Search

In this paper, we propose a new robust output feedback control approach for flexible-joint electrically driven (FJED) robots via the observer dynamic surface design technique. The proposed method only requires position measurements of the FJED robots. To estimate the link and actuator velocity information of the FJED robots with model uncertainties, we develop an adaptive observer using self-recurrent wavelet neural networks (SRWNNs). The SRWNNs are used to approximate model uncertainties in both robot (link) dynamics and actuator dynamics, and all their weights are trained online. Based on the designed observer, the link position tracking controller using the estimated states is induced from the dynamic surface design procedure. Therefore, the proposed controller can be designed more simply than the observer backstepping controller. From the Lyapunov stability analysis, it is shown that all signals in a closed-loop adaptive system are uniformly ultimately bounded. Finally, the simulation results on a three-link FJED robot are presented to validate the good position tracking performance and robustness of the proposed control system against payload uncertainties and external disturbances. Index Terms--Adaptive observer, dynamic surface control (DSC), flexible-joint electrically driven (FJED) robots, output feedback, robust control, self-recurrent wavelet neural network (SRWNN).

To diagnose the electric power apparatus, the decision tree method can be a highly recommended classification tool because it provides the if-then-rule in visible, and thus we may have a possibility to connect the physical phenomena to the observed signals. The most important point in constructing the diagnosing system is to make clear the relations between the faults and the corresponding signals. Such a database system can be built up in the laboratory using a model electric power apparatus, and we have made it. The next important thing is the feature extraction. We used [phi] - V - n patterns and POW patterns for feature variables, and feature extraction is made by the extended moments, usual moments, and the parameters in the underlying distributions such as the generalized normal distribution and the Weibuli distribution. By simple arrangements, we will be able to classify the faults and noise with high accuracy such that the misclassification rate is lower than 5%. If we set appropriate pre-processing procedure carefully, we might have a possibility of classification accuracy of less than 2%. Therefore, the decision tree with adequate feature extraction is considered to be a promising method as one of the classification tools. Index Terms--Decision tree, [phi] - V - n pattern, POW, generalized normal distribution, Weibull distribution, neural networks, GIS, model, accuracy.

This paper describes a new algorithm with neuron-by-neuron computation methods for the gradient vector and the Jacobian matrix. The algorithm can handle networks with arbitrarily connected neurons. The training speed is comparable with the Levenberg-Marquardt algorithm, which is currently considered by many as the fastest algorithm for neural network training. More importantly, it is shown that the computation of the Jacobian, which is required for second-order algorithms, has a similar computation complexity as the computation of the gradient for first-order learning methods. This new algorithm is implemented in the newly developed software, Neural Network Trainer, which has unique capabilities of handling arbitrarily connected networks. These networks with connections across layers can be more efficient than commonly used multilayer perceptron networks. Index Terms--Learning. neural network.

This paper presents a new and efficient integrated neural fuzzy approach for transformer fault diagnosis using dissolved gas analysis. The proposed approach formulates the modeling problem of higher dimensions into lower dimensions by using the input feature selection based on competitive learning and neural fuzzy model. Then, the fuzzy rule base for the identification of fault is designed by applying the subtractive clustering method which is very good at handling the noisy input data. Verification of the proposed approach has been carried out by testing on standard and practical data. In comparison to the results obtained from the existing conventional and neural fuzzy techniques, the proposed method has been shown to possess superior performance in identifying the transformer fault type. Index Terms--Cluster centers, neural-fuzzy model, self-organizing network, subtractive clustering, transformer fault diagnosis.

Presented is a new computer-aided multispectral image processing method which is used in three spatial dimensions and one spectral dimension where the dynamic, contrast enhanced magnetic resonance parameter maps derived from voxel-wise model-fitting represent the spectral dimension. The method is based on co-occurrence analysis using a 3-D window of observation which introduces an automated identification of suspicious lesions. The co-occurrence analysis defines 21 different statistical features, a subset of which were input to a neural network classifier where the assessments of the voxel-wise majority of a group of radiologist readings were used as the gold standard. The voxel-wise true positive fraction (TPF) and false positive fraction (FPF) results of the computer classifier were statistically indistinguishable from the TPF and FPF results of the readers using a one sample paired g-test. In order to observe the generality of the method, two different groups of studies were used with widely different image acquisition specifications. Index Terms--Breast, dynamic, contrast enhanced magnetic resonance imaging (DCE-MRI), multispectral image processing, neural networks, statistical co-occurrence analysis.

Labor productivity is a fundamental piece of information for estimating and scheduling a construction project. The current practice of labor productivity estimation relies primarily on either published productivity data or an individual's experience. There is a lack of a systematic approach to measuring and estimating labor productivity. Although historical project data hold important predictive productivity information, the lack of a consistent productivity measurement system and the low quality of historical data may prevent a meaningful analysis of labor productivity. In response to these problems, this paper presents an approach to measuring productivity, collecting historical data, and developing productivity models using historical data. This methodology is applied to model steel drafting and fabrication productivities. First, a consistent labor productivity measurement system was defined for steel drafting and shop fabrication activities. Second, a data acquisition system was developed to collect labor productivity data from past and current projects. Finally, the collected productivity data were used to develop labor productivity models using such techniques as artificial neural network and discrete-event simulation. These productivity models were developed and validated using actual data collected from a steel fabrication company. DOI: 10.1061/(ASCE)0733-9364(2008)134:10(786) CE Database subject headings: Productivity; Measurement; Data collection; Simulation; Neural networks; Construction management.

The use of a neural network (NN) as a feedforward control element to compensate for nonlinear system uncertainties has been investigated for over a decade. Typical NN-based controllers yield uniformly ultimately bounded (UU-B) stability results due to residual functional reconstruction inaccuracies and an inability to compensate for some system disturbances. Several researchers have proposed discontinuous feedback controllers (e.g., variable structure or sliding mode controllers) to reject the residual errors and yield asymptotic results. The research in this paper describes how a recently developed continuous robust integral of the sign of the error (RISE) feedback term can be incorporated with a NN-based feedforward term to achieve semi-global asymptotic tracking. To achieve this result, the typical stability analysis for the RISE method is modified to enable the incorporation of the NN-based feedforward terms, and a projection algorithm is developed to guarantee bounded NN weight estimates. Index Terms--Adaptive control, asymptotic stability, Lyapunov methods, neural network, nonlinear systems, RISE feedback, robust control.

In this paper, a video processing methodology for a field-programmable gate array (FPGA)-based license plate recognition (LPR) system is researched. The raster scan video is used as an input with low memory utilization. During the design, Gabor filter, threshold, and connected component labeling (CCL) algorithms are used to obtain license plate region. This region is segmented into disjoint characters for the character recognition phase, where the self-organizing map (SOM) neural network is used to identify the characters. The system is portable and relatively faster than computer-based recognition systems. The robustness of the system has been tested with a large database acquired from parking lots and a highway. The memory requirements are uniquely designed to be extremely low, which enables usage of smaller FPGAs. The resulting hardware is suitable for applications where cost, compactness, and efficiency are system design constraints. Index Terms--Character recognition, field-programmable gate arrays (FPGAs), neural network applications, neural networks, self-organizing maps (SOMs).

This brief presents an efficient and scalable online learning algorithm for recurrent neural networks (RNNs) The approach is based on the real-time recurrent learning (RTRL) algorithm, whereby the sensitivity set of each neuron is reduced to weights associated with either its input or output links. This yields a reduced storage and computational complexity of O([N.sup.2]). Stochastic meta descent (SMD), an adaptive step size scheme for stochastic gradient-descent problems, is employed as means of incorporating curvature information in order to substantially accelerate the learning process. We also introduce a clustered version of our algorithm to further improve its scalability attributes. Despite the dramatic reduction in resource requirements, it is shown through simulation results that the approach outperforms regular RTRL by almost an order of magnitude. Moreover, the scheme lends itself to parallel hardware realization by virtue of the localized property that is inherent to the learning framework. Index Terms--Constrained optimization, real-time recurrent learning (RTRL), recurrent neural networks (RNNs).

In this paper, spiking neuronal models employing means, variances, and correlations for computation are introduced. We present two approaches in the design of spiking neuronal networks, both of which are applied to engineering tasks. In exploring the input-output relationship of integrate-and-fire (IF) neurons with Poisson inputs, we are able to define mathematically robust learning rules, which can be applied to multilayer and time-series networks. We show through experimental applications that it is possible to train spike-rate networks on function approximation problems and on the dynamic task of robot arm control. Index Terms--Integrate-and-fire (IF), kernel, mean interspike interval (ISI), robot arm, variance.

When a multilayer perceptron (MLP) is trained with the split-complex backpropagation (SCBP) algorithm, one observes a relatively strong dependence of the performance on the initial values. For the effective adjustments of the weights and biases in SCBP, we propose that the range of the initial values should be greater than that of the adjustment quantities. This criterion can reduce the misadjustment of the weights and biases. Based on the this criterion, the suitable range of the initial values can be estimated. The results show that the suitable range of the initial values depends on the property of the used communication channel and the structure of the MLP (the number of layers and the number of nodes in each layer). The results are studied using the equalizer scenarios. The simulation results show that the estimated range of the initial values gives significantly improved performance. Index Terms--Backpropagation (BP), initial values, multilayer perceptron (MLP).

This brief is concerned with the stability for static neural networks with time-varying delays. Delay-independent conditions are proposed to ensure the asymptotic stability of the neural network. The delay-independent conditions are less conservative than existing ones. To further reduce the conservatism, delay-dependent conditions are also derived, which can be applied to fast time-varying delays. Expressed in linear matrix inequalities, both delay-independent and delay-dependent stability conditions can be checked using the recently developed algorithms. Examples are provided to illustrate the effectiveness and the reduced conservatism of the proposed result. Index Terms--Globally asymptotically stable, linear matrix inequality (LMI), local field neural network, Lyapunov functional, recurrent neural network (RNN), static neural network.

In this brief, a new neural network model called generalized adaptive resonance theory (GART) is introduced. GART is a hybrid model that comprises a modified Gaussian adaptive resonance theory (MGA) and the generalized regression neural network (GRNN). It is an enhanced version of the GRNN, which preserves the online learning properties of adaptive resonance theory (ART). A series of empirical studies to assess the effectiveness of GART in classification, regression, and time series prediction tasks is conducted. The results demonstrate that GART is able to produce good performances as compared with those of other methods, including the online sequential extreme learning machine (OSELM) and sequential learning radial basis function (RBF) neural network models. Index Terms--Adaptive resonance theory (ART), Bayesian theorem, generalized regression neural network (GRNN), online sequential extreme learning machine.

A new feedback-linearization-based neural network (NN) adaptive control is proposed for unknown nonaffine nonlinear discrete-time systems. An equivalent model in affine-like form is first derived for the original nonaffine discrete-time systems as feedback linearization methods cannot be implemented for such systems. Then, feedback linearization adaptive control is implemented based on the affine-like equivalent model identified with neural networks. Pretraining is not required and the weights of the neural networks used in adaptive control are directly updated online based on the input-output measurement. The dead-zone technique is used to remove the requirement of persistence excitation during the adaptation. With the proposed neural network adaptive control, stability and performance of the closed-loop system are rigorously established. Illustrated examples are provided to validate the theoretical findings. Index Terms--Feedback linearization, neural networks, nonaffine nonlinear discrete-time systems, nonlinear adaptive control.

In this paper, adaptive neural network (NN) control is investigated for a class of nonlinear pure-feedback discrete-time systems. By using prediction functions of future states, the pure-feedback system is transformed into an n-step-ahead predictor, based on which state feedback NN control is synthesized. Next, by investigating the relationship between outputs and states, the system is transformed into an input-output predictor model, and then, output feedback control is constructed. To overcome the difficulty of nonaffine appearance of the control input, implicit function theorem is exploited in the control design and NN is employed to approximate the unknown function in the control. In both state feedback and output feedback control, only a single NN is used and the controller singularity is completely avoided. The closed-loop system achieves semiglobal uniform ultimate boundedness (SGUUB) stability and the output tracking error is made within a neighborhood around zero. Simulation results are presented to show the effectiveness of the proposed control approach. Index Terms--Discrete-time system, neural network, pure-feedback system.

This paper deals with the simulation of the tire/suspension dynamics by using recurrent neural networks (RNNs). RNNs are derived from the multilayer feedforward neural networks, by adding feedback connections between output and input layers. The optimal network architecture derives from a parametric analysis based on the optimal tradeoff between network accuracy and size. The neural network can be trained with experimental data obtained in the laboratory from simulated road profiles (cleats). The results obtained from the neural network demonstrate good agreement with the experimental results over a wide range of operation conditions. The NN model can be effectively applied as a part of vehicle system model to accurately predict elastic bushings and tire dynamics behavior. Although the neural network model as a black-box model, does not provide a good insight of the physical behavior of the tire/suspension system, it is a useful tool for assessing vehicle ride and noise, vibration, harshness (NVH) performance due to its good computational efficiency and accuracy. Index Terms--Dynamics, recurrent neural network (RNN), road vehicle, suspension testing, tire.

Evolutionary algorithms are a class of stochastic search methods that attempts to emulate the biological process of evolution, incorporating concepts of selection, reproduction, and mutation. In recent years, there has been an increase in the use of evolutionary approaches in the training of artificial neural networks (ANNs). While evolutionary techniques for neural networks have shown to provide superior performance over conventional training approaches, the simultaneous optimization of network performance and architecture will almost always result in a slow training process due to the added algorithmic complexity. In this paper, we present a geometrical measure based on the singular value decomposition (SVD) to estimate the necessary number of neurons to be used in training a single-hidden-layer feedforward neural network (SLFN). In addition, we develop a new hybrid multiobjective evolutionary approach that includes the features of a variable length representation that allow for easy adaptation of neural networks structures, an architectural recombination procedure based on the geometrical measure that adapts the number of necessary hidden neurons and facilitates the exchange of neuronal information between candidate designs, and a microhybrid genetic algorithm ([mu]HGA) with an adaptive local search intensity scheme for local fine-tuning. In addition, the performances of well-known algorithms as well as the effectiveness and contributions of the proposed approach are analyzed and validated through a variety of data set types. Index Terms--Artificial neural network (ANN), evolutionary algorithms, local search, multiobjective optimization, singular value decomposition (SVD).

Generation of harmonics and the existence of waveform pollution in power system networks are important problems facing the power utilities. The determination of harmonic currents injected into a power network by a nonlinear load is complicated when the supply voltage waveform to the load is distorted by other loads and not a pure sinusoid. This paper proposes a neural network solution to this problem. A recurrent neural network trained with the backpropagation through time training algorithm is used to find a way of distinguishing between the so-called load harmonics and supply harmonics, without disconnecting the load from the network. The advantage of this method is that only waveforms of voltages and currents have to be measured. This method is applicable for both single and three phase loads and could be fabricated into a commercial instrument that could be installed in substations of large customer loads, or used as a hand-held clip on instrument. This paper is particularly useful in determining whether the utility or the customer side has a higher contribution to harmonic pollution in a network. Hence, this method would be helpful in settling utility-customer disputes over who is responsible for harmonic pollution. Index Terms--Harmonics, metering, power quality, recurrent neural networks (RNNs).

An improved hybrid particle swarm optimization (PSO)-based wavelet neural network (WNN) for Modeling the development of Fluid Dispensing for Electronic Packaging (MFD-EP) is presented in this paper. In modeling the fluid dispensing process, it is important to understand the process behavior as well as determine the optimum operating conditions of the process for a high-yield, low-cost, and robust operation. Modeling the fluid dispensing process is a complex nonlinear problem. This kind of problem is suitable to be solved by applying a neural network. Among the different kinds of neural networks, the WNN is a good choice to solve the problem. In the proposed WNN, the translation parameters are variables depending on the network inputs. Due to the variable translation parameters, the network becomes an adaptive one that provides better performance and increased learning ability than conventional WNNs. An improved hybrid PSO is applied to train the parameters of the proposed WNN. The proposed hybrid PSO incorporates a wavelet-theory-based mutation operation. It applies the wavelet theory to enhance the PSO in more effectively exploring the solution space to reach a better solution. A case study of MFD-EP is employed to demonstrate the effectiveness of the proposed method. Index Terms--Modeling, particle swarm optimization (PSO), wavelet neural network (WNN), wavelet theory.

This paper proposes a novel dynamic structure neural-fuzzy network (DSNFN) via a robust adaptive sliding-mode approach to address trajectory-tracking control of an n-link robot manipulator. In the DSNFN, a five-layer neural-fuzzy network (NFN) is used to model complex processes and compensate for structured and unstructured uncertainties. However, it is difficult to find a suitable-sized NFN to achieve the required approximation error. To deal with the mentioned problem, the number of rule nodes in the DSNFN can be either increased or decreased over time based on the tracking errors, and the adaptation laws in the sense of a projection algorithm are derived for tuning all parameters of the parameterized NFN. Using DSNFN, good tracking performance could be achieved in the system. Furthermore, the trained network avoids the problems of overfitting and underfitting. The global stability and the robustness of the overall control scheme are guaranteed, and the tracking errors converge to the required precision by the Lyapunov synthesis approach. Experiments performed on a two-link robot manipulator demonstrate the effectiveness of our scheme. Index Terms--Adaptive tuning algorithm, dynamic structure neural-fuzzy network (DSNFN), robot manipulator, stability and robustness.

Artificial neural network (ANN) and regression models are developed for the estimation of concrete slump using concrete constituent data. The concrete mix constituent and slump data from laboratory tests have been employed to develop all models. The results obtained in this study demonstrate the superiority of the ANN models. It was found that combining one or more concrete mix constituents and treating them as an independent input variable is not advantageous when using regression but can be very useful when using ANNs for modeling concrete slump. Sensitivity analyses based on the ANN models were carried out to evaluate the impact of different concrete mix constituents on the slump values. It was found that the slump attains a minimum value at the critical levels of mortar and coarse aggregates, and tends to increase with paste content and decrease with sand content in the concrete mix. DOI: 10.1061/(ASCE)0899-1561(2008)20:9(628) CE Database subject headings: Artificial networks; Neural networks; Concrete; Aggregates; Estimation; Sensitivity analysis; Regression analysis.

In this note, the global asymptotic stability for delayed recurrent neural networks is addressed with a new Lyapunov--Krasovskii function. New delay-independent linear matrix inequality (LMI)-based conditions for global asymptotic stability are derived. A key feature of the new approach is the introduction auxiliary matrices, which can provide useful and less conservative results. This feature also enables us to cast a series of previous LMI-based results into even more general framework, logic flow of ideas and comparisons are thus easily shown. Three numerical examples show the effectiveness of the proposed method. Index Terms--Delayed recurrent neural networks, global asymptotic stability, LMI.

This paper presents a Hopfield neural network (HNN) combined with estimation of distribution (EDA) for the two-page crossing number problem. In the proposed algorithm, once the network is trapped in local minima, the perturbation based on EDA can generate a new starting point for the HNN for further search, which is in a promising area characterized by a probability model and is not far away from the best solution found so far. The proposed algorithm can escape from local minima and further search better results. Simulation results show that the proposed algorithm is better than previous methods. Index Terms--Combinatorial optimization problem, estimation of distribution (EDA), Hopfield neural network (HNN), two-page crossing number.

A neutral-type delayed projection neural network is proposed to deal with nonlinear variational inequalities. Compared with the existing delayed neural networks for linear variational inequalities, the proposed approach apparently has the larger application domain. By the theory of functional differential equation, a delay-dependent sufficient stability condition is derived. This stability condition is easily checked, and can guarantee that the proposed neural network is convergent to the solution of nonlinear variational inequality problem exponentially, which improves the existing stability criteria for the neutral-type delayed neural network. Moreover, many related problems, such as the projection equation and optimization problems, can also be dealt with by the proposed method. Finally, simulation examples are given to illustrate the satisfactory performance of the proposed method. Index Terms--Neutral-type delay, projection neural network, variational inequality (VI).

One of the main problems for effective control of an uncertain system is the creation of the proper knowledge base for the control system. In this paper, the integration of fuzzy set theory and wavelet neural networks (WNNs) is proposed to alleviate the problem. The proposed fuzzy WNN is constructed on the base of a set of fuzzy rules. Each rule includes a wavelet function in the consequent part of the rule. The parameter update rules of the system are derived based on the gradient descent method. The structure is tested for the identification and the control of the dynamic plants commonly used in the literature. It is seen that the proposed structure results in a better performance despite its smaller parameter space. Index Terms--Control, fuzzy wavelet neural network (FWNN), identification, wavelet.

In this work, the hourly solar radiation data collected during the period August 1, 2005-July 30, 2006 from the solar observation station in Iki Eylul campus area of Eskisehir region are studied. A two-dimensional (2-D) representation model of the hourly solar radiation data is proposed. The model provides a unique and compact visualization of the data for inspection, and enables accurate forecasting using image processing methods. Using the hourly solar radiation data mentioned above, the image model is formed in raster scan form with rows and columns corresponding to days and hours, respectively. Logically, the between-day correlations along the same hour segment provide the vertical correlations of the image, which is not available in the regular 1-D representation. To test the forecasting efficiency of the model, nine different linear filters with various filter-tap configurations are optimized and tested. The results provide the necessary correlation model and prediction directions for obtaining the optimum prediction template for forecasting. Next, the 2-D forecasting performance is tested through feed-forward neural networks (NN) using the same data. The optimal linear filters and NN models are compared in the sense of root mean square error (RMSE). It is observed that the 2-D model has pronounced advantages over the 1-D representation for both linear and NN prediction methods. Due to the capability of depicting the nonlinear behavior of the input data, the NN models are found to achieve better forecasting results than linear prediction filters in both 1-D and 2-D. Keywords: Solar radiation; Forecasting; Linear filter; NN

In a competitive electricity market, an accurate forecasting of energy prices is an important activity for all the market participants either for developing bidding strategies or for making investment decisions. An adaptive wavelet neural network (AWNN) is proposed in this paper for short-term price forecasting (STPF) in the electricity markets. A commonly used Mexican hat wavelet has been chosen as the activation function for hidden-layer neurons of feed-forward neural network (FFNN). To demonstrate the effectiveness of the proposed approach, day-ahead prediction of market clearing price (MCP) of Spain market, which is a duopoly market with a dominant player, and locational marginal price (LMP) forecasting in PJM electricity market, are considered. The forecasted results clearly show that AWNN has good prediction properties compared to other forecasting techniques, such as wavelet-ARIMA, multi-layer perceptron (MLP) and radial basis function (RBF) neural networks as well as recently proposed fuzzy neural network (FNN). Index Terms--Adaptive wavelet neural network, electricity markets, feed-forward neural network, locational marginal price forecasting, short-term price forecasting.

With the appearance of electricity markets, the variation of the price of electricity will influence usage custom of electric energy. This will complicate short-term load forecasting and challenge the existing forecasting methods that are applied to a fixed-price environment. In regard to the influence of real-time electricity prices on short-term load, a model to forecast short. term load is established by combining the radial basis function (RBF) neural network with the adaptive neural fuzzy inference system (ANFIS). The model first makes use of the nonlinear approaching capacity of the RBF network to forecast the load on the prediction day with no account of the factor of electricity price, and then, based on the recent changes of the real-time price, it uses the ANFIS system to adjust the results of load forecasting obtained by RBF network. This system integration will improve forecasting accuracy and overcome the defects of the RBF network. As shown in this paper by the results of an example of factual forecasting, the model presented can work effectively. Index Terms--Adaptive neural fuzzy inference system, power system, radial basis function neural network, real-time price, short-term load forecasting.

Remote sensing estimation of impervious surface is significant in monitoring urban development and determining the overall environmental health of a watershed, and it has therefore attracted more interest recently in the remote sensing community. The main objective of this paper is to examine and compare the effectiveness of two advanced algorithms for estimating impervious surfaces from medium spatial resolution satellite images, namely, linear spectral mixture analysis (LSMA) and artificial neural network (ANN). Terra's Advanced Spaceborne Thermal Emission and Reflection Radiometer [(ASTER); acquired on June 16, 2001] anda Landsat Enhanced Thematic Mapper Plus (ETM+) image (acquired on June 22, 2000) of Indianapolis, IN, were used for the analysis. The LSMA was employed to generate high- and low-albedo, vegetation, and soil fraction images (endmembers), and an image of impervious surfaces was then estimated by adding high- and low-albedo fraction images. Furthermore, an ANN model, specifically the multilayer-perceptron feed-forward network with the back-propagation learning algorithm, was employed as a subpixel image classifier to estimate impervious surfaces. Accuracy assessment was performed against a high-resolution digital orthophoto. The results show that ANN was more effective than LSMA in generating impervious surfaces with high statistical accuracy. For the ASTER image, the root-mean-square error (RMSE) of the impervious surface map with the ANN model was 12.3%, and the one that resulted from LSMA was 13.2%. For the ETM+ image, the RMSE with the ANN model was 16.7%, and the one from LSMA was 18.9%. The better performance of ANN over LSMA is mainly attributable to the ANN's capability of handling the nonlinear mixing of image spectrum. In order to test the seasonal sensitivity of satellite images for estimating impervious surfaces, LSMA was applied to two additional ASTER images of the same area, which are acquired on April 5, 2004, and October 3, 2000, respectively. The results were then compared with the ASTER image acquired in June in terms of RMSE. The June image had the highest accuracy, whereas the October image was better than the one in April. Plant phenology caused changes in the variance partitioning and impacted the mixing-space characterization, leading to a less accurate estimation of impervious surfaces. Index Terms--Artificial neural network (ANN), impervious surface estimation, linear spectral mixture analysis (LSMA), medium-resolution satellite images.

We developed a method for use of accelerometers and force sensing resistors (FSRs) within an optimal controller of walking for hemiplegic individuals. The data from four dual-axis accelerometers and four FSRs were inputs, while six muscle activation profiles were outputs. The controller includes two stages: 1) estimating the target gait pattern using artificial neural networks; and 2) optimal control minimizing tracking errors (from the estimated gait pattern) and muscle efforts. The controller was tested using data collected from six healthy subjects walking at five speeds (0.6-1.4 m/s). The average root mean square errors (RMSEs) normalized by the peak-to-peak value of the target signals [normalized RMSE (NRMSE)] were below 6%, 7%, 8%, and 3% for estimation of joint angles, hip acceleration, ground reaction force, and movement of the center of pressure, respectively. Using the estimated data as inputs, the simulation generated the target healthy-like gait patterns and reproducible muscle activation profiles in 90% of 300 tested gait trials. Overall tracking NRMSE was between 2% and 9%. The optimal controller was developed for testing the feasibility of healthy-like gait patterns in hemiplegic individuals, and generating a knowledge base that is required for the synthesis of a sensory-driven control of walking assisted by functional electrical stimulation. Index Terms--Artificial neural networks (ANNs), functional electrical stimulation (FES), gait, machine learning, optimal control, sensory system.

This brief is concerned with the analysis problem of global exponential stability in the mean square sense for a class of linear discrete-time recurrent neural networks (DRNNs) with stochastic delay. Different from the prior research works, the effects of both variation range and probability distribution of the time delay are involved in the proposed method. First, a modeling method is proposed by translating the probability distribution of the time delay into parameter matrices of the transformed DRNN model, where the delay is characterized by a stochastic binary distributed variable. Based on the new method, the global exponential stability in the mean square sense for the DRNNs with stochastic delay is investigated by using the Lyapunov-Krasovskii functional and exploiting some new analysis techniques. A numerical example is provided to show the effectiveness and the applicability of the proposed method. Index Terms--Delay distribution dependent, discrete-time recurrent neural networks (DRNNs), exponential stability, linear matrix inequality (LMI), stochastic delay.

The design and implementation of an optimal neurocontroller for a flexible ac transmission device---the unified power-flow controller (UPFC)--is presented in this paper. Wide area signals in a power system are used to provide auxiliary control to a UPFC in order to achieve enhanced damping of system oscillations. The neurocontroller provides auxiliary signals to the real and reactive power references of a UPFC series inverter. The design of the optimal neurocontroller is based on an adaptive critic design approach--the heuristic dynamic programming. A system identifier, referred to as the wide-area monitor), and a critic network (performance evaluator) are designed for optimizing the neurocontroller. Real-time implementation of the optimal UPFC neurocontroller for a multimachine power system is carried out successfully on a digital signal processor. The power system is simulated on a real-time digital simulator. The performance of this neurocontroller is compared with a conventional linear damping controller. Results show enhanced damping of both inter-area and intra-area modes in the power system under different operating conditions and disturbances with the neurocontroller. The improvement in the damping is also quantified using the Prony method. Index Terms--Flexible ac transmission systems (FACTS) device, heuristic dynamic programming (HDP), optimal neurocontroller, prony analysis, unified power-flow controller (UPFC), wide-area control, wide-area monitor (WAM).

This paper deals with the problem of fault section estimation in electric power systems, undertaken at a control center level and using information about the operation of protection relays and circuit breakers. The developed methodology should be used after the occurrence of contingencies with definitive disconnections, and before beginning the process of network restoration. Due to the absence of an analytic formulation, the problem calls for the use of artificial-intelligence techniques, such as neural networks and expert systems. Neural networks are employed to model the protection systems, dealing with the uncertainties involved with relay and circuit-breaker operation messages. An expert system is used to complement the results provided by the neural networks, considering the network topology. The results show that the developed methodology is applicable to real large-scale power systems. In addition, it is capable of noise suppression in relay and circuit-breaker trip messages, treats multiple faults naturally, and infers a solution even in cases when remote backup protection action occurs. Index Terms--Fault section estimation, fuzzy expert systems, neural networks, power system protection.

The magnetic saturation and strong nonlinearity of switched reluctance machines (SRMs) makes it very difficult to derive a comprehensive mathematical model for the behavior of the machine. We propose a new method of modeling SRMs based on an adaptive neural fuzzy inference system (ANFIS). First, we use an indirect method to measure the static flux linkage and then use the co-energy method (via the principle of virtual displacement) to calculate the torque characteristics from data on flux linkage versus current and rotor position. A hybrid learning algorithm, which combines the back propagation algorithm and the linear least-squares estimation algorithm, identifies the parameters of the ANFIS. After training, the ANFIS flux linkage model and ANFIS torque model are in excellent agreement with experimental flux linkage measurements and the calculated torque data. Finally, we use an ANFIS current model and an ANFIS torque model to study SRM dynamic performance. The accuracy of the model was evaluated by comparison to laboratory measurements of the machine's current-speed and torque-speed characteristics. The model is quite accurate. Index Terms--Adaptive neural fuzzy inference system, back propagation, least-squares, switched reluctance machine.

This paper compares the output from four methods used to predict the mean overtopping discharge at the root of the South Breakwater of Povoa de Varzim Harbour (Portugal): two numerical models AMAZON, based on solving the nonlinear shallow- water equations, and COBRAS-UC, based on the Reynolds averaged Navier-Stokes equations, Pedersen's empirical formula, and the neural network tool NN_Overtopping. Output from the methods is considered alongside data from two-dimensional (2D) physical model tests. The results show that rather accurate predictions can be obtained with COBRAS-UC. Agreement between the data and the AMAZON output depends principally on the impact of porosity on overtopping: agreement is very good when the impact is limited (for high overtopping discharges) and it worsens as porosity plays a more important role (for small discharges). NN_Overtopping estimates the mean discharges with a degree of accuracy similar to AMAZON. Pedersen's results are those which deviate more from the data, substantially overpredicting the mean discharges for the conditions analyzed. DOI: 10.1061/(ASCE)0733-950X(2008)134:4(226) CE Database subject headings: Numerical models; Model tests; Empirical equations; Neural networks; Breakwaters; Portugal; Coastal structures; Coastal engineering.

Problems in geotechnical engineering are full of uncertain, vague, and incomplete information. In most instances, successfully solving such problems depends on experts' knowledge and experience. The primary object of this research was to develop an evolutionary fuzzy neural inference system (EFNIS) to imitate the decision-making processes in the human brain in order to facilitate geotechnical expert decision making. First, an evolutionary fuzzy neural inference model (EFNIM) was constructed by combining the genetic algorithm (GA), fuzzy logic (FL), and neural network (NN). In the proposed model, GA is primarily concerned with optimizing parameters required in the fuzzy neural network; FL with imprecision and approximate reasoning; and NN with learning and curve fitting. This research then integrates the EFNIM with an object-oriented computer technique to develop an EFNIS. Finally, the potential to apply the proposed system to practical geotechnical decision making is validated using two real problems, namely estimating slurry wall duration and selecting retaining wall construction methods. CE Database subject headings: Artificial intelligence; Algorithms; Neural networks; Decision making; Fuzzy sets; Geotechnical engineering.

This brief deals with the stability analysis problem for recurrent neural networks with delay. An improved stability condition is derived to guarantee the existence of the unique equilibrium point and its globally exponential stability, which is shown with novel methods. Both delay-dependent and delay-independent stability conditions are obtained. Expressed in terms of LMIs, they can be checked using the numerically efficient Matlab LMI toolbox. Examples are provided to demonstrate the effectiveness and the reduced conservatism of the analysis results. Index Terms--Delay-dependent, globally exponential stable, linear matrix inequality (LMI), local field neural networks, recurrent neural networks (RNNs), static neural networks.

Neural Networks trained with error back propagation Levenberg-Marquardt training (LM), Neural Networks trained with an algebraic method and Support Vector Machines (SVM) were evaluated to extract the position information from measured light distributions generated by the interactions of 511 keV photons in monolithic scintillator blocks. All three algorithms can achieve a similar average resolution (~1.6 mm FWHM in a 20 x 10 x 10 mm LSO block) but the LM trained neural networks do so most efficiently. When the incidence angle of the photons increases to 30[degrees], the resolution degrades slightly to 2.0 mm FWHM. A small mismatch (< [+ or -] 5[degrees]) between the true incidence angle and the angle for which a neural network was trained can be tolerated without significant resolution loss. Increasing the thickness to 20 mm and using a top-bottom readout of the block yields an average resolution of 2.2 nun FWHM. Index Terms--Monolithic scintillator, positron emission tomography, neural network.

A novel artificial neural network (NN)-based technique is proposed for enabling smart sensors to operate in harsh environments. The NN-based sensor model automatically linearizes and compensates for the adverse effects arising due to nonlinear response characteristics and nonlinear dependency of the sensor characteristics on the environmental variables. To show the potential of the proposed NN-based technique, we have provided results of a smart capacitive pressure sensor (CPS) operating under a wide range of temperature variation. A multilayer perceptron is utilized to transfer the nonlinear CPS characteristics at any operating temperature to a linearized response characteristics. Through extensive simulated experiments, we have shown that the NN-based CPS model can provide pressure readout with a maximum full-scale error of only [+ or -] 1.5% over a temperature range of -50 to 200 [degrees]C with excellent linearized response for all the three forms of nonlinear dependencies considered. Performance of the proposed technique is compared with a recently proposed computationally efficient NN-based extreme learning machine. The proposed multilayer perceptron based model is tested by using experimentally measured real sensor data, and found to have satisfactory performance. Index Terms--Artificial neural networks (NNs), auto-compensation, harsh environment, intelligent and smart sensors, linearization, pressure sensor.

This paper proposes a traffic monitoring architecture based oil a high-speed communication network whose nodes are equipped with fuzzy processors and cellular neural network (CNN) embedded systems. It implements a real-time mobility information system where visual human perceptions sent by people working on the territory and video-sequences of traffic taken from webcams are jointly processed to evaluate the fundamental traffic parameters fur every street of a metropolitan area. This paper presents the whole methodology for data collection and analysis and compares the accuracy and the processing time of the proposed soft computing techniques with other existing algorithms. Moreover, this paper discusses when and why it is recommended to fuse the visual perceptions of the traffic with the automated measurements taken from the webcams to compute the maximum traveling time that is likely needed to reach any destination in the traffic network. Index Terms--Cellular neural networks (CNNs), computing with words, fuzzy logic, intelligent systems, machine vision.

In this paper, the recurrent neural networks (RNNs) with a generalized activation function class is proposed. In this proposed model, every component of the neuron's activation function belongs to a convex hull which is bounded by two odd symmetric piecewise linear functions that are convex or concave over the real space. All of the convex hulls are composed of generalized activation function classes. The novel activation function class is not only with a more flexible and more specific description of the activation functions than other function classes but it also generalizes some traditional activation function classes. The absolute exponential stability (AEST) of the RNN with a generalized activation function class is studied through three steps. The first step is to demonstrate the global exponential stability (GES) of the equilibrium point of original RNN with a generalized activation function being equivalent to that of RNN under all vertex functions of convex hull. The second step transforms the RNN under every vertex activation function into neural networks under an array of saturated linear activation functions. Because the GES of the equilibrium point of three systems are equivalent, the next stability analysis focuses on the GES of the equilibrium point of RNN system under an array of saturated linear activation functions. The last step is to study both the existence of equilibrium point and the GES of the RNN under saturated linear activation functions using the theory of M-matrix. In the end, a two-neuron RNN with a generalized activation function is constructed to show the effectiveness of our results. Index Terms--Absolute exponential stability (AEST), convex hull, generalized activation function class, piecewise linear function, recurrent neural networks (RNNs).

To support large numbers of model neurons, neuromorphic vision systems are increasingly adopting a distributed architecture, where different arrays of neurons are located on different chips or processors. Spike-based protocols are used to communicate activity between processors. The spike activity in the arrays depends on the input statistics as well as internal parameters such as time constants and gains. In this paper, we investigate strategies for automatically adapting these parameters to maintain a constant firing rate in response to changes in the input statistics. We find that under the constraint of maintaining a fixed firing rate, a strategy based upon updating the gain alone performs as well as an optimal strategy where both the gain and the time constant are allowed to vary. We discuss how to choose the time constant and propose an adaptive gain control mechanism whose operation is robust to changes in the input statistics. Our experimental results on a mobile robotic platform validate the analysis and efficacy of the proposed strategy. Index Terms--Adaptive systems, distributed computing, neuromorphic systems, spiking neural networks.

The learning capability of neural networks is equivalent to modeling physical events that occur in the real environment. Several early works have demonstrated that neural networks belonging to some classes are universal approximators of input-output deterministic functions. Recent works extend the ability of neural networks in approximating random functions using a class of networks named stochastic neural networks (SNN). In the language of system theory, the approximation of both deterministic and stochastic functions falls within the identification of nonlinear no-memory systems. However, all the results presented so far are restricted to the case of Gaussian stochastic processes (SPs) only, or to linear transformations that guarantee this property. This paper aims at investigating the ability of stochastic neural networks to approximate nonlinear input-output random transformations, thus widening the range of applicability of these networks to nonlinear systems with memory. In particular, this study shows that networks belonging to a class named non-Gaussian stochastic approximate identity neural networks (SAINNs) are capable of approximating the solutions of large classes of nonlinear random ordinary differential transformations. The effectiveness of this approach is demonstrated and discussed by some application examples. Index Terms--Approximation, Lee-Schetzen method, neural computation, neural networks, nonlinear systems, stochastic processes (SPs).

In this paper, an optimized approximation algorithm (OAA) is proposed to address the overfitting problem in function approximation using neural networks (NNs). The optimized approximation algorithm avoids overfitting by means of a novel and effective stopping criterion based on the estimation of the signal-to-noise-ratio figure (SNRF). Using SNRF, which checks the goodness-of-fit in the approximation, overfitting can be automatically detected from the training error only without use of a separate validation set. The algorithm has been applied to problems of optimizing the number of hidden neurons in a multilayer perceptron (MLP) and optimizing the number of learning epochs in MLP's backpropagation training using both synthetic and benchmark data sets. The OAA algorithm can also be utilized in the optimization of other parameters of NNs. In addition, it can be applied to the problem of function approximation using any kind of basis functions, or to the problem of learning model selection when overfitting needs to be considered. Index Terms--Function approximation, neural network (NN) learning, overfitting.

In this paper, a supervised neural network training technique based on constrained optimization is developed for preserving prior knowledge of an input-output mapping during repeated incremental training sessions. The prior knowledge, referred to as long-term memory (LTM), is expressed in the form of equality constraints obtained by means of an algebraic training technique. Incremental training, which may be used to learn new short-term memories (STMs) online, is then formulated as an error minimization problem subject to equality constraints. The solution of this problem is simplified by implementing an adjoined error gradient that circumvents direct substitution and exploits classical backpropagation. A target application is neural network function approximation in adaptive critic designs. For illustrative purposes, constrained training is implemented to update an adaptive critic flight controller, while preserving prior knowledge of an established performance baseline that consists of classical gain-scheduled controllers. It is shown both analytically and numerically that the LTM is accurately preserved while the controller is repeatedly trained over time to assimilate new STMs. Index Terms--Adaptive critics, control, exploration, function approximation, incremental training, interference, knowledge acquisition and retention, memory, online learning, sigmoidal neural networks.

An unsupervised competitive neural network for efficient clustering of Gaussian probability density function (GPDF) data of continuous density hidden Markov models (CDHMMs) is proposed in this paper. The proposed unsupervised competitive neural network, called the divergence-based centroid neural network (DCNN), employs the divergence measure as its distance measure and utilizes the statistical characteristics of observation densities in the HMM for speech recognition problems. While the conventional clustering algorithms used for the vector quantization (VQ) codebook design utilize only the mean values of the observation densities in the HMM, the proposed DCNN utilizes both the mean and the covariance values. When compared with other conventional unsupervised neural networks, the DCNN successfully allocates more code vectors to the regions where GPDF data are densely distributed while it allocates fewer code vectors to the regions where GPDF data are sparsely distributed. When applied to Korean monophone recognition problems as a tool to reduce the size of the codebook, the DCNN reduced the number of GPDFs used for code vectors by 65.3% while preserving recognition accuracy. Experimental results with a divergence-based k-means algorithm and a divergence-based self-organizing map algorithm are also presented in this paper for a performance comparison. Index Terms--Clustering method, Gaussian distributions, hidden Markov models (HMM), neural networks, speech recognition.

In this paper, it is found that the weights of a perceptron are bounded for all initial weights if there exists a nonempty set of initial weights that the weights of the perceptron are bounded. Hence, the boundedness condition of the weights of the perceptron is independent of the initial weights. Also, a necessary and sufficient condition for the weights of the perceptron exhibiting a limit cycle behavior is derived. The range of the number of updates for the weights of the perceptron required to reach the limit cycle is estimated. Finally, it is suggested that the perceptron exhibiting the limit cycle behavior can be employed for solving a recognition problem when downsampled sets of bounded training feature vectors are linearly separable. Numerical computer simulation results show that the perceptron exhibiting the limit cycle behavior can achieve a better recognition performance compared to a multilayer perceptron. Index Terms--Boundedness, limit cycle, perceptron, time periodically varying neural network.

Cellular simultaneous recurrent neural network (SRN) has been shown to be a function approximator more powerful than the multilayer perceptron (MLP). This means that the complexity of MLP would be prohibitively large for some problems while SRN could realize the desired mapping with acceptable computational constraints. The speed of training of complex recurrent networks is crucial to their successful application. This work improves the previous results by training the network with extended Kalman filter (EKF). We implemented a generic cellular SRN (CSRN) and applied it for solving two challenging problems: 2-D maze navigation and a subset of the connectedness problem. The speed of convergence has been improved by several orders of magnitude in comparison with the earlier results in the case of maze navigation, and superior generalization has been demonstrated in the case of connectedness. The implications of this improvements are discussed. Index Terms--Backpropagation (BP), cellular neural network, dynamic programming (DP), Kalman filtering (KF), recurrent neural network.

Mosquito repellency data on acylpiperidines derived from the U.S. Department of Agriculture archives were modeled by using molecular descriptors calculated by CODESSA PRO software. An artificial neural network model was developed for the correlation of these archival results and used to predict the repellent activity of novel compounds of similar structures. A series of 34 promising N-acylpiperidine mosquito repellent candidates (4a-4q') were synthesized by reactions of acylbenzotriazoles 2a-2p with piperidines 3a-3f. Compounds (4a-4q') were screened as topically applied mosquito repellents by measuring the duration of repellency after application to cloth patches worn on the arms of human volunteers. Some compounds that were evaluated repelled mosquitoes as much as three times longer than N,N-diethyl-m-toluamide (DEET), the most widely used repellent throughout the world. The newly measured durations of repellency were used to obtain a superior correlation equation relating mosquito repellency to molecular structure. N-acylpiperidine | quantitative structure-activity relationship | CODESSA PRO | artificial neural network | Aedes aegypti