BP-ANN Models Research Articles

Prediction flow behavior of Mg-12Gd-3Y-0.6Zr alloy during hot deformation based on Arrhenius and BP-ANN models: a comparative study

Prediction flow behavior of Mg-12Gd-3Y-0.6Zr alloy during hot deformation based on Arrhenius and BP-ANN models: a comparative study

Coupling multi-sensory earth observation datasets, in-situ measurements, and machine learning algorithms for total blue C stock estimation of an estuarine mangrove forest

In recent years, research on blue carbon (C) has garnered substantial attention worldwide. Nevertheless, we observed a lack of holistic approach, in terms of measurement of total blue C (TBC) potentials. This study focuses on developing a novel approach toward blue C accounting by spatially explicit modelling and estimation of TBC stock in a mangrove wetland of eastern India. A hybrid methodology has been adopted incorporating destructive and non-destructive sampling, allometric and predictive modelling, laboratory-based elemental analysis, and multi-sensory remote sensing (RS) based datasets. Predicted TBC density has been mapped within the wetland influence zone (WIZ) of the study site. Point-specific sample data (n = 250) has been used for the determination of the soil organic C (SOC) prediction model. Spline interpolation, displaying highest R2 value (R2 = 0.74) has been chosen for spatially explicit modelling of total SOC stock. Above ground biomass (AGB) was determined using the relationship between remotely sensed data (ALOS PALSAR-2 and Pleiades-1B) and in-situ dendrometric variables (viz. wood density, tree height, and girth at breast height). Here, among the different parametric and nonparametric models to estimate AGB, the BP-ANN models, specifically model number 22 (adjusted R2 = 0.84, MSE = 1.28, AIC = 3.67, BIC = 1.60), has been identified as the best-fit one with higher adjusted R2 and lesser AIC and BIC values. Indirect allometric equations involving modelled AGB values had been used to generate spatially explicit community-specific below ground biomass values at per pixel basis (∼2 m). Above and below ground C were estimated from these raster data. Integrating all these datasets in a GIS platform, the overall TBC stock of the mangrove was recorded at 246710.91 Mg. The TBC density of mangrove WIZ had revealed considerable variations, ranging from 0.34 Mg ha−1 to 881.50 Mg ha−1. Cumulatively, the study attempted to amalgamate all facets of blue C pools with satisfactory accuracy. This holistic methodology may further aid in regional C stock inventorization, management, and policy formulation, thereby strengthening the socio-economic resilience of coastal communities through carbon trading, reduce emissions from ecosystem degradation as well as support ongoing conservation efforts.

Neural networks for predicting air gap membrane distillation performance

The hollow fiber air gap membrane distillation (AGMD) has recently attracted tremendous attention for desalination and wastewater treatment due to its high packing density, low conductive heat loss, and latent heat recovery capability. Utilizing fast and accurate modeling tools to predict MD performance can result in the further development of desalination technologies. However, simple and time-saving prediction models to assess the AGMD performance were not abundant. Herein, AGMD performance, including permeate flux (J) and gained output ratio (GOR) was predicted through multiple linear regression (MLR) model, back propagation neural network (BP ANN) and radial basis function neural network (RBF ANN) under different hot temperatures (Th), coolant temperatures (Tc), feed flow rates (F), and feed concentration (c). A total of 30 sets of data were used to train the proposed models, the other 10 external validation datasets not used for training the models were applied to validate the prediction accuracy. The results depicted that RBF ANN (SPREAD = 30, N = 30) showed greatest prediction performance (R2 = 0.99–1) compared with BP ANN and MLR models (R2 = 0.98–0.99; R2 = 0.89–0.97). The computing time consumption of RBF ANN was higher than BP ANN. According to the Mean impact value (MIV) analysis, Th had the strongest effect on J and GOR. Increasing Th and decreasing c both had positive impacts on J and GOR, but increasing Tc or F resulted in a trade-off influence. A genetic algorithm (GA) was employed to optimize J and GOR simultaneously, the optimum J and GOR could reach 6.00 kg/m2·h and 7.70 respectively. In this study, the three prediction models proved their abilities to predict AGMD performance and further provide guidance in the actual membrane distillation water treatment process.

Factors influencing indoor air pollution in buildings using PCA-LMBP neural network: A case study of a university campus

This study investigated indoor air quality (IAQ) and its association with building characteristics and environmental factors in eleven different campus buildings in Gainesville, Florida. Integrated indoor and outdoor sensor systems are built and installed to measure the levels of airborne particulate matter (PM2.5 and PM10), nitrogen dioxide (NO2), ozone (O3), relative humidity (RH), and temperature continuously in 10 min intervals for two weeks for each case building. Twenty building-related characteristics were collected through a walkthrough-based survey, HVAC historical data, and construction drawings. Through these data, a PCA-assisted Levenberg-Marquardt Backpropagation (LMBP) neural network model was developed for rapidly and accurately analyzing and predicting the interaction between IAQ and its affecting factors. Factors with significant contributions to indoor exposure were may mostly be determined by outdoor sources. This is evidenced by the strong associations that were found between indoor PM(2.5-10) and O3 values and their corresponding outdoor values and factors, including distance from the major traffic (DFMT), cracks occurring, outdoor temperature, and humidity. Indoor NO2 concentrations were affected by DFMT, indoor O3, indoor RH, number of air grilles, room volume, and window-to-wall ratio. Also, the comparison shows that the PCA-LMBP model outperforms the traditional BP-ANN and multi-linear regression methods. The average values of 1.34, 2.53, and 4.86 were obtained for the root-mean-square error (RMSE) of PCA-LMBP, BP-ANN, and MLR models, respectively. These results can be accordingly referred for the follow-up studies that analyze IAQ in similar building and environmental conditions.

A miniaturized electronic nose with artificial neural network for anti-interference detection of mixed indoor hazardous gases

Indoor air quality attracted great attention for its significant threats to human health and safety, especially the potential hazardous gases in kitchens. To meet the requirements of the anti-interference detection of multiple combustible gases, in this paper, a miniaturized electronic nose was developed using MOS sensor array for semi-quantitative and anti-interference detection of carbon monoxide and methane with the interference of hydrogen and formaldehyde. The sensor array was constructed using 6 MOS sensors and cross-reaction to target and interference gases. To implement the anti-interference capability, different models were utilized and evaluated including PCA, LDA and BP-ANN. The 10-fold cross validation results indicate that BP-ANN models have the best performance than other models with the accuracy of 93.35 % for CO and 93.22 % for CH4 without interference. With the interference of H2 and CH2O, the BP-ANN model shows the accuracies of 78.92 % for CO and 89.75 % for CH4. Adding interfering samples of H2 has a more significant impact on BP-ANN models than adding that of CH2O. The results demonstrate that the proposed e-nose with the BP-ANN model can realize semi-quantitative, simultaneous and anti-interference detection of CO and CH4 in the interference environment, which provides a promising platform for gas sensing with multiple interference.

Evaluating the performance of ANN in predicting the concentrations of ambient air pollutants in Nicosia

Air pollution is a crucial concern which impacts human health, agricultural harvests, forestry, animals as well as the environment. Indigenous environmental or health organizations regularly make daily air contamination predictions for public awareness and for use in making decisions concerning reduction methods in addition to the management air quality. Predictions are customarily based on statistical associations between meteorological conditions and ambient air contamination concentrations. Multiple linear regression (MLR) models, using similar input and output parameters, allowing a comparative study of the two methods are extensively utilized. This work aims to develop the ANN models for forecasting concentrations of PM10, NO2, and O3 in Nicosia and to compare the predictive ability of the MLR (linear method) and Artificial Neural Network (ANN) (non-linear method) models. Previous day's pollutant concentration, atmospheric pressure, wind speed, relative humidity, and temperature data from 2012 to 2015 were used as input parameters or independent variables, while observed pollutant concentration for each pollutant was used as output or dependent variable. The reliability and strength of the models were evaluated via the root mean square error, mean absolute error, and the Pearson correlation coefficient. Study results indicated that MLR did better than ANN except in a few cases. However, the Backpropagation (BP) models of all three pollutants developed in this research were found to agree with other studies in the literature proving that the BPANN models built in this study can be used for the prediction of NO2, O3, and PM10.

Research on Offshore Short-term Wind Speed Prediction Based on the CSA Modeling Improved by Random Algorithm

Aiming at improving the prediction accuracy of offshore short-term wind speed, a model based on random cuckoo search algorithm (RCSA) and artificial neural network (ANN) was proposed. Firstly, RCSA was obtained by introducing a random factor to improve CSA, and then a RCSA-ANN model for predicting offshore short-term wind speed was established. Secondly, a wind tower was built in Luchao Port, Shanghai, offshore meteorological data were measured, and the training of the model was carried out. Finally, the precision of RCSA-ANN model was verified by comparison and analysis with BP-ANN and CSA-ANN models. The results show that the improved CSA method is simple, reliable and effective, which solves the problem that the algorithm is easy to fall into local optimum. The average error of RCSA-ANN model is not only lower than that of BP-ANN model, but also much lower than that of CSA-ANN model, and the prediction accuracy of these three models decreases in turn. RCSA-ANN model has high prediction accuracy and can precisely predict fluctuating wind speed sequences, and it also has good application potential.

ARTIFICIAL INTELLIGENCE TECHNIQUES FOR PREDICTING TIDAL EFFECTS BASED ON GEOGRAPHIC LOCATIONS IN GHANA

Tidal forces as a result of attraction of external bodies (Sun, Moon and Stars) through gravity and are a source of noise in many geoscientific field observations. The solid earth tides cause deformation. This deformation results in displacement in geographic positions on the surface of the earth. The displacement due to tidal effects can result in deformation of engineering structures, loss of lives, and economic cost. Tidal forces also help in detecting other environmental and tectonic signals. This study quantifies the effects of solid earth tides on stationary survey controls in five regions of Ghana. The study is in two stages: firstly, the solid earth tides were estimated for each control by a geometric approach (combining Navier’s equation of motion and Love theories). Secondly, estimation using two artificial intelligence methods (Multivariate Adaptive Regression Splines (MARS) and Backpropagation Artificial Neural Network (ANN)). Based on statistical indices of Mean Square Error (MSE) and Correlation Coefficient (R), BPANN, and MARS models can be used as a realistic alternative technique in quantifying solid earth tides for the study area. The MSE and R (MSE; BPANN = 1.3249 × 10–04 and MSE; MARS = 2.2052 × 10–06; R; BPANN = –0.6067 and R; MARS 0.6570) values indicate that MARS outperforms BPANN in quantifying solid earth tides in the study area. BPANN and MARS can be used as an efficient tool for quantifying tidal values based on geographic positions for geodetic deformation studies within the study area.

Prediction of Oral Acute Toxicity of Organophosphates Using QSAR Methods.

Prediction of oral acute toxicity of organophosphates using QSAR methods. Prediction of oral acute toxicity of organophosphates (including some pesticides and insecticides) using GA-MLR and BPANN methods. The aim of the present study was to develop quantitative structure-activity relationship (QSAR) models, based on molecular descriptors to predict the oral acute toxicity (LD50) of organophosphate compounds. The QSAR models based on genetic algorithm-multiple linear regression (GA-MLR) and back-propagation artificial neural network (BPANN) methods were proposed. The prediction experiment showed that the BPANN method was a reliable model for screening molecular descriptors, and molecular descriptors obtained by BPANN models could well characterize the molecular structure of each compound. It was indicated that among molecular descriptors to predict the LD50 of organophosphates, ALOGP2, RDF030u, RDF065p and GATS5m descriptors have more importance than the other descriptors. Also BPANN approach with the values of root mean square error (RMSE= 0.00168), square correlation coefficient (R2 = 0.9999) and absolute average deviation (AAD=0.001675045) gave the best outcome, and the model predictions were in good agreement with experimental data. The proposed model may be useful for predicting LD50 of new compounds of similar class.

A novel model for malaria prediction based on ensemble algorithms.

Background and objectiveMost previous studies adopted single traditional time series models to predict incidences of malaria. A single model cannot effectively capture all the properties of the data structure. However, a stacking architecture can solve this problem by combining distinct algorithms and models. This study compares the performance of traditional time series models and deep learning algorithms in malaria case prediction and explores the application value of stacking methods in the field of infectious disease prediction.MethodsThe ARIMA, STL+ARIMA, BP-ANN and LSTM network models were separately applied in simulations using malaria data and meteorological data in Yunnan Province from 2011 to 2017. We compared the predictive performance of each model through evaluation measures: RMSE, MASE, MAD. In addition, gradient-boosting regression trees (GBRTs) were used to combine the above four models. We also determined whether stacking structure improved the model prediction performance.ResultsThe root mean square errors (RMSEs) of the four sub-models were 13.176, 14.543, 9.571 and 7.208; the mean absolute scaled errors (MASEs) were 0.469, 0.472, 0.296 and 0.266 and the mean absolute deviation (MAD) were 6.403, 7.658, 5.871 and 5.691. After using the stacking architecture combined with the above four models, the RMSE, MASE and MAD values of the ensemble model decreased to 6.810, 0.224 and 4.625, respectively.ConclusionsA novel ensemble model based on the robustness of structured prediction and model combination through stacking was developed. The findings suggest that the predictive performance of the final model is superior to that of the other four sub-models, indicating that stacking architecture may have significant implications in infectious disease prediction.

Predicting SuperDARN cross polar cap potential by applying regression analysis and machine learning

The cross polar cap potential (CPCP) is one of the primary parameters characterizing the electrodynamic feature of the high latitude ionosphere convection. In this study, we perform a comprehensive investigation of the Super Dual Auroral Radar Network (SuperDARN) CPCP, based on a large database of measurements from 1999 to 2009, and its relationship with various parameters of the solar wind, interplanetary magnetic field (IMF) and geomagnetic indices. Specifically, the IMF clock angle, the IMF Bz, the solar wind velocity, the plasma proton density, AE index, SymH index and Dst index are under consideration. According to the results of the correlation, the input parameters are selected and two models of the CPCP based on the multivariate regression analysis and Back Propagation Artificial Neural Network (BP ANN) algorithm are proposed respectively. The regression and BP ANN models are validated and the accuracy as well as the stability of the models is tested by using independent datasets. The result shows that the root mean square error (RMSE) between the measured and the model values ranges from 3.7 to 6.7 kV and the linear correlation coefficients are close to, or above 0.7. The ANN model is shown to have a better performance than the regression model.

Influencing factors and prediction of ambient Peroxyacetyl nitrate concentration in Beijing, China.

Peroxyacyl nitrates (PANs) are important secondary pollutants in ground-level atmosphere. Accurate prediction of atmospheric pollutant concentrations is crucial to guide effective precautions for before and during specific pollution events. In this study, four models based on the back-propagation (BP) artificial neural network (ANN) and multiple linear regression (MLR) methods were used to predict the hourly average PAN concentrations at Peking University, Beijing, in 2014. The model inputs were atmospheric pollutant data and meteorological parameters. Model 3 using a BP-ANN based on the original variables achieved the best prediction results among the four models, with a correlation coefficient (R) of 0.7089, mean bias error of -0.0043 ppb, mean absolute error of 0.4836 ppb, root mean squared error of 0.5320 ppb, and Willmott's index of agreement of 0.8214. Based on a comparison of the performance indices of the MLR and BP-ANN models, we concluded that the BP-ANN model was able to capture the highly non-linear relationships between PAN concentration and the conventional atmospheric pollutant and meteorological parameters, providing more accurate results than the traditional MLR models did, with a markedly higher goodness of R. The selected meteorological and atmospheric pollutant parameters described a sufficient amount of PAN variation, and thus provided satisfactory prediction results. More specifically, the BP-ANN model performed very well for capturing the variation pattern when PAN concentrations were low. The findings of this study address some of the existing knowledge gaps in this research field and provide a theoretical basis for future regional air pollution control.

Fractional Snow-Cover Mapping Based on MODIS and UAV Data over the Tibetan Plateau

Moderate-resolution imaging spectroradiometer (MODIS) snow-cover products have relatively low accuracy over the Tibetan Plateau because of its complex terrain and shallow, fragmented snow cover. In this study, fractional snow-cover (FSC) mapping algorithms were developed using a linear regression model (LR), a linear spectral mixture analysis model (LSMA) and a back-propagation artificial neural network model (BP-ANN) based on MODIS data (version 006) and unmanned aerial vehicle (UAV) data. The accuracies of the three models were validated against Landsat 8 Operational Land Imager (OLI) snow-cover maps (Landsat 8 FSC) and compared with the MODIS global FSC product (MOD10A1 FSC, version 005) for the purpose of finding the optimal algorithm for FSC extraction for the Tibetan Plateau. The results showed that (1) the overall retrieval results of the LR and BP-ANN models based on MODIS and UAV data were relatively similar to the OLI snow-cover maps; the accuracy and stability were greatly improved, with even some reduction in errors; compared to the Landsat 8 FSC, the correlation coefficients (r) were 0.8222 and 0.8445 respectively and the root-mean-square errors (RMSEs) were 0.2304 and 0.2201, respectively. (2) The accuracy and stability of the fully constrained LSMA model using the pixel purity index (PPI) endmember extraction method based only on MODIS data suffered the worst performance of the three models; r was only 0.7921 and the RMSE was as large as 0.3485. There were some serious omission phenomena in the study area, specifically for the largest mean absolute error (MAE = 0.2755) and positive mean error (PME = 0.3411). (3) The accuracy of the MOD10A1 FSC product was much lower than that of the LR and BP-ANN models, although its accuracy slightly better that of the LSMA based on comprehensive evaluation of six accuracy indices. (4) The optimal model was the BP-ANN model with combined inputs of surface reflectivity data (R1–R7), elevation (DEM) and temperature (LST), which can easily incorporate auxiliary information (DEM and LST) on the basis of (R1–R7) during the relationship training period and can effectively improve the accuracy of snow area monitoring—it is the ideal algorithm for retrieving FSC for the Tibetan Plateau.

Real time optimization based on a serial hybrid model for gold cyanidation leaching process

To implement optimization and control on gold cyanidation leaching process (GCLP), it is an important prerequisite to establish an accurate process model. In this paper, a hybrid model in serial structure was proposed, where a first-principle model based on mass conservation equations was presented to describe the basic process behavior and its unknown kinetic reaction rates were predicted using BP ANN models without any structures considered. The proposed serial hybrid model had been applied to the prediction of gold recovery of the GCLP in a gold treatment plant. The results indicate that the proposed serial hybrid model has better prediction performance and generalization ability than the pure mechanistic model. To further reduce the effect of prediction error (plant-model mismatch) on real time optimization (RTO), modifier adaptation approach had been investigated and implemented to the GCLP. The result shows that when model mismatches with the actual plant or larger process disturbance occurs, significant reduction of production cost can be actualized iteratively by implementing the proposed adaptive RTO strategy.

Application of particle swarm optimization-based least square support vector machine in quantitative analysis of extraction solution of yangxinshi tablet using near infrared spectroscopy

A particle swarm optimization (PSO)-based least square support vector machine (LS-SVM) method was investigated for quantitative analysis of extraction solution of Yangxinshi tablet using near infrared (NIR) spectroscopy. The usable spectral region (5400–6200 cm-1) was identified, then the first derivative spectra smoothed using a Savitzky–Golay filter were employed to establish calibration models. The PSO algorithm was applied to select the LS-SVM hyperparameters (including the regularization and kernel parameters). The calibration models of total flavonoids, puerarin, salvianolic acid B and icariin were established using the optimum hyperparameters of LS-SVM. The performance of LS-SVM models were compared with partial least squares (PLS) regression, feed-forward back-propagation network (BPANN) and support vector machine (SVM). Experimental results showed that both the calibration results and prediction accuracy of the PSO-based LS-SVM method were superior to PLS, BP-ANN and SVM. For PSO-based LS-SVM models, the determination coefficients (R2) for the calibration set were above 0.9881, and the RSEP values were controlled within 5.772%. For the validation set, the RMSEP values were close to RMSEC and less than 0.042, the RSEP values were under 8.778%, which were much lower than the PLS, BP-ANN and SVM models. The PSO-based LS-SVM algorithm employed in this study exhibited excellent calibration performance and prediction accuracy, which has definite practice significance and application value.

Nondestructive Measurement of Grape Leaf Chlorophyll Content Using Multi-Spectral Imaging Technology and Calibration Models

Nondestructive measurement of grape leaf chlorophyll content is essential for precision vineyard management. Multi-spectral imaging technology was adopted for image acquisition of grape leave. For each leaf, a color (R-G-B) image and a near-infrared (NIR) image were taken. These images were then transformed into three vegetation indices, e.g. RVI, NDVI and GNDVI. Calibration models were established, by single-variable linear regression, multi-variable linear regression and BP-ANN. Three color space systems, e.g. R-G-B, CIE XYZ and HIS, were examined with the purpose of model optimization. A total of 112 leave were divided into a calibration set(62) and an independent validation set(50). A SPAD-502 chlorophyll meter was used for reference measurement. The single-variable linear regression result shows that the NDVI index is most significant for the measurement of leaf chlorophyll content with coefficient of determination (r2) of 0.70 for calibration set and 0.69 for independent validation set. It is found that the model for R-index produces higher accuracy than those for G- and B-index, which confirms that chlorophyll content can be correlated with R-grayscale values. By comparison, the multi-variable linear regression models based on R-G-B-NIR achieves higher prediction accuracy with r2 of 0.8174. To further improve the prediction accuracy, several BP-ANN models were developed. The best result was achieved for R-G-B-NIR with r2 of 0.99 for independent validation set. It is concluded that multi-spectral imaging technology coupled with BP-ANN calibration model of R-G-B-NIR grayscales is promising for nondestructive measurement of grape leaf chlorophyll content. This method proposed in the study is worthy of being further examined for in situ determination of nutrition diagnose of grape plant.

QSAR correlation of the melting points for imidazolium bromides and imidazolium chlorides ionic liquids

The melting points of ionic liquids (ILs) of imidazolium bromides and imidazolium chlorides have been investigated by means of quantitative structure–activity relationship (QSAR) approach in order to develop prediction models for predicting the melting points of ionic liquid salts. The cationic structures of these ILs were optimized by means of Hyperchem software and MOPAC program. QSAR module of Materials Studio software and Genetic Algorithm (GA) programs were employed to calculate and select the structure descriptors of ILs, then prediction models correlating the selected structure descriptors and melting points of ionic liquids were set up by using the multiple linear regressions (MLR) method and the back-propagation artificial neural network (BP ANN) method, separately. Finally, the obtained QSAR models, including MLR model and BP ANN model, were validated by external test sets. In this work, three data sets, which were 30 imidazolium bromides, 20 imidazolium chlorides and the merging of above two data sets, respectively, were used to investigate the QSAR correlation of the melting points of ILs. The results demonstrated that the prediction mean absolute errors (MAEs) of MLR models for test sets of those three data sets were in the order of 20.52 K, 13.59 K and 21.95 K, and the prediction MAEs of BP ANN models were 8.77 K, 4.98 K and 9.31 K, respectively. It indicated that the predictions of two models for all melting points of ILs were reliable, and the prediction precision of BP ANN model was higher than that of MLR model.

Expression of P55 and P75 TNF receptors on neoplastic cells from patients with chronic lymphocytic leukemia: [formula omitted] Section of Haematology, and Institute of Cancer Research, University of Trondheim, N-7006 Trondheim, Norway

The melting points of ionic liquids (ILs) of imidazolium bromides and imidazolium chlorides have been investigated by means of quantitative structure–activity relationship (QSAR) approach in order to develop prediction models for predicting the melting points of ionic liquid salts. The cationic structures of these ILs were optimized by means of Hyperchem software and MOPAC program. QSAR module of Materials Studio software and Genetic Algorithm (GA) programs were employed to calculate and select the structure descriptors of ILs, then prediction models correlating the selected structure descriptors and melting points of ionic liquids were set up by using the multiple linear regressions (MLR) method and the back-propagation artificial neural network (BP ANN) method, separately. Finally, the obtained QSAR models, including MLR model and BP ANN model, were validated by external test sets. In this work, three data sets, which were 30 imidazolium bromides, 20 imidazolium chlorides and the merging of above two data sets, respectively, were used to investigate the QSAR correlation of the melting points of ILs. The results demonstrated that the prediction mean absolute errors (MAEs) of MLR models for test sets of those three data sets were in the order of 20.52 K, 13.59 K and 21.95 K, and the prediction MAEs of BP ANN models were 8.77 K, 4.98 K and 9.31 K, respectively. It indicated that the predictions of two models for all melting points of ILs were reliable, and the prediction precision of BP ANN model was higher than that of MLR model.