Showing papers on "Hybrid neural network published in 1995"

Real time preventive actions for transient stability enhancement with a hybrid neural network-optimization approach

Abstract: This paper reports a new approach in defining preventive control measures to assure transient stability relative to one or several contingencies that may occur separately in a power system. Generation dispatch is driven not only by economic functions but also with the derivatives of the transient energy margin value; these derivatives are obtained directly from a trained artificial neural network (ANN), using real time monitorable system values. Results obtained from computer simulations, for several contingencies in the CIGRE test system, confirm the validity of the developed approach. >

Colour bake inspection system using hybrid artificial neural networks

Abstract: The bake level of biscuits is of significant value to biscuit manufacturers as it determines the taste, texture and appearance of the products. Previous research explored and revealed the feasibility of biscuit bake inspection using feedforward neural networks (FFNN) with a backpropagation learning algorithm and monochrome images. A second study revealed the existence of a curve in colour space, called a baking curve, along which the bake colour changes during the baking process. Combining these results, the authors proposed an automated bake inspection system with artificial neural networks that utilises colour instead of monochrome images. In this paper, the authors present the implementation of the inspection system with a hybrid neural network of self-organising maps and FFNNs. The system was tested and its grading performance on biscuit bake levels was evaluated and compared to that of a trained human inspector. The authors found that the proposed colour system with a hybrid neural network performed significantly better than the human inspector.

Building a fuzzy expert system for electric load forecasting using a hybrid neural network

Abstract: This paper presents the development of a hybrid neural network to model a fuzzy expert system for time series forecasting of electric load. The hybrid neural network is trained to develop fuzzy logic rules and find optimal input/output membership values of load and weather parameters. A hybrid learning algorithm consisting of unsupervised and supervised learning phases is used for training the fuzzified neural network. In the supervised learning phase, both back propagation and linear Kalman filter algorithms are used for the adjustment of weights and membership functions. Extensive tests have been performed on a 2-year utility data for the generation of peak and average load profiles in 24 h, 48 h, and 168 h ahead time frame during summer and winter seasons. From the simulation results, it is observed that the fuzzy expert system using the Kalman filter-based algorithm gives faster convergence and more accurate prediction of a load time series.

A hybrid neural network approach to the pricing of options

Abstract: A new method of pricing options is introduced. The method is based on the augmentation of a conventional model with an artificial neural network (ANN) trained on the difference between the standard model and actual options data. The pricing accuracy has been demonstrated using the actual All Ordinaries Share Price Index (AO SPI) options on futures. This hybrid approach is shown to provide greater accuracy than either standard model or ANN used alone.

Hybrid neural network pattern recognition system for satellite measurements

Abstract: This paper presents a lower-hybrid cavity detection system (CDS). The CDS is used to analyse measurements of electron plasma density made by the FREJA satellite wave experiment. The system can reduce the amount of data to be analysed by as much as 96% and still retain more than 85% of the desired information. The CDS is a combination of a hybrid neural network (HNN), and expert rules. The HNN is a self organizing map, combined with a feedforward backpropagation neural net. The CDS can be controlled by the user to operate with various degrees of sensitivity. Maximum detection capability is as high as 95% with data reduction of about 85%.

An integrated hybrid neural network and hidden Markov model classifier for sonar signal classification

Abstract: Presents an integrated hybrid neural network and hidden Markov model (HMM) classifier that combines the time normalization property of the HMM classifier with the superior discriminative ability of the neural net (NN). Sonar signals display a strong time varying characteristic. Although the neural net has been successful in classifying transient like sonar signals, the success is achieved either by using a bigger net architecture or by incorporating a detection mechanism in the classification procedure. The present authors propose an integrated hybrid HMM and neural net classifier where a left-to-right HMM module is used first. The HMM module segments the observation sequence belonging to every exemplar into a fixed number of states starting from the left. After this segmentation, all the frames belonging to the same state are replaced by one average frame. Thus, every exemplar, irrespective of its time scale variation, is transformed into a fixed number of frames, i.e., a static pattern. The multilayer perceptron (MLP) neural net is then used as the classifier for these time normalized exemplars. For successful modeling and classification, each frame is succinctly represented by a feature vector. Two feature extraction schemes are considered-the first one is based on the FFT power spectral coefficient, and the second one is based on the quadrature mirror filter (QMF) bank based subband decomposition. Finally, some experimental results are provided to demonstrate the superiority of the hybrid integrated classifier.

Improving learning vector quantization classifier in machine fault diagnosis by adding consistency

Abstract: This paper presents a hybrid neural network system which combines the learning vector quantization (LVQ) classifier with the theory of consistency. The hybrid system employs consistency to measure the degree of matching between the input feature vectors and the output classes. In the calculation of the consistency, the probability distribution is embedded to describe the occurring frequencies of various classes in a neighborhood region associated with the input feature. This successfully avoids the case that usually occurs in complex classification problems of machine faults, that is, one or a few deviated input feature affecting the Euclidean distance and leads to misclassifications. Experiments shows that the consistency can improve the classification capability of LVQ by not only reducing the influence of distorted features but also making the boundaries of overlapped classes more discriminative. From the results of identifying faults occurred in a tapping machine, it is demonstrated that the successful rate of classification using this hybrid method outweighed the backpropagation and the conventional LVQ classifiers.

Expert Systems and Neural Networks

Abstract: Expert system and neural network technologies have developed to the point that the advantages of each can be combined into more powerful systems. In some cases, neural computing systems are replacing expert systems and other artificial intelligence solutions. In other applications, neural networks provide features not possible with conventional AI systems and may provide aspects of intelligent behavior that have thus far eluded AI.

Hybrid fuzzy neural systems for robust handwritten word recognition

Abstract: The research work described in this dissertation is aimed at developing improved handwritten character classifiers for use in off-line handwritten word recognition. In particular, our goal is to develop algorithms that can accurately reflect character class ambiguities and that can detect non-character inputs. Two specific hybrid models were developed in this study. We designed, implemented, and successfully demonstrated a hybrid neural network model for character confidence assignment using a cascade of a Kohonen self-organizing feature map (SOFM) and a multi-layer feedforward network (MLFN). The other novel hybrid neural network/fuzzy integral model for character confidence assignment use a cascade of the SOFM and a set of Choquet fuzzy integrals (FI). These new methodologies have resulted in significant improvements in handwritten word recognition performance. Recognition rates of over 90% were achieved using a single word recognizer and an average lexicon size of 100.

Hybrid neural network system for cloud classification from satellite images

Abstract: This paper presents an architecture and preliminary implementation results of a hybrid two-stage neural network system for cloud classification from satellite imagery. The system first performs pixel classification on the image spectral multi-channel data and descriptive data to discover possible areas covered by clouds and cloud contaminated pixel characteristics. Then it investigates the texture of image rectangular kernels composed of classified pixels belonging to classes recorded previously with some expected to represent clouds. The system determines cloud textures, integrates pixel information from within local image areas, and provides the final cloud classification. The method is based on the unsupervised classification approach. The hybrid neural network used consists of classical and modified learning multilayer self-organizing feature maps. The preliminary tests have been made on both artificial and satellite image data. The initial results are satisfactory and promising.

Remarks on hybrid neural network controller using different convergence speeds

Abstract: A neural network requires the partial derivative of a plant output with regard to its input However, it is unknown for an unknown nonlinear plant This paper proposes a hybrid neural network controller which overcomes this problem and which compensates online neural networks for plant fluctuation by using an identifier and a controller with different convergence speeds

A review of a hybrid neural network: Kohonen learning vector quantization and counterpropagation

Abstract: This paper reviews the background and applications of a hybrid neural network-linear vector quantization combined with backpropagation neural network. Areas for the application of LVQ-CPN include the recognition of handwriting, alphanumeric characters, symbols, patterns and pictures, and data compression.

Wavelet-based hybrid neurosystem for feature extractions, characterizations and signal classifications

Abstract: This paper presents an efficient method for signal classification from a system of multiple artificial neural networks (ANN) using wavelets. The method performs feature extraction via the wavelet transform of the underlying signal and presents the resulting coefficients to a hybrid neural network for classification. The hybrid network consists of three single neural networks; two of the networks are provided with magnitude and location information of the coefficients, and are trained with self-organizing rules. Their outputs are then presented to the third network for pattern recognition and classification. Experimental results illustrating concept feasibility for acoustic signal classifications are included.

A hybrid neural network for spatio-temporal pattern recognition

Abstract: In this paper a hybrid network is presented for spatio-temporal pattern recognition (STPR) which is called TS-LM-SOFM The top layer of TS-LM-SOFM is a single layer temporal sequence recognizer which is called TS (temporal sequence) TS can transform temporal sparse pattern sequence into abstract spatial feature representations The bottom layer of TS-LM-SOFM is a modified SOFM (self-organizing feature map) used as a spatial feature detector LM (learning matrix) is introduced as a middle layer In the experiment, some mobile robot's sonar sensor data are used for training Experiments show that the hybrid network can well capture the spatio-temporal features of input signals

Emulating human decision making process using hybrid system

Abstract: This paper presents the prototype implementation of a hybrid neural network expert system shell. The shell, structured around the concept of "network element", is aimed at preserving semantic structure of the expert system rules whilst incorporating learning capability of neural networks into the inference mechanism. Using this architecture, every rule of the knowledge base is represented by a one or two-layer neural network element. These network elements are dynamically linked up to form the rule-tree during inference process. Furthermore, the firing of netels emulate opportunistic decision making process, which is typical of human decision makers. Finally, the system is also able to adjust its inference strategy according to different users and situations.

Simulated flight control using a hybrid neural network/genetic algorithm architecture

Abstract: A controller for an agile, high-subsonic autonomous flight vehicle, incorporating neural network and genetic algorithm techniques, is presented. Simulated flight results for nominal and off-nominal vehicle configurations are reported. The results show that an inverse dynamic model neural network can offer better tracking performance and greater robustness than a conventional linear controller. However, the genetic algorithm technique employed here was found to offer no significant improvement in controller performance. >

Fuzzy Systems and Neural Networks

Abstract: Research and development in the use of fuzzy systems with neural networks has been proceeding at a rapid pace during the last few years and applications are starting to be developed. A natural integration follows the same successful path of hybrid neural network and expert systems by using fuzzy rule based systems instead of traditional expert systems. Thus all the various ways of integration are available for fuzzy connectionist systems. Additionally, neural networks can be used as tools for designing and tuning fuzzy systems. And, fuzzy principles can be used in the design of neural networks, embedding fuzziness in the internal workings of the basically neural system.

On the Hybrid Neural Network Model for Solving Optimization Problems

Abstract: A recent model of neural networks, named the Hybrid Neural Network Model (HN), for solving optimization problems appeared in [3]. In [3], the main algorithm called the Hybrid Network Updating Algorithm (HNUA) is used to drive the HN model. The best thing about the HNUA is that it reaches a feasible solution very quickly. Our argument here is that while the HNUA is very quick to satisfy the constraints, it guarantees very little in terms of the quality of the generated solution. In this paper we rewrite one of the steps in the HNUA so that the goal function is better served. we demonstrate our work using the traveling salesman problem as an example.

Novel approach to the use of neural networks to solve real-world analytical problems

Abstract: This paper describes the general architecture of a hybrid neural network used to identify noisy and extremely complex spectra. A hybrid neural network has been built for environmental monitoring, medical diagnosis, and process control applications. The hybrid neural network consists of preprocessing algorithms to enhance the features of the spectra and an interconnect weight matrix for recognition. Results suggest that the hybrid neural network, through careful design of both the preprocessing algorithms and the neural network architecture, is capable of increasing the detection limit and speed of many analytical instruments.