Predicting seismic interaction effect between soil and structure group using convolutional neural network

Heterocyclic amines (HAs) and N-nitrosamines (NAs) are formed easily during the thermal processing of food, and epidemiological studies have demonstrated that consuming HAs and NAs increases the risk of cancer. However, there are few studies on the application of back propagation artificial neural network (BP-ANN) models to simultaneously predict the content of HAs and NAs in sausages. This study aimed to investigate the effects of cooking time and temperature, smoking time and temperature, and fat-to-lean ratio on the formation of HAs and NAs in smoked sausages, and to predict their total content based on the BP-ANN model. With an increase in processing time, processing temperature and fat ratio, the content of HAs and NAs in smoked sausages increased significantly, while the content of HA precursors and nitrite residues decreased significantly. The optimal network topology of the BP-ANN model was 5-11-2, the correlation coefficient values for training, validation, testing and all datasets were 0.99228, 0.99785, 0.99520 and 0.99369, respectively, and the mean squared error value of the best validation performance was 0.11326. The bias factor and the accuracy factor were within acceptable limits, and the predicted values approximated the true values, indicating that the model has good predictive performance. The contents of HAs and NAs in smoked sausages were significantly influenced by the cooking conditions, smoking conditions and fat ratio. The BP-ANN model has high application value in predicting the contents of HAs and NAs in sausages, which provides a theoretical basis for the suppression of carcinogen formation. © 2024 Society of Chemical Industry.

Flavonoids, the most diverse class of plant secondary metabolites, exhibit high affinity toward the purified cytosolic NBD2(C-terminal nucleotide-binding domain) of P-glycoprotein (P-gp). To explore the affinity of flavonoids for P-gp, quantitative structure-activity relationships (QSARs) models were developed using back-propagation artificial neural networks (BPANN) and multiple linear regression (MLR). Molecular descriptors were calculated using PaDELDescriptor, and the number of descriptors was then reduced using a genetic algorithm (GA) and stepwise regression. The MLR (R(2)=0.855, q(2)=0.8138, Rext(2)=0.6916), 14-3-1 BPANN (R(2)=0.8514, q(2)=0.7695, Rext (2)=0.8142), 14-4-1 BPANN(R(2)=0.9199, q(2)=0.7733, Rext(2) =0.8731), and 14-5-1 BPANN (R(2)=0.8660, q(2)=0.7432, Rext(2)=0.8292) models all showed good robustness. While BPANN models exceeded significantly MLR in predictable performance for their flexible characters, could be used to predict the affinity of flavonoids for P-gp and applied in further drug screening.

Due to the similar chemical composition and matrix effect, the accurate identification of mineral pigments on wall paintings has brought great challenges. This work implemented an identification study on three mineral pigments with similar chemical compositions by combining LIBS technology with the K-nearest neighbor algorithm (KNN), random forest (RF support vector machine (SVM), back propagation artificial neural network (Bp-ANN) and convolutional neural network (CNN) to find the most suitable identification method for mural research. Using the SelectKBest algorithm, 300 characteristic lines with the largest difference among the three pigments were determined. The identification models of KNN, RF, SVM, Bp-ANN and CNN were established and optimized. The results showed that, except for the KNN model, the identification accuracy of other models for mock-up mural samples was above 99%. However, only the identification accuracy of 2D-CNN models reached above 94% for actual mural samples. Therefore, the 2D-CNN model was determined as the most suitable model for the identification and analysis of mural pigments.

Artificial neural networks for infectious diarrhea prediction using meteorological factors in Shanghai (China)

Brittleness index prediction in shale gas reservoirs based on efficient network models

Operational parameter impact and back propagation artificial neural network modeling for phosphate adsorption onto acid-activated neutralized red mud

In this study, a back propagation artificial neural network model utilizing characteristics of urban signalized intersections for occurrence prediction of intersection - related traffic crashes, along with its application for crash reduction, are proposed. The 1,593 traffic crashes reported at 62 signalized intersections in Taichung city, Taiwan during year 2000 - 2001 were analyzed. These intersections were decomposed into 636 approaching direction combinations (ADCs). Characteristics, as well as the number of crashes, corresponding to each ADC were investigated. A back propagation artificial neural network (ANN) model was then generated. It was found that the ANN model exhibited great prediction ability with 0.992 correlation coefficient, and 3.38E- 06 mean square error. With the ANN model, a proposed decision-making scheme for intersection rehabilitation was suggested. Case study indicated that the suggested scheme could serve well as a referable tool.

Application of colorimetric sensor array combined with visible near-infrared spectroscopy for the matcha classification

Crop classification is a crucial task in agricultural remote sensing, with the accuracy of such classification heavily relies on field sampling. Reusing historical samples can minimize the reliance on annual field sampling for crop classification. Although previous research primarily focused on the classification accuracy based on the reused historical samples, the underlying factors that influence the accuracy have not been adequately investigated. In this study, we employed a two-dimensional convolutional neural network (2D-CNN) model for crop classification reusing historical samples and investigated the factors influencing the classification accuracy. First, we calculated a normalized difference vegetation index (NDVI) time series from historical data to characterize crop growth patterns. Secondly, we assessed three different time-series construction methods. Subsequently, we used the 2D-CNN model to automatically extract abstract features of crop growth patterns. Finally, we designed various strategies for reusing historical samples to explore the influencing factors. Experiments were conducted in Kuitun, Xinjiang Uygur Autonomous Region, China, from 2016 to 2020, employing a long time series of Sentinel-2 images as remote sensing data. Our results indicated that optimal 2D-CNN models using an irregular satellite image time series (irSITS) outperformed random forest models in inter-annual classifications (the overall accuracies for 2016–2020 were 0.78, 0.61, 0.89, 0.89, and 0.70, respectively). In addition, we identified that the primary factors affecting classification accuracy were 1) the time-series construction method used; 2) the crop growth patterns; and 3) the sample diversity. By reusing historical samples and considering the factors influencing classification accuracy, this study provides valuable insights into high-quality crop classification mapping under limited field sample conditions.

The strong non-linearity and low prediction accuracy of defect depth inversion often lead to challenges in magnetic flux leakage (MFL) detection and evaluation. If a one-dimensional convolutional neural network (1D-CNN) model could automatically extract the features of the original MFL signal and its simple and compact configuration, then real-time and low-cost hardware implementation should be achievable in the future. Therefore, in this work, a pipeline defect size inversion method based on a 1D-CNN model is proposed and the optimal network hierarchy is explored. In order to fully fuse the three-axis MFL signal and input it into the 1D-CNN for defect size inversion, three signal input strategies suitable for MFL defect detection based on the 1D-CNN are proposed and the quantitative results of the different three-axis MFL signal input strategies for an artificial defect dataset are compared and analysed. Experimental results indicate that using a lateral connection between the three-axis MFL signals can effectively achieve high-precision quantification of defect size and especially the quantification of defect depth. In addition, the comprehensive comparison and analysis with other network structures and traditional algorithms also verify that the 1D-CNN method still has advantages in accuracy and time complexity.

Based on the data of household income of Shanghai low-rent housing families, a GM(1,1) forecast model and a Back-Propagation Artificial Neural Network (BPANN) forecast model are established respectively to predict the average household income of low-rent housing families. The comparison between the GM(1,1) and the BPANN model showed that the BPANN model is better than the GM(1,1) model at the aspects of prediction accuracy and data adaptability. The BPANN model could be applied successfully to predict the average household income of Shanghai low-rent housing families in a short-term and it will provide scientific and effective basis for formulate policy on low-rent housing.

In this study, the performance of an integrated desiccant air conditioning system (IDACS) activated by solar energy is evaluated by back propagation artificial neural network (BP-ANN). The IDACS consists of a liquid desiccant dehumidification cycle combined with a vapor compression refrigeration cycle. The integrated system performance is assessed utilizing the system coefficient of performance (COP), outlet dry air temperature (Tda-out), and specific moisture removal (SMR). The training of the BP-ANN is accomplished utilizing experimental results previously published. The results of the BP-ANN model revealed the high accuracy in predicting system performance parameters compared with experimental values. The BP-ANN model has shown relative errors in the trained mode for COP, Tda-out, and SMR within ±0.005%, ±0.006%, and ±0.05%, respectively. On the other side, the BP-ANN model is inspected in the predictive mode as well. The relative errors of the model for COP, Tda-out, and SMR in the predictive mode are within ±0.006%, ±0.006%, and ±0.004%, respectively. The influences of some selected parameters, namely regeneration temperature, desiccant solution temperature in the condenser and evaporator, and strong solution concentration on the system performance are examined and discussed as well.

Compressed chromatographic fingerprint of Artemisiae argyi Folium empowered by 1D-CNN: Reduce mobile phase consumption using chemometric algorithm.

Near infrared spectroscopy and the back propagation artificial neural network model in conjunction with backward interval partial least squares algorithm were used to estimate the purchasing price of Enshi yulu young tea shoots. The near-infrared spectra regions most relevant to the tea shoots price model (5700.5–5935.8, 7613.6–7848.9, 8091.8–8327.1, 8331–8566.2, 9287.5–9522.5, and 9526.6–9761.9 cm–1) were selected using backward interval partial least squares algorithm. The first five principal components that explained 99.96% of the variability in those selected spectral data were then used to calibrate the back propagation artificial neural tea shoots purchasing price model. The performance of this model (coefficient of determination for prediction 0.9724; root-mean-square error of prediction 4.727) was superior to those of the back propagation artificial neural model (coefficient of determination for prediction 0.8653, root-mean-square error of prediction 5.125) and the backward interval partial least squares model (coefficient of determination for prediction 0.5932, root-mean-square error of prediction 25.125). The acquisition price model with the combined backward interval partial least squares–back propagation artificial neural network algorithms can evaluate the price of Enshi yulu tea shoots accurately, quickly and objectively.

Rigid diaphragm is generally an acceptable option in most of the seismic design codes in which the in-plane deformability is not permitted because of the infinite in-plane stiffness properties. Several building configurations may exhibit significant flexibility in floor diaphragms and accordingly using the semirigid diaphragm is recommended as significant in-plane deformation does occur. In this study, a comparative study by using rigid and semirigid diaphragms is performed to identify the related effect on the reinforced concrete (RC) walls. In this study, a numerical study for a twelve-story building with dual system of RC columns and walls is performed. All geometrical and loading properties have been kept constant while using two types of diaphragm as rigid (RD) and semirigid (SRD). The seismic performance of the studied two structures was evaluated in terms of the fundamental period, maximum inter-story drift, maximum base shear and stresses on RC walls. Generally, RD produce results nearly identical to those of SRD for base shear, story displacements and inter-story drifts as the slab is sufficiently thick and membrane deformation due to lateral loading is negligible. Compared to SRD, using RD led to increase the internal moment and shear forces resulted from the seismic loads and acting on the RC walls while the resulting vertical loads are nearly identical.