Browsing by Author "Kumar, M.A."

Now showing 1 - 3 of 3

Pseudo-Dynamic Analysis of Gravity Masonry Dams
(Springer Science and Business Media Deutschland GmbH, 2024) Shalini, S.; Kumar, M.A.; Pavan, G.S.
According to USGS estimates, approximately 5 million earthquakes occur annually, of which 1 million are felt. In the north-eastern and north-western regions of India, where the Indo-Australian plate is subducted beneath the Eurasian plate, seismic activity is extremely high. In addition to the immediate damage, an earthquake can cause minor vulnerabilities that lead to future crises. Safety of important infrastructure like dams, bridges, tunnels, elevated structures, and nuclear power plants under earthquake ground motion is critical. In the past 50Â years, seismic analysis of dams has attracted considerable research interest. In this study, a pseudo-dynamic analysis of non-overflowing section of a masonry gravity dam is conducted. Invoking plane-strain condition, a 2D model of the dam is developed in Abaqus software. The dam is modeled using four node rectangular elements. The loads at various levels along the dam's height are computed for the fundamental, higher, and static modes. The effects of hydrodynamic forces acting on the dam are also incorporated. The loads are applied separately, and stress analysis is performed. Stress values are combined using the SRSS method, these stresses are compared to the material's strength properties, and the risk factor is evaluated. A comparison of the stresses obtained from FEM model and stresses obtained by considering beam idealization is also presented in this work. Â© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.
Speech enhancement using multiple deep neural networks
(2018) Karjol, P.; Kumar, M.A.; Ghosh, P.K.
In this work, we present a variant of multiple deep neural network (DNN) based speech enhancement method. We directly estimate clean speech spectrum as a weighted average of outputs from multiple DNNs. The weights are provided by a gating network. The multiple DNNs and the gating network are trained jointly. The objective function is set as the mean square logarithmic error between the target clean spectrum and the estimated spectrum. We conduct experiments using two and four DNNs using the TIMIT corpus with nine noise types (four seen noises and five unseen noises) taken from the AURORA database at four different signal-to-noise ratios (SNRs). We also compare the proposed method with a single DNN based speech enhancement scheme and existing multiple DNN schemes using segmental SNR, perceptual evaluation of speech quality (PESQ) and short-term objective intelligibility (STOI) as the evaluation metrics. These comparisons show the superiority of proposed method over baseline schemes in both seen and unseen noises. Specifically, we observe an absolute improvement of 0.07 and 0.04 in PESQ measure compared to single DNN when averaged over all noises and SNRs for seen and unseen noise cases respectively. � 2018 IEEE.
Speech enhancement using multiple deep neural networks
(Institute of Electrical and Electronics Engineers Inc., 2018) Karjol, P.; Kumar, M.A.; Ghosh, P.K.
In this work, we present a variant of multiple deep neural network (DNN) based speech enhancement method. We directly estimate clean speech spectrum as a weighted average of outputs from multiple DNNs. The weights are provided by a gating network. The multiple DNNs and the gating network are trained jointly. The objective function is set as the mean square logarithmic error between the target clean spectrum and the estimated spectrum. We conduct experiments using two and four DNNs using the TIMIT corpus with nine noise types (four seen noises and five unseen noises) taken from the AURORA database at four different signal-to-noise ratios (SNRs). We also compare the proposed method with a single DNN based speech enhancement scheme and existing multiple DNN schemes using segmental SNR, perceptual evaluation of speech quality (PESQ) and short-term objective intelligibility (STOI) as the evaluation metrics. These comparisons show the superiority of proposed method over baseline schemes in both seen and unseen noises. Specifically, we observe an absolute improvement of 0.07 and 0.04 in PESQ measure compared to single DNN when averaged over all noises and SNRs for seen and unseen noise cases respectively. Â© 2018 IEEE.