지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학위논문
- 저자정보
- 지도교수
- 최계운
- 발행연도
- 2017
- 저작권
- 인천대학교 논문은 저작권에 의해 보호받습니다.
이용수1
이 논문의 연구 히스토리 (3)
- 이 논문의 후속연구가 궁금하신가요?
- 연관 학술논문 또는 학술발표를 통해 보다 발전된 연구결과를 확인하실 수 있습니다.
- 이 논문의 연구 히스토리 확인하기
초록· 키워드
오류제보하기
The non-revenue water (NRW) ratio in the water distribution systems is the ratio of losses from unbilled authorized consumption, apparent losses and real losses among the overall amount of water supply (tap water supplied from water purification plants). The NRW ratio is one of the factors that determines priorities in the implementation of the project to improve water distribution systems, and identifying the parameters that affect this ratio is useful in maintenance planning for a district metered area (DMA). It is often difficult, however, to estimate the NRW ratio in districts with no DMA, uncertain management departments and pipe network operations that are not optimized. So selecting and determining the influencing factors of the NRW ratio according to an area’s characteristics is an essential element in the implementation of the project to improve water distribution systems.
Many studies analyze data using statistical techniques and identify the factorial relationship as a method to estimate the NRW ratio. In particular, the method of applying the regression equation through multiple regression analysis is widely applied to the estimation of the NRW ratio. This method is among the statistical analysis methods that use the main parameters of water distribution systems as the independent variable and calculates the NRW ratio as the dependent variable. Its disadvantage, however, lower accuracy compared to the measured NRW ratio.
Therefore, Methodologies were suggested to estimating the NRW ratio using the Artificial Neural Network (ANN) and Principal Component Analysis (PCA) with main parameters of water distribution systems. After analyzing the main parameters of the water distribution networks, a parameter classification system for NRW ratio estimation was established with physical, operational and socioeconomic factors. Also, the selected parameters were converted into dimensionless data through PCA and applied to an ANN, then compared to original data usage conditions to determine the accuracy of NRW ratio estimation.
As a result, the method using ANN was found to be more accurate in estimating the NRW ratio than multiple regression analysis. In addition, the most accurate estimation of the NRW ratio was possible when PCA was applied to ANN. The suitability of the selected main parameters of the water distribution systems was verified through the ANN simulation, and the optimal number of hidden neurons required for estimation of the NRW ratio in the ANN model and appropriate analysis technique were presented. The estimation accuracy of the NRW ratio in water distribution systems using the ANN with the optimal number of neurons was higher than existing statistical methods; yet such accuracy might be lowered depending on the number of hidden neurons in the ANN model. In addition, research confirmed that non-dimensional parameters considering the correlation between the physical and operational parameters in the water distribution systems were calculated through PCA, and NRW ratio estimation using principal components was statistically significant. The accuracy of such estimation was improved by using the principal component data.
In this study, the influence factors of the NRW ratio were derived from ANN, statistical techniques and existing research examples. And the methodology of NRW ratio calculation was verified after applying selected parameters to the test bed. The methodologies and guidelines for estimating the NRW ratio using ANN and PCA based on local data can be applied to decide on a project to improve management of a water distribution system. They are also expected to be helpful for devising an optimal operation method of water supply facilities for water network analysis and DMA construction.
Many studies analyze data using statistical techniques and identify the factorial relationship as a method to estimate the NRW ratio. In particular, the method of applying the regression equation through multiple regression analysis is widely applied to the estimation of the NRW ratio. This method is among the statistical analysis methods that use the main parameters of water distribution systems as the independent variable and calculates the NRW ratio as the dependent variable. Its disadvantage, however, lower accuracy compared to the measured NRW ratio.
Therefore, Methodologies were suggested to estimating the NRW ratio using the Artificial Neural Network (ANN) and Principal Component Analysis (PCA) with main parameters of water distribution systems. After analyzing the main parameters of the water distribution networks, a parameter classification system for NRW ratio estimation was established with physical, operational and socioeconomic factors. Also, the selected parameters were converted into dimensionless data through PCA and applied to an ANN, then compared to original data usage conditions to determine the accuracy of NRW ratio estimation.
As a result, the method using ANN was found to be more accurate in estimating the NRW ratio than multiple regression analysis. In addition, the most accurate estimation of the NRW ratio was possible when PCA was applied to ANN. The suitability of the selected main parameters of the water distribution systems was verified through the ANN simulation, and the optimal number of hidden neurons required for estimation of the NRW ratio in the ANN model and appropriate analysis technique were presented. The estimation accuracy of the NRW ratio in water distribution systems using the ANN with the optimal number of neurons was higher than existing statistical methods; yet such accuracy might be lowered depending on the number of hidden neurons in the ANN model. In addition, research confirmed that non-dimensional parameters considering the correlation between the physical and operational parameters in the water distribution systems were calculated through PCA, and NRW ratio estimation using principal components was statistically significant. The accuracy of such estimation was improved by using the principal component data.
In this study, the influence factors of the NRW ratio were derived from ANN, statistical techniques and existing research examples. And the methodology of NRW ratio calculation was verified after applying selected parameters to the test bed. The methodologies and guidelines for estimating the NRW ratio using ANN and PCA based on local data can be applied to decide on a project to improve management of a water distribution system. They are also expected to be helpful for devising an optimal operation method of water supply facilities for water network analysis and DMA construction.
목차
Chapter 1. Introduction 11.1. Research Background and Objectives 11.2. Research Scope and Contents 51.3. Literature Review 71.3.1 Main Influential Parameters of NRW Ratio 71.3.2 Research on Estimation of NRW Ratio 101.3.3 Application of ANN to Water Distribution Systems 12Chapter 2. Theriotical Background 172.1. Estimation of NRW Ratio in Water Distribution Systems 172.1.1 Definition of NRW Ratio in Water Distribution Systems 172.1.2 Calculaltion of NRW Ratio in Water Distribution Systems 192.2. Inclusion of Energy Consumption Parameter in NRW Ratio Estimation 212.3. Methods for NRW Ratio Estimation Using Statistical Analysis 232.3.1 Correlation Analysis Using Pearson Correlation Coefficient 232.3.2 Variation Analysis of Parameters Based on ANOVA 242.3.3 Estimation of NRW Ratio Using Multiple Regression Analysis 262.3.4 Data Standardization Based on Z-score 272.4. Principal Component Analysis for More Accurate NRW Ratio Estimation 282.5. Principle of Artificial Neural Network for the Estimation of NRW Ratio 31Chapter 3. Selection of Parameters for the Estimation of NRW Ratio 333.1 Development of Parameter Classification System for Estimation of NRW Ratio 333.1.1 Parameter Classification at Water Distribution Systems in Abroad 333.1.2 Parameter Classification at Water Distribution Systems in Korea 393.1.3 Establishment of Parameter Classification System 433.2. Parameter Classification per Data Quantification 493.2.1 Parameter Classification between Direct and Indirect Factors 493.2.2 Description of Operational Parameters 543.2.3 Description of Socioeconomic Parameters 573.3. Final Parameter Selection for Estimation of NRW Ratio 59Chapter 4. Development of Methodology for Estimation of NRW Ratio Using ANN & PCA 614.1 General Approaches to Direction of NRW Ratio Estimation 614.2. Development of Development for Application of ANN in Estimation of NRW Ratio 624.2.1 Application of ANN for Efficiency Improvement in Estimation of NRW Ratio 624.2.2 Composition of Input Layer in ANN Model 634.2.3 Composition of Output Layer in ANN Model 644.2.4 Composition of Hidden Layer in ANN Model 644.3. Methodology for Estimating NRW Ratio by Statistical Analysis with PCA 674.3.1 Analysis of Efficiency Improvement using PCA in Statistical Analysis 674.3.2 Application of PCA for Efficiency Improvement in ANN Analysis 684.3.3 Development of Methodology for Application of PCA in NRW Ratio Estimation 704.4. Development of Methodology for PCA-ANN Application in Estimation of NRW Ratio 71Chapter 5. Application of Developed Methodologies in Test Bed 755.1. General Description of Test Bed 755.1.1 General Description of Regional Characteristics 755.1.2 Description of Water Distribution Systems and DMA 785.1.3 Water Demand Data Indicated at Designed Water Distribution Systems 825.1.4 Hydraulic Characteristics at Designed Water Distribution Systems 845.1.5 Summary of Deteriorated Pipe Characteristics from Technical Diagnosis Report 865.2. Data Collection of Parameters 895.2.1 Parameter Selection for NRW Ratio Analysis 895.2.2 Data Collection for Selected Parameters 905.3 Statistical Analysis for the Relationship between Selected Parameters 945.3.1 Basic Statistical Analysis for Verification of Data Availability 945.3.2 Correlation Analysis of Selected Parameters and NRW Ratio 975.3.3 PCA of Selected Parameters for Application of Developed Methodologies 985.4. Estimation of NRW Ratio Using Multiple Regression Analysis for Application of Developed Methodologies 1005.4.1 Review for Existing Multiple Regression Analysis 1005.4.2 Results of Multiple Regression Analysis Using Original Parameters 1015.4.3 Results of Multiple Regression Analysis Using PCA Parameters 1095.4.4 Selection of Main Parameters for Application of ANN & PCA 114Chapter 6. Applicability Assessment on Developed Methodology of ANN & PCA 1176.1. Application to Test Bed Using ANN & PCA for Estimation of NRW Ratio 1176.1.1 Model Construction for ANN Simulation 1176.1.2 Simulation Cases for Estimaion of NRW Ratio 1176.1.3 Estimation of NRW Ratio by Using ANN Simulation 1196.1.4 Estimation of NRW Ratio by Using ANN & PCA 1256.2. Accuracy Assessment for Analyzing Application Method 1316.3. Discussions for Future Application 135Chapter 7. Conclusions 138References 141국문초록 148
최근 본 자료
댓글(0)
0