Recurrent Fuzzy Research Articles

A Comparative Study of SWAT, RFNN and RFNN-GA for Predicting River Runoff

Background/Objectives: Data-driven models such as Recurrent Fuzzy Neural Network (RFNN) have been proven to be great methods for modeling, characterizing and predicting various kinds of nonlinear hydrologic time series data such as rainfall, water quality and river runoff. In modeling and predicting river runoff, the most important advantage of data driven models is that they do not need as much data as do physical models such as the Soil and Water Assessment Tool (SWAT). In Vietnam, most of data which are required by SWAT are not available, thus data-driven models seem to be more suitable for predicting river runoff than SWAT. The objective of this study is to investigate the performance of SWAT, RFNN and an improvement of RFNN (RFNN-GA), which is a hybrid of RFNN and Genetic Algorithm (GA) in predicting the runoff of Srepok River in Central Highland of Vietnam. Methods/Statistical Analysis: Coefficient of correlation (R2) and mean absolute relative error (MARE) are used to analysis and compare the performance of SWAT, RFNN and RFNN-GA. Findings: The experimental results demonstrate that RFNN and RFNN-GA give the performance better than that of SWAT and they are able to be applied to real applications. Among these methods, RFNN-GA is the most superior. Application/Improvements: In the terms of MARE and R2, RFNN-GA improves RFNN 0.9% and 2.2%, respectively; and improves SWAT 27.4% and 12.5%, respectively. RFNN-GA was deployed to predict the runoff of Srepok River in Central Highland of Vietnam.

Interactively recurrent fuzzy functions with multi objective learning and its application to chaotic time series prediction

Fuzzy functions (FFs) models were introduced as an alternate representation of the fuzzy rule based approaches. This paper presents novel Interactively Recurrent Fuzzy Functions (IRFFs) for nonlinear chaotic time series prediction. Chaotic sequences are strongly dependent on their initial conditions as well as past states, therefore feed forward FFs models cannot perform properly. To overcome this weakness, recurrent structure of FFs is proposed by placing local and global feedbacks in the output parts of multidimensional subspaces. IRFFs' optimized parameters should minimize the output error and maximize clusters density. To achieve these contradictory goals, Non-dominated Sorting Genetic Algorithm II (NSGAII) is applied for simultaneously optimizing the objectives. Also, feedback loop parameters are tuned by utilizing gradient descent algorithm with line search strategy based on the strong Wolfe condition. The experimental setup includes comparative studies on prediction of benchmark chaotic sequences and real lung sound data. Further simulations demonstrate that our proposed approach effectively learns complex temporal sequences and outperforms fuzzy rule based approaches and feed forward FFs.

Fuzzy Neural Network based Real-Time Driver Fatigue Detection and Warning System using EEG Waves

Driving fatigue has been considered as a significant risk factor in transportation accidents, and the development of the human cognitive State based on Electro Encephalo Gram (EEG) has become a major focus in the field of driving safety for many drivers. In adaptive online prediction framework which is capable of capturing subject-specific predictive patterns autonomously by constructing a subject-specific pattern library, based on which a probabilistic prediction rule is established. A real-time wireless Electro Encephalo Gram (EEG) based Brain Computer Interface (BCI) system for drowsiness detection has been proposed to reduce the time delay. A new method for driver fatigue detection based on EEG waves using Recurrent Self-Evolving Fuzzy Neural Network (RSEFNN)is proposed. The model uses Mindwave to collect the attention waves of EEG from the left prefrontal lobe of the subject, and measures the data when the subject is in the State of concentration, relaxation, fatigue and sleep. The Mindwave device and Matlab is interfaced using the Think Gear Library module. The correlation coefficient of the driver attention is classified to detect driver fatigue. Thus the system using brain wave sensor warns the driver when he/she feels drowsy.

Long Term Dry Cargo Freight Rates Forecasting by Using Recurrent Fuzzy Neural Networks

Maritime transport rates are very important for planning economic strategies. Various methods have been applied to seaborne trade forecasting. This study presents a genetic algorithm based trained recurrent fuzzy neural network for long term dry cargo freight rates forecasting. The empirical results show that proposed work has better accuracy than the other approaches which have used the same data set.

Recurrent Fuzzy Wavelet Neural Network to Control a PCV

Pneumatic Control Valves (PCV) are used in the process industries. In this paper, a Recurrent Fuzzy Wavelet Neural Network (RFWNN) is constructed by using Recurrent Wavelet Neural Network (RWNN). In RWNN, temporal relations are embedded in the network by adding feedback connections on the first layer of the network, and wavelet basis n is used as fuzzy membership function. The proposed method is applied on a PCV. P, PI and PID controllers are also employed For comparison. Bondgraph method has been utilized to model the control valve, so as to compare the response characteristics of valve. Simulation results have been made for four controllers.

Fuzzy neural network-based robust adaptive control for dynamic positioning of underwater vehicles with input dead-zone

This paper proposes a design for a robust adaptive controller for the Dynamical Positioning (DP) of underwater vehicles with unknown hydrodynamic coefficients, unknown disturbances and input dead-zones. First, for convenience of controller design, the Multi-Input Multi-Output (MIMO) system is divided into several Single-Input Single-Output (SISO) systems. Next, a Dynamic Recurrent Fuzzy Neural Network (DRFNN) with feedback loops is employed to approximate the unknown portion of the controller, which can greatly reduce the number of neural network inputs. A fuzzy logic dead-zone compensator is designed to cope with the unknown dead-zone characteristics of actuators. The upper bounds of the approximation errors and disturbances of the network, which are often used in existing works, are not necessary in this paper due to the presentation of a special robust compensator. Stability analysis is conducted according to the Lyapunov theorem, and the tracking error is proved to converge to zero. Simulation results indicate that the proposed controller demonstrates good performance.

A Novel Interactively Recurrent Self-Evolving Fuzzy CMAC and Its Classification Applications

In this paper, an Interactively Recurrent Self-evolving Fuzzy Cerebellar Model Articulation Controller (IRSFCMAC) model is developed for solving classification problems. The proposed IRSFCMAC classifier consists of internal feedback and external loops, which are generated by the hypercube cell firing strength to itself and other hypercube cells. The learning process of the IRSFCMAC gets started with an empty hypercube base, and then all of hypercube cells are generated and learned online via structure and parameter learning, respectively. The structure learning algorithm is based on the degree measure to determine the number of hypercube cells. The parameter learning algorithm, based on the gradient descent method, adjusts the shapes of the membership functions and the corresponding fuzzy weights of the IRSFCMAC. Finally, the proposed IRSFCMAC model is tested by four benchmark classification problems. Experimental results show that the proposed IRSFCMAC model has superior performance than traditional FCMAC and other models.

Development of Recurrent Fuzzy Systems with External Feedback

This paper investigates the use of three external feedback connections in the development of new recurrent fuzzy models for prediction of nonlinear dynamic systems. These models are formulated by state space equations. The state transition function is a TSK recurrent fuzzy system with an external feedback connection and adaptive delayed operators, and the output function is a polynomial function of the states. The identification of the model's parameters is carried out by a canonical differential evolution algorithm. The model performance was evaluated in benchmark problems found in the literature and the results demonstrated that the aiding of external feedbacks enhanced the recurrent fuzzy system quality, yielding models with good performance.

Vehicle Suspension Control Using Recurrent Neurofuzzy Wavelet Network

The main aim of this paper is to design the controller for a vehicle suspension system to reduce the uneasiness felt by passengers which arises from road disturbances and to increase the road holding related with the movements of pitch and roll of the vehicle. This demands an accurate and quick adaptive controller to obtain such control objectives, because, the passive suspension system and semi‐active suspension cannot perform better. Therefore, an adaptive Recurrent Fuzzy Wavelet Neural Network (RFWNN) based active suspension systems are designed to give better ride comfort and vehicle stability. The proposed adaptive RFWNN model combines the traditional TSK fuzzy model and the wavelet neural networks. The RFWNN controller is highly nonlinear and robust to meet the control objectives and can handle the nonlinearities faster than other conventional controllers. An online learning algorithm, which consists of parameter learning, is also presented. The learning parameters are based on the steepest‐descent method, to train the proposed RFWNN. The proposed approach is used to minimize the vibrations of seat, heave pitch and roll of the vehicle when traveling on rough road. The performance of the proposed RFWNN control strategy is being assessed by comparing with passive and semi‐active suspension systems.

Feedforward Tracking Control of Flat Recurrent Fuzzy Systems

Flatness based feedforward control has proven to be a feasible solution for the problem of tracking control, which may be applied to a broad class of nonlinear systems. If a flat output of the system is known, the control is often based on a feedforward controller generating a nominal input in combination with a linear controller stabilizing the linearized error dynamics around the trajectory. We show in this paper that the very same idea may be incorporated for tracking control of MIMO recurrent fuzzy systems. Their dynamics is given by means of linguistic differential equations but may be converted into a hybrid system representation, which then serves as the basis for controller synthesis.

Adaptive motion/force control strategy for non-holonomic mobile manipulator robot using recurrent fuzzy wavelet neural networks

In our study, we develop an adaptive position tracking system and a force control strategy for non-holonomic mobile manipulator robot, which combine the merits of Recurrent Fuzzy Wavelet Neural Networks (RFWNNs). In order to deal with the unknown knowledge problems of the robotic system, an adaptive RFWNNs control scheme with the dynamic structure and online learning ability is utilized to approximate unknown dynamics without the requirement of prior controlled system information. In addition, an adaptive robust compensator is proposed to eliminate uncertainties that consist of approximation errors, disturbances. According to the adaptive position tracking control design, an adaptive robust controller is also considered for the non-holonomic constraint force. The design of the adaptive online learning algorithms is derived by using the Lyapunov stability theorem. Therefore, the proposed controllers prove that they not only can guarantee the stability but also the tracking performance of the mobile manipulator robot control system. The effectiveness and robustness of the proposed method are demonstrated by comparative simulation and experimental results that are implemented in an indoor cleaning crawler-type mobile manipulator robot system.

Singleton-Based Two-String Inference in Recurrent Fuzzy Systems

Two-string fuzzy inference consists of two separate inference mechanisms: One conventional fuzzy inference system that processes recommending rules, as well as a mechanism for processing negative rules, which prevent the system from outputting their associated values when their premise is fulfilled. Two-string inference has valuable applications in pattern recognition and control tasks. We present a method rendering two-string inference applicable and computationally feasible in recurrent fuzzy systems, i.e. Mamdami-type fuzzy inference systems equipped with defuzzified state feedback. We show the efficiency of our approach by means of an illustrative example from biological systems modelling and suggest application areas for recurrent two-string fuzzy systems.

Stabilization of Recurrent Fuzzy Systems via Sum of Squares-based Hybrid Control

This paper presents an approach for stabilization of equilibria in recurrent fuzzy systems. This type of dynamic fuzzy systems being defined via linguistic rules can be interpreted as interpolation between constant gradients, and therefore as hybrid dynamical system. It is shown that the latter viewpoint allows for a precise description of the system dynamics, but on the other hand lacks transparency. In order to render a given equilibrium of the recurrent fuzzy system globally asymptotically stable, local polynomial controllers are computed via sum of squares optimization to allow only for deterministic mode transitions on a micro level. In addition, the controlled recurrent fuzzy system can then be interpreted as finite deterministic automaton, thus allowing for analysis of system properties on a more abstract macro level. Relaxations are proposed in cases where recurrent fuzzy systems may not be rendered deterministic and the method is applied to two examples.

Reactive Power Control of Single-Stage Three-Phase Photovoltaic System during Grid Faults Using Recurrent Fuzzy Cerebellar Model Articulation Neural Network

This study presents a new active and reactive power control scheme for a single-stage three-phase grid-connected photovoltaic (PV) system during grid faults. The presented PV system utilizes a single-stage three-phase current-controlled voltage-source inverter to achieve the maximum power point tracking (MPPT) control of the PV panel with the function of low voltage ride through (LVRT). Moreover, a formula based on positive sequence voltage for evaluating the percentage of voltage sag is derived to determine the ratio of the injected reactive current to satisfy the LVRT regulations. To reduce the risk of overcurrent during LVRT operation, a current limit is predefined for the injection of reactive current. Furthermore, the control of active and reactive power is designed using a two-dimensional recurrent fuzzy cerebellar model articulation neural network (2D-RFCMANN). In addition, the online learning laws of 2D-RFCMANN are derived according to gradient descent method with varied learning-rate coefficients for network parameters to assure the convergence of the tracking error. Finally, some experimental tests are realized to validate the effectiveness of the proposed control scheme.

CORPORATE SOCIAL RESPONSIBILITY AND FINANCIAL PERFORMANCE: Evidence from a Panel of Korea Listed Companies

A novel recurrent neural fuzzy network is proposed in this paper. The network model is composed by two structures: a fuzzy system and a neural network. The fuzzy system contains fuzzy neurons modeled with the aid of logic and and or operations processed via t-norms and s-norms. The neural network is composed by nonlinear elements placed in series with the previous logical elements. The network model implicitly encodes a set of if-then rules and its recurrent multilayered structure performs fuzzy inference. The topology induces a clear relationship between the network structure and an associated fuzzy rule-based system. In particular we explore this structure with an heuristic learning algorithm based on associative reinforcement learning and gradient search. These learning algorithms are associated to the fuzzy system and neural network, respectively. That is, output layer weights are adjusted via an error gradient method whereas a reward and punishment scheme updates the hidden layer weights. The recurrent fuzzy neural network is particularly suitable to model nonlinear dynamic systems and to learn sequences. Computational experiments with system identification problems show that the fuzzy neural models learned are simpler and that learning is faster than its counterparts.

Embedded Differential Evolution Algorithm for Recurrent Fuzzy Neural Network Controller Optimization

A chaos concise differential evolution algorithm (CcDE) is proposed for the embedded controller with limited memory, which introduces chaotic local search based on basic differential evolution algorithm to increase exploring and prevent premature convergence. Using virtual population and Gaussian sampling, the CcDE becomes simple and reduces the memory requirements at run time. Experimental simulation on optimizing parameters of the recurrent fuzzy neural network shows that the proposed CcDE can obtain better performance than other concise algorithm.

Research on Recurrent Fuzzy Neural Networks of Unified Power Flow Controller

Research on Recurrent Fuzzy Neural Networks of Unified Power Flow Controller

Modeling Cyclic Interactions within a Production Environment using Transition Adaptive Recurrent Fuzzy Systems

Numerous dynamic influences affect producing companies and require a continuous adaptation of the production. At present, modeling these influences and their interdependencies is mainly limited to quantitative factors. In this paper, an approach is proposed that allows to model the missing qualitative influences and their interdependencies via recurrent fuzzy systems (RFS), where the transitions between the state variables are additionally weighted (transition adaptation). This allows an easy adaption of a production environment's behavior while maintaining the interpretability of the model and the model-based analysis results. To handle the complexity of the resulting models, a structured way to simplify a large fuzzy rule base, to reduce the number of required weighting coefficients, and to merge the state variable onto a single production effectiveness value is shown. All of this is directly illustrated by exemplarily modeling the influence of some relevant qualitative factors onto a production environment.

Chaotic time series prediction using brain emotional learning-based recurrent fuzzy system (BELRFS)

In this paper, an architecture based on the anatomical structure of the emotional network in the brain of mammalians is applied as a prediction model for chaotic time series studies. The architecture is called Brain Emotional Learning-based Recurrent Fuzzy System (BELRFS), which stands for: Brain Emotional Learning-based Recurrent Fuzzy System. It adopts neuro-fuzzy adaptive networks to mimic the functionality of brain emotional learning. In particular, the model is investigated to predict space storms, since the phenomenon has been recognised as a threat to critical infrastructure in modern society. To evaluate the performance of BELRFS, three benchmark time series: Lorenz time series, sunspot number time series and Auroral Electrojet (AE) index. The obtained results of BELRFS are compared with Linear Neuro-Fuzzy (LNF) with the Locally Linear Model Tree algorithm (LoLiMoT). The results indicate that the suggested model outperforms most of data driven models in terms of prediction accuracy.

MPPT Control of Photovoltaic Cells Based on Dynamic Recurrent Fuzzy Neural Network

The output characteristics of photovoltaic cells varies with the load and the external environment.In order to improve the conversion efficiency of PV cells , it is necessary to keep the system working in maximum power point(MPP) nearby.According to maximum power point tracking for PV systems,a new intelligent control method based on dynamic recurrent fuzzy neural network is proposed.The simulation results show that when the external environment changes ,the system can make rapid response and accurately track the maximum power point of PV cells.What's more,it has good control accuracy and stability.