Deep Learning- and IoT-Based Framework for Rock-Fall Early Warning

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(17), P. 9978 - 9978

Published: Sept. 4, 2023

In recent years, several strategies have been introduced to enhance early warning systems and lower the risk of rock-falls. this regard, paper introduces a deep learning- IoT-based framework for rock-fall warning, devoted reducing with high accuracy. framework, prediction accuracy was augmented by eliminating uncertainties confusion plaguing model. order achieve accuracy, fused model-based learning detection Internet Things. This study utilized parameters, namely, overall performance measures based on matrix, assess in addition its ability reduce risk. The result indicates an increase model from 86% 98.8%. addition, reduced probability 1.51 × 10−3 8.57 10−9. Our findings demonstrate which also offers reliable decision-making mechanism providing potential hazards rock falls.

Language: Английский

---

Performance of Machine Learning Based-Modelling Approach in Consolidated Bioprocessing with Microbial Consortium for Bioethanol Production

Industrial Biotechnology, Journal Year: 2024, Volume and Issue: 20(2), P. 77 - 97

Published: April 1, 2024

Owing to the complicated biomass characteristics and a variety of operating parameters, it is challenging predict bioethanol yield (Ybeth, %) from various agricultural wastes by consolidated bioprocessing with microbial consortium. In this study, Gaussian Process Regression (GPR) Artificial Neural Networks (ANN), which are powerful supervised machine learning models, were employed as predictive models that can be used estimate wastes. Ninety-six experimental data points obtained literature preprocessed remove noise or outliers dataset. The Learner App in MATLAB 2021a was on refined 50 original parallel computing cross-validation, best model selected. squared exponential GPR gave training testing results, R2 approaching 1, RMSE, MSE, MAE 0, lowest time, highest prediction speed. A larger dataset generally provides more opportunities for neural network learn improve its performance. Therefore, 3500 synthetic generated 35 seed using Gretel ACTGAN, assumptions data, reducing 1,615 points. For ANN model, MSE regression R (1,615 points) trained close 0 respectively. Since an economical method producing bioethanol, further development methods will aid predicting optimizing conditions required greater yields.

Language: Английский

---

Frost-resistance prediction model for stress-damaged lightweight aggregate concrete based on BPNN: a comparative study

Materials Research Express, Journal Year: 2024, Volume and Issue: 11(8), P. 085513 - 085513

Published: Aug. 1, 2024

Abstract With the depletion of natural resources and requirement higher strength-weight ratio, lightweight aggregate concrete has attracted more attention because its good thermal properties, fire resistance seismic performance. However, exposure to low temperature environments accelerates deterioration concrete, thereby, reduce service life concrete. Even worse, in cold arid regions, often experiences accidental impacts, wind erosion, earthquakes, other disasters during service, these damage significantly impact frost-resistance. Therefore, accurately quantitatively describing predicting frost-resistance under specific disaster conditions is crucial. In this study, take initial degree freeze-thaw cycles as input variables, while relative dynamic elastic modulus (RDEM) an out variable, a frost prediction models for stress-damaged was established based on back propagation neural network (BPNN). The results show that predicted values BPNN model are agreement with experimental values, also compared revised Loland which proposed by another author. Results demonstrate average error between only 1.69%, whereas one 9.13%, indicating can achieve relatively accurate quantitative assessment throughout entire post-disaster lifecycle it broadened idea provided reference

Language: Английский

---