Search

To detect abnormal data via semi-supervised learning, unlabeled data are generally assumed to be normal data. This assumption, however, causes inevitable performance degradation when a small fraction of abnormal data is included in the unlabeled dataset. To overcome the degradation and to maintain stable detection performance, we propose a semi-supervised anomaly detection algorithm using probabilistic labeling (SAD-PL) for unlabeled data. The proposed SAD-PL is composed of two steps: (1) estimating local outlier factor (LOF) scores of latent vectors from both labeled and unlabeled data and (2) estimating labeling probability on the unlabeled data by using the prior missing probability of the labeled data via the Neyman-Pearson (NP) criterion. The SAD-PL runs iteratively by using the proposed complementary learning functions until the rate of label changes is lower than the predefined threshold. Experimental results reveal that the SAD-PL shows superior detection probability over the existing algorithms and stable performance regardless of the normal to abnormal data ratio in unlabeled data and the ratio of change variation of unlabeled data statistics to labeled data statistics.

In this paper, an operational Dst index prediction model is developed by combining empirical and Artificial Neural Network (ANN) models. ANN algorithms are widely used to predict space weather conditions. While they require a large amount of data for machine learning, large-scale geomagnetic storms have not occurred sufficiently for the last 20 years, Advanced Composition Explorer (ACE) and Deep Space Climate Observatory (DSCOVR) mission operation period. Conversely, the empirical models are based on numerical equations derived from human intuition and are therefore applicable to extrapolate for large storms. In this study, we distinguish between Coronal Mass Ejection (CME) driven and Corotating Interaction Region (CIR) driven storms, estimate the minimum Dst values, and derive an equation for describing the recovery phase. The combined Korea Astronomy and Space Science Institute (KASI) Dst Prediction (KDP) model achieved better performance contrasted to ANN model only. This model could be used practically for space weather operation by extending prediction time to 24 h and updating the model output every hour.

The fiber optic Sagnac interferometer is well established as a sensor for detection of physical perturbations such as acoustic and vibration. In this paper acoustic signals generated in the cylindrical cavity submerged in transformer oil were measured by the fiber optic sensor array in one Sagnac loop. Two different external sound frequencies, and , were applied to the sensor array simultaneously by using piezoelectric with frequency range from 5 kHz to 90 kHz. Based on the experimental results, fiber optic sensor detected harmonic series of applied sound frequency such as , , , , , . Suggested fiber optic sensor array can be applied to monitor physical quantities such as internal sound pressure and vibration due to partial discharge in the real electric transformer system.

Surface salinity in the western sea of Jeju Island in Korea becomes low due to the inflow of the Chinese coastal waters during summer. One of the characteristics of low salinity water is the formation of a surface sound channel (SSC) due to the decrease in sound speed by salinity. However, a quantitative analysis between low salinity water and SSC has not been fully investigated yet. In this paper, a temperature-salinity (T-S) gradient diagram is introduced in order to assess SSC formation and its acoustic characteristics are also investigated through a case study of low salinity waters. Maximum angles of limiting rays were less than 4.6° and low frequency cutoffs were higher than 2.0 kHz for the SSCs formed in low salinity water. When the salinity gradients were large (0.5 psu/m), a SSC was formed more efficiently than other cases whose salinity gradients were small. On the other hand, a SSC was not formed in spite of highly positive salinity gradients when the amount of temperature gradients was negatively high enough (-0.5 °C/m). However, the acoustic energy transfer in the surface ducts was dependent on frequency and position of source.

In this paper, sound pressure sensitivity of the fiber optic acoustic sensor according to sensor direction and mandrel material were investigated experimentally. Three different directions were selected as stand, lay, and hole. Hollow cylinder type mandrel dimension is 30 mm in outer diameter, 45 mm in length, and 2 mm in thickness, and about 50 m optical fibers were wounded on the surface of the mandrel. Non-directional sound speaker was used as a sound source. Sagnac interferometer and single mode fiber, a laser with 1,550 nm in wavelength, coupler were used. Based on the experimental results, lay direction`s sensitivity is the highest in the frequency range of 2 kHz~4 kHz. `PTFE+carbon` material is more sensitive than PTFE in the frequency range of 5 kHz~20 kHz. Sound pressure detection sensitivity depends on the mandrel direction and material under certain frequency.

To substitute TLS in the hybrid system which is combined with Sagnac interferometer and fiber bragg grating (FBG) it is necessary to investigate how the laser source (TLS and CW) and sensor material variate the response of fiber optic sensor. Two different hollow cylinder type mandrel materials are proposed which are PTFE and PTFE+carbon and 18 m optical fiber is wounded at the mandrel surface. CW laser source experiments had been done in the oil tank which is filled with transformer oil in the 1 kHz~20 kHz frequency range. Also Sagnac interferometer fiber optic sensor is combined with FBG called hybrid system and TLS used as a light source. Based on the experimental results PTFE sensor showed more higher magnitude of detection signal rather than carbon sensor and this result is agreement with the McMahon`s theoretical results. Phase variation is inversely proportional to the elastic modulus of the mandrel material. In PTFE fiber sensor, tunable laser source showed more higher performance rather than CW case. Therefore, TLS fiber optic sensor can be applied to the hybrid system which is combined with Sagnac and FBG.

In the transducer arrays which are used in sonar to radiate powerful, directional sound beams, acoustic interactions between the transducers have many important effects. To predict the acoustic radiation power cancellation, 37 acoustic transducers are considered as theoretical model. Total radiation impedances and acoustic radiation power of the transducers are extracted using the equivalent electric circuit scheme. Based on the theoretical results the acoustic radiation power of the transducer has been found dependent of the mutual coupling effects.

In this paper physical quantities were measured using by the fiber optic sensor when a pendulum ball collides to the fixed ball on the wall. Both steel ball dimensions are 1 inch in diameter. The fiber optic Sagnac interferometer was used to detect the impact time. It is made a 1mm penetrated hole and optical fiber in the Sagnac loop passed through the hole. A ball was welded on the steel plate wall as a fixed ball. When the another ball collides to the fixed ball the fiber optic sensor detects the impact force in between them. Based on the experimental result impact time was measured about 0.14ms at the angle of 30 degree. This fiber optic sensor technique can be expanded to the moving bodies.

The normalized self- and mutual-radiation impedances for a square piston in a rigid infinite baffle are formulated as quadruple surface integrals by using geometric relations and the impedances are evaluated numerically. Self-radiation resistance and reactance coefficients are compared to the tabulated values of Burnett and Soroka. Mutual radiation resistance and reactance coefficients exhibit an expected oscillatory decay as functions ofkaand the frequency of oscillation is dependent on the spacing parameterd/a. The amount of interaction impedance is dependent on the orientation of the pistons with respect to each other. When separation is along the diagonal direction at 45 °C, mutual-radiation impedance is minimized. The quadruple integrals for the radiation impedances are not restricted in their range of applicability.

A theoretical and experimental approach was used to investigate the motion and effectiveness of a Self-Compensating Dynamic Balancer (SCDB). This is a device intended to minimize the effects of rotor imbalance and vibratory forces on a rotating system during normal operation. The basic concept of an automatic dynamic balancer has been described in many U.S. patents. The SCDB is composed of a circular disk with a groove containing massive balls and a low viscosity damping fluid. The objective of this research is to determine the motion of the balls and how this ball motion is related to the vibration of the rotating system using both theoretical and experimental methods. The equations of motion the balls were derived by the Lagrangian method. Static and dynamic solutions were derived from the analytic model. To consider dynamic stability of the motion, perturbation equations were investigated by two different methods: Floquet theory and direct computer simulation. On the basis of the results of the stability investigation, ball positions which result in a balance system are stable above the critical speed and unstable at critical speed and below critical speed. To determine the actual critical speed of the rotating system used in the experimental work, a modal analysis was conducted. Experimental results confirm the predicted ball positions. Based on the theoretical and experimental results, when the system operates below and near the first critical speed, the balls do not balance the system. However, when the system operates above the first critical speed the balls can balance the system.

The objective of this research is to determine how a cutting tool vibrates when chatter occurs and how this motion is related to the acoustic emission signal. Modal analysis impact tests were conducted to obtain the actual natural frequencies and mode shapes of the Cutting Tool Support Structure (CTSS) system. Cutting tests were also conducted to determine the phase relationship between feed and cutting direction accelerations. Acoustic emission tests were performed to determine the relation between AE signals and two accelerations. Bursts of AE signal were observed to occur approximately at the direction turning points of the Lissajous loop.

Dynamic behavior of the vehicle shimmy system with consideration of steering linkage clearance was analyzed. Based on nonlinear dynamic theories, a 4-DOF dynamic model of vehicle shimmy system with consideration of steering linkage clearance was presented. In which, the revolute movement pair of steering linkage with clearance between the steering tie rod and tie rod arm was described based on Hertz model. Numerical analysis on the dynamic response of the vehicle shimmy system with clearance was carried out, and the results were presented with phase plane, Poincare map, and bifurcation diagram. The dynamic behavior of the vehicle shimmy system with clearance is compared with that without clearance, and the influence of the clearance on the dynamic behavior of the system is discussed, which will provide theoretical basis for attenuation of vehicle shimmy.Copyright © 2012 by ASME

In this paper it is shown that a fundamental technology of hybrid system combining Sagnac interferometer and FBG. Fiber optic sensor array, i.e. two sensors in the Sagnac loop was suggested. Fiber optic sensor array was fabricated in the transformer oil and external sound applied to the sensor directly using by piezoelectric. Tunable laser source was used as a light source. Applied sound frequency range is from 5kHz to 90kHz. From this experiment it is confirmed that fiber optic sensor using Sagnac interferometer in the transformer oil can detect applied sound frequency. In real electric transformer system suggested fiber optic sensor array can be applied to monitoring physical quantities such as internal temperature, sound pressure and vibration due to partial discharge.© (2012) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.

For condition monitoring of large scale electrical power transformers, a fiber-optic multi-stress sensor system was constructed by combining fiber-optic acoustic sensors and fiber Bragg grating temperature sensors in a fiber-optic Sagnac interferometer. To separate the grating signals from the interferometer output, an attenuator was placed at an asymmetrical position in the Sagnac loop. By balancing the counter propagating light intensities with the attenuator, the background noises could be suppressed to obtain grating signals with enough signal-to-noise ratio. With the preliminary experiments, the temperature and the vibration information at multiple locations could be measured simultaneously with single optical circuit and signal processing unit.

Fiber Bragg grating (FBG) sensors are employed in a fiber-optic Sagnac interferometer sensor to measure multi-stress information of electric power systems. By using the hybrid sensor configuration, it was possible to measure the temperature and the vibration signal in an insulating transformer oil bath at the same time. A novel fiber-optic Sagnac interferometer design and a signal processing technique were used to separate the FBG sensor signals from the interference signal. The preliminary experimental results are presented to show the feasibility of the sensor system.

Salinity is usually neglected in underwater acoustics because of its minimal contribution on sound speed variation. However, it is required to consider in calculation of sound speed and acoustic propagation for low salinity water freshened by continental run-off. Furthermore, the acoustic characteristics of low salinity environment are not fully investigated yet. Therefore, regional difference of acoustic propagation in low salinity water was studied by comparing the acoustic characteristics of the East China Sea with those of the tropical Atlantic Ocean in this paper. Frequency dependency of sound propagation in the haline channel was analyzed with transmission losses and low frequency cut-off using oceanic data from NODC (National Oceanographic Data Center). The tropical Atlantic Ocean showed larger channel depth, critical angles and less transmission loss in the haline channel than the East China Sea. The cut-off frequency in the haline channel was computed as around 1 kHz and 5 kHz in the tropical Atlantic Ocean and the East China Sea, respectively. The effects of low salinity water on sound propagation showed regional characteristics, and need to be considered in sonar operation near sea surface. [This work was supported by Defense Acquisition Program Administration and Agency for Defense Development under the contract UD130007DD.]

It is interesting to measure the impact force when the moving pendulum ball collides to the fixed body. In this paper physical quantities were measured using by fiber optic sensor when a pendulum ball collides to the fixed ball on the wall. Both steel ball dimensions are 1 inch in diameter. The fiber optic Sagnac interferometer is well established as a sensor for acoustic and vibration. It is made a 1mm penetrated hole and optical fiber in the Saganac loop passed through the hole. The ball was welded on the wall as a fixed ball. When the external force applied to the fixed ball the optical fiber in the Sagnac loop detects the impact force. The output signal is proportional to the output voltage in the oscilloscope. Based on the result suggested fiber optic sensor can measure the impact intensity and this technique can be expanded to the moving bodies.

In this paper, an optical fiber sensor (OF sensor) utilizing the principal of the Sagnac interferometer was proposed to detect the discharge acoustic signals which generated from a needle‐sphere electrode system in insulating oil. The performance of the OF sensor was checked by a sinusoidal calibration signal generated by a PZT actuator at 198 kHz. The detected discharge signals consisted of the acoustic signal and electrical noise. The noise signal could be removed by a digital low‐pass filter. It was demonstrated that the OF sensor in this research had a possibility to detect the discharge acoustic signals in the power apparatus.