Faculty of Engineering

Permanent URI for this communityhttps://rda.sliit.lk/handle/123456789/4203

Browse

Filters

2022 - 20266

Advanced Search

Filter by

Settings

Subject: search.filters.subject.Machine learning

Search Results

Now showing 1 - 6 of 6
  • Thumbnail Image
    PublicationEmbargo
    Investigation of inelastic response ratios for buildings with damping subjected to near-fault ground motions using numerical simulations and transformer-based models
    (Elsevier Ltd, 2026-03-09) Konara, L; Deshika, T; Gobirahavan, R; Alahakoon, Y; Ekanayake I.U; Meddage D.P.P
    Inelastic responses are used in seismic design to estimate inelastic seismic demand from known elastic demand, yet current provisions remain limited, especially when damping and displacement ductility are considered. This study investigated the inelastic displacement ratio and inelastic velocity ratio for single degree of freedom (SDOF) systems subjected to near-fault ground motions, with particular focus on the effects of fling-step and forward-directivity motions. For numerical modeling and analysis, an extensive nonlinear response history analysis (NLRHA) was conducted on SDOF systems incorporating parametric variations in dynamic characteristics of structural systems such as elastic period, displacement ductility, and viscous damping under different ground motion conditions. From numerical modeling, empirical equations are proposed to express the inelastic displacement ratio ((Formula presented) ) and inelastic velocity ratio ((Formula presented) ) using elastic period, viscous damping ratio, displacement ductility, and the type of ground motion. In parallel, neural networks are trained on a dataset of 36,456 samples using additional variables, including the predominant period of the ground motion, moment magnitude, and closest rupture distance. Neural network models achieved (Formula presented) (for (Formula presented) ) and (Formula presented) (for (Formula presented) ) for unseen data, indicating the highest accuracy. Model explanations indicated that the predictions adhere to the domain knowledge. Comparative assessments reveal that while empirical equations capture general trends for design purposes, neural network models accurately predict even minor variations in inelastic responses. These data-driven methods provide a complementary approach in predicting the inelastic response compared to empirical equations.
  • Thumbnail Image
    PublicationOpen Access
    Uncertainty Reduction in Near Real-time Satellite Precipitation Estimates by Integrating Soil Moisture and Potential Evapotranspiration Using a Machine Learning Approach
    (Springer Science and Business Media, 2026) Wanniarachchi, S; Sarukkalige, R; Hapuarachchi, H. A. P; Gomes, P.I.A; Rathnayake, U
    Near-real-time (NRT) satellite precipitation data inherits complex and random errors due to various reasons. The primary objective of this research is to utilize satellite-based precipitation data for hydrological modelling in ungauged areas. The novelty of this study lies in the development of a hybrid stacking-based machine learning framework that integrates hydrologically meaningful predictors: root-zone soil moisture, potential evapotranspiration (PET), and their time-lagged representations to reduce uncertainty in near-real-time satellite precipitation (GSMaP-NRT). Unlike conventional bias-correction approaches that rely primarily on statistical adjustment between satellite and gauge rainfall, this study incorporates physically relevant catchment-state variables to improve predictive skill, with a focus on the Ovens River basin in Australia. A calibrated GR4H hydrological model was used to simulate the runoff of the catchment. Six objective functions were used to evaluate the performance of the approach. The results demonstrate that stacking machine learning algorithms reduces the Mean Absolute Error of GSMaP-NRT satellite precipitation data by 36% and the corresponding modelled streamflow error by 44% for lower precipitation events (< 2 mm/hour). All six objective functions achieved optimal performances within the low precipitation events. However, RMSE remained high for intermediate and heavy precipitation events. The model-estimated major streamflow peaks for the years 2010 and 2016, based on gauged precipitation and ML-corrected satellite precipitation, are 41% and 48% lower than the observed streamflow peaks, respectively. The reasons were the inability of the GR4H model to capture the perfect initial conditions and the x4 time parameter during the calibration process.
  • Thumbnail Image
    PublicationOpen Access
    Forecasting renewable energy for microgrids using machine learning
    (Springer Nature, 2025-05-03) Sudasinghe, P; Herath, D; Karunarathne, I; Weeratunge, H; Jayasuriya, L
    Microgrids, comprised of interconnected loads and distributed energy resources, function as single controllable entities with respect to the main grid. However, the inherent variability of distributed wind and solar generation within microgrids presents operational stability challenges concerning voltage regulation and frequency stability. Accurate forecasting of renewable generation is crucial for mitigating these challenges. This work proposes a one-dimensional Convolutional Neural Network (1-D CNN) based approach to forecast photovoltaic (PV) generation and wind energy, using data from the University of California, San Diego microgrid and San Diego Airport weather records. The proposed method is evaluated against various forecasting methods using key metrics: Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-squared value. Results show that the 1-D CNN model achieves an improvement of up to 229.8 times in MSE and a 24.47 fold improvement in MAE compared to baseline models that use traditional statistical methods in forecasting. This demonstrates the potential of machine learning for enhancing microgrid management, particularly in short-term forecasting of renewable generation. © The Author(s) 2025. Evaluated ML-based renewable energy forecasting models by implementing 1-D CNN and LSTM models using real-world data. Proposed 1-D CNN performs better than LSTM and baseline models, achieving higher accuracy and computational efficiency. Accurate forecasting of PV and wind energy generation enhances grid stability, reduces backup power dependency, and supports sustainable energy integration.
  • Thumbnail Image
    PublicationOpen Access
    Predicting adhesion strength of micropatterned surfaces using gradient boosting models and explainable artificial intelligence visualizations
    (Elsevier, 2023-06-27) Ekanayake, I.U; Palitha, S; Gamage, S; Meddage, D.P.P.; Wijesooriya, K; Mohotti, D
    Fibrillar dry adhesives are widely used due to their effectiveness in air and vacuum conditions. However, their performance depends on various factors. Previous studies have proposed analytical methods to predict adhesion strength on micro-patterned surfaces. However, the method lacks interpretation on which parameters are critical. This research utilizes gradient-boosting machine learning (ML) algorithms to accurately predict adhesion strength. Additionally, explainable machine learning (XML) methods are employed to interpret the underlying reasoning behind the predictions. The analysis demonstrates that gradient boosting models achieve a high correlation coefficient (R > 0.95) in accurately predicting pull-off force on micro-patterned surfaces. The use of XML methods provides insights into the importance of features, their interactions, and their contributions to specific predictions. This novel, explainable, and data-driven approach holds potential for real-time applications, aiding in the identification of critical features that govern the performance of fibrillar adhesives. Furthermore, it improves end-users’ confidence by offering human-comprehensible explanations and facilitates understanding among non-technical audiences
  • Thumbnail Image
    PublicationEmbargo
    Machine learning based classification of ripening and decay stages of Mango (Mangifera indica L.) cv. Tom EJC
    (IEEE, 2022-06-21) Hippola, H. M. W. M; WaduMesthri, D. P; Rajakaruna, R. M. T. P; Yasakethu, L; Rajapaksha, M
    om EJC is a variety of Mango grown in tropical countries like Sri Lanka and India which has a very large demand and hence expensive. From the early stage of ripening, until the senescence stage, the process takes around 10–14 days. The fruit shows a yellowish color starting from the very early stage of ripening, throughout the period until it reaches the senescence stage. Unlike the other Mango varieties, it is difficult for a regular customer to determine the stage of ripening of the Tom EJC fruit, by observation. This paper focuses towards developing a vision-based classifier to rapidly identify ripening and decay stages of Tom EJC mango using surface image captures. A dataset of Tom EJC mango images was collated at different maturity levels. A Convolutional Neural Network (CNN) is proposed and tested using over 6000 Tom EJC images. The proposed model is shown to have a 76% testing accuracy in identifying four stages of maturity.
  • Thumbnail Image
    PublicationOpen Access
    Explainable Machine Learning (XML) to predict external wind pressure of a low-rise building in urban-like settings
    (2022-07) Meddage, D. P. P; Ekanayake, I; Weerasuriya, A; Lewangamage, C. S; Ramanayaka, C. D. E; Miyanawala, T
    This study used explainable machine learning (XML), a new branch of Machine Learning (ML), to elucidate how ML models make predictions. Three tree-based regression models, Decision Tree (DT), Random Forest (RF), and Extreme Gradient Boost (XGB), were used to predict the normalized mean (Cp,mean), fluctuating (Cp,rms), minimum (Cp,min), and maximum (Cp,max) external wind pressure coefficients of a low-rise building with fixed dimensions in urban-like settings for several wind incidence angles. Two types of XML were used — first, an intrinsic explainable method, which relies on the DT structure to explain the inner workings of the model, and second, SHAP (SHapley Additive exPlanations), a post-hoc explanation technique used particularly for the structurally complex XGB. The intrinsic explainable method proved incapable of explaining the deep tree structure of the DT, but SHAP provided valuable insights by revealing various degrees of positive and negative contributions of certain geometric parameters, the wind incidence angle, and the density of buildings that surround a low-rise building. SHAP also illustrated the relationships between the above factors and wind pressure, and its explanations were in line with what is generally accepted in wind engineering, thus confirming the causality of the ML model’s predictions.