Generalized Regression Neural Network Method Research Articles

Real-time retrieval of high-precision ZTD maps using GNSS observation

Zenith Tropospheric Delay (ZTD) is an important factor that restricts the high-precision positioning of global navigation satellite system (GNSS), and it is of great significance in establishing a real-time and high-precision ZTD model. However, existing ZTD models only consider the impact of linear terms on ZTD estimation, whereas the nonlinear factors have rarely been investigated before and thus become the focus of this study. A real-time and high-precision ZTD model for large height difference area is proposed by considering the linear and nonlinear characteristics of ZTD spatiotemporal variations and is called the real-time linear and nonlinearity ZTD (RLNZ) model. This model uses the ZTD estimated from the Global Pressure and Temperature 3 (GPT3) model as the initial value. The linear impacts of periodic term and height on the estimation of ZTD difference between GNSS and GPT3 model is first considered. In addition, nonlinear factors such as geographical location and time are further used to fit the remaining nonlinear ZTD residuals using the general regression neural network method. Finally, the RLNZ-derived ZTD is obtained at an arbitrary location. The western United States, with height difference ranging from −500 to 4000 m, is selected, and the hourly ZTD of 484 GNSS stations provided by the Nevada Geodetic Laboratory (NGL) and the data of 9 radiosonde (RS) stations in the year 2021 are used. Experiment results show that a better performance of ZTD estimation can be retrieved from the proposed RLNZ model when compared with the GPT3 model. Statistical results show the averaged root mean square (RMS), Bias, and mean absolute error (MAE) of ZTD from GPT3 and RLNZ models are 33.7/0.8/25.7 mm and 22.6/0.1/17.4 mm, respectively. The average improvement rate of the RLNZ model is 33 % when compared to the GPT3 model. Finally, the application of the proposed RLNZ model in simulated real-time Precise Point Positioning (PPP) indicates that the accuracy of PPP in N, E and U components is improved by 8 %, 2 %, and 6 % when compared with that from the GPT3-based PPP. Meanwhile, the convergence time in N and U components is improved by 23 % and 7 %, respectively. Such results verify the superiority of the proposed RLNZ model in retrieving real-time ZTD maps for GNSS positioning and navigation applications.

Application of General Regression Neural Network Algorithm in Data Mining for Predicting Glass Sales and Inventory Quantity

FF Jaya Glass is a shop that supplies and installs 3 mm to 12 mm glass. The store obtained glass from suppliers to be processed in shape and size according to customers’ order. After completing the customer's order, the shop worker will install the glass at the requested location. Unfortunately, currently stores do not utilize sales data to predict sales either manually or by utilizing technology. As a result, the store cannot predict when the number of glass orders will increase or decrease. In addition, errors often occur when ordering glass for the next period. As a result, stores often run out of glass supplies due to the large number of glass orders so that the achievement of profits is not optimal. This study aims to identify sales variables in glass sales data and build a general regression neural network model as a data mining method. In addition, this study aims to iterate to find the best value in the sales data training process, design and create applications according to user needs, and conduct system validation tests. The general regression neural network method is used to predict sales. The results of this study indicate that the application of general regression neural networks can be used to predict sales. This will make it easier for the store to provide glass supplies in the coming months with an accuracy of 98.1%.

Fractal Localized EBG Structure Intelligently Optimized by K-GRNN and GA Method for Signal and Power Integrity

In this article, a miniaturized fractal localized electromagnetic bandgap (EBG) structure for signal and power integrities is intelligently modeled and optimized with a K-means generalized regression neural network (GRNN) and genetic algorithm (GA). The proposed EBG structure is realized with a combination of L-shaped bridge and slit patch, which is inserted by a Koch fractal resonator with a smaller electric length. To minimize the influence of the period EBG structure on the signal integrity, the fractal structure is only designed on the noise source and sensitive area with localized topology. In comparison with the traditional GRNN method, the modeling precision for discontinuous points such as resonance and antiresonance peaks of the scattering parameter curve can be improved with K-means-GRNN. Finally, a power distribution network sample with the proposed localized fractal EBG structure is designed with a steep suppression level of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$-$</tex-math> </inline-formula> 50 dB from 0.36 to 20 GHz.

Application of Soft Computing, Statistical and Multi-Criteria Decision-Making Methods to Develop a Predictive Equation for Prediction of Flyrock Distance in Open-Pit Mining

Blasting operations in open-pit mines generally have various management strategies relating to flyrock. There are empirical models for calculating the flyrock distance, but due to the complexity and uncertainty of rock properties and their interactions with blasting properties, there are still no models that can predict the flyrock distance that may be applicable across mining operations in general. In this regard, the Jajarm bauxite mine complex was used as a case study. The purpose of this study was to develop and evaluate different methods that can predict flyrock distance. For this purpose, soft computing models were developed using generalized regression neural network (GRNN), gene expression programming (GEP) and genetic-algorithm-based GRNN (GA-GRNN) methods. To obtain statistical models, multivariable regression was applied in the form of linear and nonlinear equations. A flyrock index was introduced using a classification system developed by incorporating fuzzy decision-making trial and evaluation methods (fuzzy DEMATEL). In order to achieve this goal, the data of 118 blasts in eight mines of the Jajarm bauxite complex were collected and used. Following this, four performance benchmarks were applied: the coefficient of determination (R2), variance accounted for (VAF), root-mean-square error (RMSE) and mean absolute percentage error (MAPE). The performance of the models was evaluated, and they were compared with each other as well as with the most common previous empirical models. The obtained results indicate that the GA-GRNN model has a higher performance in predicting the flyrock distance in actual cases compared to the other models. At first, data on factors that were the main cause of flyrock (and had a direct impact on it) were collected and classified from different blasts. Then, using the collected data, 19 different combinations were established, which can be used to provide the appropriate predictive equation. The purpose of this work is to more accurately predict flyrock and prevent heavy damage to buildings and mining machines across the mining complex.

The relative roles of different land-use types in bike-sharing demand: A machine learning-based multiple interpolation fusion method

Land use plays a crucial role in promoting the bike-sharing demand. Traditionally, studies on bike-sharing demand (BSD) are mainly focused on its prediction through regression methods, but the influence of MAUP (modifiable areal unit problem) in modeling is ignored. This paper aims to model spatial BSD distribution and prove the driving forces of different land use types to BSD through a machine-learning-based multiple interpolation fusion method. The hotspot detection model is employed to establish sample points covering different land use types in urban areas. In order to capture the differences in adaptations among different urban regions and for different data sizes, six machine learning methods are applied and evaluated to improve BSD estimation by fusing five spatial interpolation algorithms, including Inverse Distance Weight, Spline, Kriging, Natural Neighborhood and Trend. The methodological verification of Beijing City shows that the fusion models improve the estimation performance compared with individual interpolation algorithms, and that GRNN (generalized regression neural network) method is superior to all the others. According to fitting results of all POIs based on the GRNN fusion model, we identify which types of facilities correspond to customers that will have a stronger preference for bike-sharing and demonstrate which facility names are more prominent in each land use type. The conclusions presented here enrich our understanding relationships between land-use and BSD, which provide a valuable foundation for the bike-sharing development. Compared with implementing regression in an analysis zone or a square grid, troubles caused by the MAUP are effectively solved through this method.

An efficient model to estimate the soil profile and stratigraphic uncertainty quantification

Comprehensive geological information is very important and needed in underground construction and underwater infrastructure. However, it is quite difficult to reveal the configuration of different soil features of underground with sparsely measured data. In this study, a new algorithm named General Regression-Markov Chain (GR-MC) was proposed by coupling the general regression neural network and the Markov chain method together to reproduce the characteristics of underground stratification based on rare borehole information. The general regression neural network method was introduced to evaluate the geological variation along the inclined orientation, while the Markov chain method was applied to examine the change in soil types along the depth. The spatial coordinates of the treated construction site were designed as the input variables, while a vector including the probability information of different soil types was set as the output. The proposed GR-MC method is very intuitive and can be efficiently implemented to predict the geological model using only the coordinates as inputs. Moreover, the proposed method can be smoothly extended into a three-dimensional case with low computational resources. Compared to the existing methods, the proposed GR-MC algorithm is computationally efficient and robust. Three cases from real practice were evaluated and the calculated results were consistent with the known borehole information. Besides, the stratigraphic uncertainty was quantified based on information entropy theory. The results can visually reveal the zones of the estimated soil profile with relatively large uncertainty. The practical application is that possible additional borehole locations were identified to reduce the stratigraphic uncertainty of a construction site.

Spatiotemporal variations, influence factors, and simulation of global cooling degree days.

The cooling degree days (CDDs) can indicate the hot climatic impacts on energy consumption and thermal environment comfort effectively. Nevertheless, seldom studies focused on the spatiotemporal characteristics, influence factors, and simulation of global CDDs. This study analyzed the spatial-temporal characteristics of global CDDs from 1970 to 2018 and in the future, explored five determinants, and simulated CDDs and their interannual changing rates. The results showed that the global CDDs were generally higher at lower latitudes and altitudes. Many places experienced significant positive changes of CDDs (p < 0.05), and the rates became larger at lower latitudes and attitudes. In the future, most CDDs had the sustainability trends. Besides, significant negative partial correlations existed between not only CDDs but also their variation rates with latitude, altitude, and average enhanced vegetation index in the summer, while positive with the annual PM2.5, distance to large waterbodies (p = 0.000). Moreover, the values and variation rates of CDDs can be deduced using the generalized regression neural network method. The root-mean-square errors were 231.73°C * days and 1.71°C * days * year-1, respectively. These conclusions were helpful for the energy-saving for cooling with the climate change and optimization of thermal environment.

Geoid undulation prediction using ANNs (RBFNN and GRNN), multiple linear regression (MLR), and interpolation methods: A comparative study

The present work aimed to develop a prediction model to estimate geoid undulation and to compare its efficiency with other methods including radial basis function neural network (RBFNN), generalized regression neural network (GRNN), multiple linear regression (MLR) and, ten different interpolation methods. In this study, the k-fold cross-validation method was used to evaluate the model and its behavior on the independent dataset. With this validation method, each of a k number of groups has the chance to be divided into training and testing data. The performances of the methods were evaluated in terms of the root mean square error (RMSE) mean absolute error (MAE), Nash–Sutcliffe efficiency coefficient (NSE), and correlation coefficient (R2) and using graphical indicators. The evaluation of the performance of the datasets obtained using cross-validation was done in two ways. If we accept the method having the minimum error result as the most appropriate method, the natural neighbor (NN) method in the DS#5 dataset gave better results than the other methods (RMSE=0.14173 m, MAE=0.09729 m, NSE=0.98986, and R2=0.99011. On the other hand, it has been observed that, the GRNN method exhibited the best performance, on average, with RMSE=0.18539 m, MAE=0.13676 m, NSE=0.98229, and R2=0.98249.

Generalized regression neural network application for fault type detection in distribution transformer windings considering statistical indices

PurposeThis study aims to use frequency response analysis, a powerful tool to detect the location and types of transformer winding faults. Proposing an effective intelligent approach for interpreting the frequency responses is the most crucial problem of this method and has created many challenges.Design/methodology/approachHeat maps based on appropriate statistical indices have been supplied to depict the variations in the frequency responses associated with each fault type, fault location and fault extent along the windings. Also, after analyzing the results of artificial neural network (ANN) techniques, the generalized regression neural network method is introduced as the most effective solution for the classification of transformer winding faults.FindingsUsing a comparative approach, the performance of the used indices and ANN techniques are evaluated. The results showed the proper performance of Lin’s concordance coefficient (LCC) index and the amplitude (Amp) part of the frequency response. The proposed fitting percentage (FP) index can assist the intelligent classifiers in diagnosing the radial deformation (RD) fault with the highest accuracy considering all frequency response components in the classification procedure of winding faults.Practical implicationsVarious ANN techniques are used to detect and determine the type of four important faults of transformer winding, i.e. axial displacement, RD, disc space variation and short circuit. Various statistical indices, such as cross-correlation factor, LCC, standard difference area, sum of errors, normalized root-mean-square deviation and FP, are used to extract the features of the frequency responses to consider as the ANN inputs. In addition, different components of the frequency response, such as Amp, argument, real and imaginary parts are examined in this paper. To implement the proposed procedure, step by step, various types of winding faults with different locations and extents are applied on the 20 kV winding of a 1.6 MVA distribution transformer.Originality/valueContributions have been made in identifying and diagnosing transformer winding defects through the use of appropriate algorithms for future research.

Novel neural network optimization approach for modeling scattering and noise parameters of microwave transistor

AbstractThis study performed modeling of the scattering (S) and noise (N) parameters of the ATF53189 using the General Regression Neural Network (GRNN) and Multi Layer Perceptron Neural Network (MLPNN) methods based on Artificial Neural Network (ANN). For modeling the linear behavior of the transistor, the optimum design parameters of the GRNN and the MLPNN methods were determined using four different optimization algorithms. These are whale optimization algorithm (WOA), artificial bee colony (ABC), particle swarm optimization (PSO) and ant lion optimizer (ALO) algorithms. With the help of these algorithms, the sigma parameter of the GRNN and the number of hidden layers, numbers of neurons in the hidden layers and the activation functions of the hidden layers of the MLPNN were optimized. This way, the best models required for prediction of the S and N parameters of the ATF53189 were obtained. Different models that provided each of the angles and magnitudes of the S11, S21, S12, S22 parameters and the Fmin, Γmin, Γopt magnitude, Γopt angle and Rn noise parameters as the output were created. The experimental results showed that the GRNN method should be used in linear behavior modeling of S parameters of the ATF53189 and the MLPNN method should be used in linear behavior modeling of N parameters of the ATF53189. It is understood that the best algorithm for optimizing the design parameters of the GRNN and the MLPNN methods is the PSO. As a result, the modeling of the S and N parameters of the ATF53189 transistor was successfully carried out with the methods used in this study.

Using the General Regression Neural Network Method to Calibrate the Parameters of a Sub-Catchment

Computer software is an effective tool for simulating urban rainfall–runoff. In hydrological analyses, the storm water management model (SWMM) is widely used throughout the world. However, this model is ineffective for parameter calibration and verification owing to the complexity associated with monitoring data onsite. In the present study, the general regression neural network (GRNN) is used to predict the parameters of the catchment directly, which cannot be achieved using SWMM. Then, the runoff curve is simulated using SWMM, employing predicted parameters based on actual rainfall events. Finally, the simulated and observed runoff curves are compared. The results demonstrate that using GRNN to predict parameters is helpful for achieving simulation results with high accuracy. Thus, combining GRNN and SWMM creates an effective tool for rainfall–runoff simulation.

Stall characteristics analyses and stall lift robustness inverse design for high-lift devices of a wide-body commercial aircraft

Stall is an important cause closely associated with the flight accidents of commercial aircrafts, and stall lift robustness of HLD (high-lift devices) plays a significant role in flight safety issues. In this study, stall characteristics of the HLD of a wide-body commercial aircraft are simulated and analyzed via DES (detached eddy simulation) method, and the impacts of geometric parameters of the HLD on stall lift characteristics are investigated. Afterwards, stall lift robustness design for the HLD is performed via a proposed inverse method which combines the GRNN (generalized regression neural network) method and PCA (principal component analysis) method. Then the inverse design model is established with the aerodynamic parameters as input and the geometric parameters as output. The PCA method is introduced to reduce the dimension of the input data, while the GRNN is employed to predict the geometric parameters. The design process is iterated with new sample points added in. The validated results via DES method indicate that the finally designed HLD configuration conforms with the design targets very well, and thus the stall lift robustness is improved obviously. More specifically, the average lift coefficient near the stall AoA (angle of attack) is improved by 1.4%, and the variance of the lift coefficients near the stall AoA is reduced by 79%, while the stall AoA remains unchanged.

Local geoid determination using a generalized regression neural network and interpolation methods: A case study in Kars, Turkey

This study aimed to determine the most suitable local geoid model based on 641 GNSS/leveling points within the borders of Kars Province in eastern Turkey using the generalized regression neural network (GRNN), weighted average (WA), multiquadric (MQ), inverse multiquadric (IMQ) function, and local polynomial (LP) method. Among these methods used in local geoid determination, the studies conducted with the GRNN method are very limited in the literature. To test the performance of the model, 169 GNSS/leveling points were selected as test data. When selecting the reference and test points, care was taken to ensure that the distribution of the points was homogeneous. The criteria of root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2) were used to assess the accuracy and error rates of the results achieved using the different methods. The analysis showed that the GRNN yielded better results than other interpolation methods (RMSE = 1.215 cm, MAE = 0.467 cm, R2 = 0.99980).

Specialty Coffee Cupping Score Prediction with General Regression Neural Network (GRNN)

Coffee is a plant that can be processed into beverages. The cupping score is a score for a coffee quality graded with an expert called Q grader. The cupping score will decide if a coffee may be called as specialty coffee. In this research, the cupping score will be predicted by the coffee properties and did not involve the Q grader for giving the score. The prediction of the score is obtained by using the GRNN method. The experiment consists of finding when the MAE and the MSE are converged and find the neuron's best number. The model's performance is measured with MSE and MAE with the best MSE value of 0.097 and MAE value 0.245.

Spatio-temporal variation and daily prediction of PM2.5 concentration in world-class urban agglomerations of China.

The contradiction between the development of urban agglomerations and ecological protection has long been a challenging issue. China has experienced an astonishing expansion of its urban scale in the past 40years, and nearly 783 million of the nation's people now live in cities. Beijing-Tianjin-Hebei, the Yangtze River Delta and the Pearl River Delta have been prioritized to become world-class clusters by 2020. The health effects of air pollution in these three urban agglomerations are becoming increasingly formidable. Given these conditions, using the daily mean PM2.5 concentration in 40 cities from January 2014 to December 2016, this research explored the spatial-temporal characteristics of PM2.5 concentrations in these three urban agglomerations. The annual mean PM2.5 concentrations in Beijing-Tianjin-Hebei, the Yangtze River Delta and the Pearl River Delta are 35.39µg/m3, 53.72µg/m3 and 78.54µg/m3, respectively. Compared with the other two urban agglomerations, abundant rainfall causes the Pearl River Delta to have the lowest PM2.5 level. Furthermore, a general regression neural network (GRNN) method is developed to predict the PM2.5 concentration in these clusters on the second day, with inputs including the average, maximum and minimum temperature; average, maximum and minimum atmosphere; total rainfall; average humidity; average and maximum wind speed; and the PM2.5 concentration measured 1day ahead. The results indicate that the GRNN method can precisely predict the concentration level in these clusters, and it is especially useful for the Pearl River Delta, as the underlying influence mechanism is more specified in this cluster than in the others. Importantly, this 1-day-ahead forecasting of PM2.5 concentrations can raise awareness among the public to improve their precautionary behaviours and help urban planners to provide corresponding support.

Objective approach for rainstorm based on dual-factor feature extraction and generalized regression neural network

Rainstorm often causes inland flooding and mudslides that threaten lives and properties. In this study, rainstorm is used as a forecasting object, and an interpretation prediction model for rainstorm based on the European Center for medium-range weather forecasting (ECMWF) numerical prediction model is constructed through the generalized regression neural network method. Model inputs are forecasted through principal component analysis, and dual-factor feature extraction is performed on the primary predictors to obtain new irrelevant variables and optimize network structures. The experimental forecast results of the 24 h aging test using an independent sample of large-scale rainstorm in Guangxi, China from 2012 to 2016, the actual forecast results of selected rainstorm cases with great influence on Guangxi, and different influencing systems show that the new prediction scheme is sophisticated. Thus, the scheme has a certain universal applicability. The results of the comparative analysis between the new program and ECMWF show that the forecasting ability of the new method is more accurate than that of the direct numerical forecasting model. The threat score of the new forecast model for 5 years has a 58.4% increase relative to that of the ECMWF. The forecasting skills are positive and good and can thus improve the rainstorm forecasting ability of ECMWF and provide a better guidance for forecasters.

Modelling of Bektas Creek Daily Streamflow with Generalized Regression Neural Network Method

The dependence of agricultural production on water resources is a known fact. Therefore, understanding hydrological processesand events in agricultural production form the basis of effective and reliable management of water resources. Many traditionalmethods used for the analysis of time-dependent variables in hydrology and meteorology assume linear relationships betweenthese variables. However, the temporal changes of these parameters are quite complex and therefore cannot be easily modeled by conventional estimation methods. Artificial neural networks (ANNs), on the other hand, allow the analysis of nonlinear relationships or processes whose statistical or mathematical calculations cannot be determined in such systems. ANNs have beenaccepted as a successful model in multidimensional research involving dynamic processes in the field of hydrology. This study, it is aimed at the modeling of stream flows in Bektas Creek. With the aim of modeling, daily meteorological parameters (precipitation, temperature, sunbathing time, relative humidity) measured in Kangal Region and one day delayed flows were used. Streamflow forecasts are simulated with the Generalized Regression Neural Network (GRNN). To reveal the difference of the GRNN model from other ANNs, the same data were also used in the feed-forward neural network (FFNN) model. Model performances were evaluated taking into account the Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Determination coefficient (R).

Inversion of aerosol extinction coefficient by Raman-Mie scattering lidar

An observation experiment of atmospheric aerosol based on Raman-Mie scattering lidar was introduced. Raman scattering lidar signals were smoothed by generalized regression neural network(GRNN) combined with wavelet denoising. Based on Raman-Mie scattering lidar, several groups of aerosol extinction coefficient profiles were retrieved by GRNN method. And the variation characteristics of aerosol extinction coefficient were analyzed by combining with meteorological data.

Parametric uncertainty assessment of hydrological models: coupling UNEEC-P and a fuzzy general regression neural network

ABSTRACTDue to the complicated nature of environmental processes, consideration of uncertainty is an important part of environmental modelling. In this paper, a new variant of the machine learning-based method for residual estimation and parametric model uncertainty is presented. This method is based on the UNEEC-P (UNcertainty Estimation based on local Errors and Clustering – Parameter) method, but instead of multilayer perceptron uses a “fuzzified” version of the general regression neural network (GRNN). Two hydrological models are chosen and the proposed method is used to evaluate their parametric uncertainty. The approach can be classified as a hybrid uncertainty estimation method, and is compared to the group method of data handling (GMDH) and ordinary kriging with linear external drift (OKLED) methods. It is shown that, in terms of inherent complexity, measured by Akaike information criterion (AIC), the proposed fuzzy GRNN method has advantages over other techniques, while its accuracy is comparable. Statistical metrics on verification datasets demonstrate the capability and appropriate efficiency of the proposed method to estimate the uncertainty of environmental models.

Prediction of microfibril angle in Dahurian larch wood using visible–near infrared spectroscopy and chemometric techniques

The microfibril angle of the S2 layer in the secondary cell wall of the tracheid is important for molecular and microscopic properties that influence collapse resistance, longitudinal modulus of elasticity and other lateral properties of conifers at the macroscopic level. This research aimed to investigate the feasibility of using a fruit fly optimization algorithm for visible and near infrared modeling optimization of Dahurian larch wood microfibril angle prediction. Originally, the linear relationship between microfibril angle and their raw spectra and visible and near infrared spectra pretreated by wavelet transform was established. Then, a nonlinear coupled model was built by combining the stepwise regression analysis and generalized regression neural network methods. As a final point, fruit fly optimization algorithm was used for optimizing stepwise regression analysis–generalized regression neural network coupled model. It was found that stepwise regression analysis–generalized regression neural network coupled model coupled model based on the optimization of fruit fly optimization algorithm simplify visible and near infrared spectral data and its prediction results ([Formula: see text] = 0.90, RMSEP = 0.75, mean average percentage error ([Formula: see text]) = 0.05) outperforms original partial least squares model ([Formula: see text] = 0.86, RMSEP = 0.88, [Formula: see text] = 0.06). This work demonstrated the feasibility of using improved chemometric techniques for improving the precision of visible and near infrared spectra in the prediction of microfibril angle.