采用谱分析建模和基于人工神经网络的短期负荷预测方案

提出了一种基于谱分析法进行建模的短期负荷预测方案,该方案利用负荷历史数据的谱分析结果进行人工神经网络(ANN)模式分类和选择输入变量.方案采用快速傅立叶变换(FFT)进行负荷数据预处理,运用滤波算法及小时负荷曲线的频谱分析来研究电网负荷的周期特性,所得结果表明四季负荷的谱特性具有明显差异,应采用不同的模型和方案进行预测.谱分析有助于各时段预测方案提取输入变量.利用该思路构造的基于人工神经网络的负荷预测方案被用于预测广东省网的负荷,与其他普遍采用的输入变量预测结果的对比表明,所提方案在短期负荷预测中的性能良好.     

Recurrent neural networks for short-term load forecasting

Forecasting the short-term load entails the construction of a model, and, using the information available, estimating the parameters of the model to optimize the prediction performance. It follows that the more closely the chosen model approximates the actual physical generating process, the higher the expected performance of the forecasting system. In this paper it is postulated that the load can be modeled as the output of some dynamic system, influenced by a number of weather, time and other environmental variables. Recurrent neural networks, being members of a class of connectionist models exhibiting inherent dynamic behavior, can thus be used to construct empirical models for this dynamic system. Because of the nonlinear dynamic nature of these models, the behavior of the load prediction system can be captured in a compact and robust representation. This is illustrated by the performance of recurrent models on the short-term forecasting of the nation-wide load for the South African utility, ESKOM. A comparison with feedforward neural networks is also given     

Unit commitment using hybrid models: a comparative study for dynamic programming, expert system, fuzzy system and genetic algorithms

Hybrid models for solving unit commitment problem have been proposed in this paper. To incorporate the changes due to the addition of new constraints automatically, an expert system (ES) has been proposed. The ES combines both schedules of units to be committed based on any classical or traditional algorithms and the knowledge of experienced power system operators. A solution database, i.e. information contained in the previous schedule is used to facilitate the current solution process. The proposed ES receives the input, i.e. the unit commitment solutions from a fuzzy-neural network. The unit commitment solutions from the artificial neural network cannot offer good performance if the load patterns are dissimilar to those of the trained data. Hence, the load demands, i.e. the input to the fuzzy-neural network is considered as fuzzy variables. To take into account the uncertainty in load demands, a fuzzy decision making approach has also been developed to solve the unit commitment problem and to train the artificial neural network. Due to the mathematical complexity of traditional techniques for solving unit commitment problem and also to facilitate comparison genetic algorithm, a non-traditional optimization technique has also been proposed. To demonstrate the effectiveness of the models proposed, extensive studies have been performed for different power systems consisting of 10, 26 and 34 generating units. The generation cost obtained and the computational time required by the proposed model has been compared with the existing traditional techniques such as dynamic programming (DP), ES, fuzzy system (FS) and genetic algorithms (GA).     

Feature extraction of one‐step‐ahead daily maximum load with regression tree

In this paper, a new efficient feature extraction method is proposed to handle the one‐step‐ahead daily maximum load forecasting. In recent years, power systems become more complicated under the deregulated and competitive environment. As a result, it is not easy to understand the cause and effect of short‐term load forecasting with a bunch of data. This paper analyzes load data from the standpoint of data mining. By it we mean a technique that finds out rules or knowledge through large database. As a data mining method for load forecasting, this paper focuses on the regression tree that handles continuous variables and expresses a knowledge rule as if‐then rules. Investigating the variable importance of the regression tree gives information on the transition of the load forecasting models. This paper proposes a feature extraction method for examining the variable importance. The proposed method allows to classify the transition of the variable importance through actual data. © 2006 Wiley Periodicals, Inc. Electr Eng Jpn, 156(2): 43–51, 2006; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.20104     

用自适应模糊推理系统预测电力短期负荷

为寻求有效的电力系统负荷预测方法以提高预测结果的准确度,提出了基于Takagi-Sugeno(T-S)模型的自适应神经模糊推理系统(ANFIS)。该系统采用减法聚类初始化模糊推理,把神经网络学习机制引入到逻辑推理中,并用混合学习算法调整前件参数和结论参数,自动产生模糊规则。考虑气象、日期类型等因素后将学习样本分为3组进行训练和检测。该方法对于受天气影响较明显的电网,能有效防止不合理预测结果的出现。对于武汉地区实际负荷的预测结果的分析表明该方法有较高的预测准确度,取得了令人满意的结果。     

基于神经网络的电力系统短期负荷预测研究

电力系统负荷预测是电力生产部门的重要工作之一，作者利用BP神经网络进行电力系统短期负荷预测，在保证有足够的训练样本的前提下，对预测模型进行合理分类，构造了相应于不同季节的周预测，日预测模型，并对输入变量的选择，特别是温度的选取问题，进行了讨论，在神经网络训练的过程中，往往会出现过拟合的现象，给预测的结果带来不利的影响，为此在训练过程中，将样本随机地分离为训练集和测试集来防止这个问题，典型算例的计算表明，该方法是有效的。     

基于专家评分机制的模糊综合评判法在电力系统负荷预测中的应用

针对目前电力系统单一负荷预测模型的不足之处,研究了结合各种单一预测模型优点组合预测方法。对基于模糊综合评判的组合预测模型进行了改进,从不同评价因素出发,把专家评分机制和模糊隶属函数结合起来用于各个单项模型的综合评价,可以充分利用专家的知识和预测人员的经验,有效地处理电力系统负荷预测的不确定性和模糊性,提高电力系统负荷预测的精度。最后用一个某地区的实例验证了该方案的可行性和正确性。     

神经网络短期负荷预测的输入变量选择研究

短期负荷预测中输入变量的选择直接关系到神经网络的预测性能。本文将自相关函数的概念应用于神经网络短期负荷预测中的输入变量集选择,对输入变量集的选择提出了一种比较科学系统的方法。通过采用FFT来实现对自相关函数的快速计算,增加了该方法的可操作性,并通过具体的实例验证了该方法的有效性。     

月前日负荷曲线的概率预测和随机场景模拟

针对现有中长期日负荷曲线预测方法大多为点预测,难以满足电力系统不确定性分析的不足,提出了一种基于因子分析和神经网络分位数回归的月前日负荷曲线概率预测和随机场景模拟方法。采用因子分析技术,在保留日内负荷时序相关性的前提下,对日负荷序列向量降维;提取出少数相互独立的负荷公共因子作为预测变量,以日气象因素、星期类型和前一日公共因子值为输入特征,建立计及相邻日负荷相关性的神经网络分位数回归概率预测模型;以此为基础,利用中期气象预报信息,逐日预测和模拟未来30日的负荷曲线,并生成未来月负荷曲线的随机模拟场景。实际算例结果验证了所提概率预测方法的准确性和高效性,其生成的日负荷曲线模拟场景更好地体现了负荷的时序相关性,能为调度人员提供更准确、全面的月前负荷预测信息。     

电力系统中长期负荷预测软件包的开发

作者开发了基于Windows操作平台的电力系统中长期负荷预测软件包。该软件包分为原始数据处理、负荷预测、预测结果处理三个模块，各模块相互独立，易于扩充。在该软件包预测结果处理三个模块，各模块相互独立，易于扩充。在该软件包建立的负荷预测的模型库具有分析、计算快速，检验方法多、结果输出形式多样的特点。模型库具有智能化的专家系统，可以针对不同地区的具体情况进行调整，适用范围广泛。最后以某地区的需电量历史数据为例，选择了5种不同模型进行了预测，并与综合模型的预测结果进行了分析和比较。分析表明该软件包人机界面友好、选用模型合理、预测结果精度较高，能够满足当今快速发展的电力市场对于电力负荷预测的精度的要求，并解决了当前电力系统中一些负荷预测软件可操作性不强，模型适用范围小的问题。     

电力系统短期负荷预测模型研究

给出电力系统短期负荷预测的两种模型——变差分析模型和年度分解模型。变差分析模型将负荷分解为基准值、年度变差、月份变差以及随机变差,通过对其分别估计便可得到负荷的预测值;年度分解法先通过预测年度负荷,再预测每月的负荷贡献率,即可得到负荷预测值。实例分析表明,这两种方法都是有效的、实用的     

Medium term system load forecasting with a dynamic artificial neural network model

This paper presents the development of a dynamic artificial neural network model (DAN2) for medium term electrical load forecasting (MTLF). Accurate MTLF provides utilities information to better plan power generation expansion (or purchase), schedule maintenance activities, perform system improvements, negotiate forward contracts and develop cost efficient fuel purchasing strategies. We present a yearly model that uses past monthly system loads to forecast future electrical demands. We also show that the inclusion of weather information improves load forecasting accuracy. Such models, however, require accurate weather forecasts, which are often difficult to obtain. Therefore, we have developed an alternative: seasonal models that provide excellent fit and forecasts without reliance upon weather variables. All models are validated using actual system load data from the Taiwan Power Company. Both the yearly and seasonal models produce mean absolute percent error (MAPE) values below 1%, demonstrating the effectiveness of DAN2 in forecasting medium term loads. Finally, we compare our results with those of multiple linear regressions (MLR), ARIMA and a traditional neural network model.     

基于模糊粗糙集和神经网络的短期负荷预测方法

针对采用神经网络进行电力系统短期负荷预测时其网络输入变量的选择是影响预测效果的关键问题，该文提出使用模糊粗糙集理论解决这一问题：对采集到的信息进行特征提取、形成决策表；利用模糊粗糙集理论进行属性约简、去除冗余信息；用得到的属性作为BP网络的输入进行训练预测。该方法既全面考虑了影响负荷预测的历史时间序列、气象等各种因素，为合理地选择神经网络的输入变量提供了一种新的方法，又避免了由于输入变量过多而导致神经网络拓扑结构复杂、训练时间长等不足。计算实例表明，文中提出的方法是有效且可行的。     

基于最大信息系数和多目标Stacking集成学习的综合能源系统多元负荷预测

精确的多元负荷预测对于综合能源系统的能源调度与运行规划起到重要的作用.对电、热、冷负荷单独进行预测的传统方法会忽略多元负荷间的耦合关系.针对这一问题,提出一种基于多目标Stacking集成学习的多元负荷协同预测模型.引入最大信息系数对多元负荷及天气因素进行相关性分析,并提出负荷耦合形态指标来深度挖掘多元负荷间的耦合关系...     

数据挖掘与非正常日的负荷预测

提高非正常日的负荷预测精度是当前负荷预测工作的难点。文中提出了一种基于知识库的事先判别突变并做出适当处理的预测流程,介绍了利用数据挖掘的决策树技术建立知识库的方法,并给出了几种典型的非正常日修正模型。最后,通过对长时期负荷预测数据的统计分析,说明了新方法的有效性和实用性。     

短期负荷预测中实时气象因素的影响分析及其处理策略

短期负荷预测对于电力系统安全经济运行有着重要的作用,因此,人们一直致力于研究新的预测模型,提高预测精度。目前,实现提高预测精度这个目标的关键是如何更加合理地考虑气象因素对负荷的影响,因为气象敏感负荷在总负荷中所所占的比重越来越大。长期以来,鉴于气象部门无法提供实时温度等气象预测结果,电力系统所建立的预测模型绝大多数都是基于日特征气象因素,诸如日最高温度、最低温度等。针对短期负荷预测,作者剖析了气象因素的影响和作用,分析了处理不同阶段气象因素的策略,并提出了考虑实时气象因素的短期负荷预测新模型,该模型基于神经网络,力图寻求温度、湿度等实时气象因素与负荷曲线之间的相关关系和变化规律。实际应用表明,文中的预测模型和处理策略可以得到更加精确的预测结果。此短期负荷预测新模型也适用于超短期负荷预测。     

基于神经网络的短期负荷预测

针对电力系统短期负荷的变化与影响因素间的复杂非线性关系,首先,提出用BP神经网络进行负荷预测,接着,在输入变量的选择上引入了负荷日期和气象温度,对于日期变量分为工作日和休息日,对于气温变量进行分段处理。最后通过实例仿真表明该方法可以取得较高的预测精度。     

基于回归分析与神经网络的短期负荷预测

针对电力系统短期负荷预测中神经网络输入变量选择与网络训练问题,提出了一种基于回归分析与神经网络相结合的短期负荷预测方法,利用回归分析选择神经网络的输入变量,利用遗传算法训练神经网络.实例研究结果表明该方法可以取得较高的预测精度.     

Application of switched adaptive system to load forecasting

In this paper, we propose an approach for next day peak load forecasting for electrical companies. First a nonlinear model for the peak load is proposed taking into account the historical load and the temperature. Based on this model time-varying local models are obtained for some temperature intervals. The peak load forecasting system is constructed based on these local models which parameters are estimated using an on-line recursive algorithm. We remark that in this methodology it is not necessary to know precisely the temperature of the days since the proposed system is based on an interval for the future temperature instead of a number. An application example illustrates the proposed approach.     

An explainable framework for load forecasting of a regional integrated energy system based on coupled features and multi-task learning

To extract strong correlations between different energy loads and improve the interpretability and accuracy for load forecasting of a regional integrated energy system (RIES), an explainable framework for load forecasting of an RIES is proposed. This includes the load forecasting model of RIES and its interpretation. A coupled feature extracting strat egy is adopted to construct coupled features between loads as the input variables of the model. It is designed based on multi-task learning (MTL) with a long short-term memory (LSTM) model as the sharing layer. Based on SHapley Additive exPlanations (SHAP), this explainable framework combines global and local interpretations to improve the interpretability of load forecasting of the RIES. In addition, an input variable selection strategy based on the global SHAP value is proposed to select input feature variables of the model. A case study is given to verify the effectiveness of the proposed model, constructed coupled features, and input variable selection strategy. The results show that the explainable framework intuitively improves the interpretability of the prediction model.