• Journal of Computing Science and Engineering
• /
• 제5권2호
• /
• pp.131-140
• /
• 2011

Different cores typically share the last-level cache in a multi-core processor. Threads running on different cores may interfere with each other. Therefore, the multi-core worst-case execution time (WCET) analyzer must be able to safely and accurately estimate the worst-case inter-thread cache interference. This is not supported by current WCET analysis techniques that manly focus on single thread analysis. This paper presents a novel approach to analyze the worst-case cache interference and bounding the WCET for threads running on multi-core processors with shared L2 instruction caches. We propose to use an interference matrix to model inter-thread interference, on which basis we can calculate the worst-case inter-thread cache interference. Our experiments indicate that the proposed approach can give a worst-case bound less than 1%, as in benchmark fib-call, and an average 16.4% overestimate for threads running on a dual-core processor with shared-L2 cache. Our approach dramatically improves the accuracy of WCET overestimatation by on average 20.0% compared to work.

• 한국전산구조공학회논문집
• /
• 제35권6호
• /
• pp.375-380
• /
• 2022

고속철도 교량은 열차 하중에 의한 공진으로 인한 동적응답 증폭의 위험이 존재하므로 설계기준에 따른 동적해석을 통한 주행안전성 및 승차감 검토를 반드시 수행하여야 한다. 그러나 주행안전성 및 승차감 산정 절차는 열차의 종류별로 임계속도를 포함하여 설계속도의 110km/h까지 10km/h 간격으로 동적해석을 일일이 수행해야 하므로 많은 시간과 경비가 소요된다. 이 연구에서는 딥러닝 알고리즘을 활용하여 별도의 동적해석 없이 주행안전성 및 승차감을 사전에 예측할 수 있는 딥러닝 기반 예측 시스템 개발하였다. 제안된 시스템은 철도교량의 열차별, 속도별 동적해석 결과를 학습한 후 학습 완료된 신경망을 기반으로 한 예측 시스템이며, 열차속도, 교량 특성 등의 입력파라미터에 따른 주행안전성 및 승차감 산정 결과를 사전에 예측할 수 있다. 제안된 시스템의 성능을 확인하기 위하여 단경간 직선 단순보 교량을 대상으로 한 주행안전성 및 승차감 예측을 수행하였고, 주행안전성 및 승차감 산정을 위한 상판 연직변위 및 상판 연직가속도를 높은 정확도로 예측할 수 있음을 확인하였다.

• 한국자동차공학회논문집
• /
• 제3권3호
• /
• pp.1-8
• /
• 1995

In the companion paper of the present paper, a coast down test method to determine the resistance forces on running vehicle based on the distance-time measurement was explained along with the suggestions to improve its accuracy and testing methodology. In the present paper some of the suggestions discussed previously are implemented and actually road tested to see the applicability of the improved method(short distance method) in the field. From the results. it is shown that the short distance method which requires only 600m long proving ground road gives at least comparable results on the accuracy compared to the original S-t method which requires 2000m. It is hoped that the present method be further refiend to give more accurate results.

• 한국생산제조학회지
• /
• 제8권1호
• /
• pp.28-34
• /
• 1999

According to the development of tool material and the improvement of machinability of cutting material like aluminium alloy, the higher spindle speed is needed. However, the higher speed causes the heat generation of bearings, the deformation of spindle unit parts, and the rotational accuracy of spindle to be worse. Therefore, it is essential to analyze and control the heat generation and the thermal behavior of spindle unit in order to have higher speed and better rotational accuracy. This paper shows the analogy between the analyzation of heat generation and thermal behavior of high speed spindle system by finite element method and the test results of actual temperature rise through running test, and shows the necessity of cooling the spindle and inner ring side of bearings for the thermal balance of high speed spindle system.

• 전기학회논문지
• /
• 제56권7호
• /
• pp.1324-1330
• /
• 2007

In this paper, we propose pattern recognition algorithm for activities of daily living by adopting extreme learning machine based on single layer feedforward networks(SLFNs) to the signal from bidirectional accelerometer. For activity classification, 20 persons are participated and we acquire 6, types of signals at standing, walking, running, sitting, lying, and falling. Then, we design input vector using reduced model for ELM input. In ELM classification results, we can find accuracy change by increasing the number of hidden neurons. As a result, we find the accuracy is increased by increasing the number of hidden neuron. ELM is able to classify more than 80 % accuracy for experimental data set when the number of hidden is more than 20.

• 한국정밀공학회지
• /
• 제31권2호
• /
• pp.133-139
• /
• 2014

Magnetic Sensors can be employed to localize the unmanned vehicle which is running a predefined path where magnets are embedded for certain spaces. Among various sensor types, sensor arrays of 1-dimensional magnetic sensor have the merit of easy elimination of external magnetic component such as terrestrial magnetism. However, interpolation should be considered in the array sensors in order to increase the precision level because there is a limit in arranging sensors in close interval. We propose the novel interpolation method which can be performed with simple computation and represents the improved accuracy by increasing the linearity of the interaction formula. Demonstration of the linearity and simulation results show the proposed method exhibits the improved accuracy compared to the conventional method.

• 한국자동차공학회논문집
• /
• 제3권2호
• /
• pp.51-61
• /
• 1995

A coast down test mothod has been used to determine the resistance forces on running vehicle due to the aerodynamic drag, rolling resistance and driveline resistance. Most of the tests, however, are based on the Velocity-Time measurements, which require a sophisticated velocity measuring device and contain much error by nature. In the present study a coast down test method based on Distance-Time measurements is introduced, which contains the original idea of Russian scientist Prof. Petrushov along with the suggestions for improvement of the accuracy.

• Journal of Electrical Engineering and Technology
• /
• 제6권6호
• /
• pp.786-792
• /
• 2011

The present paper presents an optimal capacitor placement (OCP) algorithm for voltagestability enhancement. The OCP issue is represented using a mixed-integer problem and a highly nonlinear problem. The hybrid particle swarm optimization (HPSO) algorithm is proposed to solve the OCP problem. The HPSO algorithm combines the optimal power flow (OPF) with the primal-dual interior-point method (PDIPM) and ordinary PSO. It takes advantage of the global search ability of PSO and the very fast simulation running time of the OPF algorithm with PDIPM. In addition, OPF gives intelligence to PSO through the information provided by the dual variable of the OPF. Numerical results illustrate that the HPSO algorithm can improve the accuracy and reduce the simulation running time. Test results evaluated with the three-bus, New England 39-bus, and Korea Electric Power Corporation systems show the applicability of the proposed algorithm.

• 한국기계가공학회지
• /
• 제8권1호
• /
• pp.71-76
• /
• 2009

In a running blower, there are various exciting forces which can generate noise. The noise of an running industrial fan or blower depends on the structure of a fan, the machining accuracy of each element, and assembled conditions. Many studies have been carried out to reduce the noise of it. In this study, 3-hole buffs are used in pipe of blower to study the influence the number and position of buffs on the noise reduction at inlet and outlet in pipe. Commercial engineering software ANSYS and SYSNOISE were employed to analyze the characteristics and reduction ratio of pressure. It is concluded that optimal position and number of buffs in pipe of blower to show the least reduction ratio of pressure.

• Journal of Korean Neurosurgical Society
• /
• 제46권1호
• /
• pp.11-15
• /
• 2009

Objective: Two findings easily found at coronal source images of MR myelography (MRM) were evaluated : dorsal root ganglion (DRG) swelling and running course abnormality (RCA) of L5 exiting root at foramen or extraforamen. We tried to find the sensitivity of each finding when root was compressed. Methods: From 2004 July to 2006, one hundred and ten patients underwent one side paraspinal decompression for their L5 root foraminal or extraforaminal compression at L5-S1 level. All kinds of conservative treatments failed to improve leg symptom for several months. Before surgery, MRI, CT and MRM were done. Retrospective radiologic analysis for their preoperative MRM coronal source images was done to specify root compression sites and L5 root morphologic changes. Results: DRG swelling was found in 66 (60%) of 110 patients. DRG swelling has statistically valuable meaning in foraminal root compression (chi-square test, p<0.0001). Seventy-two (66%) in 110 patients showed abnormal alteration of running course. Abnormal running course has statistically valuable meaning in foraminal or extraforaminal root compression (chi-square test, p<0.0001). Conclusion: Three-dimensional MRM provides precise thin sliced coronal images which are most close to real operative views. DRG swelling and running course abnormality of L5 exiting root are two useful findings in diagnosing L5 root compression at L5-S1 foramen or extraforamen. MRM is thought to provide additional diagnostic accuracy expecially in L5-S1 foraminal and extraforaminal area.