Artificial Neutral Network Model Research Articles

Flood hazard assessment in Yemen using a novel hybrid approach of Grey Wolf and Levenberg Marquardt optimizers

This study aims to map flood susceptibility in the Qaa’Jahran watersheds located in Dhamar, Yemen, using geoprocessing and computational techniques. Historical flood data and SAR imagery were used to monitor and create a flood inventory map. The artificial neutral network (ANN) was trained using a novel algorithm called GWO_LM, which is a hybridization between the Levenberg-Marquardt algorithm (LM) and Grey Wolf Optimizer (GWO) meta-heuristic algorithm and compared the results with state of art machine learning algorithms. The GWO_LM_ANN model exhibited excellent performance in the evaluation, achieving a precision of 97.92%, sensitivity of 100%, specificity of 100%, F1 score of 98.95%, accuracy of 98.75% and AUC of 98.48. This indicates that using GWO_LM for training ANN enhanced the searching process for the optimal weights, resulting in outperforming other state-of-the-art models. The findings hold significant implications for disaster preparedness and response in the Qaa’Jahran watersheds, enabling targeted and efficient non-structural solutions to mitigate the detrimental effects of flash floods in particularly sensitive locations. The use of the previously unexplored GWO_LM model represents a notable advancement in flood susceptibility assessment, surpassing traditional methods and offering novel insights to the existing literature.

Relative contribution of multi-source water recharge to riparian wetlands along the lower Yellow River

Rivers play a vital role in both the formation and maintenance of riparian wetland hydrology. However, few studies have focused on the response of water recharge of riparian wetlands to altered hydrological processes induced by water-sediment regulation practices. To fill this gap, our study investigated the contribution of multi-source water recharge of riparian wetlands in the lower Yellow River, as well as its influence both during and before the water-sediment regulation scheme of Xiaolangdi Dam. Our study is based on hydrochemistry and isotopic methods, using a Bayesian mixing model and artificial neutral network model. The results showed that riparian wetlands were fed by mixed sources, including groundwater, canals, the Yellow River, and precipitation. However, seasonal evaporation introduced additional variation, which affected the relative contribution of these sources across seasons. Among these sources, the Yellow River served as the main water source for recharging riparian wetlands, and its contribution varied both spatially and temporally (across seasons). Specifically, proximity of riparian wetlands was the primary factor explaining spatial variation in the contribution of Yellow River, while climatic (12.38%) and hydrological variabilities (87.62%) explained seasonal variation. Among these climatic and hydrological variables, suspended sediment content was the most important factor—with a relative contribution of 36.33%. By determining the contribution of the Yellow River to the recharge of riparian wetlands, our study has provided information which is beneficial to adaptive management of river-fed riparian wetlands, especially under the implementation of water-sediment regulation practices.

Machine learning prediction of the optimal carrier concentration and band gap of quaternary thermoelectric materials via element feature descriptors

AbstractWhile the thermoelectric (TE) materials have attracted significant attention in recent years, the design and discovery of new TE materials with optimal carrier concentration and band gap remains a great challenge. Herein, we report the development of machine learning (ML) methods to predict TE materials with introducing physically meaningful simple descriptors. Specifically, we use the number of electrons, Pauling electronegativity and relative atomic mass as the basic physical variables and compute 242 descriptors in 64 categories to characterize the molecular information of a TE material. Multiple stepwise regression is employed to reduce the dimensionality in the developed ML models, and 5 and 4 important features for the band gap and carrier concentration is selected, respectively. The important features are used as input of a total number of 19 ML methods to select the optimal ML models for the prediction of band gap and carrier concentration, respectively. It is shown that the least square support vector machines method is the best model for the prediction of the band gap, while the back propagating artificial neutral network model exhibits the best performance in predicting the carrier concentration values. This work provides novel theoretical guidance for the rapid prediction properties of TE materials. The simple descriptors we defined can accurately predict the band gap and carrier concentration of quaternary TE materials.

Prediction of Distal Aortic Enlargement after Proximal Repair of Aortic Dissection Using Machine Learning

This study aimed to construct a risk prediction model for distal aortic enlargement in patients with type B aortic dissection (TBAD) treated with proximal thoracic endovascular aortic repair (TEVAR). From June 2010 to June 2016, patients with TBAD who underwent proximal TEVAR were retrospectively analyzed. A total of 38 clinical and imaging variables were collected. Univariable logistic regression was conducted to explore potential risk factors associated with distal aortic enlargement. Elastic net regression was employed to select significantly influential variables. Then, machine learning algorithms (logistic regression (LR), artificial neutral network (ANN), random forest and support vector machine) were applied to build risk prediction models. The area under the receiver operating characteristic curve (AUC), sensitivity and specificity were used to evaluate the performance of these models. A total of 503 patients were enrolled in this study. During the follow-up, 105 (20.9%) patients were identified as having distal aortic enlargement, and 69 (13.7%) patients were found to have distal aortic aneurysm formation. Five patients were identified with aortic rupture. True lumen collapse and multi-false lumens were two potential risk factors for distal aortic enlargement after proximal repair of TBAD. The LR model performed the best in predicting distal aortic enlargement, with the highest sensitivity (96.7%) and an AUC of 0.773. The best model for predicting distal aneurysm formation was the ANN model, which yielded the highest AUC (0.876) and a specificity of 79.1%. Machine learning approaches can produce accurate predictions of distal aortic enlargement after proximal repair of TBAD, which potentially benefits subsequent management.

Bond Strength Assessment of Concrete-Corroded Rebar Interface Using Artificial Neutral Network

Bond strength assessment is important for reinforced concrete structures with rebar corrosion since the bond degradation can threaten the structural safety. In this study, to assess the bond strength in concrete-corroded rebar interface, one of the machine learning techniques, artificial neutral network (ANN), was utilized for the application. From existing literature, data related to the bond strength of concrete and corroded rebar were collected. The ANN model was applied to understand the factors on bond property degradation. For the input in the ANN model, the following factors were considered the relative bond strength: (1) corrosion level; (2) crack width; (3) cover-to-diameter ratio; and (4) concrete strength. For the cases with confinement (stirrups), (5) the diameter/stirrups spacing ratio was also considered. The assessment was conducted from input with single parameter to multiple parameters. The scaled feed-forward multi-layer perception ANN model with the error back-propagation algorithm of gradient descent and momentum was found to match the experimental and computed results. The correlation of each parameter to the bond strength degradation was clarified. In cases without confinement, the relative importance was (1) > (2) > (4) > (3), while it was (2) > (1) > (3) > (5) > (4) for the cases with confinement.

Research on modelling vestibular rehabilitation decision based on machine learning

Objective:To evaluate the effect of the support vector machine（SVM） and artificial neutral network（ANN） on the treatment choice of vestibular rehabilitation. Method:Total scores COMP and three ratios of sensory analysis: somatosensory（SOM）, visual（VIS）, vestibular（VEST） from the sensory organization test（SOT）, and physical score（DHI-P）, emotional score（DHI-E）, functional score（DHI-F） from the dizziness handicap inventory（DHI） were chosen as input of SVM and ANN, rehabilitation program as output. According to the data source of the literatures, we constructed simulation database used as the sample set to conduct model training, and part of the clinical data was used to train the model accuracy. Result:After trainings, the accuracy rate of ANN model was 52.3%, and that of SVM model was 83.4%. The error mainly comes from the serious overlap of each score data interval under the three diagnostic schemes, which easily leads to the misclassification of boundary sample points, which is also a difficult problem to overcome in clinical diagnosis. Conclusion:Vestibular rehabilitation decision based SVM is more accurate than ANN. The use of machine learning to assist decision-making of vestibular rehabilitation scheme has important prospective reference significance in promoting clinical medical informatization and improving medical quality.

Chemical characterization of aromas in beer and their effect on consumers liking

Identification of volatiles in beer is important for consumers acceptability. In this study, triplicates of 24 beers from three types of fermentation (top/bottom/spontaneous) were analyzed using Gas Chromatograph with Mass-Selective Detector (GC-MSD) employing solid-phase microextraction (SPME). Principal components analysis was conducted for each type of fermentation. Multiple regression analysis, and an artificial neutral network model (ANN) were developed with the peak-areas of 10 volatiles to evaluate/predict aroma, flavor and overall liking. There were no hops-derived volatiles in bottom-fermentation beers, but they were present in top and spontaneous. Top and spontaneous had more volatiles than bottom-fermentation. 4-Ethyguaiacol and trans-β-ionone were positive towards aroma, flavor and overall liking. Styrene had a negative effect on aroma, flavor and overall liking. An ANN model with high accuracy (R = 0.98) was obtained to predict aroma, flavor and overall liking. The use of SPME-GC-MSD is an effective method to detect volatiles in beers that contribute to acceptability.

Particulate pollution status and its characteristics during 2015–2016 in Hunan, China

Severe air pollution, especially particulate pollution, is a distressful problem for many regions. This study focused on a respective inland area-Hunan province China, which was surrounded by mountains in three sides. This unique characteristic of terrain contributed largely to the heavy pollution level. We used Multiple Linear Regression (MLR) model, Artificial Neutral Network (ANN) model and the Hybrid Single Paricle Lagrangian Integrated Trajectory Model to study pollution characteristics of PM2.5. The results showed that the meteorological parameters were correlated with PM2.5 concentration and ANN model performed better than MLR model in predicting PM2.5 concentrations in Hunan. According to the trajectory, several peaks were observed in January. The first one occurred at the beginning of January. We supposed that the firework used to celebrate the New Year accounted for the climb of PM2.5 concentration. Moreover, peaks were found in the same day in 14 cities during heavy polluted period. Based on the trajectory, we found that there were successive air masses coming from the northern or northwest China entered into Hunan in January. The pocket-like terrain of Hunan led to the stagnation of air masses causing the air pollution. Besides, the trajectories also implied that the air masses would bypass Changsha, Zhuzhou and Xiangtan, which were regarded as pollution sources, before reaching other cities. Therefore, the pollutants emission should be strictly regulated during the polluted period. Moreover, ozone pollution is becoming more obvious during summer time, which might be a major pollutant after particulate pollution is under control.

Prediction of Photovoltaic (PV) Output Using Artificial Neutral Network (ANN) Based on Ambient Factors

This paper is proposed an artificial neural network (ANN) to apply in the system of prediction of power output from photovoltaic (PV) panel system. In order to test the efficiency and reliability of a proposed ANN model experimental output will be comparing with mathematical equation. The objectives of this project are to develop the ANN model that capable of predicting power output. The activation functions using for the hidden layer is hyperbolic tangent. The training algorithm is used Levenberg-Marquardt backpropagation. The meteorology data as input data was obtained from RET screen database in the period from 1st January 2015 until 31st August 2016. There were two locations in Malaysia to be the subject test; Melaka and Kuala Lumpur. From the result, for Melaka, Malaysia the outputs Vm (MAPE = 0.0003% and RMSE = 8.5%) and Im (MAPE = 4.3% and RMSE = 26.8%). Then, for Kuala Lumpur, Malaysia have a less error and good correlation with Vm (RMSE = 0.2%) and Im(MAPE = 0.008% and RMSE = 0.3%). Hence, average power output was high level in January to March for both locations. The conclusion shows that the performance of power output is depending on the level of solar radiation on the day.

A new method to predict short-term wind power in the renewable energy systems

The production of energy from renewable energy systems RES arrays has the properties of nonlinearity and uncertainty and usually fluctuates with the changes of weather and other factors. In this paper, an integrated method based on case-based reasoning CBR, empirical mode decomposition EMD and artificial neural network ANN is proposed to predict the productivity of the RES. At first, the non-stationary time series is pre-processed by CBR to find out the ten days with the most similarity to the predicted day. Then uses EMD decomposed times series of ten days we selected into a series of IMFs Intrinsic Mode Functions with features of stationarity and multiple time scale, for each IMF component, constructing models of ANN to predict. Finally would be straight line fit to final predict result. The simulation results show that the productivity predicted by using this integrated algorithm is more accurate than that of single artificial neutral network model.

Approximate and Three-Dimensional Modeling of Brightness Levels in Interior Spaces by Using Artificial Neural Networks

In this study, artificial neural networks were used to determine the intensity of brightness in interior spaces. The illumination elements to illuminate indoor spaces were considered, not individually, but as a system. So, during the planned maintenance periods of an illumination system, after its design and installation, simple brightness level measurements must be taken. For a three-dimensional evaluation of the brightness level in indoor spaces in a speedy and accurate manner, the obtained brightness level measurement results and artificial neural network model were used. Upon estimation of the most suitable brightness level for indoor spaces by using the artificial neutral network model, the energy demands required by the illumination elements decreased. Consequently, in this study, with estimations of brightness levels, the extent to which the artificial neutral networks become successful was observed and more correct results have been obtained in terms of both economy and usage.

Predicting Equivalent Static Density of Fuzzy Ball Drilling Fluid by BP Artificial Neutral Network

A back-propagation artificial neutral network model is built based on 220 groups of PVT experimental data to predict the equivalent static density versus depth for fuzzy ball drilling fluid which is a kind of gas-liquid two-phase material. The model is applied in the Mo80-C well located in Sichuan Province of China; the maximum relative error between calculated results and measured data is less than 2%. By comparing with the multiple regression model, the present model has a higher precision and flexibility. The equivalent static density of fuzzy ball drilling fluid from ground to the depth of 6000 m is predicted by the present model, and the results show that the equivalent static density of fuzzy ball drilling fluid will decrease slowly with the growth of depth, which indicates that the gas cores of the fuzzy balls still can exist as deep as 6000 m.

Retrieval of soil moisture by artificial neural network using X-band ground based data

Monitoring of soil moisture is very important to environmental studies, including hydrology, meteorology and their interactive fields. Today back propagation artificial neural networking is a well known and widely applied mathematical model for the remote sensing applications. For the soil moisture estimation an artificial neutral network (ANN) based algorithm is implemented and tested. The ANN model is calibrated (trained) and tested with the experimentally obtained data. The experimentally data is obtained by using X-band (9.5 GHz) scatterometer for different soil moistures viz. 10, 12, 18 and 22%. The measurement of the scattering coefficient was carried out over a range of incidence angle from 20° to 70° at the step of 5° for both the HH and VV polarization. Surface roughness (i.e. root mean square height) is taken constant as 0.5 cm for the whole experimentation. The performance of the ANN model is evaluated by the direct measured soil moisture and by the soil moisture estimated by the ANN model. Our work suggests that ANN modeling for such experimentation is a promising alternative for soil moisture estimation. The advantage of the ANN approach for soil moisture estimation is that it has potential for worldwide coverage.

Advances in the modelling and simulation of emulsion polymerisation

Several different modelling strategies describing homopolymerisation, copolymerisation and terpolymerisation are presented, including zero-one kinetic model, pseudo-bulk kinetic model, level-one model and level-two model, heterogeneous polymerisation and pseudo-homopolymerisation, lumped model and population balance equations (PBEs) model, artificial neutral network model, first-principle model and Monte Carlo model. The mathematical model of the process was solved to obtain several typical product quality indexes of characterising a latex, such as particle size distribution (PSD), molecular weight distribution (MWD), monomer conversion, mean particle diameter and solid content. Micelle nucleation mechanism, particle growth mechanism and coagulation mechanism are presented. Dialectical relations between model prediction conciseness and model quality are discussed based on process control technology. The corporation of chemical engineering model with kinetic model ensures the addition of information about a number of additional phenomena that might affect the PSD into the dynamic model. The perspective and methods of modelling and simulating the process is proposed.

A novel visual image sensor for CO2 short circuiting arc welding and its application in weld reinforcement detection

In this paper, a novel visual image sensor was designed to detect the weld pool images during CO2 short circuiting arc welding. Based on a charge coupled device (CCD) camera, the new sensor was smaller than the previous one. As a result, larger operation space is achieved for the welding torch which is attached to it. Moreover, the mismatching problem between the fixed shooting schedule of the CCD camera and the random short circuiting metal transfer during welding was successfully resolved. By eliminating the influences from welding arc and spatter, clear and complete welding pool images were acquired from different angles around the welding torch. Based on these images, the shape parameters of the welding pool were defined and the relationship between these parameters and weld reinforcement was analysed. In addition, the multiple linear regression (MLR) model and the artificial neutral network (ANN) model were proposed and trained, respectively, to detect weld reinforcement from weld pool images. Results of both the MLR model and the ANN model show that weld reinforcement can successfully be detected.

Artificial neural network vs linear discriminant analysis in credit ratings forecast

PurposeThe purpose of this paper is to perform a comparative study of prediction performances of an artificial neutral network (ANN) model against a linear prediction model like a linear discriminant analysis (LDA) with regards to forecasting corporate credit ratings from financial statement data.Design/methodology/approachThe ANN model used in the study is a fully connected back‐propagation model with three layers of neurons. The paper uses a comparative approach whereby two prediction models – one based on ANN and the other based on LDA are developed using identically partitioned data set.FindingsThe study found that the ANN model comprehensively outperformed the LDA model in both training and test partitions of the data set. While the LDA model may have been hindered by omitted variables; this actually lends further credence to the ANN model showing that the latter is more robust in dealing with missing data.Research limitations/implicationsA possible drawback in the model implementation probably lies in the selection of the various accounting ratios. Perhaps future replications of this study should look more carefully at choosing the ratios after duly addressing the problems of collinearity and duplications more rigorously.Practical implicationsThe findings of this study imply that since ANN models can better deal with complex data sets and do not require restraining assumptions like linearity and normality, it may be overall a better approach in corporate credit rating forecasts that uses large financial data sets.Originality/valueThis study brings out the effectiveness of non‐linear pattern learning models as compared to linear ones in forecasts of financial solvency. This goes on to further highlight the practical importance of the new breed of computational tools available to techno‐savvy financial analysts and also to the providers of corporate credit.

비점원오염모델을 이용한 오염총량모의시스템의 개발 및 적용

본 연구에서는 소유역에서의 오염총량을 추정하기 위하여 위성영상 카테고리분류 인공신경망 모형과 지리정보시스템 기반의 오염총량모의시스템(Total maximum daily Loads simulation System, TOLOS)을 개발하였으며, 발안유역의 HP#6 소유역을 시험유역으로 선정하여 유역 수문·수질 모니터링을 수행하였고, 시험유역의 도형 자료를 구축하여 TOLOS의 적용성을 평가하였다. TOLOS의 오염총량추정 모듈인 SWAT 모형은 논에서의 지표배수량을 고려하여 구성하였다. TOLOS을 이용하여 일별 측정 자료인 유출량, 유사량, 그리고 영양물질에 대하여 SWAT 모형의 보정과 검정을 실시하였으며, 그 결과 적용 가능성이 있는 것으로 나타났다. The objectives of this study are to develop the total maximum daily loads simulation system, TOLOS that is capable of estimating annual nonpoint source pollution from small watersheds, to monitor the hydrology and water quality of the Balkan HP#6 watershed, and to validate TOLOS with the field data. TOLOS consists of three subsystems: the input data processor based on a geographic information system, the models, and the post processor. Land use pattern at the tested watershed was classified from the Landsat TM data using the artificial neutral network model that adopts an error back propagation algorithm. Paddy field components were added to SWAT model to simulate water balance at irrigated paddy blocks. SWAT model parameters were obtained from the GIS data base, and additional parameters calibrated with field data. TOLOS was then tested with ungauged conditions. The simulated runoff was reasonably good as compared with the observed data. And simulated water quality parameters appear to be reasonably comparable to the field data.

Prediction of extrusion pressure using an artificial neural network

A three layer feed-forward artificial neutral network (ANN) model was used for the description of extrusion pressure. The studies employ experimental data obtained from capillary flow experiments using a paste containing 5A zeolite, bentonite and water. On comparing the experimental data, the predictions using the Benbow–Bridgwater equation and the ANN model predictions, it is found that the ANN model is capable of predicting the extrusion pressure well. The neural network model shows how the significant parameters influencing extrusion pressure can be found.