Feed-forward Artificial Neural Network Research Articles

Artificial neural network-empowered projected future rainfall intensity-duration-frequency curves under changing climate

With the increasing trend of greenhouse gases in the atmosphere, by 2052 the temperature is expected to rise by 1.5 °C from the Pre-industrial Period, affecting future extreme rainfall events. This necessitates quantifying extreme hydrologic events to plan and design hydrologic and hydraulic structures using rainfall Intensity-Duration-Frequency (IDF) curves to adapt to future climate scenarios. This study developed future projected IDF curves for the Southeast United States using disaggregated sub-hourly (15-, 30-, and 45-min) monthly maximum rainfall from 2030 to 2059 using five climate models under the Representative Concentration Pathway 8.5 scenario. A computationally efficient feed-forward back-propagation Artificial Neural Network (ANN)-based approach was found to be significantly superior for disaggregating rainfall to a stochastic model with an average Nash–Sutcliffe efficiency (NSE) ranging from 0.67 to 0.84. The study found that there is an increasing rate of future projected annual maximum rainfall intensities in the range of 7% to 36% with reference to the historical period. The spatial variation in future projected extreme rainfall depths showed that the Gulf-Atlantic coast and the Appalachian Mountains are expected to receive more extreme rainfalls.

Numerical study of TC4-NiCr/EG+Water hybrid nanofluid over a porous cylinder with Thompson and Troian slip boundary condition: Artificial neural network model

The analysis is intended to demonstrate two-phase flow of hybrid dusty Eyring–Powell nanofluid flow over a porous cylinder with Thompson and Troian slip boundary condition. The combination of (C2H6O2) ethylene glycol (40 %) and (H2O) water (60 %) -based TC4 (TC4 is an alloy of titanium consisting of 90 % titanium, 6 % aluminium, and 4 % vanadium) was utilized to preparing the hybrid nanofluids. Ti–6Al–4V is another name for this alloy) and nickel-chromium (NiCr) nanoparticles flow properties after the mechanism has been filled with dusty particles. A new approach to the execution of an intelligent numerical computing solution is shown in this paper, which uses the Levenberg-Marquard algorithm to drive MLP feed-forward back-propagation ANN. Data were collected for the testing, certification, and training of the ANN model. The bvp4c with MATLAB solver is used to resolve nonlinear versions of well-known mathematical problems, such as those involving the skin friction coefficient, heat transfer rates, velocity, temperature, and other variables. The ANN model is designed to perform the following tasks: selecting data, building networks, training networks, and assessing their effectiveness via the use of the mean square error measure. The influence of key factors on the fluid transport characteristics is demonstrated via tables and graphs. The velocity profile is higher values, resulting in an amplification of the curvature parameter. The temperature rises when the radiation parameter values decline, and the specific heat ratio parameter exhibits a similar pattern. Higher values of the velocity slip parameter and melting factor both are decays on the Nusselt number. This type of theoretical investigation is a necessary aspect of many aircraft applications, including power production, cooling of electrical and propulsion reactors.

Evaluation of Feature Selection Methods in Estimation of Precipitation Based on Deep Learning Artificial Neural Networks

Precipitation is the most important element of the water cycle and an indispensable element of water resources management. This paper’s aim is to model the monthly precipitation in 8 precipitation observation stations in the province of Hamadan, Iran. The effects and role of different feature weights pre-processing methods (Weight by deviation, Weight by PCA, Weight by correlation and Weight by Support Vector Machine) on artificial intelligence modeling were investigated. Deep learning method based on a multi-layer feed-forward artificial neural network that is trained with Stochastic Gradient Descent using back-propagation (DL-SGD) and Convolutional Neural Networks (CNN) modelling were applied. The precipitation of each station is modeled using the precipitation values of the other stations. The best result, among all scenarios, at the Vasaj station according to the DL-SGD method (CC = 0.9845, NS = 0.9543 and RMSE = 10.4169 mm) and at the Varayineh station according to the CNN method (CC = 0.9679, NS = 0.9362 and RMSE = 16.0988 mm) were estimated.

Development of fragility functions of low-rise steel moment frame by artificial neural networks and identifying effective parameters using SHAP theory

Estimating analytical fragility functions requires high computational costs due to numerous incremental non-linear dynamic analyses. This study employs a soft computing approach to reduce this process and increase estimation accuracy. Utilized a fragility database of low-rise steel moment structures designed for different versions of Iranian seismic design code, and six feedforward Artificial Neural Networks (ANN) with various architectures were trained to estimate the fragility function’s parameters to determine the most desired architectures. The neural networks used the design parameters (frames story, code edition of design, ductility, soil type, near- and far-field earthquake, and seismic zone factor and importance factor) as inputs. For each of the fragility function’s parameters (Median PGA and βs of damages: slight, moderate, extensive, and complete) as outputs, an ANN was trained. The best architect of ANN was selected based on the accuracy of the models from different statistical indicators and the most balanced of the inputs’ importance on the outputs using SHapley Additive explanations (SHAP) analysis. To demonstrate the accuracy of the selected architect, the coefficient of determination was estimated for all database values, and the fragility functions of four randomly chosen frames were compared. The high R2 value in most cases and the closeness of the predicted fragility curved with the actual one indicates the reliability of the ANN method for predicting the fragility function.

Detailed experimentation and prediction of thermophysical properties in lauric acid-based nanocomposite phase change material using artificial neural network

The thermal conductivity (TC) of a nanocomposite phase change material (NPCM) may be improved by adding nanostructured materials to a Phase Change Material (PCM). To assess the heat transfer rate during the process of phase change, such as melting and freezing, an accurate TC prediction of NPCM is required. A Field Emission Scanning Electron Microscope (FESEM) was utilized to examine the nanoparticle morphological study, and X-Ray Diffraction (XRD) analysis evaluated the crystalline structure. NPCMs were verified using Fourier Transform Infrared Spectroscopy (FTIR). The goal of this research is to create an Artificial Neural Network (ANN) that guesses the TC and viscosity of Lauric Acid (LA) embedded with dispersed copper oxide (CuO) and aluminium oxide (Al2O3). A multi-layered feed-forward ANN (MLFFANN) is trained using the Levenberg-Marquardt (LM) backpropagation algorithm. There are 130 experimental datasets in total, obtained from experiments with nanoparticle mass fractions ranging from 1.25 to 10 wt%. The minimum mean square error (MSE) for TC and viscosity is 4.6815 × 10−5 and 2.4681 × 10−5, respectively. The average absolute deviation (AAD) for TC and viscosity is 0.004249 and 0.003596, respectively, while the mean absolute percentage error (MAPE) is 2.1835 % and 4.8197 % for TC and viscosity, and the correlation coefficients(R) are 0.992 and 0.975, respectively. The largest percentage variation between experimental values and ANN computed values for the liquid and solid phases, respectively, is 3.54 % and 0.832 %. It demonstrates that the constructed ANN prediction model predicts the increased TC and viscosity of NPCM for different nanoparticle loadings, temperatures, and oxide nanoparticles.

Enhancing river health monitoring: Developing a reliable predictive model and mitigation plan

The escalating environmental harm inflicted upon rivers is an unavoidable outcome resulting from climate fluctuations and anthropogenic activities, leading to a catastrophic impact on water quality and thousands of individuals succumb to waterborne diseases. Consequently, the water quality monitoring stations have been established worldwide. Regrettably, the real-time evaluation of Water Quality Index (WQI) is hindered by the intricate nature of off-site water quality parameters. Thus, there is a pressing need to create a precise and robust water quality prediction model. The dynamic and non-linear characteristics of water quality parameters pose significant challenges for conventional machine learning algorithms like multi-linear regression, as they struggle to capture these complexities. In this particular investigation, machine learning model called Feedforward Artificial Neural Networks (FANNs) was employed to develop WQI prediction model of Batu Pahat River, Malaysia exclusively utilizing on-site parameters. The proposed method involves a consideration of whether to include or exclude parameters such as BOD and COD, which are not measured in real time and can be costly to monitor as model inputs. Validation accuracy values of 99.53%, 97.99%, and 91.03% were achieved in three different scenarios: the first scenario utilized the full input, the second scenario excluded BOD, and the third scenario excluded both BOD and COD. It was suggested that the model has better predictive power between input variables and output variables. Factor contributed to river pollution has been identified and mitigation plan for Batu Pahat river pollution has been proposed. This could provide an effective alternative to compute the pollution, better manage water resources and mitigate negative impacts of climate change of river ecosystems.

Entropy analysis on EMHD 3D micropolar tri-hybrid nanofluid flow of solar radiative slendering sheet by a machine learning algorithm

The purpose of this paper is to analyze the heat transfer behavior of the electromagnetic 3D micropolar tri-hybrid nanofluid flow of a solar radiative slendering sheet with non-Fourier heat flux model. The conversion of solar radiation into thermal energy is an area of significant interest as the demand for renewable heat and power continues to grow. Due to their enhanced ability to promote heat transmission, nanofluids can significantly contribute to enhancing the efficiency of solar-thermal systems. The combination of silicon oil-based silicon (Si), magnesium oxide (MgO), and titanium (Ti) nanofluids has attracted attention for their ability to improve the performance of solar-thermal systems. The present study discloses a new approach for intelligent numerical computing solving, which utilizes an MLP feed-forward back-propagation ANN and the Levenberg-Marquard algorithm. The collection of data was conducted for the purpose of testing, certifying, and training the ANN model. The Bvp4c solver in MATLAB is utilized to solve the nonlinear equations governing the momentum, temperature, skin-friction coefficient, and Nusselt number. The characteristics of numerous dimensionless parameters such as porosity parameter left(K={0.0,2.0,4.0}right), vortex viscosity parameter left({R}_{1}={0.5,1.0,1.5}right), electric field parameter left(E={0.0,0.1,0.2}right), thermal relaxation time left(Lambda ={0.01,0.10,0.20}right), heat source/sink parameter, left(Q=-{0.3,0.0,0.3}right) thermal radiation parameter left(R={0.5,1}.{0,1.5}right), temperature ratio parameter left({theta }_{w}={0.5,1.0,1.5}right),nanoparticle volume fraction left(phi ={0.00,0.02,0.04}right) on Si + MgO + Ti/silicon oil micropolar tri-hybrid nanofluida are analyzed. The ANN model engages in a process of data selection, network construction, training, and evaluation of its effectiveness through the utilization of mean square error. Tables and graphs are used to show how essential parameters affect fluid transport properties. The velocity profile is decreased by higher values of the porosity parameter, whereas the temperature profile is increased. The temperature profile is inversely proportional to higher values of the electric field parameter. The micro-rotation profiles reduced by expanding values vortex viscosity parameter. It has been determined that entropy generation and Bejan number intensifications for enlarged nanoparticle volume fraction.

Estimation of Continuous Distribution of Iterated Fission Probability Using an Artificial Neural Network with Monte Carlo-Based Training Data

The Monte Carlo neutron transport method is used to accurately estimate various quantities, such as k-eigenvalue and integral neutron flux. However, in the case of estimating a distribution of a desired quantity, the Monte Carlo method does not typically provide continuous distribution. Recently, the functional expansion tally (FET) and kernel density estimation (KDE) methods have been developed to provide a continuous distribution of a Monte Carlo tally. In this paper, we propose a method to estimate a continuous distribution of a quantity in all phase-space variables using a fully connected feedforward artificial neural network (ANN) model with Monte Carlo-based training data. As a proof of concept, a continuous distribution of iterated fission probability (IFP) was estimated by ANN models in two distinct fissile systems. The ANN models were trained on the training data created using the Monte Carlo IFP method. The estimated IFP distributions by the ANN models were compared with the Monte Carlo-based data that include the training data. Additionally, the IFP distributions by the ANN models were also compared with the adjoint angular neutron flux distributions obtained with the deterministic neutron transport code PARTISN. The comparisons showed varying degrees of agreement or discrepancy; however, it was observed that the ANN models learned the general trend of the IFP distributions from the Monte Carlo-based training data.

Machine Learning based IoT Flood Rediction Using Data Modeling and Decision Support System

An essential step in supplying data for climate impact studies and evaluations of hydrological processes is rainfall prediction. However, rainfall events are complex phenomenon’s that continue to be difficult to forecast. In this paper , we present unique hybrid models for the prediction of monthly precipitation that include Seasonal Artificial Neural Networks and Discrete wavelet transforms are two pre-processing methods, together with Artificial Neural Networks have two feed forward neural networks. The temporal series of observed monthly rainfall from Vietnam’s Ca Mau hydrological station were decomposed into three subsets by seasonal decomposition and five sub signals and four levels by wavelet analysis. The methods for predicting rainfall that use feed forward artificial neural networks (ANN) and seasonal artificial neural network (SANN) were fed with the processed data. The classic genetic method and simulated annealing method backed by using an integrated moving average and autoregressive moving was contrasted with the predicted models for model evaluation. The results showed that non-stationary regarding issues with non-linear time series, such forecasting rainfall could be satisfactorily simulated. The SANN model was integrated with the wavelet transform and seasonal decomposition are both used. Techniques, however the wavelet transform method produced the most accurate monthly rainfall data, Predictions. Due to the effects of climate change, nations including the Japan, China, the United States of America, and Taiwan, etc., have recently experienced severe and devastating natural disasters. One of the biggest causes of the destruction in Asian nations like china, India, Bangladesh, Sri Lanka, etc. is the flood. The danger of fatality from these floods is increased by 78% as information technology advances; there is a demand for simple access to massive amounts of cloud storage and computing capacity.

Development of prediction models for interlayer shear strength in asphalt pavement using machine learning and SHAP techniques

The interlayer bonding condition in asphalt pavement significantly affects pavement performance. This study employed machine learning techniques to predict interlayer shear strength (ISS). Feed-forward artificial neural networks (ANN) and random forest (RF) models were developed and compared with traditional multiple linear regression (MLR). Utilizing 156 datasets, divided into 70% training and 30% testing, model performance was assessed using R-squared, mean squared error (MSE), root mean squared error (RMSE), and mean absolute error (MAE). SHapley Additive exPlanations (SHAP) was utilized for model interpretation. The results indicated that the ANN and RF models outperformed MLR, explaining over 95% of experimental data. RF exhibited superior performance with lowest MSE, RMSE, and MAE (0.0029, 0.0538, and 0.0376 MPa). SHAP analysis highlighted the significance of temperature, normal stress, shear deformation rate, and curing time as influential variables in ISS prediction. Elevated temperature adversely influenced ISS, while normal stress, shear deformation rate, and curing time positively contributed to ISS.

A feed-forward artificial neural network with several training methods and various geometries for predicting the rheological behavior of Al2O3/ethylene glycol-water hybrid nanofluid

In this study, the effect of volume fraction of nanoparticles (φ), temperature (Tem.), and shear rates (SR) on the viscosity (μnf) of Al2O3/Ethylene Glycol-Water hybrid nanofluid is investigated using an artificial neural network (ANN). To this end, the μnf is computed for all 180 different combinations of factors with experimental tests, including 7 different φ ranging from 0 ∼ 1.3 % (φ= 0, 0.15, 0.25, 0.5, 0.75, 1, and 1.3 %), 8 different Tem. varying from 25 ∼ 60 °C (T = 25, 30, 35, 40, 45, 50, 55, and 60 °C), and 5 values for shear rates from 20 ∼ 100 rpm (SR = 20, 30, 50, 60, and 100). The results show that two trainbr and trainlm methods show the highest performance for data prediction among all training methods. The trained ANN by employing the trainbr function (and trainlm in the next level) for training has the best performance and shows 0.041 and 0.997 for MSE value and correlation coefficient, respectively. Based on the obtained data, the φ has the highest influence on μnf for this material. Increasing this parameter from 0 to 1.3 % grows the μnf from 2.5 cP to around 12 cP. The next influential parameter is the SR, which has a moderate effect on μnf that changes the μnf around 125 % by changing the SR from 20 to 100. Also, increasing the Tem. to 60 °C will cause a variation of 40 % in the μnf. Tem. and SR are inversely related to μnf, i.e., increasing these variables reduces the output. In general, the results show that the φ has the greatest effect on μnf.

Coarse-Grained Artificial Intelligence for Design of Brush Networks.

The ability to synthesize elastomeric materials with programmable mechanical properties is vital for advanced soft matter applications. Due to the inherent complexity of hierarchical structure-property correlations in brush-like polymer networks, the application of conventional theory-based, so-called Human Intelligence (HI) approaches becomes increasingly difficult. Herein we developed a design strategy based on synergistic combination of HI and AI tools which allows precise encoding of mechanical properties with three architectural parameters: degrees of polymerization (DP) of network strands, nx, side chains, nsc, backbone spacers between side chains, ng. Implementing a multilayer feedforward artificial neural network (ANN), we took advantage of model-predicted structure-property cross-correlations between coarse-grained system code including chemistry specific characteristics S = [l, v, b] defined by monomer projection length l and excluded volume v, Kuhn length b of bare backbone and side chains, and architecture A = [nsc, ng, nx] of polymer networks and their equilibrium mechanical properties P = [G, β] including the structural shear modulus G and firmness parameter β. The ANN was trained by minimizing the mean-square error with Bayesian regularization to avoid overfitting using a data set of experimental stress-deformation curves of networks with brush-like strands of poly(n-butyl acrylate), poly(isobutylene), and poly(dimethylsiloxane) having structural modulus G < 50 kPa and 0.01 ≤ β ≤ 0.3. The trained ANN predicts network mechanical properties with 95% confidence. The developed ANN was implemented for synthesis of model networks with identical mechanical properties but different chemistries of network strands.

Screening COVID-19 by Swaasa AI platform using cough sounds: a cross-sectional study

The Advent of Artificial Intelligence (AI) has led to the use of auditory data for detecting various diseases, including COVID-19. SARS-CoV-2 infection has claimed more than six million lives to date and therefore, needs a robust screening technique to control the disease spread. In the present study we created and validated the Swaasa AI platform, which uses the signature cough sound and symptoms presented by patients to screen and prioritize COVID-19 patients. We collected cough data from 234 COVID-19 suspects to validate our Convolutional Neural Network (CNN) architecture and Feedforward Artificial Neural Network (FFANN) (tabular features) based algorithm. The final output from both models was combined to predict the likelihood of having the disease. During the clinical validation phase, our model showed a 75.54% accuracy rate in detecting the likely presence of COVID-19, with 95.45% sensitivity and 73.46% specificity. We conducted pilot testing on 183 presumptive COVID subjects, of which 58 were truly COVID-19 positive, resulting in a Positive Predictive Value of 70.73%. Due to the high cost and technical expertise required for currently available rapid screening methods, there is a need for a cost-effective and remote monitoring tool that can serve as a preliminary screening method for potential COVID-19 subjects. Therefore, Swaasa would be highly beneficial in detecting the disease and could have a significant impact in reducing its spread.

Appraisal of rock dynamic, physical, and mechanical properties and forecasting shear wave velocity using machine learning and statistical methods

Direct determination of shear wave velocity requires time, cost, and high accuracy due to the complexity of the rock texture. In current research statistical and intelligent approaches have been used to predict the shear wave velocity of rock samples. Also, a new correlation between dynamic and static rock properties was established and the shear wave velocity was estimated based on index tests using Gaussian process regression, multivariate linear regression, feedforward back-propagation artificial neural network, and K-nearest neighbor methods. In total, 120 data related to limestone and sandstone samples of the main projects were used for modeling. Water absorption, compressional wave velocity, and density were used as inputs. The outcomes revealed that the PW/SW ratio is equal to 1.69. Various statistics were used to check the method results. The statistical results showed that it is possible to forecast Ed, and SW with high accuracy. Also, the precision of the GPR was higher than the FBP-ANN, statistical analysis, and KNN. Estimation of SW by GPR showed R of 0.992, and RMSE of 0.06, respectively. These four methods were able to estimate the SW with a mean variation percentage of +0.19%. It's important to consider that these models are best suited for predicting SW when the predictor indicators fall within the same range as this study.

Using machine learning techniques to predict ammonium concentration in membrane contactors for nitrogen recovery as a valuable resource

The availability of real-time measurements of primary quality water variables is one of the key challenges in the wastewater treatment industry. However, due to the cost and maintenance requirements of sensors and probes for on-line measurement of primary quality variables, the prediction of these variables via data-driven approaches using as inputs easy-to-measure process variables has attracted research interest. In this paper, different machine learning techniques: feed-forward artificial neural network, random forest, support vector machine, gaussian process regression and partial least squares were used to predict in real-time the total ammonium nitrogen concentration during the operation of a hollow fibre membrane contactor. This recently developed technology allows the recovery of nitrogen from nitrogen rich streams (i.e. supernatant of anaerobic digesters in wastewater treatment plants) as ammonium sulphate (a marketable fertilizer). These contactors are usually operated in batch mode, pumping the high nitrogen concentration feed from the storage tank, where the total ammonium nitrogen concentration decreases progressively as the fertilizer is produced. Knowing the real-time concentration of total ammonium nitrogen in the storage tank would enable the optimization of the process operation, avoiding its operation with conservative fixed-time batch duration. The pH is an easy-to-measure process variable usually available in wastewater treatment plants that was used as input of the tested data-driven models, together with two extracted features from this variable (its derivative and increments after each reagent dosing). The number of total ammonium nitrogen measurements in the collected database is 2350 data points (corresponding to 8 complete batches, which were divided into 6 for training the data-driven models and 2 for testing them), ranging from 987 to 2.5 mg NH4+-N/L which covers almost the complete range of total ammonium nitrogen concentration values in the membrane contactor. The predictive ability of the developed predictive models was evaluated on the test data set by four indices, namely: the root-mean-square error, the slope and the intercept of the linear fit between the measured and predicted concentrations and the determination coefficient. The results showed a strong predictive ability of the fitted ANN that outperformed the other approaches exhibiting a determination coefficient of 0.99 and the lowest root-mean-square error (19.87 mg/L) in the test set. Permutation variable importance demonstrated that all machine learning techniques depended mainly on the two variables extracted from the pH: its derivative and increments, which resulted to be more important than the pH itself to predict the total ammonium nitrogen concentration.

Damage detection in variable temperature conditions using artificial intelligence

When considering damage detection using the natural frequencies of structures, small frequency drops can indicate either the presence of cracks or a temperature change. This change can lead to additional stress affecting the modal parameters for specific structures, making it much harder to detect, locate, and evaluate damage accurately. The current research aims to describe a method for detecting transverse cracks in beams, considering temperature variations. The considered beam is fixed at both ends, thus inducing axial forces when the temperature is increased. The influence of temperature is considered using adjustment coefficients developed for each vibration mode. This coefficient can be used to accurately calculate the natural frequency for an intact or damaged beam. An analytical method for determining the natural frequencies caused by the changing temperature and the presence of a transverse crack is described and used to generate data for training a feedforward artificial neural network (ANN). The ANN’s capability of determining the position of transverse cracks in double-clamped beams subjected to small temperature changes is proven by creating numerical simulations with known crack positions and thermal conditions for testing the developed method.

Plastic hinge modeling of reinforced concrete Beam-Column joints using artificial neural networks

Beam-column joints play a critical role in transferring forces between beam and column elements and maintaining structural integrity during severe loading. While the nonlinear behaviors of beams and columns are commonly modeled in global frame analyses through the use of plastic hinges, the behavior of joints is often omitted through the use of rigid end offsets. The objective of this study is to develop an artificial neural network and derive the plastic hinge curves required for modeling reinforced concrete beam-column joints in global frame analyses. In pursuit of this objective, a feed-forward artificial neural network (FFNN) is developed to predict the shear strengths of beam-column joints. A comprehensive dataset of 598 experimental joint specimens is compiled from 153 previously published research studies. The 555 data points, which passed the exploratory data analysis, are used to train, test, and validate the proposed network for applicability to a wide range of input variables and joint configurations. The accuracy and reliability of the proposed FFNN are evaluated using a comprehensive set of evaluation metrics in comparison with three existing networks from the literature. The proposed FFNN is used to derive the shear stress–strain and moment-rotation curves required for defining joint hinges in global frame analyses. In addition, a spreadsheet tool is developed to execute the network formulations, calculate the joint shear strength, and derive the joint hinge curves for the use of engineers and researchers.

On the Computability of Primitive Recursive Functions by Feedforward Artificial Neural Networks

We show that, for a primitive recursive function h(x,t), where x is a n-tuple of natural numbers and t is a natural number, there exists a feedforward artificial neural network N(x,t), such that for any n-tuple of natural numbers z and a positive natural number m, the first m+1 terms of the sequence {h(z,t)} are the same as the terms of the tuple (N(z,0),…,N(z,m)).

Intelligent Network Maintenance Modeling for Fixed Broadband Networks in Sustainable Smart Homes

Due to the emergence of sustainable smart homes, each smart device requires more bandwidth putting pressure on the existing home networks. A very good solution to ensure high-bandwidth home networks is the Fiber-to-the-Home (FTTH) technology. FTTH delivers high-speed Internet from a central point directly to the home through fiber optic cables. This fixed broadband network can transmit information at virtually unlimited speed and capacity enabling homes to be smarter. Hence, a well-monitored and well-maintained FTTH broadband network is necessary to obtain a high level of service availability and sustainability in smart homes. This study aims to develop a predictive model that will proactively monitor and maintain FTTH networks through the use of sophisticated modeling techniques such as machine learning (ML). The predictive model targets to classify the proposed technician resolution based on the historical FTTH field dataset. The results show that the K-Nearest Neighbors (KNN)-based model obtained the highest accuracy of 89% followed by the Feed-Forward Artificial Neural Network (FF-ANN)-based model with 86%. In addition, the identified anomalies from the dataset affecting service degradation and performance include FTTH access issues, ONU issues, and faults in customer premises equipment.

Daily scale streamflow forecasting in multiple stream orders of Cauvery River, India: Application of advanced ensemble and deep learning models

Accurate forecasts of streamflow (Qflow) are crucial for optimal management of water reservoir systems and preparing for catastrophic events such as floods. Although several machine learning (ML) techniques have been documented to forecast Qflow with notable accuracy, no model could do so with the utmost certainty since Qflow data can be noisy and sometimes incomplete. Despite these challenges, there has been significant progress in the field of Qflow forecasting. Hence, novel techniques need to be explored or applied to forecast Qflow with improved accuracy and reduced data dependency. Recently, optimized input selection techniques were shown to increase the performance of forecast models, although they have yet to be thoroughly investigated in Qflow forecasting applications. This study aims at the application of deep-learning and advanced ML techniques, namely the (i) Convolutional Neural Network (CNN), (ii) Random Forest (RF), and (iii) Gradient Tree Boosting (GTB), to forecast daily Qflow in multiple stream orders of the Cauvery River, India. The performance of these models was compared with that of conventional multilayer perceptron (MLP), a class of feedforward artificial neural network (ANN). Antecedent daily Qflow time series were used as model inputs, with input selection guided by Akaike’s Information Criteria (AIC) and Bayesian Information Criteria (BIC). The models were also calibrated using inputs selected from commonly used cross-correlation function for comparative evaluation with those models developed using AIC and BIC. The results show that applying AIC and BIC to select the ideal model input parameters marginally increase the model’s performance. The deep-learning and advanced ML techniques clearly outperformed the conventional ANN models in forecasting one- and three-day lead daily Qflow at multiple gauging stations considered along the Cauvery River. It was found that the deep learning (CNN) model had superior forecasting accuracy with respect to all 3 stations for the 1st Input-Output (I/O) combination. An NNSE, WI and KGE indices of 0.85 or higher indicates that the forecast is highly efficient.