Department Publications

Results: ANFIS training takes about 30 minutes/epoch on a Pentium IV, 1.5 GHz, 256MB RAM system. And, it could be sufficiently trained in very few epochs (less than 10) provided good data-driven initialization. The trained neuro-fuzzy controller was tested by 10hr simulation under random disturbances and random reference changes at all power levels. The first 1hr simulation results are shown in Fig. 3 for each power

Precise temporal localization of named entities in speech is crucial for privacy-preserving audio processing. However, prevailing cascaded pipelines propagate transcription errors and end‐to‐end models lack the temporal granularity required for reliable frame‐level detection. To address these limitations, we introduce DEnSNEL (Dense End‐to‐end Spoken Named Entity Localizer), the first end-to-end model to perform direct, frame-level spoken named entity localization, without intermediate character or text representations. We reformulate the task as a dense, frame-wise binary classification ("entity" vs. "non-entity"), employing a lightweight encoder-classifier architecture. To improve boundary delineation, DEnSNEL incorporates a learnable complex filter bank to capture phonetic information, and employs a boundary-focused loss that explicitly optimizes span precision. On the SLUE Phase 2 benchmark …

Externally bonded Carbon Fiber Reinforced Polymer (CFRP) systems have emerged as a reliable solution for upgrading the mechanical performance and durability of reinforced concrete structures. However, the interfacial bond strength between CFRP and concrete is a critical factor governing the overall performance of these strengthening systems. Inadequate bonding can result in premature debonding, thereby diminishing the load-bearing capacity and compromising structural integrity. This study examines the impact of key geometric and material parameters including bond length, bond width, CFRP thickness, concrete compressive strength and CFRP stiffness on the interfacial bond strength of CFRP–concrete systems. A detailed literature review was undertaken to identify the most influential bond parameters, and a comprehensive dataset of 855 experimental results was compiled from previous studies. Four supervised machine learning algorithms Support Vector Machine (SVM), Gamma Likelihood (GL), Random Forest (RF) and Extreme Gradient Boosting (XGB) were employed to develop predictive models. Model performance was assessed using the Coefficient of Determination (R²), Mean Absolute Error (MAE) and Mean Squared Error (MSE). The results indicate that the ML models, particularly XGB and RF, achieved superior prediction accuracy, demonstrating their potential as effective tools for practical structural engineering applications. Overall, this research underscores the importance of data-driven methods in enhancing the evaluation and design of CFRP-strengthened concrete structures.

The use of Carbon Fiber Reinforced Polymer (CFRP) for the strengthening and restoration of concrete structures has gained a significant rise in recent years. The effectiveness of CFRP as a strengthening material is highly dependent on the quality of the bond between the CFRP and the concrete substrate. Extensive research has been conducted on evaluating bond performance for a variety of geometries and materials. However, the identification of variations in bond performance with respect to bond sensitive parameters, such as geometric dimensions of the bond and material properties of CFRP, epoxy and concrete, will provide clear guidance to strengthen future applications. A comprehensive data set was collected from previous studies on bond shear performance. The data was analyzed for critical bond parameters. The results emphasized the importance of careful design considerations when implementing CFRP systems in concrete structures. Considering future development, integrating Machine Learning (ML) approaches to enhance the understanding of these complex interactions is highlighted.