An intelligent heart disease prediction system based on swarm-artificial neural network
- Authors: Nandy, Sudarshan , Adhikari, Mainak , Balasubramanian, Venki , Menon, Varun , Li, Xingwang , Zakarya, Muhammad
- Date: 2023
- Type: Text , Journal article
- Relation: Neural Computing and Applications Vol. 35, no. 20 (2023), p. 14723-14737
- Full Text: false
- Reviewed:
- Description: The accurate prediction of cardiovascular disease is an essential and challenging task to treat a patient efficiently before occurring a heart attack. In recent times, various intelligent healthcare frameworks have been designed with different machine learning and swarm optimization techniques for cardiovascular disease prediction. However, most of the existing strategies failed to achieve higher accuracy for cardiovascular disease prediction due to the lack of data-recognized techniques and proper prediction methodology. Motivated by the existing challenges, in this paper, we propose an intelligent healthcare framework for predicting cardiovascular heart disease based on Swarm-Artificial Neural Network (Swarm-ANN) strategy. Initially, the proposed Swarm-ANN strategy randomly generates predefined numbers of Neural Networks (NNs) for training and evaluating the framework based on their solution consistency. Additionally, the NN populations are trained by two stages of weight changes and their weight is adjusted by a newly designed heuristic formulation. Finally, the weight of the neurons is modified by sharing the global best weight with other neurons and predicts the accuracy of cardiovascular disease. The proposed Swarm-ANN strategy achieves 95.78% accuracy while predicting the cardiovascular disease of the patients from a benchmark dataset. The simulation results exhibit that the proposed Swarm-ANN strategy outperforms the standard learning techniques in terms of various performance matrices. © 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.
Forced oscillation management in a microgrid with distributed converter-based resources using hierarchical deep-learning neural network
- Authors: Surinkaew, Tossaporn , Emami, Kianoush , Shah, Rakibuzzaman , Islam, Md Rabiul , Islam, Syed
- Date: 2023
- Type: Text , Journal article
- Relation: Electric Power Systems Research Vol. 222, no. (2023), p.
- Full Text: false
- Reviewed:
- Description: In future microgrids (MGs), increasing penetration of distributed converter-based resources (DCRs) has inevitably resulted in the problem of inertia scarcity. The interaction, combination, and/or resonance among converter control loops of DCRs, forced inputs, grid parameters, parasitic elements in networks, and system dominant modes can lead to major forced oscillations (FOs). Previous research works mostly focused the problem of FOs on large-scale power systems. However, the effects of FOs in MGs may be more severe than large-scale power systems due to the lower system inertia. With different characteristics of each DCR, conventional FO management methods applied in large-scale power systems may be ineffective. In this paper, a unified AI-framework named hierarchical deep-learning neural network (HiDeNN) is proposed to effectively handle the FOs in a MG with DCRs. To properly managing the FOs, the HiDeNN is divided into three levels for FO detection, identification, and mitigation, respectively. By considering big data produced from DCRs, the HiDeNN is used to solve complicated FO management problems with a low computational demand. By comparison to conventional FO management methods, performances of the proposed HiDeNN are verified in the modified IEEE 13-node feeder with DCRs under various operating points and FO conditions. © 2023
Modeling the effects of particle shape on damping ratio of dry sand by simple shear testing and artificial intelligence
- Authors: Baghbani, Abolfazl , Costa, Susanga , Faradonbeh, Roohoollah , Soltani, Amin , Baghbani, Hasan
- Date: 2023
- Type: Text , Journal article
- Relation: Applied Sciences (Switzerland) Vol. 13, no. 7 (2023), p.
- Full Text:
- Reviewed:
- Description: This study investigates the effects of sand particle shape, in terms of roundness, sphericity and regularity, on the damping ratio of a dry sand material. Twelve different cyclic simple shear testing scenarios were considered and applied using vertical stresses of 50, 150 and 250 kPa and cyclic stress ratios (CSR) of 0.2, 0.3, 0.4 and 0.5 in both constant- and controlled-stress modes. Each testing scenario involved five tests, using the same sand that was reconstructed from its previous cyclic test. On completion of the cyclic tests, corresponding hysteresis loops were established to determine the damping ratio. The results indicated that the minimum and maximum damping ratios for this sand material were 6.9 and 25.5, respectively. It was observed that the shape of the sand particles changed during cyclic loading, becoming progressively more rounded and spherical with an increasing number of loading cycles, thereby resulting in an increase in the damping ratio. The second part of this investigation involved the development of artificial intelligence models, namely an artificial neural network (ANN) and a support vector machine (SVM), to predict the effects of sand particle shape on the damping ratio. The proposed ANN and SVM models were found to be effective in predicting the damping ratio as a function of the particle shape descriptors (i.e., roundness, sphericity and regularity), vertical stress, CSR and number of loading cycles. Finally, a sensitivity analysis was conducted to identify the importance of the input variables; the vertical stress and regularity were, respectively, ranked as first and second in terms of importance, while the CSR was found to be the least important parameter. © 2023 by the authors.
Prediction of secant shear modulus and damping ratio for an extremely dilative silica sand based on machine learning techniques
- Authors: Baghbani, Abolfazl , Choudhury, Tanveer , Samui, Pijush , Costa, Susanga
- Date: 2023
- Type: Text , Journal article
- Relation: Soil Dynamics and Earthquake Engineering Vol. 165, no. (2023), p.
- Full Text: false
- Reviewed:
- Description: A better understanding of soil response to dynamic loads, including earthquakes, can result in safer designs that can reduce casualties. The damping ratio and shear modulus are critical parameters in soil dynamics, and several factors affect these parameters, including density and moisture content. The non-linear and multiple influences on these two parameters make their estimation difficult. Machine learning techniques are very powerful mapping tools with a remarkable capacity to perform nonlinear multivariate function approximations. In this study, to predict sand secant shear modulus and damping ratio from input variables, artificial neural networks (ANN) and classification and regression random forests (CRRF) were used as alternative estimators. The database was created using a series of simple shear tests that accurately assessed damping ratios and secant shear modulus to predict these two dynamic parameters. The input variables of the proposed predictive models included vertical stress, relative density and cyclic stress ratio, and its outputs included secant shear modulus and damping ratio. The Bayesian Regularization (BR) back-propagation ANN model produced correlation coefficient (R) and mean absolute error (MAE) values of 0.998 and 0.006, respectively, while CRRF models gave R and MAE values of 0.995 and 66.051, respectively. Additionally, sensitivity analysis of artificial intelligence (AI) models demonstrated that vertical stress and relative density played a vital role in predicting damping ratio, while all three parameters were important in predicting secant shear modulus. In this study, two developed artificial intelligence models were compared with existing literature models. According to the results, for test database, the existing models were able to predict the shear modulus and damping ratio with R of 0.911 and 0.918, respectively. However, the proposed ANN and CRRF models were able to predict shear modulus with R of 0.993 and 0.996, and damping ratios with R of 0.992 and 0.990, respectively. The results showed that ANNs and CRRFs were more robust than existing models for predicting damping ratio and shear modulus, as well as identifying the influence of input variables on sand dynamic properties. © 2022 Elsevier Ltd
Application of artificial intelligence in geotechnical engineering : a state-of-the-art review
- Authors: Baghbani, Abolfazl , Choudhury, Tanveer , Costa, Susanga , Reiner, Johannes
- Date: 2022
- Type: Text , Journal article , Review
- Relation: Earth-Science Reviews Vol. 228, no. (2022), p.
- Full Text: false
- Reviewed:
- Description: Geotechnical engineering deals with soils and rocks and their use in engineering constructions. By their nature, soils and rocks exhibit complex behaviours and a high level of uncertainty in material modelling. Artificial intelligence (AI) methods have been developed and used by an increasing number of researchers in the field of geotechnical engineering in the last three decades. These methods have been considered successful due to their ability to predict complex nonlinear relationships. Based on more than one thousand (i.e. 1235) published literatures, this paper presents a detailed review of the performance of AI methods and algorithms used in geotechnical engineering. Nine key areas where the application of AI methods is prominent were identified: frozen soils and soil thermal properties, rock mechanics, subgrade soil and pavements, landslide and soil liquefaction, slope stability, shallow and piles foundations, tunnelling and tunnel boring machine, dams, and unsaturated soils. Artificial Neural Network (ANN) emerged as the most widely used and preferred AI method with 52% of studies relying on it. Other methods that were used to a lesser extent were FIS, ANFIS, SVM, LSTM, CNN, ResNet and GAN. The analysis shows that the success and accuracy of AI applications depends on the number and type of datasets and selection of input parameters. The paper also provides statistical information on research incorporating AI methods and discusses the opportunities and challenges for future research and practical applications in geotechnical engineering. © 2022 Elsevier B.V.
Stability evaluation of dump slope using artificial neural network and multiple regression
- Authors: Bharati, , Ashutosh , Ray, Arunava , Khandelwal, Manoj , Rai, Rajesha , Jaiswal, , Ashok
- Date: 2022
- Type: Text , Journal article
- Relation: Engineering with Computers Vol. 38, no. (2022), p. 1835-1843
- Full Text:
- Reviewed:
- Description: The present paper focuses on designing an artificial neural network (ANN) model and a multiple regression analysis (MRA) that could be used to predict factor of safety of dragline dump slope. To implement these two models, the dataset was utilized from the numerical simulation results of dragline dump slopes, wherein 216 dragline dump slope models were simulated using a numerical modeling technique employed with the finite element method. The finite element model was incorporated a combination of three geometrical parameters, namely, coal-rib height (Crh), dragline dump slope height (Sh), and dragline dump slope angle (Sa) of the dump slope. The predicted results derived from the MRA and ANN models were compared with the results obtained from the numerical simulation of the dump slope models. Moreover, to compare the validity of both the models, various performance indicators, such as variance account for (VAF), determination coefficient (R2), root mean square error (RMSE), and residual error were calculated. Based on these performance indicators, the ANN model has shown a higher prediction accuracy than the MRA model. The study reveals that the ANN model developed in this research could be handy in designing the dragline dump slopes at the preliminary stage. © 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd. part of Springer Nature.
Intelligent energy prediction techniques for fog computing networks
- Authors: Farooq, Umar , Shabir, Muhammad , Javed, Muhammad , Imran, Muhammad
- Date: 2021
- Type: Text , Journal article
- Relation: Applied Soft Computing Vol. 111, no. (2021), p.
- Full Text:
- Reviewed:
- Description: Energy Efficiency is a key concern for future fog-enabled Internet of Things (IoT). Since Fog Nodes (FNs) are energy-constrained devices, task offloading techniques must consider the energy consumption of the FNs to maximize the performance of IoT applications. In this context, accurate energy prediction can enable the development of intelligent energy-aware task offloading techniques. In this paper, we present two energy prediction techniques, the first one is based on the Recursive Least Square (RLS) filter and the second one uses the Artificial Neural Network (ANN). Both techniques use inputs such as the number of tasks and size of the tasks to predict the energy consumption at different fog nodes. Simulation results show that both techniques have a root mean square error of less than 3%. However, the ANN-based technique shows up to 20% less root mean square error as compared to the RLS-based technique. © 2021 Elsevier B.V.
Short-term forecasting of load and renewable energy using artifical neural network
- Authors: Srinivasan, Ram , Balasubramanian, Venki , Selvaraj, Buvana
- Date: 2021
- Type: Text , Journal article
- Relation: International journal of engineering trends and technology Vol. 69, no. 6 (2021), p. 175-181
- Full Text:
- Reviewed:
- Description: Load forecasting is a technique used for the prediction of electrical load demands in battery management. In general, the aggregated level used for short-term electrical load forecasting (STLF) consists of either numerical or non-numerical information collected from multiple sources, which helps in obtaining accurate data and efficient forecasting. However, the aggregated level cannot precisely forecast the validation and testing phases of numerical data, including the real-time measurements of irradiance level (W/m2) and photovoltaic output power (W). Forecasting is also a challenge due to the fluctuations caused by the random usage of appliances in the existing weekly, diurnal, and annual cycle load data. In this study, we have overcome this challenge by using Artificial Neural Network (ANN) methods such as Bayesian Regularisation (BR) and Levenberg-Marquardt (LM) algorithms. The STLF achieved by ANN-based methods can improve the forecast accuracy. The overall performance of the BR and LM algorithms were analyzed during the development phases of the ANN. The input layer, hidden layer and output layer used to train and test the ANN together predict the 24-hour electricity demand. The results show that utilizing the LM and BR algorithms delivers a highly efficient architecture for renewable power estimation demand. © 2021 Seventh Sense Research Group®
Stability prediction of Himalayan residual soil slope using artificial neural network
- Authors: Ray, Arunava , Kumar, Vikash , Kumar, Amit , Rai, Rajesh , Khandelwal, Manoj , Singh, T.
- Date: 2020
- Type: Text , Journal article
- Relation: Natural Hazards Vol. 103, no. 3 (2020), p. 3523-3540
- Full Text:
- Reviewed:
- Description: In the past decade, advances in machine learning (ML) techniques have resulted in developing sophisticated models that are capable of modelling extremely complex multi-factorial problems like slope stability analysis. The literature review indicates that considerable works have been done in slope stability using ML, but none of them covers the analysis of residual soil slope. The present study aims to develop an artificial neural network (ANN) model that can be employed for evaluating the factor of safety of Shiwalik Slopes in the Himalayan Region. Data obtained from numerical analysis of a residual soil slope were used to develop two ANN models (ANN1 and ANN2 utilising eleven input parameters, and scaled-down number of parameters based on correlation coefficient, respectively). A four-layer, feed-forward back-propagation neural network having the optimum number of hidden neurons is developed based on trial-and-error method. The results derived from ANN models were compared with those achieved from numerical analysis. Additionally, several performance indices such as coefficient of determination (R2), root mean square error, variance account for, and residual error were employed to evaluate the predictive performance of the developed ANN models. Both the ANN models have shown good prediction performance; however, the overall performance of the ANN2 model is better than the ANN1 model. It is concluded that the ANN models are reliable, valid, and straightforward computational tools that can be employed for slope stability analysis during the preliminary stage of designing infrastructure projects in residual soil slope. © 2020, Springer Nature B.V.
A feature agnostic approach for glaucoma detection in OCT volumes
- Authors: Maetschke, Stefan , Antony, Bhavna , Ishikawa, Hiroshi , Wollstein, Gadi , Schuman, Joel , Garnavi, Rahil
- Date: 2019
- Type: Text , Journal article
- Relation: PLoS One Vol. 14, no. 7 (2019), p. e0219126
- Full Text:
- Reviewed:
- Description: Optical coherence tomography (OCT) based measurements of retinal layer thickness, such as the retinal nerve fibre layer (RNFL) and the ganglion cell with inner plexiform layer (GCIPL) are commonly employed for the diagnosis and monitoring of glaucoma. Previously, machine learning techniques have relied on segmentation-based imaging features such as the peripapillary RNFL thickness and the cup-to-disc ratio. Here, we propose a deep learning technique that classifies eyes as healthy or glaucomatous directly from raw, unsegmented OCT volumes of the optic nerve head (ONH) using a 3D Convolutional Neural Network (CNN). We compared the accuracy of this technique with various feature-based machine learning algorithms and demonstrated the superiority of the proposed deep learning based method. Logistic regression was found to be the best performing classical machine learning technique with an AUC of 0.89. In direct comparison, the deep learning approach achieved a substantially higher AUC of 0.94 with the additional advantage of providing insight into which regions of an OCT volume are important for glaucoma detection. Computing Class Activation Maps (CAM), we found that the CNN identified neuroretinal rim and optic disc cupping as well as the lamina cribrosa (LC) and its surrounding areas as the regions significantly associated with the glaucoma classification. These regions anatomically correspond to the well established and commonly used clinical markers for glaucoma diagnosis such as increased cup volume, cup diameter, and neuroretinal rim thinning at the superior and inferior segments.
Deep deterministic learning for pattern recognition of different cardiac diseases through the internet of medical things
- Authors: Iqbal, Uzair , Wah, Teh , Habib ur Rehman, Muhammad , Mujtaba, Ghulam , Imran, Muhammad , Shoaib, Muhammad
- Date: 2018
- Type: Text , Journal article
- Relation: Journal of Medical Systems Vol. 42, no. 12 (2018), p.
- Full Text: false
- Reviewed:
- Description: Electrocardiography (ECG) sensors play a vital role in the Internet of Medical Things, and these sensors help in monitoring the electrical activity of the heart. ECG signal analysis can improve human life in many ways, from diagnosing diseases among cardiac patients to managing the lifestyles of diabetic patients. Abnormalities in heart activities lead to different cardiac diseases and arrhythmia. However, some cardiac diseases, such as myocardial infarction (MI) and atrial fibrillation (Af), require special attention due to their direct impact on human life. The classification of flattened T wave cases of MI in ECG signals and how much of these cases are similar to ST-T changes in MI remain an open issue for researchers. This article presents a novel contribution to classify MI and Af. To this end, we propose a new approach called deep deterministic learning (DDL), which works by combining predefined heart activities with fused datasets. In this research, we used two datasets. The first dataset, Massachusetts Institute of Technology–Beth Israel Hospital, is publicly available, and we exclusively obtained the second dataset from the University of Malaya Medical Center, Kuala Lumpur Malaysia. We first initiated predefined activities on each individual dataset to recognize patterns between the ST-T change and flattened T wave cases and then used the data fusion approach to merge both datasets in a manner that delivers the most accurate pattern recognition results. The proposed DDL approach is a systematic stage-wise methodology that relies on accurate detection of R peaks in ECG signals, time domain features of ECG signals, and fine tune-up of artificial neural networks. The empirical evaluation shows high accuracy (i.e., ≤99.97%) in pattern matching ST-T changes and flattened T waves using the proposed DDL approach. The proposed pattern recognition approach is a significant contribution to the diagnosis of special cases of MI. © 2018, Springer Science+Business Media, LLC, part of Springer Nature.
Exploring the application of artificial neural network in rural streamflow prediction - A feasibility study
- Authors: Choudhury, Tanveer , Wei, Jackie , Barton, Andrew , Kandra, Harpreet , Aziz, Abdul
- Date: 2018
- Type: Text , Conference proceedings
- Relation: 27th IEEE International Symposium on Industrial Electronics, ISIE 2018; Cairns, Australia; 13th-15th June 2018 Vol. 2018-June, p. 753-758
- Full Text:
- Reviewed:
- Description: Streams and rivers play a critical role in the hydrologic cycle with their management being essential to maintaining a balance across social, economic and environmental outcomes. Accurate streamflow predictions can provide benefits in many different ways such as water allocation decision making, flood forecasting and environmental watering regimes. This is particularly important in regional areas of Australia where rivers can play a critical role in irrigated agriculture, recreation and social wellbeing, major floods and sustainable environments. There are several hydrological parameters that effect stream flows in rivers and a major challenge with any prediction methodology, is to understand these parameter interdependencies, correlations and their individual effects. A robust methodology is, thus, required for accurate prediction of streamflow under usually unique, waterway-specific conditions using available data. This research employs an approach based on Artificial Neural Network (ANN) to provide this robust methodology. Data from readily available sources has been selected to provide appropriate input and output parameters to train, validate and optimise the neural network. The optimisation steps of the methodology are discussed and the predicted outputs are compared and analysed with respect to the actual collected values. © 2018 IEEE.
- Description: IEEE International Symposium on Industrial Electronics
Prediction of clogging in stormwater filters using artificial neural network
- Authors: Lin, Junlin , Kandra, Harpreet , Choudhury, Tanveer , Barton, Andrew
- Date: 2018
- Type: Text , Conference proceedings
- Relation: 27th IEEE International Symposium on Industrial Electronics, ISIE 2018; Cairns, Australia; 13th-15th June 2018 Vol. 2018-June, p. 771-776
- Full Text: false
- Reviewed:
- Description: Stormwater filtration technologies play a significant role in improving water quality and making treated water available for non-potable uses. However, during treatment processes, contaminants such as suspended solids would lead to clogging in storm water filters, especially those with high infiltration rates. There are several parameters that affect clogging of filters, and a major challenge is to understand the parameter interdependencies, correlations and their individual effects. A robust methodology is, thus, required to accurately predict clogging which would contribute to the development of filtration technologies and in predictive maintenance. This research employs the use of Artificial Neural Network (ANN) model to predict clogging performance of stormwater filters under different operational conditions using experimental data from previous work. A single hidden layer ANN model with 19 hidden layer neurons was developed in this preliminary work.
- Description: IEEE International Symposium on Industrial Electronics
Tire size identification using extreme learning machine algorithm
- Authors: Kahandawa, Gayan , Choudhury, Tanveer , Ibrahim, Yousef
- Date: 2018
- Type: Text , Conference proceedings
- Relation: 27th IEEE International Symposium on Industrial Electronics, ISIE 2018; Cairns, Australia; 13th-15th June 2018 Vol. 2018-June, p. 571-576
- Full Text: false
- Reviewed:
- Description: Precise tire size identification is needed to increase the efficiency and the reliability of tire inflators and to minimize the inflation cycle time. On the other hand the correct inflation pressure improve the road safety and tire life as well. A single hidden layer feed forward neural network (SLFN) is used in this study to precisely identify a tire size to enhance the tire inflation cycle. The training times of traditional back propagation algorithms, mostly used to model such tire identification processes, are far slower than desired for implementation of an on-line control system. Use of slow gradient based learning methods and iterative tuning of all network parameters during the learning process are the two major causes for such slower learning speed. An extreme learning machine (ELM) algorithm, which randomly selects the input weights and biases and analytically determines the output weights, is used in this work to train the SLFNs. It is found that networks trained with ELM have relatively good generalization performance, much shorter training times and stable performance with regard to the changes in number of hidden layer neurons. The result represents robustness of the trained networks and enhance reliability of the mode. Together with short training time, the algorithm has valuable application in tire identification process.
- Description: IEEE International Symposium on Industrial Electronics
An expert system based on hybrid ICA-ANN technique to estimate macerals contents of Indian coals
- Authors: Khandelwal, Manoj , Mahdiyar, Amir , Armaghani, Danial , Singh, Trilok , Fahimifar, Ahmad , Faradonbeh, Roohollah
- Date: 2017
- Type: Text , Journal article
- Relation: Environmental Earth Sciences Vol. 76, no. 11 (2017), p. 1-14
- Full Text: false
- Reviewed:
- Description: Coal, as an initial source of energy, requires a detailed investigation in terms of ultimate analysis, proximate analysis, and its biological constituents (macerals). The rank and calorific value of each type of coal are managed by the mentioned properties. In contrast to ultimate and proximate analyses, determining the macerals in coal requires sophisticated microscopic instrumentation and expertise. This study emphasizes the estimation of the concentration of macerals of Indian coals based on a hybrid imperialism competitive algorithm (ICA)–artificial neural network (ANN). Here, ICA is utilized to adjust the weight and bias of ANNs for enhancing their performance capacity. For comparison purposes, a pre-developed ANN model is also proposed. Checking the performance prediction of the developed models is performed through several performance indices, i.e., coefficient of determination (R2), root mean square error and variance account for. The obtained results revealed higher accuracy of the proposed hybrid ICA-ANN model in estimating macerals contents of Indian coals compared to the pre-developed ANN technique. Results of the developed ANN model based on R2 values of training datasets were obtained as 0.961, 0.955, and 0.961 for predicting vitrinite, liptinite, and inertinite, respectively, whereas these values were achieved as 0.948, 0.947, and 0.957, respectively, for testing datasets. Similarly, R2 values of 0.988, 0.983, and 0.991 for training datasets and 0.989, 0.982, and 0.985 for testing datasets were obtained from developed ICA-ANN model. © 2017, Springer-Verlag Berlin Heidelberg.
Novel tire inflating system using extreme learning machine algorithm for efficient tire identification
- Authors: Choudhury, Tanveer , Kahandawa, Gayan , Ibrahim, Yousef , Dzitac, Pavel , Mazid, Abdul Md , Man, Zhihong
- Date: 2017
- Type: Text , Conference proceedings , Conference paper
- Relation: 2017 IEEE International Conference on Mechatronics, ICM 2017; Gippsland, Victoria; 13th-15th February 2017 p. 404-409
- Full Text: false
- Reviewed:
- Description: Tire inflators are widely used all around the word and the efficient and accurate operation is essential. The main difficulty in improving the inflation cycle of a tire inflator is the identification of the tire connected for inflation. A robust single hidden layer feed forward neural network (SLFN) is, thus, used in this study to model and predict the correct tire size. The tire size is directly related to the tire inflation cycle. Once the tire size is identified, the inflation process can be optimized to improve performance, speed and accuracy of the inflation system. Properly inflated tire and tire condition is critical to vehicle safety, stability and controllability. The training times of traditional back propagation algorithms, mostly used to model such tire identification processes, are far slower than desired for implementation of an on-line control system. Use of slow gradient based learning methods and iterative tuning of all network parameters during the learning process are the two major causes for such slower learning speed. An extreme learning machine (ELM) algorithm, which randomly selects the input weights and biases and analytically determines the output weights, is used in this work to train the SLFNs. It is found that networks trained with ELM have relatively good generalization performance, much shorter training times and stable performance with regard to the changes in number of hidden layer neurons. The result represents robustness of the trained networks and enhance reliability of the mode. Together with short training time, the algorithm has valuable application in tire identification process. © 2017 IEEE.
- Description: Proceedings - 2017 IEEE International Conference on Mechatronics, ICM 2017
Comparison of multiple surrogates for 3D CFD model in tidal farm optimisation
- Authors: Moore, William , Mala-Jetmarova, Helena , Gebreslassie, Mulualem , Tabor, Gavin , Belmont, Michael , Savic, Dragan
- Date: 2016
- Type: Text , Conference proceedings
- Relation: 12th International Conference on Hydroinformatics - Smart Water for the Future, HIC 2016; Songdo Convensialncheon, South Korea; 21st-26th August 2016; published in Procedia Engineering Vol. 154, p. 1132-1139
- Full Text:
- Reviewed:
- Description: Marine currents have been identified as a considerable renewable energy source. Therefore, in recent years, research on optimising tidal stream farm layouts in order to maximise power output has emerged. Traditionally, computational fluid dynamics (CFD) models are used to model power output, but their computational cost is prohibitive within an optimisation algorithm. This paper uses surrogate models in place of CFD simulations to optimise the layout of tidal stream farm layouts. Surrogates are functions which are designed to emulate the behaviour of other models with radically reduced computational expense. Two surrogate models are applied and compared: artificial neural network (ANN) and k-nearest neighbours regression (k-NN). We measure their suitability by four criteria: accuracy, efficiency, robustness and performance within an optimisation algorithm. The results reveal that the ANN surrogate is superior in every criteria to the k-NN surrogate. However, the k-NN surrogate is also able to perform adequate optimisation. Finally, we demonstrate that optimisation relying solely on surrogate models is a viable approach, with dramatically reduced computational expense of optimisation. © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license.
- Description: Procedia Engineering
Prediction of drillability of rocks with strength properties using a hybrid GA-ANN technique
- Authors: Khandelwal, Manoj , Armaghani, Danial
- Date: 2016
- Type: Text , Journal article
- Relation: Geotechnical and Geological Engineering Vol. 34, no. 2 (2016), p. 605-620
- Full Text:
- Reviewed:
- Description: The purpose of this paper is to provide a proper, practical and convenient drilling rate index (DRI) prediction model based on rock material properties. In order to obtain this purpose, 47 DRI tests were used. In addition, the relevant strength properties i.e. uniaxial compressive strength and Brazilian tensile strength were also used and selected as input parameters to predict DRI. Examined simple regression analysis showed that the relationships between the DRI and predictors are statistically meaningful but not good enough for DRI estimation in practice. Moreover, multiple regression, artificial neural network (ANN) and hybrid genetic algorithm (GA)-ANN models were constructed to estimate DRI. Several performance indices i.e. coefficient of determination (R2), root mean square error and variance account for were used for evaluation of performance prediction the proposed methods. Based on these results and the use of simple ranking procedure, the best models were chosen. It was found that the hybrid GA-ANN technique can performed better in predicting DRI compared to other developed models. This is because of the fact that the proposed hybrid model can update the biases and weights of the network connection to train by ANN.
- Description: The purpose of this paper is to provide a proper, practical and convenient drilling rate index (DRI) prediction model based on rock material properties. In order to obtain this purpose, 47 DRI tests were used. In addition, the relevant strength properties i.e. uniaxial compressive strength and Brazilian tensile strength were also used and selected as input parameters to predict DRI. Examined simple regression analysis showed that the relationships between the DRI and predictors are statistically meaningful but not good enough for DRI estimation in practice. Moreover, multiple regression, artificial neural network (ANN) and hybrid genetic algorithm (GA)-ANN models were constructed to estimate DRI. Several performance indices i.e. coefficient of determination (R2), root mean square error and variance account for were used for evaluation of performance prediction the proposed methods. Based on these results and the use of simple ranking procedure, the best models were chosen. It was found that the hybrid GA-ANN technique can performed better in predicting DRI compared to other developed models. This is because of the fact that the proposed hybrid model can update the biases and weights of the network connection to train by ANN. © 2015 Springer International Publishing Switzerland
A new technique to measure interfacial tension of transformer oil using UV-Vis spectroscopy
- Authors: Abu Bakar, Norazhar , Abu-Siada, Ahmed , Islam, Syed , El-Naggar, Mohammed
- Date: 2015
- Type: Text , Journal article
- Relation: IEEE Transactions on Dielectrics and Electrical Insulation Vol. 22, no. 2 (2015), p. 1275-1282
- Full Text:
- Reviewed:
- Description: Interfacial tension (IFT) and acid numbers of insulating oil are correlated with the number of years that a transformer has been in service and are used as a signal for transformer oil reclamation. Oil sampling for IFT measurement calls for extra precautions due to its high sensitivity to various oil parameters and environmental conditions. The current used technique to measure IFT of transformer oil is relatively expensive, requires an expert to conduct the test and it takes long time since the extraction of oil sample, sending it to external laboratory and getting the results back. This paper introduces a new technique to estimate the IFT of transformer oil using ultraviolet-to-visible (UV-Vis) spectroscopy. UV-Vis spectral response of transformer oil can be measured instantly with relatively cheap equipment, does not need an expert person to conduct the test and has the potential to be implemented online. Results show that there is a good correlation between oil spectral response and its IFT value. Artificial neural network (ANN) approach is proposed to model this correlation.
An experimental study on the relationship between localised zones and borehole instability in poorly cemented sands
- Authors: Hashemi, Sam , Taheri, Abbas , Melkoumian, Nouné
- Date: 2015
- Type: Text , Journal article
- Relation: Journal of Petroleum Science and Engineering Vol. 135, no. (2015), p. 101-117
- Full Text: false
- Reviewed:
- Description: Poorly cemented sands are mainly located in areas where layers of unconsolidated formations exist. Drilling a borehole in the ground causes stress perturbation and induces tangential stresses on the borehole wall. If the cohesion between sand particles generated by existing cementation is not high enough, the tensile stress concentration may cause grain debonding and, consequently, borehole breakout. In this study a series of solid and thick-walled hollow cylinder (TWHC) laboratory tests was performed on synthetic poorly cemented sand specimens. The applied stresses were high enough to generate breakout on the borehole wall. Simultaneous real-time monitoring and deformation measurement identified the development of localised breakout zones and compaction bands at the borehole wall during the tests. The results from the video recording of the tests showed that a narrow localised zone develops in the direction of the horizontal stress, where stress concentration causes the full breakout in specimens. Dilation occurred at lower confining pressures in TWHC specimens and contracting behaviour was observed during the onset of shear bands at higher pressures. Scanning electron microscopy (SEM) studies showed that sand particles stayed intact under the applied stresses and micro- and macrocracks develops along their boundaries. The SEM imaging was also used to investigate and characterize pre-existing microcracks on the borehole wall developed due to the specimen preparation. It showed that boring the solid specimen in order to produce a TWHC specimen could generate microcracks on the borehole wall prior to testing which affects the process of borehole failure development during the test. © 2015 Elsevier B.V.