Weather sensitivity analysis and adjustment for holiday load forecasting and incentive-based demand side management program planning

다국어 초록 (Multilingual Abstract)

In deregulated power systems, lack of investment in power systems infrastructure makes it difficult to cope with an increasing electrical load. In addition, owing to the increase in electricity demand for cooling and heating, the power system peak load is sensitive to sudden weather changes. To overcome these problems, there has been increased interest in the improvement of load forecasting accuracy and demand side management (DSM).
This dissertation describes a holiday load forecasting model using fuzzy polynomial regression that enables weather feature selection and adjustment. At the same time, a stochastic planning method for an incentive-based DSM program using load forecasting with weather sensitivity analysis and adjustment is also proposed.
Load forecasting is a nonlinear problem associated with social phenomena, economic factors, and weather variations. In particular, holiday load forecasting is a challenging task because there is little historical data for holidays compared with that for normal weekdays and weekends. For holiday load forecasting, this dissertation proposes a fuzzy polynomial regression model with weather feature selection and adjustment. This forecasting model utilizes the relationship between weekdays and holidays, using weekday electric loads as input and determining the holiday electric loads as output. Therefore, the selection and adjustment of past weekday data relevant to a given holiday is extremely important for improving the accuracy of holiday load forecasting. When preprocessing data for the proposed method, the similarity to previous weather and a sensitivity analysis using dominant weather features are used to select and adjust historical weekday data. The dominant weather feature is identified by evaluating mutual information between various weather features and loads from season to season. The results of case studies are presented to show the effectiveness of the proposed method.
DSM can be used by electricity companies to reduce the system peak load at a particular time and by an amount specified by the utility. For its efficient operation, a DSM planning method is necessary. This dissertation focuses on a planning technique for incentive-based DSM program. In particular, the estimation of the operating days, time duration, and total capacity required for incentive-based DSM program to meet the target peak load determined by the utility is studied. This dissertation describes a systematic planning method for incentive-based DSM program using load forecasting with stochastic simulation. The stochastic temperature time series obtained from weather derivatives is used for the stochastic simulation. To improve the accuracy of load forecasting in the proposed method, a temperature sensitivity analysis is applied. The generalized extreme value distribution is also proposed for estimating stochastic results. Several numerical tests show the feasibility and efficiency of the proposed method.
The methods for holiday load forecasting and incentive-based DSM program planning proposed in this dissertation are expected to provide practical assistance for the efficient planning and operation of power systems

목차 (Table of Contents)

• TABLE OF CONTENTS
• ABSTRACT I
• TABLE OF CONTENTS IV
• LIST OF TABLES VI
• LIST OF FIGURES VII
• CHAPTER 1. INTRODUCTIONS 1
• 1.1 Motivation and Objective of Dissertation 1
• 1.2 Literuture Reviews 5
• 1.2.1 Load Forecasting 5
• 1.2.2 Demand Side Management 7
• 1.3 Contribution of Dissertation 11
• 1.4 Outline of Dissertation 13
• CHAPTER 2. FUZZY POLYNOMIAL REGRESSION MODEL FOR HOLIDAY LOAD FORECASTING WITH WEATHER FEATURE SELECTION AND ADJUSTMENT 14
• 2.1 Mahalanobis Distance-based Weekday Data Selection and Weather Adjustment for Holiday Load Forecasting 15
• 2.1.1 Weather Feature Selection Using Mutual Information 15
• 2.1.2 Mahalanobis Distance-based Weekday Data Selection 16
• 2.1.3 Adjustment of Weekday Data Using Weather Sensitivity Analysis 17
• 2.2 Fuzzy Polynomial Regression for Holiday Load Forecasting 21
• CHAPTER 3. STOCHASTIC PLANNING METHOD FOR THE INCENTIVE-BASED DEMAND SIDE MANAGEMENT PROGRAM USING LOAD FORECASTING WITH WEATHER SENSITIVITY ANALYSIS 26
• 3.1 Load Forecasting with Stochastic Simulation 27
• 3.1.1 Historical Data Acquisition in Accordance with User-preference Criteria 27
• 3.1.2 Reference Value Forecasting 27
• 3.1.3 Temperature Stochastic Modeling 29
• 3.1.4 Temperature Adjustment Using Weather Sensitivity Analysis 32
• 3.1.5 24-Hour Load Forecasting 33
• 3.2 Probabilistic Estimation of Operating Days, Time Duration, and Total Capacity for the Incentive-based Demand Side Management Program 35
• CHAPTER 4. NUMERICAL RESULTS 37
• 4.1 Case Study I: Holiday Load Forecastimg 38
• 4.2 Case Study II: Incentive-based Demand Side Management Program Planning 44
• CHAPTER 5. CONCLUSIONS 58
• REFERENCES 60
• KOREAN ABSTRACT 65