Abstract: The use of deep learning for species identification in camera trap images has revolutionised our ability to study, conserve and monitor species in a highly efficient and unobtrusive manner, with state-of-the-art models achieving accuracies surpassing the accuracy of manual human classification. The high imbalance of camera trap datasets, however, results in poor accuracies for minority (rare or endangered) species due to their relative insignificance to the overall model accuracy. This paper investigates the use of Focal Loss, in comparison to the traditional Cross Entropy Loss function, to improve the identification of minority species in the “255 Bird Species” dataset from Kaggle. The results show that, although Focal Loss slightly decreased the accuracy of the majority species, it was able to increase the F1-score by 0.06 and improve the identification of the bottom two, five and ten (minority) species by 37.5%, 15.7% and 10.8%, respectively, as well as resulting in an improved overall accuracy of 2.96%.
Abstract: The Failure Modes and Effect Analysis (FMEA) is an efficient evaluation technique to identify potential failures in products, processes, and services. FMEA is designed to identify and prioritize failure modes. It proves to be a useful method for identifying and correcting possible failures at its earliest possible level so that one can avoid consequences of poor performance. In this paper, FMEA tool is used in detection of failures of various components of heat exchanger cycle and to identify critical failures of the components which may hamper the system’s performance. Further, a detailed Pareto analysis is done to find out the most critical components of the cycle, the causes of its failures, and possible recommended actions. This paper can be used as a checklist which will help in maintainability of the system.
Abstract: Non-parametric reliability technique is useful for assessment of reliability of systems for which failure rates are not available. This is useful when detection of malfunctioning of any component is the key purpose during ongoing operation of the system. The main purpose of the Heat Exchanger Cycle discussed in this paper is to provide hot water at a constant temperature for longer periods of time. In such a cycle, certain components play a crucial role and this paper presents an effective way to predict the malfunctioning of the components by determination of system reliability. The method discussed in the paper is feasible and this is clarified with the help of various test cases.
Abstract: Insulation performance of a gas insulated system is severely affected by particle contaminants. These metallic particles adversely affect the characteristics of insulating system. These particles can produce surface charges due to partial discharge activities. These particles which are free to move enhance the local electric fields. This paper deals with the influence of conducting particle placed in a co-axial duct on the discharge characteristics of gas mixtures. Co-axial duct placed in a high pressure chamber is used for the purpose. A gas pressure of 0.1, 0.2 and 0.3 MPa have been considered with a 10:90 SF6 and N2 gas mixtures. The 2D and 3D histograms of clean duct and duct with copper particle are discussed in this paper.
Abstract: Partial Discharge measurement is a very important
means of assessing the integrity of insulation systems in a High
Voltage apparatus. In compressed gas insulation systems, floating
particles can initiate partial discharge activities which adversely
affect the working of insulation. Partial Discharges below the
inception voltage also plays a crucial in damaging the integrity of
insulation over a period of time. This paper discusses the effect of
loose and fixed Copper and Nichrome wire particles on the PD
characteristics in SF6-N2 (10:90) gas mixtures at a pressure of
0.4MPa. The Partial Discharge statistical parameters and their
correlation to the observed results are discussed.
Abstract: The fundamental aim of extended expansion concept is
to achieve higher work done which in turn leads to higher thermal
efficiency. This concept is compatible with the application of
turbocharger and LHR engine. The Low Heat Rejection engine was
developed by coating the piston crown, cylinder head inside with
valves and cylinder liner with partially stabilized zirconia coating of
0.5 mm thickness. Extended expansion in diesel engines is termed as
Miller cycle in which the expansion ratio is increased by reducing the
compression ratio by modifying the inlet cam for late inlet valve
closing. The specific fuel consumption reduces to an appreciable level
and the thermal efficiency of the extended expansion turbocharged
LHR engine is improved.
In this work, a thermodynamic model was formulated and
developed to simulate the LHR based extended expansion
turbocharged direct injection diesel engine. It includes a gas flow
model, a heat transfer model, and a two zone combustion model. Gas
exchange model is modified by incorporating the Miller cycle, by
delaying inlet valve closing timing which had resulted in considerable
improvement in thermal efficiency of turbocharged LHR engines. The
heat transfer model, calculates the convective and radiative heat
transfer between the gas and wall by taking into account of the
combustion chamber surface temperature swings. Using the two-zone
combustion model, the combustion parameters and the chemical
equilibrium compositions were determined. The chemical equilibrium
compositions were used to calculate the Nitric oxide formation rate by
assuming a modified Zeldovich mechanism. The accuracy of this
model is scrutinized against actual test results from the engine. The
factors which affect thermal efficiency and exhaust emissions were
deduced and their influences were discussed. In the final analysis it is
seen that there is an excellent agreement in all of these evaluations.
Abstract: Laboratory classes in Electrical Engineering are often hampered by safety issues, as students have to work on high voltage lines. One solution is to make use of virtual laboratory simulations, to help students understand the concepts taught in their coursework. In this context, we have conceived and implemented virtual lab experiments in connection with the study of earthing arrangements. In this work, software was developed, which aid student in understanding the working of a residual current device (RCD) in a TT earthing system. Various parameters, such as the earthing resistances, leakage currents and harmonics were included for a TT system with RCD connection.
Abstract: Automated operations based on voice commands will become more and more important in many applications, including robotics, maintenance operations, etc. However, voice command recognition rates drop quite a lot under non-stationary and chaotic noise environments. In this paper, we tried to significantly improve the speech recognition rates under non-stationary noise environments. First, 298 Navy acronyms have been selected for automatic speech recognition. Data sets were collected under 4 types of noisy environments: factory, buccaneer jet, babble noise in a canteen, and destroyer. Within each noisy environment, 4 levels (5 dB, 15 dB, 25 dB, and clean) of Signal-to-Noise Ratio (SNR) were introduced to corrupt the speech. Second, a new algorithm to estimate speech or no speech regions has been developed, implemented, and evaluated. Third, extensive simulations were carried out. It was found that the combination of the new algorithm, the proper selection of language model and a customized training of the speech recognizer based on clean speech yielded very high recognition rates, which are between 80% and 90% for the four different noisy conditions. Fourth, extensive comparative studies have also been carried out.
Abstract: The current speech interfaces in many military
applications may be adequate for native speakers. However,
the recognition rate drops quite a lot for non-native speakers
(people with foreign accents). This is mainly because the nonnative
speakers have large temporal and intra-phoneme
variations when they pronounce the same words. This
problem is also complicated by the presence of large
environmental noise such as tank noise, helicopter noise, etc.
In this paper, we proposed a novel continuous acoustic feature
adaptation algorithm for on-line accent and environmental
adaptation. Implemented by incremental singular value
decomposition (SVD), the algorithm captures local acoustic
variation and runs in real-time. This feature-based adaptation
method is then integrated with conventional model-based
maximum likelihood linear regression (MLLR) algorithm.
Extensive experiments have been performed on the NATO
non-native speech corpus with baseline acoustic model trained
on native American English. The proposed feature-based
adaptation algorithm improved the average recognition
accuracy by 15%, while the MLLR model based adaptation
achieved 11% improvement. The corresponding word error
rate (WER) reduction was 25.8% and 2.73%, as compared to
that without adaptation. The combined adaptation achieved
overall recognition accuracy improvement of 29.5%, and
WER reduction of 31.8%, as compared to that without
adaptation.