정수장 전염소 공정제어를 위한 침전지 잔류염소농도 예측 머신러닝 모형

김주환,이경혁,김수전,김경훈,Kim, Juhwan,Lee, Kyunghyuk,Kim, Soojun,Kim, Kyunghun 한국 수자원 학회 2022 한국수자원학회논문집 Vol.55 No.-

본 연구는 정수장의 수처리 공정에서 계측되고 있는 수량 및 수질데이터의 활용과 수처리 공정제어의 지능화를 위한 것으로 정수장에서 전염소 공정이 수반되는 처리공정에서 침전지 유출수 잔류염소농도 안정화를 위하여 이를 추정할 수 있는 모형을 구축하고자 하였다. 정수장 침전지 유출수의 잔류염소농도를 예측하기 위하여 중회귀모형과 인공지능 알고리즘 중 다층퍼셉트론 신경망, 랜덤포레스트 및 장단기기억(Long Short Term Memory; LSTM) 모형을 활용하였고 그 결과를 비교, 평가하였다. 모형의 입력변수로는 전염소 공정이 도입된 정수장에서의 잔류염소농도, 수온, 탁도, pH, 전기전도도, 유량, 알칼리도 등이 사용되었고 전염소에 따른 침전지의 안정적 운영을 위해 요구되는 침전지 잔류염소농도를 출력변수로 구성하였다. 적용 결과에서는 랜덤포레스트 모형이 가장 양호한 결과를 보여 주었으며 다음으로 LSTM, 다층퍼셈트론 신경망 순으로 나타났다. 수학적 모형인 중회귀모형은 적합도 측면에서 가장 낮은 결과를 보여 주었는데, 이는 수량과 수질데이터의 수치적인 규모나 차원의 차이뿐만 아니라 계절별 수질특성에 따라 염소소비 특성이 매우 다양하게 반응하기 때문으로 판단된다. 따라서 정수장 수처리 공정에서 인공지능 알고리즘의 적용을 위해서는 랜덤포레스트와 같이 의사결정 트리구조의 도입과 적용이 타당한 것으로 나타났다. 본 연구에서 분석된 결과를 근거로 전염소 공정이 도입된 정수장 수처리 공정에서 염소주입량을 실시간으로 예측 가능하게 함으로써 침전지 유출수에서 잔류염소농도를 일정하게 유지하는데 기여할 수 있을 것으로 기대된다. The purpose of this study is to predict residual chlorine in order to maintain stable residual chlorine concentration in sedimentation basin by using artificial intelligence algorithms in water treatment process employing pre-chlorination. Available water quantity and quality data are collected and analyzed statistically to apply into mathematical multiple regression and artificial intelligence models including multi-layer perceptron neural network, random forest, long short term memory (LSTM) algorithms. Water temperature, turbidity, pH, conductivity, flow rate, alkalinity and pre-chlorination dosage data are used as the input parameters to develop prediction models. As results, it is presented that the random forest algorithm shows the most moderate prediction result among four cases, which are long short term memory, multi-layer perceptron, multiple regression including random forest. Especially, it is result that the multiple regression model can not represent the residual chlorine with the input parameters which varies independently with seasonal change, numerical scale and dimension difference between quantity and quality. For this reason, random forest model is more appropriate for predict water qualities than other algorithms, which is classified into decision tree type algorithm. Also, it is expected that real time prediction by artificial intelligence models can play role of the stable operation of residual chlorine in water treatment plant including pre-chlorination process.

하악전치하방 ( 下顎前齒下方 ) 에 발생한 cementifying - fibroma의 1예보고

김주환,김종열,김창현 ( Ju Whan Kim,Chong Youl Kim,Chang Hyun Kim ) 대한예방치과·구강보건학회 1967 大韓口腔保健學會誌 Vol.1 No.2

발치 ( 拔齒 ) 로 인한 파상풍의 1예 보고

김주환,김종배,고진수,김신규 ( Ju Whan Kim,Jong Bai Kim,Jin Soo Kauh,Shin Kyu Kim ) 대한예방치과·구강보건학회 1967 大韓口腔保健學會誌 Vol.1 No.2

생쥐 난소에서 Preantral Follice의 단순 분리법

김주환,박기상,송해범,전상식,Kim, Ju-Hwan,Park, Kee-Sang,Song, Hai-Bum,Chun, Sang-Sik 대한생식의학회 2000 Clinical and Experimental Reproductive Medicine Vol.27 No.3

Objective: Our present studies were conducted to examine more effective isolating method of preantral follicles from mouse ovaries. Methods: ICR mice (3-6 weeks old) were sacrificed through cervical dislocation and their ovaries were removed and put into watch glasses containing Hams F-10 supplemented with 10% fetal bovine serum (FBS). Preantral follicles were isolated by three different methods; 1) enzymatical method and 2) mincing method, and 3) scraping method. Enzymatical method was carried out as following. Ovaries were bisected with a pair of fine 30G needles. Bisected ovaries were incubated at $37^{\circ}C$ and 5% $CO_2$ incubator in 2-well dish containing Hams F-10 supplemented with collagenase 600 lU/ml and DNAse 20 lU/ml. After 20 min., follicles were isolated by repeated pipetting. Isolated preantral follicles were collected, and the remnant of tissues was placed in incubator and previous procedure was repeated. Mincing method was carried out with a pair of fine 30G needles attached to 1 ml syringes and minced ovary. Scraping method was carried out with a pair of fine 30G needles and scratched to surface of ovary. The differences between isolating methods were analyzed using Student's t-test and Chi-square. Results were considered statistically significant when ${\rho}$ value was less than 0.05. Results: In handling time, mincing or scraping method ($28{\pm}3.42$ min or $16{\pm}1.58$ min) were significantly (p<0.00001) shorter than enzymatical method ($72{\pm}1.69$ min), and scraping method was significantly (p<0.01) shorter than mincing method. Total number of isolated follicles was significantly (p<0.0001) higher in enzymatical method ($49.8{\pm}3.91$) than in mincing or scraping method ($25.3{\pm}2.33$ or $20.5{\pm}1.75$). Isolated follicles in ${\leq}$90${\mu}m$ were significantly (p<0.005) higher in enzymatical method ($15{\pm}1.71$) than in mincing or scraping method ($7.8{\pm}0.98$ or $8.1{\pm}1.31$). In 91~130 ${\mu}m$, isolated follicles were significantly (p<0.0005) higher in enzymatical method ($33{\pm}3.27$) than in mincing or scraping method ($16.3{\pm}1.82$ or $10.7{\pm}1.38$). In ${\geq}$ 131 ${\mu}m$, isolated follicles were not significantly differences between all groups. In equal sizes, the rate of isolated follicles in ${\leq}$ 90 ${\mu}m$ was highest in scraping method (39.6% vs. enzymatical method: 30.1%, p<0.05; mincing method: 30.9%, p=0.11719, NS). Rate of follicles in $91{\sim}130$ ${\mu}m$ was significantly (p<0.05) lower in scraping method (52.7%) than in enzymatical or mincing method (66.3% or 64.5%). Rate of follicles in ${\geq}$131 ${\mu}m$ was highest in scraping method (8.3% vs. enzymatical or scraping method: 3.6%, p<0.05 or 4.6%, p=0.19053, NS). Conclusions: This study suggests that scraping method is simple and useful for isolation of preantral follicles, because this method reduced handling time and recovered enough follicles. The recovered rate of isolated follicles in diameter of 91 ~ 130 ${\mu}m$ was highest in all methods.

해운선사 안전관리 평가지표 개발에 관한 연구(2) : 안전경영 지표 평가항목 개발

김주환,김화영,Kim, Joo Hwan,Kim, Hwa Young 해양환경안전학회 2015 해양환경안전학회지 Vol.21 No.6

본 연구는 '해운선사 안전관리 평가지표 개발에 관한 연구'의 후속연구로서 선행연구에서 도출한 '안전관리 평가지표'의 '안전가점' 지수에 해당하는 안전경영 지표의 평가항목을 개발하고, 적정성을 평가하는데 목적이 있다. 이를 위해 해상화물운송사업장에서 일하는 안전관리 담당자를 대상으로 의견을 수렴하여 안전경영 지표별 11개의 평가항목을 개발하였다. 식별된 11개 평가항목에 대한 적정성 검토를 위해 설문조사를 실시하고, 사업형태별(내항해상화물운송, 외항해상화물운송), 사업규모별(보유선박수)로 응답자의 적정성 인식의 차이를 통계적 기법을 활용하여 분석하였다. 그 결과, 평가항목에 대한 적정성 인식의 차이는 사업형태와는 무관하였다. 그러나 사업규모에 따라서는 소(小) 중(中) 대(大) 3개의 그룹 간 평가항목의 적정성에 인식의 차이가 존재했다. 따라서 향후 안전경영 지표를 활용하여 선사를 평가할 경우, 사업형태별로 동일하게 평가하되 사업규모별로는 보유선박수에 따라 그룹화하여 평가하는 것이 타당하다는 결론을 얻었다.

인바합금 도금층의 물성에 영향을 미치는 도금인자에 관한 연구

김주환,정명원,임태홍,이재호,Kim, Ju-Hwan,Jung, Myung-Won,Yim, TaiHong,Lee, Jae-Ho 한국마이크로전자및패키징학회 2013 마이크로전자 및 패키징학회지 Vol.20 No.1

The experiments were carried out in the variation of current density, pH, temperature, and duty cycle to investigate the influence of electroplating parameters on the properties of Ni-Fe invar alloys. When the current density and temperature were changed, the composition of invar alloy was varied, however, duty cycle and pH hardly affected on the composition of electrodeposited alloys. However, as the duty cycle was increased, microstructure was changed and the decrease of hardness was also observed.

구강내에 방생한 악성흑색종 1예

김주환,정성창,김황중 ( Ju Whan Kim,Sung Chang Chung,Whang Joong Kim ) 대한예방치과·구강보건학회 1967 大韓口腔保健學會誌 Vol.1 No.2

최적화 기법을 이용한 광역상수도 관로시스템 설계

김주환,김종우,박재홍,Kim, Ju Hwan,Kim, Zong Woo,Park, Jae Hong 대한상하수도학회 1999 상하수도학회지 Vol.13 No.1

In this research, it is proposed that optimization method is introduced and applied to the design of pipeline system in multi-regional water supply project, which has been constructed to settle the regional unbalance problems of available water resources. For the purpose, interface programs are developed to integrate linear programming model and KYPIPE model which is used for optimization and hydraulic analysis, respectively. The developed program is applied to the pipeline system design of multi-regional water supply project. The optimal diameters from the application of linear programming technique are compared with those from conventional method that is time-consuming and tedious trail and error process. Since the conventional design largely depends upon the experience of designers and the results of general hydraulic analysis, it can not be reasonable and consistent. The application of linear programming technique can make it possible to design pipeline system optimally by using same design factors of general hydraulic models. The model can select commercial discrete pipe diameter as optimal size by using pipe length as decision variables. The developed model is applied to Pohang multi-regional water supply system design with two different objective functions, which are initial construction cost and annual cost including electric cost. As results, it is calculated that the initial construction cost of 1,449,740 thousand won is saved and annual cost of 128,951 thousand won is saved for a year within study year. Also, the optimal site of pump station is selected on 5th pipe, which is located between the diverging junction to Kangdong(2) province and the diverging junction to Cheonbuk province. It is explained that pump cost is less than pipe cost in this application case study due to little pump station scale. In the case of water supply with large pump capacity, it is reasonal that the increase of pipe size is more efficient instead the increase of pump station capacity to save annual cost.

Markov Chain Model for Synthetic Generation by Classification of Daily Precipitation Amount into Multi-State

김주환,박찬영,강관원,Kim, Ju-Hwan,Park, Chan-Yeong,Kang, Kwan-Won Korea Water Resources Association 1996 물과 미래(한국수자원학회지) Vol.29 No.6

수자원의 주공급원인 강수는 현상의 발생여부에 따라 건조일과 습윤일이 교대로 반복되는 과정으로 구성되어 있으며, 특히, 일강수계열의 습윤일에 발생되는 강수량의 크기는 매우 다양한 형태를 지니고 있어 이 과정을 모형화 하는데는 복잡한 확률과정이 수반된다. 본 연구에서는 일강수계열의 발생과정을 건조일, 습윤일로 구분하고 습윤일의 강수량을 상태별로 분류하여 각 상태별 천이확률을 계산함으로써 이를 장래에 발생 가능한 강수사상의 모의발생에 이용하였다. 본 모형은 수문사상의 발생과 비발생만을 구분하던 2-state Markov 연쇄모형에 강수의 발생시 강수량의 크기에 따라 상태를 여러 개로 구분하여 강수량을 추정할 수 있도록 수정한 것으로 간헐 수문사상인 일강수계열의 구성성분인 건조일과 습윤일, 건조, 습윤 지속기간 및 습윤일의 강수량을 Markov 연쇄에 의해 동시에 발생있도록 한 것이며 다른 모형에 비해 사용이 비교적 용이하다. 본 연구에서 제안한 multi-state Markov 연쇄모형의 적용 가능성을 검토하기 위하여 비교적 장기간의 자료를 보유하고 있는 관측소의 강수자료를 이용하였으며 그 결과를 강수량, 건조, 습윤일수 및 건조, 습윤계속기간의 분포를 실제자료와 비교하여 모형의 적합도를 평가하였다. 이를 토대로 홍수 및 한발기간의 추정과 모의발생에 의한 자료 확장으로 중장기 수자원 계획 및 운영에 효율적으로 이용될 수 있을 것으로 판단된다. The chronical sequences of daily precipitation are of great practical importance in the planning and operational processes of water resources system. A sequence of days with alternate dry day and wet day can be generated by two state Markov chain model that establish the subsequent daily state as wet or dry by previously calculated vconditional probabilities depending on the state of previous day. In this study, a synthetic generation model for obtaining the daily precipitation series is presented by classifying the precipitation amount in wet days into multi-states. To apply multi-state Markov chain model, the daily precipitation amounts for wet day are rearranged by grouping into thirty states with intervals for each state. Conditional probabilities as transition probability matrix are estimated from the computational scheme for stepping from the precipitation on one day to that on the following day. Statistical comparisons were made between the historical and synthesized chracteristics of daily precipitation series. From the results, it is shown that the proposed method is available to generate and simulate the daily precipitation series with fair accuracy and conserve the general statistical properties of historical precipitation series.

강수계열의 상태분류에 의한 Markov 연쇄 모의발생모형

김주환,박찬영,Kim, Ju-Hwan,Park, Chan-Yeong 한국수자원학회 1996 물과 미래(한국수자원학회지) Vol.29 No.6

개수로내의 검변 및 급변 부정류 해석을 위해서 dynamic wave식을 기본방정식으로 하고 이를 불연속 보간함수와 upstream weighting 을 도입한 Petrov-Galerkin 기법에 의해 해석하는 유한요소모형을 개발하였다. 매트릭스 안정성 해석 결과 Petrov-Galerkin기법은 단파장에서의 선택적 감쇠능력과 위상오차에 있어 우수한 것으로 나타났다. 반면에 Preissmann기법은 단파장에서의 선택적 감쇠능력과 위상오차에 있어 열등한 것으로 나타났고, Bubnov-Galerkin 기법은 비감쇠특성을 나타내고 있어 단파장 영역에서 발산해를 일으키는 주요원인임을 확인할 수 있었다. Petrov-Galerkin 방법은 Courant수의 넓은 범위에서 높은 주파수를 가진 진행파에 대한 선택적인 감쇠와 작은 Courant 수의 범위에서 양호한 위상정도를 가지는 이상적인 조합을 나타내고 있어 점변 및 급변 부정류 해석에 있어 이상적인 기법으로 활용될 수 있을 것으로 판단되었다. A finite element model for simulating gradually and rapidly varied unsteady flow in open channel is developed based on dynamic wave equation using Petrov-Galerkin method. A matrix stability analysis shows the selective damping of short wave lengths and excellent phase accuracies achived by Petrov-Galerkin method. Whereas the Preissmann scheme displays less selective damping and poor phase accuracies, and Bubnov-Galerkin method shows nondissipative characteristics whicn causes a divergence problem in short wave length. The analysis also shows that the Petrov-Galerkin method displays the desirable combination of selective damping of high frequency progressive waves over a wide range of Courant number and good phase accuracy at low Courant number. Therefore, the Petrov-Galerkin can be effectively applied to gradually and rapidly varied unsteady flow.