Wyniki wyszukiwania - BazTech

1

A Comparative Evaluation of the Use of Artificial Neural Networks for Modeling the Rainfall-Runoff Relationship in Water Resources Management

Turhan Evren

Journal of Ecological Engineering

|

2021

|

Vol. 22, nr 5

166--178

EN

Recently, Artificial Neural Network (ANN) methods, which have been successfully applied in many fields, have been considered for a large number of reliable streamflow estimation and modeling studies for the design and project planning of hydraulic structures. The present study aimed to model the rainfall-runoff relationship using different ANN methods. The Nergizlik Dam, located in the Seyhan sub-basin and one of the important basins in Turkey, was chosen as the study area. Analyses were carried out based on streamflow estimation with the help of observed precipitation and runoff data at certain time intervals. Feed Forward Backpropagation Neural Network (FFBPNN) and Generalized Regression Neural Network (GRNN) methods were adopted, and obtained results were compared with Multiple Linear Regression (MLR) method, which is accepted as the traditional method. Also, the models were performed using three different transfer functions to create optimum ANN modeling. As a result of the study, it was seen that ANN methods showed statistically good results in rainfall-runoff modeling, and the developed models can be successfully applied in the estimation of average monthly flows.

2

Streamflow prediction using data-driven models: Case study of Wadi Hounet, northwestern Algeria

Beddal Dalila, Achite Mohammed, Baahmed Djelloul

Journal of Water and Land Development

|

2020

|

no. 47

16--24

EN

Streamflow modelling is a very important process in the management and planning of water resources. However, complex processes associated with the hydro-meteorological variables, such as non-stationarity, non-linearity, and randomness, make the streamflow prediction chaotic. The study developed multi linear regression (MLR) and back propagation neural network (BPNN) models to predict the streamflow of Wadi Hounet sub-basin in north-western Algeria using monthly hydrometric data recorded between July 1983 and May 2016. The climatological inputs data are rainfall (P) and reference evapotranspiration (ETo) on a monthly scale. The outcomes for both BPNN and MLR models were evaluated using three statistical measurements: Nash–Sutcliffe efficiency coefficient (NSE), the coefficient of correlation (R) and root mean square error (RMSE). Predictive results revealed that the BPNN model exhibited good performance and accuracy in the prediction of streamflow over the MLR model during both training and validation phases. The outcomes demonstrated that BPNN-4 is the best performing model with the values of 0.885, 0.941 and 0.05 for NSE, R and RMSE, respectively. The highest NSE and R values and the lowest RMSE for both training and validation are an indication of the best network. Therefore, the BPNN model provides better prediction of the Hounet streamflow due to its capability to deal with complex nonlinearity procedures.

3

Prediction of electrical conductivity using ANN and MLR: a case study from Turkey

Ekemen Keskin Tülay, Özler Emre, Şander Emrah, Düğenci Muharrem, Mohammed Yadgar Ahmed

Acta Geophysica

|

2020

|

Vol. 68, no. 3

811--820

EN

The study areas are located in Turkey (Kastamonu, Bartın, Karabük, Sivas) and contain very diferent rock types, various mining and agricultural activity opportunities. So, the areas have groundwaters that have diferent chemical compositions and electrical conductivity (EC) values. The EC can be measured using EC meter, and it must be measured in situ. But, the measurement of EC in situ is laborious, time-consuming, expensive, and difcult in arduous terrain environments. In recent years, machine learning models have been a primary focus of interest for a lot of study by providing often highly accurate forecast for solutions of such problems. The aim of the study is to forecast EC of groundwater using artifcial neural networks (ANN) and multiple linear regressions (MLR). Twelve diferent hydrochemical parameters, which afect the EC, such as major/minor ions and trace elements, were used in the analysis. Multilayer feed-forward ANN trained with backpropagation in Python machine learning libraries was used in this study. In order to obtain the most appropriate ANN architecture, trialand-error procedure was used and diferent numbers of hidden layers, neurons, activation functions, optimizers, and test sizes were constructed. This study also tests the usability of input parameters in EC prediction studies. As a result, comparisons between the measured and predicted values indicated that the machine learning models could be successfully applied and provide high accuracy and reliability for EC and similar parameters forecasting.

4

Multiple regression analysis model to predict and simulate winter rapeseed yield

Niedbała G., Piekutowska M., Adamski M.

Journal of Research and Applications in Agricultural Engineering

|

2018

|

Vol. 63, nr 4

139--144

EN

The aim of the work is to create a model for prediction and simulation of winter rapeseed yield. The model made it possible to perform a yield forecast on 30 June, directly before harvest in the current agrotechnical season. The prediction model was built using the multiple regression method (MLR). The model was based on meteorological data (air temperature and precipitation) and information about mineral fertilization. The data were collected from the years 2008-2017 from 291 production fields located in Poland, in the southern Opole region. The assessment of the quality of forecasts generated on the basis of the regression model was verified by determining prediction errors using RAE, RMS, MAE and MAPE error meters. An important feature of the created prediction model concerns the possibility of making the forecast in the current agrotechnical year on the basis of the current weather and fertilizer information.

PL

Celem pracy było zbudowanie modelu do predykcji i symulacji plonu rzepaku ozimego. Model ten umożliwił wykonanie prognozy plonu na dzień 30 czerwca, bezpośrednio przed zbiorem w aktualnie trwającym sezonie agrotechnicznym. Do budowy modelu predykcyjnego użyto metody regresji wielorakiej (MLR). Model powstał w oparciu o dane meteorologiczne (temperatura powietrza i opady atmosferyczne) oraz informacje o nawożeniu mineralnym. Dane zostały zebrane z lat 2008- 2017 z 291 pól produkcyjnych zlokalizowanych w Polsce, na obszarze południowej Opolszczyzny. Ocena jakości prognoz wytworzonych na bazie modelu regresyjnego została zweryfikowana poprzez określenie błędów prognozy za pomocą mierników błędów RAE, RMS, MAE oraz MAPE. Ważną cechą wytworzonego modelu predykcyjnego jest możliwość wykonania prognozy w bieżącym roku agrotechnicznym w oparciu o aktualne informacje pogodowe i nawozowe.

5

Application of multiple linear regression for multi-criteria yield prediction of winter wheat

Niedbała G.

Journal of Research and Applications in Agricultural Engineering

|

2018

|

Vol. 63, nr 4

125--131

EN

The aim of the work was to produce three independent models for prediction and simulation of winter wheat yield, which were marked in the following way: ReWW15_04, ReWW31_05 and ReWW30_06. The produced models enable to make yield forecasts for April 15, May 31 and June 30, directly before harvest in the current agrotechnical season. For the construction of prediction models the Multiple Linear Regression (MLR) method was used. The models are based on meteorological data (air temperature and rainfall) and information on mineral fertilisation. The data were collected from 2008- 2015 from 301 production fields located in Poland, in the Wielkopolskie Voivodeship. Evaluation of the quality of forecasts based on MLR models was verified by determining forecast errors using RAE, RMS, MAE and MAPE error gauges. An important feature of the produced prediction model consists in the possibility of making a prediction in the current agrotechnical year on the basis of current weather and fertilizer information.

PL

Celem pracy było wytworzenie trzech niezależnych modeli do predykcji i symulacji plonu pszenicy ozimej, które oznaczono w następujący sposób: ReWW15_04, ReWW31_05 and ReWW30_06. Wytworzone modele umożliwiają wykonanie prognozy plonu na dzień 15 kwietnia, 31 maja i 30 czerwca, bezpośrednio przed zbiorem w aktualnie trwającym sezonie agrotechnicznym. Do budowy modeli predykcyjnych użyto metody liniowej regresji wielorakiej (MLR). Modele powstały w oparciu o dane meteorologiczne (temperatura powietrza i opady atmosferyczne) oraz informacje o nawożeniu mineralnym. Dane zostały zebrane z lat 2008-2015 z 301 pól produkcyjnych zlokalizowanych w Polsce, na terenie województwa Wielkopolskiego. Ocena jakości prognoz wytworzonych na bazie modeli MLR została zweryfikowana poprzez określenie błędów prognozy za pomocą mierników błędów RAE, RMS, MAE oraz MAPE. Ważną cechą wytworzonego modelu predykcyjnego jest możliwość wykonania prognozy w bieżącym roku agrotechnicznym w oparciu o aktualne informacje pogodowe i nawozowe.

6

Procedure of Building and Analysis of the Information Database of the Resistance of Existing Bridge Structures to Mining Tremors

Rusek J.

Geomatics and Environmental Engineering

|

2017

|

Vol. 11, no. 4

111--123

PL

Jednym z przejawów negatywnego wpływu działalności przemysłowej kopalń na środowisko są wstrząsy górnicze. Ochrona obiektów budowlanych przed szkodliwym działaniem wynikających stąd drgań podłoża gruntowego wymaga ustalenia ich odporności dynamicznej. Problem ten nabiera szczególnego znaczenia w przypadku istniejących obiektów mostowych, przy których projektowaniu nie uwzględniono możliwości wystąpienia wstrząsów górniczych. W pracy przedstawiono metodę pozyskiwana danych o odporności dynamicznej istniejących obiektów mostowych usytuowanych na terenach górniczy w wyniku obliczeń numerycznych metodą elementów skończonych (MES). Odporność obiektów opisano za pomocą granicznych wartości przyspieszeń drgań gruntu w płaszczyźnie pionowej i poziomej, które istniejąca konstrukcja może przejąć bez zagrożenia bezpieczeństwa. Uwzględniając zróżnicowanie geometryczne i materiałowe, utworzono 3000 modeli numerycznych żelbetowych mostów drogowych. Następnie w odniesieniu do każdego obiektu przeprowadzono obliczenia numeryczne MES, w wyniku których dla każdego przypadku wyznaczono dopuszczalne wartości przyspieszeń drgań gruntu określających ich odporność dynamiczną. Utworzoną bazę danych poddano wstępnej analizie w celu wykrycia linowych relacji wiążących dane opisujące geometrię i właściwości materiałowe poszczególnych obiektów z ich odpornością dynamiczną na wpływ wstrząsów górniczych. W efekcie tych badań wyselekcjonowano zmienne, na podstawie których utworzono model wielorakiej regresji liniowej (MLR). Analiza uzyskanych wyników pozwoliła ocenić możliwości stosowania modeli liniowych do ustalania odporności dynamicznej obiektów mostowych poddanych wstrząsom górniczym.

EN

Mining tremors are one of the manifestations of negative impacts of the mining industry on the environment. In order to protect building structures against the damaging effects of ground vibrations, it is required that their dynamic resistance be determined. This problem is of particular importance for the existing bridge structures that were not designed for the potential occurrence of mining tremors. This paper presents the assumptions of and a method for acquiring data on the dynamic resistance of existing bridge structures located in mining areas as a result of numerical calculations using the Finite Element Method (FEM). Object resistance was described by the limit values of the acceleration of ground vibrations in the vertical and horizontal planes that can be carried by the existing structure without compromising safety. Taking into account the geometrical and material diversity, 3,000 numerical models of reinforced concrete overpasses were created. Then, for each object, numerical calculations using the FEM were performed, which resulted in the determination of permissible values of the acceleration of ground vibrations defining their dynamic resistance. The created database was subjected to a preliminary analysis in order to detect linear relationships binding the data that describe the geometry and material properties of individual structures with their dynamic resistance to the impact of mining tremors. As a result of these studies, variables were selected that formed the basis for creating a multiple linear regression model (MLR). Analysis of the obtained results allowed us to assess the possibilities of using linear models to determine the dynamic resistance of bridge structures subjected to mining tremors.

7

Computational Model For Chromatographic Relative Retention Time of Polychlorinated Biphenyls Using Sub-structural Molecular Fragments

Saaidpour S.

Computational Methods in Science and Technology

|

2016

|

Vol. 22, No. 1

41--53

EN

Quantitative structure-retention relationship (QSRR) analysis is a useful technique capable of relating chromatographic retention time to the chemical structure of a solute. Using the sub-structural molecular fragments (SMF) derived directly from the molecular structures, the gas chromatographic relative retention times (RRTs) of 209 polychlorinated biphenyls (PCBs) on the SE-54 stationary phase were calculated. An eight-variable regression equation with the correlation coefficient of 0.9945 and the root mean square errors of 0.0134 was developed. Forward and backward stepwise regression variable selection and multi-linear regression analysis (MLRA) are combined to describe the effect of molecular structure on the RRT of PCB according to the QSRR method. To quantitatively relate RRT with the molecular structure MLR analysis is performed on the set of 163 sub-structural molecular fragments (SMF) provided by the ISIDA software. The eight fragments selected by variable subset selection, all belonging to the sub-fragments, adequately represent the structural factors influencing the affinity of PCB to SE-54 stationary phase in the separation process. Finally, a QSRR model is selected based on leave-one-out cross-validation and its prediction ability is further tested on 42 representative compounds excluded from model calibration. The prediction results from the MLR model are in good agreement with the experimental values. By applying the MLR method we can predict the test set with squared cross validated correlation coefficient (Q2 ext) of 0.9913 and root mean square error (RMSE) of 0.0169.

8

Prediction of the Density of Energetic Materials on the Basis of their Molecular Structures

Rahimi R., Keshavarz M. H., Akbarzadeh A. R.

Central European Journal of Energetic Materials

|

2016

|

Vol. 13, no. 1

73--101

EN

The density of an energetic compound is an essential parameter for the assessment of its performance. A simple method based on quantitative structure-property relationship (QSPR) has been developed to give an accurate prediction of the crystal density of more than 170 polynitroarenes, polynitroheteroarenes, nitroaliphatics, nitrate esters and nitramines as important classes of energetic compounds, by suitable molecular descriptors. The evaluation techniques included cross-validation, validation through an external test set, and Y-randomization for multiple linear regression (MLR) and training state analysis for artificial neural network (ANN), and were used to illustrate the accuracy of the proposed models. The predicted MLR results are close to the experimental data for both the training and the test molecular sets, and for all of the molecular sets, but not as close as the ANN results. The ANN model was also used with 20 hidden neurons that gave good result. The results showed high quality for nonlinear modelling according to the squared regression coefficients for all of the training, validation and the test sets (R2 = 0.999, 0.914 and 0.931, respectively). The calculated results have also been compared with those from several of the best available predictive methods, and were found to give more reliable estimates.

9

Prediction the Normal Boiling Points of Primary, Secondary and Tertiary Liquid Amines from their Molecular Structure Descriptors

Saaidpour S., Bahmani A., Rostami A.

Computational Methods in Science and Technology

|

2015

|

Vol. 21, No. 4

201--210

EN

In this article, at first, a quantitative structure–property relationship (QSPR) model for estimation of the normal boiling point of liquid amines is developed. QSPR study based multiple linear regression was applied to predict the boiling points of primary, secondary and tertiary amines. The geometry of all amines was optimized by the semi-empirical method AM1 and used to calculate different types of molecular descriptors. The molecular descriptors of structures were calculated using Molecular Modeling Pro plus software. Stepwise regression was used for selection of relevance descriptors. The linear models developed with Molegro Data Modeller (MDM) allow accurate estimate of the boiling points of amines using molar mass (MM), Hansen dispersion forces (DF), molar refractivity (MR) and hydrogen bonding (HB) (1◦ and 2◦ amines) descriptors. The information encoded in the descriptors allows an interpretation of the boiling point studied based on the intermolecular interactions. Multiple linear regression (MLR) was used to develop three linear models for 1◦ , 2◦ and 3◦ amines containing four and three variables with a high precision root mean squares error, 15.92 K, 9.89 K and 15.76 K and a good correlation with the squared correlation coefficient 0.96, 0.98 and 0.96, respectively. The predictive power and robustness of the QSPR models were characterized by the statistical validation and applicability domain (AD).

10

Structure Activity Relationships, QSAR Modeling and Drug-like calculations of TP inhibition of 1,3,4-oxadiazoline-2-thione Derivatives

Almi Z., Belaidi S., Lanez T., Tchouar N.

International Letters of Chemistry, Physics and Astronomy

|

2014

|

Vol. 18

113--124

EN

QSAR studies have been performed on twenty-one molecules of 1,3,4-oxadiazoline-2-thiones. The compounds used are among the most thymidine phosphorylase (TP) inhibitors. A multiple linear regression (MLR) procedure was used to design the relationships between molecular descriptor and TP inhibition of the 1,3,4-oxadiazoline-2-thione derivatives. The predictivity of the model was estimated by cross-validation with the leave-one-out method. Our results suggest a QSAR model based of the following descriptors: logP, HE, Pol, MR, MV, and MW, qO1, SAG, for the TP inhibitory activity. To confirm the predictive power of the models, an external set of molecules was used. High correlation between experimental and predicted activity values was observed, indicating the validation and the good quality of the derived QSAR models.

11

Grain boundary effect on the anisotropy piezoresistance of laser-recrystallized polysilicon layers in SOI-structures

Pankov Y., Druzhinin A.

Journal of Telecommunications and Information Technology

|

2001

|

nr 1

46-48

EN

A physical model of grain boundary influence on the piezoresistive effect of p-type conductivity of polysilicon layers in SOI-structures is developed. Software calculating piezoresistive properties of boron-doped p-type polysilicon layers has been developed. These properties may be calculated over wide concentration and temperature ranges with anisotropy taken into account and with the average grain size as a parameter. The potential barrier regions around the grain boundaries influence the deformation changes of anisotropy resistance in the fine-grained non-recrystallized SOI-structures doped with boron up to 3ź10(19)cm(-3) only.