• Journal of Biosystems Engineering
• /
• v.34 no.2
• /
• pp.114-119
• /
• 2009

Field test was conducted to investigate primary factors reducing rapeseed harvesting using a reciprocating cutter-bar of combine. The results showed that the correlation between crop moisture content and yield loss had a U-type, which indicated that the yield reduction increased at too high and too low crop moisture contents. The proper ranges of crop moisture contents were 27${\sim}$35%, 21${\sim}$56%, and 62${\sim}$73% in case of grain, pod and stem, respectively. Crop moisture content was negatively correlated with header loss, but positively correlated with threshing loss. In contrary, stem moisture content showed positive correlations with total loss, threshing loss and separation loss. Working speed was positively correlated with header loss. Total flow rate, pod flow rate and stem flow rate were highly correlated with threshing loss and separation loss. However, grain flow rate did not show any correlation with total loss. According to the principal component analysis, two principal components were derived as components with eigenvalues greater than 1.0. The contribution rates of the first and the second components were 52.7% and 38.9%, which accounted for 91.6% of total variance. As a contributive factor influencing total loss of rapeseed mechanical harvesting, a crop moisture content factor was greater than a crop flow rate factor. The stepwise multiple regression analysis for total loss was conducted using crop moisture content factor, crop flow rate factor and coefficient. However, the model did not show any correlation among independent and dependent factors ($R^2$=0.060).

• Proceedings of the KIEE Conference
• /
• 1998.07b
• /
• pp.575-577
• /
• 1998

In this paper, we propose a hybrid target tracking algorithm which combine the conventional Kalman filter algorithm and the fuzzy neural network. Conventional methods are degraded in the presence of uncertanties and the environmental noise. These problems can be resolved by the proposed method. The training data for the proposed target tracker is obtained by the off-line simulation. Unlike other target trackers usging fuzzy inference system, our method can be easily integrated into the existing system. A numerical simulation is included to show the effectiveness and the feasibility of the proposed method.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.4 no.1
• /
• pp.35-49
• /
• 1994

It is well known that, for nominimum phase systems, a conventional linear controller of PID type or an adaptive controller of this structure shows limitation in achieving a satisfactory performance under tight specifications. In this paper, we combine a neuro-controller with a PI-controller with off-line learning capability provided by the Genetic Algorithm to propose a novel neuro-controller to control nonminimum phase systems effectively. The simulation results show that our proposed model is more efficient with faster rising time and less undershoot effect when the performances of the proposed controller and a conventional form are compared.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.67 no.12
• /
• pp.1699-1704
• /
• 2018

In this paper, proposed a method to improve of control characteristics for power facility using the adaptive sizing of fuzzy inference method. In the use of the controller based the fuzzy logic, a basic mamdani fuzzy controller is applied. However, when the maximum value and the minimum value have to taken, the fuzzy controller can not take a normal value because of formalized grouping form. In this paper, we combine the conventional methods with single valued sets to compensate for the disadvantage caused by the mamdani method control. Simulation results show that the proposed method has better overshoot and steady state arrival time than the conventional control method.

• Journal of KIISE:Computer Systems and Theory
• /
• v.33 no.9
• /
• pp.658-665
• /
• 2006

In these days, the demand of users to extend available period of mobile systems is increased according as the functions of mobile systems have been varied and the use of multimedia application has been increased. This paper proposes an integrated power management framework that considers executed workload types for effective energy management. The conventional methods use DVFS technique for CPU and DPM technique for WNIC separately or simply combine them based on the assumption that they are orthogonal one another. However, the proposed mechanism determines the kind of workload under analysis of the characteristics of workloads incoming through a WNIC. The proposed method can reduce energy consumption of system level effectively by controlling CPU and WNIC to proper power mode based on analyzed characteristics of workload. The experimental result shows the proposed method reduces energy consumption by 9% for BE (Best Effort) workload, CBR (Constant Bit Rate) workload, and Interactive workload on average and by 16% to maximum when compared with the conventional methods which simply combine DVFS technique for CPU and DPM technique for WNIC.

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 2000.11b
• /
• pp.287-294
• /
• 2000

Calculation of the flexural rigidity value (EI) is indispensable for prescription of deflection characteristics of crop stalks in harvesting□Conventionally□EI has been determined by either average EI of the whole stalk or average EI of each stems divided into node through the calculation method of cantilever with homogeneous section□However□deflection characteristics of crop stalks caused by mechanical operation such as combine harvester were not exactly presumed by these conventional EI through the experiment by authors. Further, actual EI of a stalk changes in company with a change of moisture contents as time passes during the experiment. Finally, efficient calculation method for determining EI is needed in order to improve these problems. In this study, mechanical model based on actual structure of the crop stalk with variety sectional area was proposed. This mechanical model is calculated by the theory of cantilever with continuous stages. Therefore, improvement of both calculating accuracy on EI and efficiency of measuring system was tried. At first, this calculation method was applied to piano wire of which EI was recognized in advance. As a result, EI calculated from this new method coincided approximately with piano wire's EI. Next, applying to crop stalks as same as piano wire, relationship between loads acting on crop stalks and deflection values calculated by EI using this new calculation method was exactly presumed in comparison with conventional method. Further, measuring time of deflection test was greatly reduced. Finally, new calculation method of EI will be available for estimating mechanical characteristics of so many kinds of crop stalks in harvesting operation. Further, in this study, new deflection test using image-processing apparatus by computer will be introduced.

• Asia pacific journal of information systems
• /
• v.7 no.1
• /
• pp.85-97
• /
• 1997

This paper is concerned with knowledge representation schemes best suited for human resource management (HRM) problem domains including human resource planing, selection, placement, compensations, performance evaluation, training and labor-management relations. In order to suggest the scheme we consider two research gods. First, we evaluate and prioritize. The knowledge representation techniques of frames rules, semantic nets and predicate logic that hove been recommended to managerial domains. The combined Analytic Hierarchy Process technique is employed to combine individual judgments effectively between two different expert groups. As a result if we are to select a single knowledge representation technique, a frame representation is best for most HRM domains and to combine frames with others is another choice. Second as a strategy for knowledge representation schemes we show some examples for each damn in terms of labeled semantic nets and two types of rules derived from the semantic nets. We propose nine knowledge components as ontologies. The labeled semantic nets con provide some benefits compared with conventional one. More clearly definea node rode information maces it easy to find the ac information. In the rule sets, the variables are the node of the semantic nets. The consistency of rules is validated by the relationship of the knowledge components.

• Journal of Advanced Marine Engineering and Technology
• /
• v.16 no.5
• /
• pp.60-66
• /
• 1992

In most cases a structure consists of the assembly of some substructures, we assemble them with various joints, and the structure is fixed to a foundation through mounts. In case of the structure with rigid joints like welding, the Finite Element Mothod could be easily used to analyize the structure's characteristics, but in case of the structure with elastic joints like bolts or rivets, it might be difficult to analyize it by taking account of joint's rigidities, with the conventional method. This paper proposes the method to identify the joint rigidities by the Sensitive Analysis Method and the Optimization Techniques. And the proposed method applied to identify the rigidities of 4 bolts to combine 2 plates(500mm long, 100mm wide, 3.15mm thich). The analized results were well coincident with the experimental results. To confirm the reliability 0 the rigidities identified, another trial was done for the stucture to combine other 2 plates with same joints. The results were good too. This paper is proposin the identifying method of the joint rigidity of a structure, and it could be used for the data base of the joint rigidity and for the guidance to select joint stiffness.

• ETRI Journal
• /
• v.45 no.5
• /
• pp.822-835
• /
• 2023

Recent semantic segmentation frameworks usually combine low-level and high-level context information to achieve improved performance. In addition, postlevel context information is also considered. In this study, we present a Context ReFinement Network (CRFNet) and its training method to improve the semantic predictions of segmentation models of the encoder-decoder structure. Our study is based on postprocessing, which directly considers the relationship between spatially neighboring pixels of a label map, such as Markov and conditional random fields. CRFNet comprises two modules: a refiner and a combiner that, respectively, refine the context information from the output features of the conventional semantic segmentation network model and combine the refined features with the intermediate features from the decoding process of the segmentation model to produce the final output. To train CRFNet to refine the semantic predictions more accurately, we proposed a sequential training scheme. Using various backbone networks (ENet, ERFNet, and HyperSeg), we extensively evaluated our model on three large-scale, real-world datasets to demonstrate the effectiveness of our approach.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.6 no.6
• /
• pp.556-563
• /
• 2001

In this paper a new zero current transition pulse width modulation (ZCT-PWM) buck converter is proposed to combine the desirable feature of both the conventional buck converter and resonant converters. In this proposed scheme an auxiliary circuit is added to the conventional buck converter and used to achieve soft-switching for both the main switch and the freewheeling diode while not incurring any additional losses due to auxiliary circuit And this converter operates exactly like the conventional PWM converter except for a short particular time interval. The operation of the proposed converter is explained and analyzed. and design guidelines are presented. To validate the feasibility of the proposed converter, a 100KHz 180-W prototype is built and tested.