Short-term load forecasting with local ANN predictors

A new technique for artificial neural network (ANN) based short-term load forecasting (STLF) is presented in this paper. The technique implemented active selection of training data, employing the k-nearest neighbors concept. A novel concept of pilot simulation was used to determine the number of hidden units for the ANNs. The ensemble of local ANN predictors was used to produce the final forecast, whereby the iterative forecasting procedure used a simple average of ensemble ANNs. Results obtained using data from two US utilities showed forecasting accuracy comparable to those using similar techniques. Excellent forecasts for one-hour-ahead and five-days-ahead forecasting, robust behavior for sudden and large weather changes, low maximum errors and accurate peak-load predictions are some of the findings discussed in the paper     

An adaptive modular artificial neural network hourly loadforecaster and its implementation at electric utilities

This paper describes a modular artificial neural network (ANN) based hourly load forecaster which has already been implemented at 20 electric utilities across the US and is being used on-line by several of them. The behavior or the load and its correlation with parameters affecting it (e.g. weather variables) are decomposed into three distinct trends of weekly, daily, and hourly. Each trend is modeled by a separate module containing several multi-layer feed-forward ANNs trained by the back-propagation learning rule. The forecasts produced by each module are then combined by adaptive filters to arrive at the final forecast. During the forecasting phase, the parameters of the ANNs within each module are adaptively changed in response to the system's latest forecast accuracy. The performance of the forecaster has been tested on data from these 20 utilities with excellent results. The on-line performance of the system has also been quite satisfactory and superior to other forecasting packages used by the utilities. Moreover, the forecaster is robust, easy to use, and produces accurate results in the case of rapid weather changes     

Cascaded artificial neural networks for short-term load forecasting

An application of artificial neural networks (ANNs) to short-term load forecasting is presented in this paper. An algorithm using cascaded learning together with historical load and weather data is proposed to forecast half-hourly power system load for the next 24 hours. This cascaded neural network algorithm (CANNs) includes peak, minimum and daily energy prediction as additional input data for the final forecast stage. These additional input data are predicted using the first (ANNs) model. The networks are trained and tested on the electric power system of Kuwait. The absolute average forecasting error is reduced from 3.367% to 2.707% by applying CANNs as compared to the conventional ANNs. Simulation results indicate that the developed forecasting approach is effective and point to the potential of the methodology for economic applications     

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。     

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.     

Short-term electric load forecasting based on Kalman filtering algorithm with moving window weather and load model

This paper presents a novel time-varying weather and load model for solving the short-term electric load-forecasting problem. The model utilizes moving window of current values of weather data as well as recent past history of load and weather data. The load forecasting is based on state space and Kalman filter approach. Time-varying state space model is used to model the load demand on hourly basis. Kalman filter is used recursively to estimate the optimal load forecast parameters for each hour of the day. The results indicate that the new forecasting model produces robust and accurate load forecasts compared to other approaches. Better results are obtained compared to other techniques published earlier in the literature.     

Short-term load forecasting using neural networks

An artificial neural network (ANN) model for short-term load forecasting (STLF) is presented. The proposed model is capable of forecasting the next 24-hour load profile at one time, as opposed to the usual ‘next one hour’ ANN models. The inputs to the ANN are load profiles of the two previous days and daily maximum and minimum temperature forecasts. The network is trained to learn the next day's load profile. Testing of the model with one year of data from the Greek interconnected power system resulted in a 2.66% average absolute forecast error.     

基于改进型自适应白噪声完备集成经验模态分解的工业用户负荷预测方法

工业用户的负荷通常由多种负荷类型共同组成,结构较为复杂,并且常常含有较大的冲击性负荷.传统的负荷预测方法难以准确预测负荷突变,导致预测精度不高.将负荷分解成不同频率的分量再分别进行预测是较为可行的解决方式.提出了基于改进型自适应白噪声完备集成经验模态分解的工业用户负荷预测方法.首先,采用ICEEMDAN算法将工业用户的...     

A real-time short-term load forecasting system using functionallink network

This paper presents a new functional-link network based short-term electric load forecasting system for real-time implementation. The load and weather parameters are modelled as a nonlinear ARMA process and parameters of this model are obtained using the functional approximation capabilities of an auto-enhanced functional link net. The adaptive mechanism with a nonlinear learning rule is used to train the link network on-line. The results indicate that the functional link net based load forecasting system produces robust and more accurate load forecasts in comparison to simple adaptive neural network or statistical based approaches. Testing the algorithm with load and weather data for a period of two years reveals satisfactory performance with mean absolute percentage error (MAPE) mostly less than 2% for a 24-hour ahead forecast and less than 2.5% for a 168-hour ahead forecast     

基于人工神经网络和模糊集的电力系统短期负荷预测方法

综合考虑到温度、日期类型和天气等因素对短期电力负荷的影响,提出了一种将人工神经网络(ANN)RBF模型和模糊逻辑相结合的短期负荷预测方法.该方法将电力负荷分为周期性的基本负荷和受多种因素影响的变动负荷两部分,对于周期负荷用ANN进行预测,采用负荷预测中比较精确的RBF算法;变动负荷采用模糊逻辑对天气因素、温度、日期类型分别做不同的模糊处理,然后利用模糊推理规则对基本负荷预测结果进行修正.通过典型算例与普通BP法预测结果相比较,结果表明该方法具有较高的预测精度.     

Enhancement, implementation, and performance of an adaptiveshort-term load forecasting algorithm

An adaptive load forecasting algorithm that was originally developed for the one-hour time period has been extensively enhanced and implemented. The major enhancement is the ability to forecast total system hourly load as far ahead as five days. The focus is on the enhancements and implementation of the algorithm. The purpose is to describe the final form of the forecasting model and the overall forecasting procedure; the procedure for utilizing minimum/maximum daily weather forecasts made by the US Weather Bureau: the offline forecasting accuracy based on four years of historical hourly load and weather data; and the implementation of the algorithm on a desktop computer     

Risk quantification for ANN based short‐term load forecasting

A new risk assessment method for short‐term load forecasting is proposed. The proposed method makes use of an artificial neural network (ANN) to forecast one‐step‐ahead daily maximum loads and evaluate uncertainty of load forecasting. With ANN as the model, the radial basis function (RBF) network is employed to forecast loads due to its good performance. Sufficient realistic pseudo‐scenarios are required to carry out quantitative risk analysis. The multivariate normal distribution with the correlation between input variables is used to give more realistic results to ANN. In addition, the method of moment matching is used to improve the accuracy of the multivariate normal distribution. The peak over threshold (POT) approach is used to evaluate risk that exceeds the upper bounds of generation capacity. The proposed method is successfully applied to real data of daily maximum load forecasting. © 2008 Wiley Periodicals, Inc. Electr Eng Jpn, 166(2): 54– 62, 2009; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.20464     

Short-term load forecasting for special days in anomalous loadconditions using neural networks and fuzzy inference method

Conventional artificial neural network (ANN) based short-term load forecasting techniques have limitations in their use on holidays. This is due to dissimilar load behaviors of holidays compared with those of ordinary weekdays during the year and to insufficiency of training patterns. The purpose of this paper is to propose a new short-term load forecasting method for special days in anomalous load conditions. These days include public holidays, consecutive holidays, and days preceding and following holidays. The proposed method uses a hybrid approach of ANN based technique and fuzzy inference method to forecast the hourly loads of special days. In this method, special days are classified into five different day-types. Five ANN models for each day-type are used to forecast the scaled load curves of special days, and two fuzzy inference models are used to forecast the maximum and the minimum loads of special days. Finally, the results of the ANN and the fuzzy inference models are combined to forecast the 24 hourly loads of special days. The proposed method was tested with actual load data of special days for the years of 1996-1997. The test results showed very accurate forecasting with the average percentage relative error of 1.78%     

Input variable selection for ANN-based short-term load forecasting

This paper describes a novel method for input variable selection for artificial neural network (ANN) based short-term load forecasting (STLF). The method is based on the phase-space embedding of a load time-series. The accuracy of the method is enhanced by the addition of temperature and cycle variables. To test the viability of the method, real load data for two US-based electric utilities were used. Only 15 input variables were identified in both cases and used for 24-hour ahead load forecasting. Results compare favorably to the ones reported in the literature, indicating that more parsimonious set of input variables can be used in STLF without sacrificing the accuracy of the forecast. This allows more compact ANNs, smaller training sets and easier training. Consequently, the method represents a step forward in determining a general procedure for input variable selection for ANN-based STLF     

Smooth approximation and forecasting of load duration curves

The purpose of this paper is to develop a methodology for forecasting a load duration curve. The approach adopted in the development is to estimate a load duration curve as a combined linear-exponential function and relate the coefficients of this function to a set of economic and weather related variables. This relationship allows one to forecast the coefficients which are in turn used to provide a forecast of the coefficients of interest.The model presented, for a specific set of data, forecasts well and presents a substantial improvement over existing approaches to obtaining future load duration curves.     

Results of Egyptian unified grid hourly load forecasting using an artificial neural network with expert system interface

This paper presents the hourly load forecasting results of the Egyptian unified grid (EUG). The technique is based on a generalized model combining the features of ANN and an expert system. The above methodology makes the technique robust, updatable and provides for operator intervention when necessary. This property makes it especially suitable for the EUG where the load patterns are influenced mostly because of social activities, and weather contributes very little to load forecast. For example, many social occasions depend on religious preferences which cannot be decided well in advance.This technique has been tested with one year data of EUG during 1993. The results clearly demonstrate the advantage of the above methodology over statistical based techniques. The average absolute forecast errors for the proposed methodology is 2.63% with a standard deviation of 2.62% whereas, the conventional multiple regression method scores an average absolute error of 4.69% with a standard deviation of 4.03%.     

An adaptable automated procedure for short-term electricity loadforecasting

The Irish Electricity Supply Board requires forecasts of system demand or electrical load for: (a) one day ahead; and (b) 7-10 days ahead. Here, the authors concentrate on and give results only for one day ahead forecasts although the method is also applicable for 7-10 days ahead. A forecasting model has been developed which identifies a `normal' or weather-insensitive load component and a weather-sensitive load component. Linear regression analysis of past load and weather data is used to identify the normal load model. The weather-sensitive component of the load is estimated using the parameters of regression analysis. Certain design features of the short-term load forecasting system are important for its successful operation over time. These include adaptability to changing operational conditions, computational economy and robustness. An automated load forecasting system is presented here that includes these design features. A fully automated algorithm for updating the model is described in detail as are the techniques employed in both the identification and treatment of influential points in the data base and the selection of predictors for the weather-load model. Monthly error statistics of forecast load for only one day ahead are presented for recorded weather conditions     

以神经网络与模糊逻辑互补的电力系统短期负荷预测模型及方法

根据电力系统短期负荷预测的特点,采用神经网络与模糊逻辑互补的方法建立了负荷预测模型.通过粗糙集理论中的信息熵概念对神经网络的输入参数进行了筛选,以与待预测量相关性大的参数作为输入,不仅减少了神经网络的工作量,缩短了计算时间,而且提高了预测的准确性;在神经网络中,通过引进动量系数和遗忘系数优化网络,提高了ANN的收敛速度;在模糊逻辑中,充分利用了人们对负荷变化取得的主观经验,引进不平均隶属函数,来反映负荷对温度的敏感性.     

基于多模型随机组合的水文集合预报方法研究

准确、可靠的水文预报是水资源开发利用的基础.集合预报以概率或区间的形式表征预报的不确定性,是未来水文预报研究的重点发展方向.本文提出了一种基于多模型随机组合的水文集合预报方法.首先通过加权形式将多种预报模型进行组合;再采用多目标优化算法率定各成员模型权重的上、下限;最后在优化的上、下限内随机生成权重以构建集合预报.以汉...     

Short-term load forecasting using an artificial neural network

An artificial neural network (ANN) method is applied to forecast the short-term load for a large power system. The load has two distinct patterns: weekday and weekend-day patterns. The weekend-day pattern includes Saturday, Sunday, and Monday loads. A nonlinear load model is proposed and several structures of an ANN for short-term load forecasting were tested. Inputs to the ANN are past loads and the output of the ANN is the load forecast for a given day. The network with one or two hidden layers was tested with various combinations of neurons, and results are compared in terms of forecasting error. The neural network, when grouped into different load patterns, gives a good load forecast