• Proceedings of The Korean Society of Agricultural and Forest Meteorology Conference
• /
• 2018.06a
• /
• pp.128-128
• /
• 2018

• ETRI Journal
• /
• v.28 no.1
• /
• pp.59-66
• /
• 2006

Learning behaviors of a radial basis function network (RBFN) using a singular value decomposition (SVD) and stochastic gradient (SG) algorithm, together named RBF-SVD-SG, for odor sensing systems are analyzed, and a fast training method is proposed. RBF input data is from a conducting polymer sensor array. It is revealed in this paper that the SG algorithm for the fine-tuning of centers and widths still shows ill-behaving learning results when a sufficiently small convergence coefficient is not used. Since the tuning of centers in RBFN plays a dominant role in the performance of RBFN odor sensing systems, our analysis is focused on the center-gradient variance of the RBFN-SVD-SG algorithm. We found analytically that the steadystate weight fluctuation and large values of a convergence coefficient can lead to an increase in variance of the center-gradient estimate. Based on this analysis, we propose to use the least mean square algorithm instead of SVD in adjusting the weight for stable steady-state weight behavior. Experimental results of the proposed algorithm have shown faster learning speed and better classification performance.

• Journal of Sensor Science and Technology
• /
• v.16 no.3
• /
• pp.179-186
• /
• 2007

Uroflowmetry is non-invasive and easily performed to diagnose benign prostate hypertrophy (BPH) frequent in aged men. Weight change during urination is usually measured to estimate the urinary flow rate by a load cell, but sensitive to any impacts against the bottom of the container, leading to unnecessary noise generation. Moreover, load cells are relatively expensive raising the production cost. The present study proposed a new technique, measuring hydraulic pressure on the bottom of the urine container to evaluate the urinary flow rate. Low cost pressure transducer enabled almost perfectly linear relationship between the urine volume and the hydraulic pressure. During both the simulated and human urination experiment, variance of the pressure signal was more than 50 % smaller than the weight signal acquired by a load cell, which demonstrated that the impact noise was decreased to a great degree by pressure compared to weight measurement.

• Journal of Biosystems Engineering
• /
• v.21 no.2
• /
• pp.123-133
• /
• 1996

An automatic drum seeder was developed to improve the seeding operation. It consisted of a conveyor to transfer seedling trays, a seed-hopper to supply seeds, a drum to drop seeds on the tray, and an air blower to remove extra seeds. A photo sensor was used to detect the transfer of seedling trays, and its signal was fed into microcomputer which operated a stepping motor driving the drum. The seeds were adhered to the surface of drum by vacuum pressure, and were dropped into tray cells by compressed air. An air connection unit was devised to alternate between vacuum pressure and compressed air. A control program for the system, written in C language, could operate the drum at the given number of revolutions and revolutions per minute. The results showed that the air connection unit could operate well and the seeds were dropped satisfactorily into tray cells. In case of cabbage and perilla seeds, which are regular and spherical shape, the missing rate was low and the single seeding rate was more than 97%. Low missing rate and high multiple seeding rate were observed in lettuce seeds which have narrow ends with tight weight. The missing rate of pepper seed was very high because of heavy weight and irregular shape. To improve the performance of the seeder, adjustment of vacuum pressure based upon shape and weight of the seeds, careful selection of the material of drum, maintenance of consistent air blower pressure, and replacement of stepping motor to DC motor are recommended.

• International journal of advanced smart convergence
• /
• v.8 no.3
• /
• pp.78-86
• /
• 2019

Recently, IoT researches using sensor data based on embedded networks in various fields including healthcare and sports have been continuously attempted. This study analyzes golf ball mobility to support IoT application in golf sports field. Generally, since the difference in density occurs due to the condition of the inner material and the abnormal state at the time of the outer skin joining during the manufacturing of the golf ball, the weight of each subset is equal for any two points with the same radius in the sphere cannot be guaranteed. For this reason, the deflected weight of the sphere has the undesirable effect of hitting the ball in a direction in which the weight of the ball is heavy. In this study, it is assumed that there is a unique center of gravity of the ball, and even if the golf ball cannot be manufactured perfectly, it wants to establish the basic principle to accurately recognize or mark the putting line based on the center of gravity. In addition, it is evaluated how the mobility of the golf ball with a deviation from the center of gravity of the golf ball affects the progress path (or movement direction) and the moving distance (or carry distance) after the golfer hits. The basic model of the mobility of the golf ball can help the golfer exercise model and the correlation analysis. The basic model of the mobility of the golf ball can help the golfer exercise model and the correlation analysis.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.60 no.12
• /
• pp.2352-2360
• /
• 2011

In this paper, we introduce a comprehensive design methodology of Radial Basis Function Neural Networks (RBFNN) that is based on mechanism of clustering and optimization algorithm. We can divide some clusters based on similarity of input dataset by using clustering algorithm. As a result, the number of clusters is equal to the number of nodes in the hidden layer. Moreover, the centers of each cluster are used into the centers of each receptive field in the hidden layer. In this study, we have applied Fuzzy-C Means(FCM) and K-Means(KM) clustering algorithm, respectively and compared between them. The weight connections of model are expanded into the type of polynomial functions such as linear and quadratic. In this reason, the output of model consists of relation between input and output. In order to get the optimal structure and better performance, Particle Swarm Optimization(PSO) is used. We can obtain optimized parameters such as both the number of clusters and the polynomial order of weights connection through structural optimization as well as the widths of receptive fields through parametric optimization. To evaluate the performance of proposed model, NXT equipment offered by National Instrument(NI) is exploited. The situation awareness system-related intelligent model was built up by the experimental dataset of distance information measured between object and diverse sensor such as sound sensor, light sensor, and ultrasonic sensor of NXT equipment.

• Journal of Sensor Science and Technology
• /
• v.8 no.1
• /
• pp.24-31
• /
• 1999

Ammonia gas sensors were fabricated with ZnO-based thin films grown by RF-magnetron sputtering method. The films which were doped with $MoO_3$ catalysts of various weight percents were grown in different sputtering gases to fabricate the sensors with a high sensitivity, low working temperature and rapid response time. To improve electrical stability, the films were aged in various conditions. The sensors doped with the catalysts and grown in oxygen sputtering gas showed the improvement of sensitivity. These exhibited the increase of surface carrier concentration and electron mobility. The sensor with 0.875wt.% $MoO_3$ catalysts showed the maximum sensitivity of 70 in ammonia gas concentration of 160 ppm at a working temperature of $300^{\circ}C$. The sensor which is aged at $330^{\circ}C$ for 72hrs in oxygen ambient exhibited tourer sensitivity of 57, but more stable properties, excellent linearity.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.10 no.5
• /
• pp.364-371
• /
• 2017

TPMS is a vehicle electric system that measures the air pressure of a tire, and informs the driver of current tire states. The TPMS sensor typically uses unidirectional communication for small size, light weight, and low power. The transmission period of the sensor indicates the service quality of monitoring the tire. In order to determine the optimal transmission period, frame collision probability and the life time of the sensor should be analyzed. In this paper, collision probability model using Venn diagram is designed in low power TPMS with the normal and warning mode. And the life time and a collision probability were analyzed with the ratio(n) of the normal mode to warning mode transmission period. As a result, $T_{nP}=31sec$ and $T_{wP}=2.4sec$ at 5 years, and $T_{nP}=71sec$ and $T_{wP}=2.5sec$ at 7 years.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.7
• /
• pp.2414-2434
• /
• 2015

Among numerous localization schemes proposed specifically for Wireless Sensor Network (WSN), the range-free localization algorithms based on the received signal strength indication (RSSI) have attracted considerable research interest for their simplicity and low cost. As a typical range-free algorithm, Approximate Point In Triangulation test (APIT) suffers from significant estimation errors due to its theoretical defects and RSSI inaccuracy. To address these problems, a novel localization method called FIAPIT, which is a combination of an improved APIT (IAPIT) and a fuzzy logic system, is proposed. The proposed IAPIT addresses the theoretical defects of APIT in near (it's defined as a point adjacent to a sensor is closer to three vertexes of a triangle area where the sensor resides simultaneously) and far (the opposite case of the near case) cases partly. To compensate for negative effects of RSSI inaccuracy, a fuzzy system, whose logic inference is based on IAPIT, is applied. Finally, the sensor's coordinates are estimated as the weighted average of centers of gravity (COGs) of triangles' intersection areas. Each COG has a different weight inferred by FIAPIT. Numerical simulations were performed to compare four algorithms with varying system parameters. The results show that IAPIT corrects the defects of APIT when adjacent nodes are enough, and FIAPIT is better than others when RSSI is inaccuracy.

• Journal of rehabilitation welfare engineering & assistive technology
• /
• v.8 no.4
• /
• pp.259-265
• /
• 2014

The aims of this study are to investigate the walking intention detection of a rollator based on Infraed (IR) sensor measuring knee joint anterior displacement and leg muscle activities. We used Active Walker attached IR sensor to measure the knee joint anterior displacement and EMG signal of leg muscles(rectus femoris, biceps femoris, tibialis anterior, gastrocnemius) were taken by Delsys bagnli-8ch. Subjects were eight healthy males(age $23.7{\pm}0.5years$, height $175.4{\pm}2.3cm$, weight $70.6{\pm}5.6kg$) and they were involved in experiments which had been proceeded 30 minutes a week, during 3 weeks. This system indicates that the knee joint anterior displacement had the distinction increases according to the gait slope and velocity. We showed the increase of the femoral muscle activities along the anterior tilt and the increase of the crural muscle activities along the posterior tilt.