Group Method Of Data Research Articles

Composite Reliability Assessment Based on Monte Carlo Simulation and Artificial Neural Networks

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> This paper presents a new methodology for reliability evaluation of composite generation and transmission systems, based on nonsequential Monte Carlo simulation (MCS) and artificial neural network (ANN) concepts. ANN techniques are used to classify the operating states during the Monte Carlo sampling. A polynomial network, named Group Method Data Handling (GMDH), is used, and the states analyzed during the beginning of the simulation process are adequately selected as input data for training and test sets. Based on this procedure, a great number of <emphasis emphasistype="boldital">success</emphasis> states are classified by a simple polynomial function, given by the ANN model, providing significant reductions in the computational cost. Moreover, all types of composite reliability indices (i.e., loss of load probability, frequency, duration, and energy/power not supplied) can be assessed not only for the overall system but also for areas and buses. The proposed methodology is applied to the IEEE Reliability Test System (IEEE-RTS), to the IEEE-RTS 96, and to a configuration of the Brazilian South-Southeastern System. </para>

Logistic GMDH-Type Neural Network and its Application to Identification of X-Ray Film Characteristic Curve

The logistic Group Method of Data Handing (GMDH)-type neural network identifying a complex nonlinear system we propose is automatically organized using heuristic self-organization that is basic to GMDH algorithm. In this neural network, structural parameters such as the number of layers, the number of neurons per layer, useful input variables, and optimum neuron architectures are automatically determined using a prediction error criterion defined as Akaike’s Information Criterion (AIC) to produce an optimum neural network architecture suiting the complexity of the nonlinear system. In applying this neural network to the identification problem of the X-ray film characteristic curve, we found that modeling with such a neural network is more accurate than applying multiple regression analysis, a conventional neural network, and GMDH algorithm.

Fuzzy GMDH-type neural network model and its application to forecasting of mobile communication

In this paper, the fuzzy group method data handling-type (GMDH) neural networks and their application to the forecasting of mobile communication systems are described. At present, the GMDH family of modeling algorithms discovers the structure of empirical models and it gives only the way to get the most accurate identification and demand forecasts in case of noised and short input sampling. In distinction to neural networks, the results are explicit mathematical models, obtained in a relative short time. In this paper, an adaptive learning network is proposed as a kind of neural-fuzzy GMDH. The proposed method can be reinterpreted as a multi-stage fuzzy decision rule which is called the neural-fuzzy GMDH. The GMDH-type neural networks have several advantages compared with conventional multi-layered GMDH models. Therefore, many types of nonlinear systems can be automatically modeled by using the neuro-fuzzy GMDH. A computer program is developed and successful applications are shown in the field of estimating problems of mobile communication with a number of factors considered.

Modeling and Optimization of Process Parameters for Defect Toleranced Drilling of GFRP Composites

Drilling is a frequently practiced machining process for fiber-reinforced plastics (FRP) in industry owing to the need for component assembly in mechanical structures. Drilling experiments are performed on a (0/±45/90)2s 3-mm-thick glass fiber-reinforced laminate using 4-, 6- and 8-mm-diameter HSS drills. The machining response of the quasi-isotropic laminate was studied by monitoring the thrust and torque. The performance of the HSS drills for different cutting conditions was studied by measuring the tool wear. Delamination due to drilling is a major concern in machining a composite laminate and is analyzed by using linear elastic fracture mechanics, classical plate bending theory, and the mechanics of composites. A mechanical model for evaluating the critical thrust at which delamination is initiated at different ply locations has been used, and the critical thrust force at the onset of delamination has been found to be 70 N. The present work analyzes data on the thrust force, torque, and tool life by using a group method data handling (GMDH) algorithm. An optimization algorithm using simulated annealing with a performance index is then applied to search for the optimal process parameters for delaminatien constrained drilling.

Conditional Asset Pricing - Predicting Time Varying Beta-Factors with Group Method of Data Handling Methods

Allowing for time-varying risk premia yields sophisticated asset pricing models, but the search for adequate model specifications is more challenging. We introduce, to our knowledge, previously in conditional asset pricing not used Group Method of Data Handling (GMDH) that rests on sorting out requiring statsitical models for complex problems of unknown structure but does not require a model to predict conditional variation in betas. We find that lagged instruments used to proxy for expected returns in conditional asset pricing provide a challenge not only for the unconditional CAPM but also the Fama-French-model. Thereby non-linear GMDH-algorithms challenge traditional models of conditional asset pricing as we find a highly non-linear influence of lagged instruments on both conditional alphas and betas. Therefore, predetermining a structure for functional relationships between conditional alphas as well as betas and lagged instruments may lead to a significant misspecification of asset pricing models.

S-scheme of sliding examination for optimal set features determination in discriminant analysis by the group method of data handling

It is shown analytically that the traditional scheme of sliding examination of the Group Method of Data Handling (GMDH) allows to solve the problem of discriminant analysis in a broad sense on the basis of finite samples of observations. It is shown that there exists an optimal set of features corresponding to the maximum mathematical expectation of some generalized distance between the observations from two general sets. It is shown analytically that parameters of the general sets and samples sizes influence on a complexity of optimal discriminant function, and conditions under which the optimal discriminant function is simplified are exhibited.

The Fuzzy Group Method of Data handling and its Application to the Tasks of the Macroeconomic Indexes Forecasting

The task of synthesis of the predicting models on the basis of a fuzzy group method of data handling is considered. Problem of adapting of coefficients of the predicting model also is considered and the algorithm of its solution based on a method of stochastic approximation is offered. The application of the method is illustrated at the problem of prediction of macroeconomic indexes of Ukrainian economy.

Application of the Group Method of Data Handling for Constructing Collective Estimates of Properties of Organic Compounds on the Basis of Inductive Search for Their "Structural Spectra"

Application of the Group Method of Data Handling for Constructing Collective Estimates of Properties of Organic Compounds on the Basis of Inductive Search for Their "Structural Spectra"

Searching for the electric load—weather temperature function by using the group method of data handling

Searching for the electric load—weather temperature function by using the group method of data handling

The Modified GMDH: Sigmoidal and Polynomial Neural Net

Two modificatios of the Group Method Data Handling neural nets are introduced. The first one consist of introducing sigmoidal transfer function to the neuron which originally produced quadratic approximation. The other modification introduces new stopping rule for overfitting prevention. Using this rule no splitting of the learning set to the training and testing set is needed. Some experience with application of the net for credit risk rating is described.

Tool wear estimation by group method of data handling in turning

Tool wear monitoring and estimation are essential for improved productivity of manufacturing systems. Multi-sensory approaches based on force, vibration and Acoustic Emission (AE) signals have been recognized as potential methods for tool wear monitoring. In the present work, steady-state components of force, dynamics of the main cutting force and vibration in the direction of the main cutting force have been used for on-line tool wear estimation in a turning process. The group method of data handling (GMDH), a heuristic self-organizing method of modelling, has been used to integrate information from different sensors and the cutting conditions to obtain estimates of tool wear. Different methods of preprocessing the forces have been attempted to determine the best method to suit the data. Various heuristics of GMDH have been analysed to obtain the appropriate models for tool wear estimation. The results show that GMDH can be effectively used to integrate sensor information and obtain reliable estimates of...

Group Method of Data Handling: A New Method for Stock Identification

Abstract The group method of data handling (GMDH) is described and applied to the problem of fish stock separation (classification). The method, which increases model complexity in a stepwise fashion, finds an empirical polynomial model of a complex system that satisfies any of several optimality criteria. By basing the model's structure largely on the data, rather than a priori assumptions, GMDH can reduce the number of assumptions that must be made about the system under study. I analyzed data from two previously published data sets with GMDH to arrive at stock identification models for American shad Alosa sapidissima and striped bass Morone saxatilis. Each model was verified with data not used in model construction. The GMDH-based models were as good as conventional discriminant models at classifying the verification samples of fish. In addition, GMDH was able to identify excellent models that used fewer characteristics (features) of the samples.

GMDH utilizing BDS with insufficient modelling data

Abstract It has been pointed out that the group method of data handling algorithm has a disadvantage that predicts an abnormal result in the event of shortage of modelling data. This paper presents a method of modelling that avoids the emergence of abnormal predicted values. For estimation of partial polynomial parameters, a bad data suppression estimator is employed. It is based on a non-quadratic cost function that reduces to the weighted least squares estimator in the absence of bad data. Real data are used to illustrate that this method prevents abnormal results.

Prediction of monthly mean chlorophylla concentration in Mikawa Bay using GMDH

Data on organic and inorganic nitrogen and phosphorus, and meteorological conditions observed a few times a month at a buoy station in Mikawa Bay (Japan) are averaged on a monthly basis, and monthly transitions of chlorophylla at the station are predicted by GMDH (Group Method of Data Handling, one of the methods of nonlinear system identification) using these monthly mean data. The prediction is successful. GMDH is useful in describing the ecosystem and in predicting transitions within the bay.

Self-Organization GMDH Computer Aided Design for Modelling of Cyclic Processes

The Group Method Data Handling (GMDH) is a specific selforganization computer-aided design which has been proposed by Ivakhnenko (1970, 1971) and which has been described in several recent articles (Ivakhnenko et al, 1978, 1979a) and also in Soviet Automatic Control (USA). One of the several interesting properties of this CAD is its apparent ability to make high accuracy predictions of future states of complex systems. The development presents a procedure for constructing models in a simple and systematic way for complex systems which are cyclic in nature. In the present paper the structures of finite-difference models for cyclic processes are described. The criterion of prediction balance is used for analysing the optimal model. System of combined models constructed in this manner is used recursively for predicting the future states of the cyclic system. Three examples are given. One example used the data generated from a known model and the others the airtemperature data with large number of points and the tea-crop production data with less number of points. The system of combined models is compared with the single overall model. The former has given reasonably better results. The models are checked for stability in long-range actions.

Identification of a dynamic lake model by the group method of data handling: An application to lake ontario

Abstract A revised algorithm of the group method of data handling (GMDH) is used to model and identify water quality dynamics in Lake Ontario. The state variables in the model are corrected chlorophyll a, volume-averaged over the top 20 m, and soluble reactive phosphorus in three layers, 0–20 m, 20–40 m and from 40 m to bottom. Inputs are the average quarterly temperatures for each layer and the total phosphorus loadings. Two kinds of model are developed: recursive and iterative; they are used to simulate long-term predictions for three different hypothetical rates of phosphorus loading. Results show that GMDH can be usefully employed to develop lake models with a very low expense of manpower and computer time. This exercise compares well with other large-scale modelling efforts of Lake Ontario, even if necessarily more limited in scope.