Gaussian Neural Network Research Articles

Signal processing and Gaussian neural networks for the edge and damage detection in immersed metal plate-like structures

The present study concerns the remote monitoring of immersed plate-like structures as the ones used for marine current turbines. The innovation of this work is the remote damage detection based on a systematic analysis of a small set of ultrasonic measurements limited by the backscattered echoes from the structure edges. The detection and localization are performed by combination of signal processing tools as Hilbert transform, principal component analysis, and thresholding methods and artificial intelligence tools as Gaussian neural networks. The edges of the structure are detected with a Gaussian neural network classifier, and the useful ranges of the measurements are extracted. These ranges are compared with reference signals in order to compute residuals. Finally damage detection is obtained from the magnitude of the residuals. In addition, some geometric parameters such as the incidence angle, the distance between the structure and the emission---reception device, and eventually the damage localization are estimated. The proposed method is validated with laboratory experimental measurements, and the performance is discussed with respect to some significant parameters.

Detection of freezing of gait for Parkinson’s disease patients with multi-sensor device and Gaussian neural networks

Freezing of Gait (FoG) in Parkinson Disease (PD) is a sudden episode characterized by a brief failure to walk. The aim of this study is to detect FoG episodes using a multi-sensor device for data acquisition, and Gaussian neural networks as a classification tool. Thus we have built a multi sensor prototype that detects FoG using new indicators like the variation of the inter-foot distance or the knee angle. Data are acquired from PD patients having FoG as a major symptom. The major social challenge is obtaining the acknowledgment of patients to participate in our study, whereas the main technical difficulty is extracting efficient features from various walking behaviors. For that purpose, the acquired signals are analyzed in order to extract both time and frequency domain features that separate the FoG class from the other gaits modes. Due to the complexity of FoG episodes, the optimal features are then extracted using Principal Component Analysis technique. Another contribution is to introduce the combined data into the Gaussian Neural Network (GNN) classification method, that is a new technique used for FoG detection, and has been developed in our previous works. Moreover, the classical thresholding method is implemented to compare and validate the GNN method. Results showed the feasibility of integrating the chosen sensors, in addition to the effectiveness of combining data from different types of sensors on the classification rate. The efficiency rate of classification in the proposed method is about 87 %.

DATA ACQUISITION AND DATA PROCESSING CHALLENGES IN HEAVY METAL MEASUREMENTS

Water quality is a key factor to preserve human life quality, as well as, environmental and biological ecosystems. This paper highlights specific issues related with the acquisition and data processing of heavy metal measurement systems. Between the main challenges related with this kind of measurements, such as the ones related with the very low signal amplitudes to be measured, since very low concentrations, in the order of tens of p.p.b., of some heavy metals, can be very dangerous for human life and for ecosystems sustainability. Additional challenges, that are associated with online heavy metals measurements, are related with the capability to obtain accurate results using a low number of measurement points. Thus, the main goal of this paper is a comparison of different data processing algorithms that can be used to improve heavy metal measurement accuracy when a segmented voltammetric voltage scan is performed with a low number of measurement points. Regarding data processing of the measurement data, B-Spline, Gaussian and artificial neural network based techniques are compared with traditional least mean square techniques based on polynomial curve fitting. The performance of each technique is evaluated in terms of the required number of measurement points, for a given root mean square deviation between curve fitted and experimental data. A brief comparison of the different techniques, in terms of insensitivity to errors caused by measurement data outliers, is also presented.

A Noncontact Method for the Detection and Diagnosis of Surface Damage in Immersed Structures

Detection and diagnosis method is proposed for surface damage in immersed structures. It is based on noncontact ultrasonic echography measurements, signal processing tools, and artificial intelligence methods. Significant features are extracted from the measured signals and a classification method is developed to detect the echoes resulting from surface damage in an immersed structure. The identification of the damage is also provided. Gaussian neural networks trained with a specific learning algorithm are developed for this purpose. The performance of the method is validated by laboratory experiments which indicate that this method could be suitable for the monitoring of inaccessible systems like marine turbines whose unavailability causes severe economic losses.

Improved adaptive neural network control for humanoid robot hand in workspace

In order to improve the control performance of humanoid robot hand in workspace, an adaptive control method based on improved neural network was proposed. rival-penalized competitive learning and recursive orthogonal least-squares algorithms were applied to reinforce the learning capability of Gaussian radial basis function neural network and realize the real-time of neural network. Moreover, an improved neural network model for humanoid robot hand was established with Ge-Lee matrix and its operator, and a controller was designed. Finally, an example of humanoid robot hand finger was provided. The results showed that the proposed control method could effectively control the unknown nonlinear dynamic properties and load disturbances of the finger with a much smaller tracking errors.

User Authentication Based on Keystroke Dynamics

ABSTRACTThis paper presents a technique to verify user identity using keystroke dynamics from short text, namely the computer login string. The keystroke behavioural pattern is obtained when a person types with a QWERTY keyboard. Two features hold time of an individual key and the latency of the consecutive keystrokes is used for authentication. Using a small training sample, accuracies of 90% and 99% are achieved for the data-set of 220 login strings per user (40 strings from legal user + 180 strings from nine intruders) using Gaussian mixture model and two-layer feed-forward neural network, respectively, as classifier. The paper then proceeds to a comprehensive study to explain how the accuracy varies with the length of the input string and with negative data in the training set.

Automated classification of brain tumours from short echo time in vivo MRS data using Gaussian Decomposition and Bayesian Neural Networks

Neuro-oncologists must ultimately rely on their acquired knowledge and accumulated experience to undertake the sensitive task of brain tumour diagnosis. This task strongly depends on indirect, non-invasive measurements, which are the source of valuable data in the form of signals and images. Expert radiologists should benefit from their use as part of an at least partially automated computer-based medical decision support system. This paper focuses on Magnetic Resonance Spectroscopy signal analysis and illustrates a method that combines Gaussian Decomposition, dimensionality reduction by Moving Window with Variance Analysis and classification using adaptively regularized Artificial Neural Networks. The method yields encouraging results in the task of binary classification of human brain tumours, even for tumour types that have seldom been analyzed from this viewpoint.

Anisotropic Probabilistic Neural Network for Image Interpolation

This study proposes a novel image interpolation method based on an anisotropic probabilistic neural network (APNN). The proposed method uses an anisotropic Gaussian kernel to improve image interpolation, which causes blurred edges. The objective of this anisotropic Gaussian kernel-based probabilistic neural network is to provide high adaptivity of smoothness/sharpness during image/video interpolation. This APNN interpolation method adjusts the smoothing parameters for varied smooth/edge regions, and considers edge direction. This APNN uses a single neuron to estimate sharpness/smoothness. The proposed method achieves better sharpness enhancement at edge regions, and reveals the noise reduction at smooth region. This study also uses interpolating a slanted-edge image to reveal blurring and blocking effects. Finally, this study compares the performance of these proposed methods with other image interpolation methods.

Forecasting Energy Commodity Prices Using Neural Networks

A new machine learning approach for price modeling is proposed. The use of neural networks as an advanced signal processing tool may be successfully used to model and forecast energy commodity prices, such as crude oil, coal, natural gas, and electricity prices. Energy commodities have shown explosive growth in the last decade. They have become a new asset class used also for investment purposes. This creates a huge demand for better modeling as what occurred in the stock markets in the 1970s. Their price behavior presents unique features causing complex dynamics whose prediction is regarded as a challenging task. The use of a Mixture of Gaussian neural network may provide significant improvements with respect to other well-known models. We propose a computationally efficient learning of this neural network using the maximum likelihood estimation approach to calibrate the parameters. The optimal model is identified using a hierarchical constructive procedure that progressively increases the model complexity. Extensive computer simulations validate the proposed approach and provide an accurate description of commodities prices dynamics.

The rate of approximation of Gaussian radial basis neural networks in continuous function space

There have been many studies on the dense theorem of approximation by radial basis feedforword neural networks, and some approximation problems by Gaussian radial basis feedforward neural networks (GRBFNs) in some special function space have also been investigated. This paper considers the approximation by the GRBFNs in continuous function space. It is proved that the rate of approximation by GRNFNs with n d neurons to any continuous function f defined on a compact subset K ⊂ ℝ d can be controlled by ω(f,n −1/2), where ω(f, t) is the modulus of continuity of the function f.

CONSTRUCTIVE APPROXIMATION BY GAUSSIAN NEURAL NETWORKS

In this paper, we discuss a constructive approximation by Gaussian neural networks. We show that it is possible to construct Gaussian neural networks with integer weights that approximate arbitrarily well for functions in <TEX>$C_c(\mathbb{R}^s)$</TEX>. We demonstrate numerical experiments to support our theoretical results.

Presenting a Hybrid Method in Order to Predict the 2009 Pandemic Influenza A (H1N1)

By the emergence and rapid spread of 2009 pandemic influenza A (H1N1) virus through the world, several methods have been developed to predict and prevent this lethal disease. Although many efforts have been made by statistical and traditional intelligent methods to anticipate this disease, but none of them could satisfy the expectations of specialists. This paper aims to present an efficient hybrid method to predict H1N1 with a reliable sensitivity. In this way, three methods including Gaussian mixture model (GMM), neural network (NN), and fuzzy rule-based system (FRBS) have been fused in order to provide an accurate and reliable prediction scheme to anticipate presence of H1N1 influenza. In this study, 230 individuals were participated and their clinical data were collected. The proposed hybrid scheme was implicated and the results showed to be superior to using each of the decision components containing NN, FRBS, and GMM classifiers. The achieved results produced 95.83% sensitivity and 80.95% specificity on unseen data which support the effectiveness of the hybrid method to predict the influenza in its golden time.

Transmission Limits Analysis of a 40 Channels DWDM System at 10 Gb/s Without Amplification

In this paper, a trajectory tracking control for a nonholonomic mobile robot subjected to kinematic disturbances is proposed. A variable structure controller based on the sliding mode theory is designed, and applied to compensate these disturbances. To minimize the problems found in practical implementations of the classical variable structure controllers, and eliminate the chattering phenomenon, is used a neural compensator, which is nonlinear and continuous, in lieu of the discontinuous portion of the control signals present in classical forms. This proposed neural compensator is designed by the Gaussian radial basis function neural networks modeling technique and it does not require the time-consuming training process. Stability analysis is guaranteed based on the Lyapunov method. Simulation results are provided to show the effectiveness of the proposed approach.

Gaussian radial basis function neural network controller of a synchronous reluctance motor in electric motorcycle applications

In this article, a sliding mode control (SMC) design based on a Gaussian radial basis function neural network (GRBFNN) is proposed for the synchronous reluctance motor (SynRM) system in electrical motorcycle applications. The conventional SMC assumes that the upper lumped boundaries of parameter variations and external disturbances are known, and the sign function is used. This causes high-frequency chattering and the high-gain phenomenon. In order to avoid these drawbacks, the proposed method utilizes the Lyapunov stability method and the steep descent rule to guarantee the convergence asymptotically, and reduce the magnitude of the chattering or avoid it completely. Finally, numerical simulations are shown to illustrate the good performance of our controller design.

Identification of nonwoven uniformity using generalized Gaussian density and fuzzy neural network

A joint method to identify nonwoven uniformity by combining wavelet transform, generalized Gaussian density (GGD) and generalized dynamic fuzzy (GDF) neural network is presented in this paper. Six hundred and twenty‐five nonwoven images of five different grades, 125 images of each grade, are decomposed at three different levels with coif4 wavelet base. Wavelet coefficients in each subband are independently modeled by GGD model, while the scale and shape parameters of that are extracted as input features of GDF neural network. For comparison, two energy‐based features are also extracted from wavelet coefficients directly, the number of which is the same as the scale and shape parameters estimated from GGD model with maximum likelihood (ML) estimator. Experimental results on the 625 nonwoven images indicate the GGD model parameters are more expressive and powerful in characterizing textures than the energy‐based ones. The proposed method has high identification accuracy, such as when the images are decomposed at Level 3 and described with GGD model parameters, the identification accuracies of five grades are all 100%. Additionally, to reduce the redundancy of the generated fuzzy rules, an effective complementary approach, fuzzy rule base reduction based on ‘CityBlock’ distance is proposed.

Advances in biological time series prediction by neural networks

Abstract In this paper, we consider some different aspects involved in the prediction of biological time series. This is often a difficult task, which can be accomplished by defining a suited function approximation problem whose inputs are determined using past samples of the sequence to be predicted. We demonstrate that neural networks can be very useful in this regard. For instance, we illustrate how Mixture of Gaussian neural networks can be efficiently used for the prediction of glucose temporal evolution in Diabetes Mellitus Type 2, instead of using the well-known MINMOD model. We also focus on the chaotic nature of biological time series, which can be contaminated by spurious noise. Neural networks represent a powerful tool for the prediction of these sequences, when the past samples to be used for prediction are chosen by suitable embedding techniques. A new embedding approach is proposed in the paper, using a genetic algorithm where each individual represents a possible embedding solution. We demonstrate that the proposed technique yields better prediction accuracies, considering as a particular application the prediction of those biological parameters (glucose, heart rate, and skin conductivity) that are strictly related to road safety, since they may evidence stress conditions or possible loss of consciousness.

Nonholonomic Mobile Robot with Kinematic Disturbances in the Trajectory Tracking: A Variable Structure Controller

In this paper, a trajectory tracking control for a nonholonomic mobile robot subjected to kinematic disturbances is proposed. A variable structure controller based on the sliding mode theory is used, and applied to compensate these disturbances. To minimize the problems found in practical implementations of the classical variable structure controllers, and eliminate the chattering phenomenon a neural compensator is used, which is nonlinear and continuous, in lieu of the discontinuous portion of the control signals present in classical forms. The proposed neural compensator is designed by the Gaussian radial basis function neural networks modeling technique and does not require the time-consuming training process. Stability analysis is guaranteed based on the Lyapunov method. Simulation results are provided to show the effectiveness of the proposed approach.

Synchronous reluctance motor speed drive using sliding mode controller based on Gaussian radial basis function neural network

In this article, a sliding mode control (SMC) design based on a Gaussian radial basis function neural network (GRBFNN) is proposed for a synchronous reluctance motor (SynRM) system robust stabilization and disturbance rejection. This method utilizes the Lyapunov function and the steep descent rule to guarantee the convergence of the SynRM drive system asymptotically. Finally, we employ experiments to validate the proposed method.

Intelligent controller design with UUB stability using Gaussian wavelet neural networks

PurposeA chaotic system is a nonlinear deterministic system that displays complex, noisy‐like and unpredictable behavior. The interest in chaotic systems lies mostly upon their complex, unpredictable behavior, and extreme sensitivity to initial conditions as well as parameter variations. Based on wavelet neural network's (WNN) online approximation ability, the purpose of this paper is to propose an adaptive Gaussian wavelet neural control (AGWNC) system to control a chaotic system.Design/methodology/approachThe proposed AGWNC system is composed of a wavelet neural controller and a compensation tangent controller. The wavelet neural controller utilizes a Gaussian WNN to mimic an ideal controller, and the compensation tangent controller is designed to compensate the approximation error between the ideal and the wavelet neural controllers. The controller parameters of the proposed AGWNC can online tune in the Lyapunov sense, thus the uniformly ultimately bounded stability of closed‐loop system can be guaranteed.FindingsThe proposed AGWNC system is applied to a chaotic system. Simulation results are used to demonstrate the effectiveness and performance of the proposed AGWNC scheme. Simulation results show that not only the favorable control performance can be achieved but also the control efforts without any chattering phenomena. Moreover, all controller parameters can be online tuning by the derived adaptive laws based on the Lyapunov function.Originality/valueThe proposed AGWNC approach is interesting for the design of an intelligent control scheme. The main contributions of this paper are: the overall closed‐loop control system is globally stable in uniform ultimate boundedness; the tracking error can be asymptotically attenuated to a desired small level around zero by appropriate chosen parameters and learning rates; and the AGWNC system can achieve favorable tracking performance.

NEURAL ADAPTIVE CONTROL FOR VIBRATION SUPPRESSION IN COMPOSITE FIN-TIP OF AIRCRAFT

In this paper, we present a neural adaptive control scheme for active vibration suppression of a composite aircraft fin tip. The mathematical model of a composite aircraft fin tip is derived using the finite element approach. The finite element model is updated experimentally to reflect the natural frequencies and mode shapes very accurately. Piezo-electric actuators and sensors are placed at optimal locations such that the vibration suppression is a maximum. Model-reference direct adaptive neural network control scheme is proposed to force the vibration level within the minimum acceptable limit. In this scheme, Gaussian neural network with linear filters is used to approximate the inverse dynamics of the system and the parameters of the neural controller are estimated using Lyapunov based update law. In order to reduce the computational burden, which is critical for real-time applications, the number of hidden neurons is also estimated in the proposed scheme. The global asymptotic stability of the overall system is ensured using the principles of Lyapunov approach. Simulation studies are carried-out using sinusoidal force functions of varying frequency. Experimental results show that the proposed neural adaptive control scheme is capable of providing significant vibration suppression in the multiple bending modes of interest. The performance of the proposed scheme is better than the H(infinity) control scheme.