of network nodes in hidden layers. For the parameter learning, an iterative linear SVR algorithm is proposed to tune free parameters in the rule consequent part and feedback loops. The motivation for using SVR for parameter learning is to improve the LRFNN-SVR generalization ability. This paper demonstrates LRFNN-SVR capabilities by conducting simulations in dynamic system prediction and identification problems with noiseless and noisy data. In addition, this paper compares simulation results from the LRFNN-SVR with other recurrent fuzzy models.

This paper proposes a type-2 self-organizing neural fuzzy system (T2SONFS) and its hardware implementation. The antecedent parts in each T2SONFS fuzzy rule are interval type-2 fuzzy sets, and the consequent part is of Mamdani type. Using interval type-2 fuzzy sets in T2SONFS enables it to be more robust than type-1 fuzzy systems. T2SONFS learning consists of structure and parameter identification. For structure identification, an online clustering algorithm is proposed to generate rules automatically and flexibly distribute them in the input space. For parameter identification, a rule-ordered Kalman filter algorithm is proposed to tune the consequent-part parameters. The learned T2SONFS is hardware implemented, and implementation techniques are proposed to simplify the complex computation process of a type-2 fuzzy system. The T2SONFS is applied to nonlinear system identification and truck backing control problems with clean and noisy training data. Comparisons between type-1 and type-2 neural fuzzy systems verify the learning ability and robustness of the T2SONFS. The learned T2SONFS is hardware implemented in a field-programmable gate array chip to verify functionality of the designed circuits.

In the development of remote labs and virtual engineering tools the focus has rightly been on the technical challenges to be overcome to provide useful and usable tools and experimentation. However, the utilization of such facilities in educational settings is not simply a case of making students and faculty aware of their existence. In fact, there are significant pedagogical issues in the blending of remote and virtual facilities with cohort and location-based teaching and learning, and a number of research findings have highlighted student issues with both traditional teaching methods and the use of predominantly on-line materials. The authors have considerable experience in the development, production and use of eLearning materials in academic and industrial environments, and in tool virtualization and remote labs, and here propose a model for a distributed Masters program that supports students on a location-neutral basis utilizing online eLearning materials, virtual tools and remote lab facilities, combined with location-specific specialist teaching and learning facilities. The program described is already in operation between three European Universities, with the intention to expand both within Europe and beyond, utilizing the Erasmus Mundus scheme. The program is based on a constructivist pedagogic model that demands considerable independence of study and research on the part of the students, within a rich environment of high-quality specialist materials offered in a wide variety of modes. The authors believe that this approach optimizes the benefits of individual academic specialization in research and teaching, combined with effective use of eLearning materials, remote labs and virtual tools in a distributed environment, and thereby addresses a number of the issues identified from the research while also offering a high quality program to educate and develop next generation Engineers. I.

This research focuses on three iconic Indonesian batik patterns-Kawung, Mega Mendung, and Parang-due to their cultural significance and recognition. Kawung symbolizes harmony, Mega Mendung represents power, and Parang signifies protection and spiritual power. Using the YOLOv5 deep learning model, the study aimed to accurately identify these patterns. Results showed mean average precision (mAP) scores of 77% for Kawung, 80% for Parang, and an impressive 99% for Mega Mendung. The highest precision results were 91% for Kawung, 88% for Parang, and 77% for Mega Mendung. These findings highlight the potential of pattern recognition in preserving cultural heritage. Understanding these designs contributes to the appreciation of Indonesia s culture. The research suggests applications in cultural studies, digital archiving, and the textile industry, ensuring the legacy of these patterns endures.

watts can operate currently in a common power net. Also, the use of different types of energy sources in one power net can result in power dispersal. This combination of different energy sources makes it possible to instantly respond to load changes at individual sites on the one hand and poses problems in its management on the other hand. The Energorynok State Enterprise is responsible for support of the system ensuring functioning of the wholesale energy market in Ukraine [3, 4]. The task of this enterprise is to perform hourly and round-the-clock planning of the operating regime of the entire combined system on the basis of optimization of operating regimes of power plants. In order to prevent peak loads, hourly and round-the-clock electricity generation planning is carried out for optimal work of energy producers. And therefore, the more accurate the forecast, the less likely is occurrence of overloads and error conditions. Intelligent power systems with an active-adaptive SMART GRID network are increasingly widely used in the European Union member-countries [5, 6]. One of the tasks of such systems is planning of load distribution throughout the country, in particular, based on an analysis of all energy producers, both industrial and private.

Bank failure prediction is important for the regulators (such as the central banks and the finance ministries) of the banking industries. The collapse and failure of a bank could have devastating consequences to the entire banking system and a widespread repercussion effect on other banks and financial institutions. Some of the negative impacts are the massive bail out cost for a failing bank and the negative sentiments and loss of confidence developed by investors and depositors. Very often, bank failures do not occur over night and are usually due to a prolonged period of financial distress. Hence, it is desirable to have an early warning system that identifies potential failing or high-risk banks through financial distress. Various traditional statistical models have been employed to study bank failures [1]-[4]. However, these models have not identified the symptoms of financial distress leading to eventual bank failure. This paper attempts to identify the financial distress (the symptoms) that leads to a bank failure using financial covariates derived from publicly available financial statements using a novel neural fuzzy system named the Generic Selforganising Fuzzy Neural Network (GenSoFNN) [5]. Subsequently, the performance of the Cox proportional hazards model [3][4] is benchmarked against that of the GenSoFNN in predicting bank failures based on a population of 3635 US banks observed over a 21 years period. In addition, it is believed that the event of financial distress does not develop out of the blue. The deterioration of the financial conditions of distressed banks can be observed over time. Thus, the performance of a bank can be tracked and studied from its annual financial statements over a period of time, which essentially is time-series modeling. However, it may not be possible to obtain all the financial statements or there may be missing information in the observed period of a bank. Hence, as part of the study, the Pseudo-Outer-Product Fuzzy Neural Network (POPFNN) [6] is used to reconstruct missing financial data that tracks the solvency (financial health) of banking institutions. The performances of both the GenSoFNN as a bank failure classification system and the POPFNN network as a tool to reconstruct missing financial data are encouraging.

To extract the maximum power from a photovoltaic (PV) energy system, the real-time maximum power point (MPP) of the PV array must be tracked closely. The non-linear and time-variant characteristics of the PV array and the non-linear and non-minimum phase characteristics of a boost converter make it difficult to track the MPP for traditional control strategies. We propose a fuzzy neural network controller (FNNC), which combines the reasoning capability of fuzzy logical systems and the learning capability of neural networks, to track the MPP. With a derived learning algorithm, the parameters of the FNNC are updated adaptively. A gradient estimator based on a radial basis function neural network is developed to provide the reference information to the FNNC. Simulation results show that the proposed control algorithm provides much better tracking performance compared with the fuzzy logic control algorithm.

Ever since the assemblage of the first computer, efforts have been made to improve the way people could use machines. Recently, the usage of smart environments has become popular in order to make everyday living more comfortable and to improve the quality of living of humans. In this paper, a hand posture and gesture modeling and recognition system is introduced, which can be used as an interface to make possible communication with smart environment (intelligent space) by simple hand gestures. The system transforms preprocessed data of the detected hand into a fuzzy hand-posture feature model by using fuzzy neural networks and based on this model determines the actual hand posture applying fuzzy inference. Finally, from the sequence of detected hand postures, the system can recognize the hand gesture of the user. Index Terms-Fuzzy neural networks, hand-gesture recognition, hand-posture modeling, intelligent space, iSpace, smart environment. I. INTRODUCTION T ODAY, computers play an increasing role in our everyday life. The applications of intelligent systems that aim to improve the living conditions and quality of everyday life are also gaining more and more importance. In ideal case, these systems are realized in such a way that the usage of the systems becomes as easy as possible, while the presence of the system does not bother its user at all. This leads to the basic idea of "ubiquitous computing," which is proposed by [1]. One example for ubiquitous computing applications is the intelligent space (iSpace) [2], which has been developed at the University of Tokyo. iSpace is a room or area that has built-in intelligence: It can monitor the events and actions taking place

A fuzzy neural network method is proposed to predict minimum daily dissolved oxygen concentration in the Bow River, in Calgary, Canada. Owing to the highly complex and uncertain physical system, a data-driven and fuzzy number based approach is preferred over traditional approaches. The inputs to the model are abiotic factors, namely water temperature and flow rate. An approach to select the optimum architecture of the neural network is proposed. The total uncertainty of the system is captured in the fuzzy numbers weights and biases of the neural network. Model predictions are compared to the traditional, non-fuzzy approach, which shows that the proposed method captures more low DO events. Model output is then used to quantify the risk of low DO for different conditions.

This paper deals with the application of Fuzzy-Neural Networks (FNNs) in multi-machine system control applied on hot steel rolling. The electrical drives that used in rolling system are a set of three-phase induction motors (IM) controlled by indirect field-oriented control (IFO). The fundamental goal of this type of control is to eliminate the coupling influence though the coordinate transformation in order to make the AC motor behaves like a separately excited DC motor. Then use Fuzzy-Neural Network in control the IM speed and the rolling plant. In this work MATLAB/SIMULINK models are proposed and implemented for the entire structures. Simulation results are presented to verify the effectiveness of the proposed control schemes. It is found that the proposed system is robust in that it eliminates the disturbances considerably.

In this paper a hierarchical structure of fuzzy neural networks (FNNs) and how to train it for fault isolation given an appropriate data patterns, are presented. Fault symptoms concerning multiple simultaneous faults are harder to learn than those associated with single faults. Furthermore, the larger the set of faults, the larger the set of fault symptoms will be and, hence, the longer and less certain the training outcome. In order to overcome this problem, the proposed approach has a hierarchical structure of three levels where several FNNs are used. Thus, a large number of patterns are divided into many smaller subsets so that the classification can be carried out more efficiently. One of the advantages of this approach is that multiple faults can be detected in new data even if the network is trained only with data representing single abrupt faults. A continuous binary distillation column having several actuated valves with PID loops has been used as test bed for the proposed a...

This paper focuses on the study of short term load forecasting (STELF) using interval Type-2 Fuzzy Logic (IT2FL) and feed-forward Neural Network with back-propagation (NN-BP) tuning algorithm to improve their approximation capability, flexibility and adaptiveness. IT2FL for STELF is presented which provides additional degrees of freedom for handling more uncertainties for improving prediction accuracy and reducing cost. The IT2FL comprises five components which include; the fuzzification unit, the knowledge base, the inference engine, the type reducer and the defuzzification unit. Gaussian membership function is used to show the degree of membership of the input variables. The lower and upper membership functions (fired rules) as well as their consequent coefficients of IT2FL are fed into a (NN) which produces a crisp value coresponding to the optimal defuzzified output of IT2FLSs. The NN type reducer is trained to optimize parameters of membership function (MF) so as to produce an output with minimum error function with the purpose of improving forecasting performance of IT2FLS models. The IT2FNN system has the ability to overcome the limitations of individual technique and enhances their strengths to handle electric load forecasting. The IT2FNN is applied for STELF in Akwa Ibom State-Nigeria. The result of performance of IT2FNN is compared with IT2FLS and T1FLS methods for short term load forecasting with MSE of 0.00123, 0.00185 and 0.00247 respectively. Also, the results of forecasting are compared using RMSE of 0.035, 0.043 and 0.035 respectively, indicating a best accurate forecasting with IT2FNN. In addition, the result of performance of IT2FNN is compared with IT2FLS and T1FLS methods for short term load forecasting with MAPE of 1.5%, 3% and 4.5% respectively. Simulation results show that the IT2FNN approach takes advantages of accuracy and efficiency and performs better in prediction than IT2FL and T1FL methods in power load forecasting task. .

In this paper we propose a robust adaptive fuzzy controller for a class of nonlinear system with unknown dynamic. The method is based on type-2 fuzzy logic system to approximate unknown non-linear function. The design of the on-line adaptive scheme of the proposed controller is based on Lyapunov technique. Simulation results are given to illustrate the effectiveness of the proposed approach.

Segmentation is a crucial task in medical image processing. One of the greatest significant usages of image processing in medicine is recognizing cancerous or leisured tissues in MRI images. In this paper a new Clustering-Based algorithm is proposed for finding leisured segments in cancerous breast MRI images. The algorithm recursively detect the similar pixels to the current pixel and place them in the desired cluster. A threshold is defined due to the train data and if two adjacent pixels' colors differ less than the threshold, they are considered as similar. Experimental evaluations show that the algorithm is able to recognize the leisured areas with 94.3% accuracy and high speed.

SUMMARYThis paper presents an intelligent control approach that incorporates sliding mode control (SMC) and fuzzy neural network (FNN) into the implementation of back‐stepping control for a path tracking problem of a dual‐arm wheeled mobile manipulator subject to dynamic uncertainties and nonholonomic constraints. By using the back‐stepping technique, the system equations are reformulated into two levels: the kinematic level and the dynamic level. A sliding manifold is constructed by considering the disturbance free kinematic level equations only. With all the system uncertainties concentrated in the dynamic level, an FNN controller associated with a switching type of control law is employed to enforce sliding mode on the prescribed manifold. All parameter adjustment rules for the proposed controller are derived from the Lyapunov theory such that uniform ultimate boundedness for both the tracking error and the FNN weighting updates is ensured. A simulation study, which compares diff...

In fuzzy classi"er systems the classi"cation is obtained by a number of fuzzy If}Then rules including linguistic terms such as Low and High that fuzzify each feature. This paper presents a method by which a reduced linguistic (fuzzy) set of a labeled multi-dimensional data set can be identi"ed automatically. After the projection of the original data set onto a fuzzy space, the optimal subset of fuzzy features is determined using conventional search techniques. The applicability of this method has been demonstrated by reducing the number of features used for the classi"cation of four real-world data sets. This method can also be used to generate an initial rule set for a fuzzy neural network.

This paper introduces the use of the adaptive particle swarm optimization (APSO) for adapting the weights of fuzzy neural networks (FNN) on line. The fuzzy neural network is used for identification of the dynamics of a DC motor with nonlinear load torque. Then the motor speed is controlled using an inverse controller to follow a required speed trajectory. The parameters of the DC motor are assumed unknown as well as the nonlinear load torque characteristics. In the first stage a nonlinear fuzzy neural network (FNN) is used to approximate the motor control voltage as a function of the motor speed samples. In the second stage, the above mentioned approximator is used to calculate the control signal (the motor voltage) as a function of the speed samples and the required reference trajectory. Unlike the conventional back-propagation technique, the adaptation of the weights of the FNN approximator is done on-line using adaptive particle swarm optimization (APSO). The APSO is based on the least squares error minimization with random initial condition and without any off-line pre-training. Simulation results are presented to prove the effectiveness of the proposed control technique in achieving the tracking performance.

this paper presents a Wavelet-based Recurrent Fuzzy Neural Networks (WRFNN) trained with a stochastic searchbased adaptation algorithm. A WRFNN represents a recurrent network of neurons employing wavelet functions whose outputs are combined using fuzzy rules. In this paper an earlier WRFNN model proposed by Lin, and Chin, [1], is modified by application of Simultaneously Perturbed Stochastic Approximation (SPSA) method for training the network. The model includes TSK-type fuzzy implication to compute output of each layer. The SPSA algorithm was shown to be a stable global optimization technique that is applicable to WRFNN models with demonstrated computational advantages over other optimization algorithms.

In this paper, a neural network (NN)-based inverse kinematics problem of redundant manipulators subject to joint limits is presented. The Widrow-Hoff NN with an adaptive learning algorithm derived by applying Lyapunov stability theory is introduced. Since the inverse kinematics has an infinite number of joint angle vectors, a fuzzy neural network (FNN) is designed to provide an approximate value for that vector. This vector is fed into the NN as a hint input vector to guide the output of the NN within the self motion. This FNN is designed on the basis of cooperatively controlling each joint angle in the sense that it stops the motion on the critical axis at its limit at the expense of greater compensation from the most relaxed joint to accomplish the task. Physical constraints such as the joint velocity limits as well as the joint angle limits are incorporated into the method. Experiments are conducted for the PA-10 redundant manipulator to show the effectiveness of the proposed control system. A comparative study is carried out with the gradient projection method. Index Terms-Fuzzy neural network (FNN), inverse kinematics, joint limits avoidance, neural network (NN), redundant manipulators.

Although, computational Grid has been initially developed to solve large-scale scientific research problems, it is extended for commercial and industrial applications. An interesting commercial application with a wide impact on a variety of fields, is 3D rendering. In order to implement, however, 3D rendering to a grid infrastructure, we should develop appropriate scheduling and resource allocation mechanisms so that the negotiated Quality of Service (QoS) requirements are met. Efficient scheduling schemes require modeling and prediction of rendering workload. This is addressed in this paper, based on a combined fuzzy classification and neural network model. Initially, appropriate descriptors are extracted to represent the synthetic world. Fuzzy classification is used for organizing rendering descriptor so that a reliable representation is accomplished which increases the prediction accuracy. Neural network performs workload prediction by modeling the non-linear input-output relationship between rendering descriptors and the respective computational complexity. To increase the prediction accuracy, a constructive algorithm is adopted in this paper to train the neural network so that network weights and size are simultaneously estimated.