Defence Science Journal
Not a member yet
3794 research outputs found
Sort by
Deep Learning for Unearthing Emotions in Twitter A Hybrid Emotional Recognition Model
Full text linkWith the intensification of new classes of media such as Twitter, the Internet has become a primary route for individual and interpersonal messaging. Many individuals share their thoughts regarding news-related topics on Twitter, an established SNS network built on people’s relationships. It offers us with a Source of data from which we can dig people’s thoughts, which is useful for product reviews and community monitoring. A Hybrid Emotional Recognition Model (HERM) is proposed in this research. Hashtags are recognized as the tag for emotional cataloging based on gathered posts from Twitter. Meanwhile, emoji and the N-grams are dug and used to classify the gathered topic comments into four distinct sentiment groups using the distorted emotional models. Machine learning approaches are applied of categorizing the emotional information set, yielding an 92 % accuracy result. Furthermore, entities underlying emotions might be obtained using the deep learning model SENNA
Investigation of Transonic Flow Features in an Axial Flow Compressor Rotor
Full text linkNumerical investigation of an axial flow rotor is carried out for its performance characterization and aerodynamic behavior during the design and off-design operating conditions. The study focuses on capturing the transonic flow features from the choke point (CP) to the near stall (NS) point in the rotor. This includes the analysis of passage shock structure, its movement in the blade passage with varying back-pressure, shock boundary layer interaction, tip leakage flow structure, and resulting losses. The study is carried out from 60 % to 100 % of the design speed using steady and unsteady RANS simulations. Three turbulence models, namely; SST, k-ε, and Reynolds stress models, are employed. The SST model predicted the closest approximation to the experimental data. The rotor aerodynamic performance is predicted in terms of total pressure ratio, efficiency, and flow contours. Unsteady analysis revealed that the primary and secondary tip leakage vortices, combined with the suction side tip corner separation, are the major instabilities near the stall region
Cyclic Loading on Composite Repair of Corroded Steel Pipelines
Full text linkThe study aims to determine how cyclic loading affects the structural integrity and lifespan of composite repair systems used to restore corroded steel pipelines. As specified in Annexure C of the ISO 24817 repair code, pipe specimens are machined to produce flaws with 80% wall loss. Testing under static and cyclic pressure loading is done as per ASTM D2992 and ASTM D2143. Static pressure loading is accomplished by continually pressurizing the pipe specimen, and burst pressure is assessed. Various Rc-ratios or levels of cyclic loading severity are used in cyclic pressure loading tests. Each case's number of cycles before failure is determined experimentally, and the service de-rating factor is assessed in accordance with ISO 24817. The 235 bar pressure was sustained by the static-loaded repaired pipe specimens with 80% wall loss, and the failure was catastrophic. At around 7000 cycles, the cyclically loaded repaired samples with 80% wall loss failed, and the failure manifests as debonding or a leak
Studying the Interaction of Waves to Determine the Impact Response of a Layered Elastic Medium
Full text linkWhen an impactor strikes a layered target, both the impactor and the target experience waves. The waves produced travel and engage in interactions with other waves as well as the interfaces in the impactor-target system. For the impact problems on a layered medium with periodic properties and layered elastic media of Goupillaud-type (each layer has the same wave travel time), researchers have presented an analytical solution for stress variation with position and time within the target. However, the solution for an elastic media not satisfying the above conditions is not available in the literature. The present study fills this gap and finds the behaviour of a generalized layered medium to an impact problem. The response of the material at any position inside the layered medium is found by solving the interaction between waves, interfaces, and boundaries. The mass, momentum balance and constitutive relationship are solved to get the exact analytical expressions for particle velocity and stress for each possible wave interaction happening in the impactor and the layered medium. The expressions are utilized in a computer program to study the impact behaviour of a layered media. The code tracks each wave as it travels through the system and identifies those interactions that occur in the shortest time, uses the stress and velocity expression for that interaction, and updates the state of the material. When stress produced at the impact surface is tensile in nature, the impactor and target can be separated. The work can be applied to both finite and semi-infinite impactors and targets, and the layered medium does not necessarily have to be a periodic layered media or a Goupillaud-type medium
FuzzyBandit An Autonomous Personalized Model Based on Contextual Multi Arm Bandits Using Explainable AI
Full text linkIn the era of artificial cognizance, context-aware decision-making problems have attracted significant attention. Contextual bandit addresses these problems by solving the exploration versus exploitation dilemma faced to provide customized solutions as per the user’s liking. However, a high level of accountability is required, and there is a need to understand the underlying mechanism of the black box nature of the contextual bandit algorithms proposed in the literature. To overcome these shortcomings, an explainable AI (XAI) based FuzzyBandit model is proposed, which maximizes the cumulative reward by optimizing the decision at each trial based on the rewards received in previous observations and, at the same time, generates explanations for the decision made. The proposed model uses an adaptive neuro-fuzzy inference system (ANFIS) to address the vague nature of arm selection in contextual bandits and uses a feedback mechanism to adjust its parameters based on the relevance and diversity of the features to maximize reward generation. The FuzzyBandit model has also been empirically compared with the existing seven most popular art of literature models on four benchmark datasets over nine criteria, namely recall, specificity, precision, prevalence, F1 score, Matthews Correlation Coefficient (MCC), Fowlkes–Mallows index (FM), Critical Success Index (CSI) and accuracy
A Novel Method to Estimate Base Drag and Burn Time from Flight Data Using Extended Kalman Filter
Full text linkEstimating highly spinning projectiles’ base drag and burn time by static trials is challenging. Drag profiles obtained using numerical methods or experimental or wind tunnel estimations should be updated using the estimates obtained from dynamic flight test data. This paper proposes a novel methodology to estimate the base drag and burn time from flight data using an extended Kalman Filter. Trajectory positional data is used to calculate base drag and indirect measurement of burn time. The simulation is carried out for two cases, artillery shell and rocket. The proposed method works well for both cases. Exhaustive simulation results indicate that the technique can be used for any configuration. Estimating both base drag and burn time is within 5 % accuracy
MILP Based Differential Cryptanalysis on IVLBC and Eslice 64
Full text linkLightweight block ciphers provide security to resource-limited devices. However, many of these ciphers lack security analysis against basic attacks. This paper provides a detailed security analysis of two lightweight block ciphers, IVLBC and Eslice-64, against differential attack. The designers of IVLBC and Eslice-64 claimed that their ciphers were secure against differential attack. In this paper, to substantiate existing cryptanalysis’s claims, we perform differential attack on these two ciphers using the mixed-integer linear programming (MILP) method. We incorporate the difference distribution table (DDT) probabilities into MILP models. We discover differential distinguishers up to seven and 15 rounds for IVLBC and Eslice-64, respectively. We improve the known distinguishers for Eslice-64 by one round. Further, we mount the key recovery attack on an eight-round IVLBC and a 16-round Eslice-64 with data/memory/time complexities of 249/ 250.59/249 and 263/212.58/263 respectively
Modelling of Human Factors in Aviation Maintenance Using HFACS ME Human Factors Analysis and Classification System Maintenance Extension and Bayesian Network
Full text linkAircraft maintenance is a complex task involving a skilled human workforce, spare parts, and various other resources. Human factors are an inherent element of the human workforce. Human factors analysis, therefore, becomes an essential aspect of aviation maintenance. Human factors have been identified and classified using various methods in existing literature. However, there is a gap in the study of the interdependency of critical human factors including subfactors, and measuring them effectively to reduce incidents and accidents. This research work proposed a novel approach for human factors modeling using human factors analysis and classification system maintenance extension (HFACS-ME), and bayesian network (BN). Inadequate maintenance processes, inadequate documentation, inadequate supervision, Judgement decision, and attention memory were identified as some of the critical human factors in aircraft maintenance. These critical human factors were further analysed and divided into subfactors. The main contribution of the present research work is the methodology of developing a dependency model of the human factors and subfactors to analyse their measured effects on aircraft maintenance. The proposed BN model demonstrated the estimation of the probability of effective maintenance by considering the critical human factors with available facilities, and resources in an aviation maintenance setup
Flow Analyses of Integrated Liquid Fuel RAMJET Propulsion System
Full text linkA CFD study is performed to check the seamless flow behaviour of the Liquid Fuel Ramjet Propulsion system which includes, air intakes, combustor and nozzle. Resolving both supersonic and subsonic flow scales in the same domain makes the simulations complex. Addition of combustion with stiff chemistry makes the simulations more difficult. CFD simulations are carried out using commercially available CFD software. Liquid fuel is injected as discrete phase and the flow turbulence is modelled using Realizable k-ε turbulence model. Jet-A + air combustion has been simulated using combined finite rate / eddy dissipation model. Finite rate chemistry was modelled using three step chemistry which was obtained from the published literature. Flow structures such as oblique shocks, normal shocks and combustion are observed