基于人工神经网络与支持向量机的负荷预测比较研究
发布时间:2021-01-02 05:04
电力负荷预测已成为电力工程的重要研究内容之一,其是一个复杂的多变量、多维度的估计问题。智能电表及传感器的应用,提供了不同形式分布式能源的大量历史数据,进一步增加了问题的复杂性。传统的负荷预测方法无法准确跟踪负荷随机变化,准确性较差,但基于人工智能的机器学习算法因具有数据建模的灵活性和精确性,具有提高负荷预测准确性的潜力。本文给出了人工神经网络(ANN)短期预测方法和支持向量机(SVM)的短期负荷预测方法。利用支持向量机的核函数对负荷预测中的复杂非线性关系进行建模,建立了多项式核、无线基函数核和皮尔逊函数核函数的三种支持向量机模型。本文比较了支持向量机模型采用不同核函数时的性能,明确了适用于负荷预测的最佳核函数。通过软件WEKA的仿真结果验证了所提方法的性能,确立了支持向量机参数、核的优化以及神经网络模型的参数。基于上述任务,本文收集了巴基斯坦配电网6周的历史数据,并对该数据进行预处理,删除了丢失值和异常值,使数据规范化以获得更好的性能;再利用所得数据对支持向量机模型和神经网络模型进行检验,并基于错误度量指标比较了基于不同核函数的支持向量机模型与ANN模型的负荷预测优劣。研究结果表明,基...
【文章来源】:华北电力大学(北京)北京市 211工程院校 教育部直属院校
【文章页数】:78 页
【学位级别】:硕士
【文章目录】:
摘要
ABSTRACT
CHAPTER 1 INTRODUCTION
1.1 MACHINE LEARNING AND DATA ANALYTICS FOR POWER GRIDS
1.2 RESEARCH STATUS OF LOAD FORECASTING BY ML METHODS
1.3 MOTIVATION TOWARDS TOPIC
1.4 THESIS OBJECTIVE
1.5 SOME NATURAL QUESTIONS
1.6 SOLUTION OVERVIEW
1.7 CONTRIBUTION OF THIS THESIS
1.8 THESIS OUTLINE
CHAPTER 2 LOAD FORECASTING AND MACHINE LEARNING-LITERATURE REVIEW
2.1 IMPORTANCE OF FORECASTING IN UTILITIES
2.2 LOAD FORECASTING
2.2.1 Very Short-Term Load Forecasting(VSTLF):
2.2.2 Short-Term Load Forecasting (STLF):
2.2.3 Medium Term Load Forecasting (MTLF):
2.2.4 Long-Term Load Forecasting(LTLF):
2.3 SHORT-TERM LOAD FORECASTING
2.4 STATISTICAL METHODE
2.5 MACHINE LEARNING METHODS
2.6 SUMMART
CHAPTER 3 RESEARCH MODELING
3.1 DATA COLLECTION
3.2 ATTRIBUTE SELECTION
3.3 DATA PREPARATION
3.4 LIMITATIONS
3.5 PREPROCESSING OF THE DATASET
3.6 NORMALIZATION
3.7 DATA FORECASTING MODELS:
3.8 MODEL PARAMETERS FOR SVR AND ANN
3.9 ERROR METRICS
3.9.1 Popular Error Metrics
3.9.2 Mean Absolute Error (MAE)
3.9.3 Mean Absolute Percentage Error (MAPE)
3.9.4 Root Mean Squared Error (RMSE)
3.9.5 Mean Squared Error (MSE)
3.10 WEKA
3.11 SUMMARY
CHAPTER 4 LOAD FORECASTING MODEL BASED ON SVM
4.1 SUPPORT VECTOR MACHINES
4.2 SUPPORT VECTOR CLASSIfiCATION
4.3 SVC FOR LINEARLY SFPARABLE SET
4.4 SVC FOR NON-LINEARLY SEPARABLE SETS
4.5 SUPPORT VECTOR REGRESSION
4.6 KERNEL FUNCTIONS
4.7 PEARSON VII KERNEL
4.8 EXPERIMENTATION FOR KERNEL SELECTION
4.9 SUMMARY
CHAPTER 5 COMPARISON OF LOAD FORECASTING MODELS
5.1 ARTIFICIAL NEURAL NETWORK
5.1.1 Artificial Neurons
5.1.2 Layers of Neurons
5.1.3 Backpropagation
5.1.4 Multi-Layer Perceptron
5.2 EVALUATION OF THE MODELS
5.2.1 ANN Model
5.2.2 SVR Model
5.3 PERFORMANCE COMPARISON
5.4 SUMMARY
CHAPTER 6 CONCLUSIONS AND FUTURE WORKS
6.1 CONCLUSIONS
6.2 FUTURE WORKS
ACKNOWLEDGEMENT
CHAPTER 7 REFERENCES
本文编号:2952708
【文章来源】:华北电力大学(北京)北京市 211工程院校 教育部直属院校
【文章页数】:78 页
【学位级别】:硕士
【文章目录】:
摘要
ABSTRACT
CHAPTER 1 INTRODUCTION
1.1 MACHINE LEARNING AND DATA ANALYTICS FOR POWER GRIDS
1.2 RESEARCH STATUS OF LOAD FORECASTING BY ML METHODS
1.3 MOTIVATION TOWARDS TOPIC
1.4 THESIS OBJECTIVE
1.5 SOME NATURAL QUESTIONS
1.6 SOLUTION OVERVIEW
1.7 CONTRIBUTION OF THIS THESIS
1.8 THESIS OUTLINE
CHAPTER 2 LOAD FORECASTING AND MACHINE LEARNING-LITERATURE REVIEW
2.1 IMPORTANCE OF FORECASTING IN UTILITIES
2.2 LOAD FORECASTING
2.2.1 Very Short-Term Load Forecasting(VSTLF):
2.2.2 Short-Term Load Forecasting (STLF):
2.2.3 Medium Term Load Forecasting (MTLF):
2.2.4 Long-Term Load Forecasting(LTLF):
2.3 SHORT-TERM LOAD FORECASTING
2.4 STATISTICAL METHODE
2.5 MACHINE LEARNING METHODS
2.6 SUMMART
CHAPTER 3 RESEARCH MODELING
3.1 DATA COLLECTION
3.2 ATTRIBUTE SELECTION
3.3 DATA PREPARATION
3.4 LIMITATIONS
3.5 PREPROCESSING OF THE DATASET
3.6 NORMALIZATION
3.7 DATA FORECASTING MODELS:
3.8 MODEL PARAMETERS FOR SVR AND ANN
3.9 ERROR METRICS
3.9.1 Popular Error Metrics
3.9.2 Mean Absolute Error (MAE)
3.9.3 Mean Absolute Percentage Error (MAPE)
3.9.4 Root Mean Squared Error (RMSE)
3.9.5 Mean Squared Error (MSE)
3.10 WEKA
3.11 SUMMARY
CHAPTER 4 LOAD FORECASTING MODEL BASED ON SVM
4.1 SUPPORT VECTOR MACHINES
4.2 SUPPORT VECTOR CLASSIfiCATION
4.3 SVC FOR LINEARLY SFPARABLE SET
4.4 SVC FOR NON-LINEARLY SEPARABLE SETS
4.5 SUPPORT VECTOR REGRESSION
4.6 KERNEL FUNCTIONS
4.7 PEARSON VII KERNEL
4.8 EXPERIMENTATION FOR KERNEL SELECTION
4.9 SUMMARY
CHAPTER 5 COMPARISON OF LOAD FORECASTING MODELS
5.1 ARTIFICIAL NEURAL NETWORK
5.1.1 Artificial Neurons
5.1.2 Layers of Neurons
5.1.3 Backpropagation
5.1.4 Multi-Layer Perceptron
5.2 EVALUATION OF THE MODELS
5.2.1 ANN Model
5.2.2 SVR Model
5.3 PERFORMANCE COMPARISON
5.4 SUMMARY
CHAPTER 6 CONCLUSIONS AND FUTURE WORKS
6.1 CONCLUSIONS
6.2 FUTURE WORKS
ACKNOWLEDGEMENT
CHAPTER 7 REFERENCES
本文编号:2952708
本文链接:https://www.wllwen.com/kejilunwen/dianlidianqilunwen/2952708.html