Medium-Term Electric Load Forecasting Using Multivariable Linear and Non-Linear Regression

Medium-term forecasting is an important category of electric load forecasting that covers a time span of up to one year ahead. It suits outage and maintenance planning, as well as load switching operation. We propose a new methodol-ogy that uses hourly daily loads to predict the next year hourly loads, and hence predict the peak loads expected to be reached in the next coming year. The technique is based on implementing multivariable regression on previous year's hourly loads. Three regression models are investigated in this research: the linear, the polynomial, and the exponential power. The proposed models are applied to real loads of the Jordanian power system. Results obtained using the pro-posed methods showed that their performance is close and they outperform results obtained using the widely used ex-ponential regression technique. Moreover, peak load prediction has about 90% accuracy using the proposed method-ology. The methods are generic and simple and can be implemented to hourly loads of any power system. No extra in-formation other than the hourly loads is required.

Long Term Load Forecasting and Recommendations for China Based on Support Vector Regression

Long-term load forecasting (LTLF) is a challenging task because of the complex relationships between load and factors affecting load. However, it is crucial for the economic growth of fast developing countries like China as the growth rate of gross domestic product (GDP) is expected to be 7.5%, according to China’s 11th Five-Year Plan (2006-2010). In this paper, LTLF with an economic factor, GDP, is implemented. A support vector regression (SVR) is applied as the training algorithm to obtain the nonlinear relationship between load and the economic factor GDP to improve the accuracy of forecasting.

Long-Term Electrical Load Forecasting in Rwanda Based on Support Vector Machine Enhanced with Q-SVM Optimization Kernel Function

In recent years, Rwanda’s rapid economic development has created the “Rwanda Africa Wonder”, but it has also led to a substantial increase in energy consumption with the ambitious goal of reaching universal access by 2024. Meanwhile, on the basis of the rapid and dynamic connection of new households, there is uncertainty about generating, importing, and exporting energy whichever imposes a significant barrier. Long-Term Load Forecasting (LTLF) will be a key to the country’s utility plan to examine the dynamic electrical load demand growth patterns and facilitate long-term planning for better and more accurate power system master plan expansion. However, a Support Vector Machine (SVM) for long-term electric load forecasting is presented in this paper for accurate load mix planning. Considering that an individual forecasting model usually cannot work properly for LTLF, a hybrid Q-SVM will be introduced to improve forecasting accuracy. Finally, effectively assess model performance and efficiency, error metrics, and model benchmark parameters there assessed. The case study demonstrates that the new strategy is quite useful to improve LTLF accuracy. The historical electric load data of Rwanda Energy Group (REG), a national utility company from 1998 to 2020 was used to test the forecast model. The simulation results demonstrate the proposed algorithm enhanced better forecasting accuracy.

Long-Term Load Forecasting of Southern Governorates of Jordan Distribution Electric System

Load forecasting is vitally important for electric industry in the deregulated economy. This paper aims to face the power crisis and to achieve energy security in Jordan. Our participation is localized in the southern parts of Jordan including, Ma’an, Karak and Aqaba. The available statistical data about the load of southern part of Jordan are supplied by electricity Distribution Company. Mathematical and statistical methods attempted to forecast future demand by determining trends of past results and use the trends to extrapolate the curve demand in the future.

Medium Term Load Forecasting for Jordan Electric Power System Using Particle Swarm Optimization Algorithm Based on Least Square Regression Methods

This paper presents a technique for Medium Term Load Forecasting (MTLF) using Particle Swarm Optimization (PSO) algorithm based on Least Squares Regression Methods to forecast the electric loads of the Jordanian grid for year of 2015. Linear, quadratic and exponential forecast models have been examined to perform this study and compared with the Auto Regressive (AR) model. MTLF models were influenced by the weather which should be considered when predicting the future peak load demand in terms of months and weeks. The main contribution for this paper is the conduction of MTLF study for Jordan on weekly and monthly basis using real data obtained from National Electric Power Company NEPCO. This study is aimed to develop practical models and algorithm techniques for MTLF to be used by the operators of Jordan power grid. The results are compared with the actual peak load data to attain minimum percentage error. The value of the forecasted weekly and monthly peak loads obtained from these models is examined using Least Square Error (LSE). Actual reported data from NEPCO are used to analyze the performance of the proposed approach and the results are reported and compared with the results obtained from PSO algorithm and AR model.

Optimal Scheme with Load Forecasting for Demand Side Management (DSM) in Residential Areas

Utilities around the world have been considering Demand Side Management (DSM) in their strategic planning. The costs of constructing and operating a new capacity generation unit are increasing everyday as well as Transmission and distribution and land issues for new generation plants, which force the utilities to search for another alternatives without any additional constraints on customers comfort level or quality of delivered product. De can be defined as the selection, planning, and implementation of measures intended to have an influence on the demand or customer-side of the electric meter, either caused directly or stimulated indirectly by the utility. DSM programs are peak clipping, Valley filling, Load shifting, Load building, energy conservation and flexible load shape. The main Target of this paper is to show the relation between DSM and Load Forecasting. Moreover, it highlights on the effect of applying DSM on Forecasted demands and how this affects the planning strategies for utility companies. This target will be clearly illustrated through applying the developed algorithm in this paper on an existing residential compound in Cairo-Egypt.

Deep Learning Network for Energy Storage Scheduling in Power Market Environment Short-Term Load Forecasting Model

In the electricity market,fluctuations in real-time prices are unstable,and changes in short-term load are determined by many factors.By studying the timing of charging and discharging,as well as the economic benefits of energy storage in the process of participating in the power market,this paper takes energy storage scheduling as merely one factor affecting short-term power load,which affects short-term load time series along with time-of-use price,holidays,and temperature.A deep learning network is used to predict the short-term load,a convolutional neural network(CNN)is used to extract the features,and a long short-term memory(LSTM)network is used to learn the temporal characteristics of the load value,which can effectively improve prediction accuracy.Taking the load data of a certain region as an example,the CNN-LSTM prediction model is compared with the single LSTM prediction model.The experimental results show that the CNN-LSTM deep learning network with the participation of energy storage in dispatching can have high prediction accuracy for short-term power load forecasting.

Long-term system load forecasting based on data-driven linear clustering method

In this paper, a data-driven linear clustering(DLC) method is proposed to solve the long-term system load forecasting problem caused by load fluctuation in some developed cities. A large substation load dataset with annual interval is utilized and firstly preprocessed by the proposed linear clustering method to prepare for modelling.Then optimal autoregressive integrated moving average(ARIMA) models are constructed for the sum series of each obtained cluster to forecast their respective future load. Finally, the system load forecasting result is obtained by summing up all the ARIMA forecasts. From error analysis and application results, it is both theoretically and practically proved that the proposed DLC method can reduce random forecasting errors while guaranteeing modelling accuracy, so that a more stable and precise system load forecasting result can be obtained.

基于多模型融合的中长期径流集成预测方法

中长期水文预报是流域水资源规划与合理配置的重要依据。为提高中长期径流预测精度,提出了一种基于多模型融合的水库中长期径流集成预测方法。该方法将ARMA、BP、LSTM、RF和SVR等5个异质预测模型进行融合,同时采用超参数优化方法确定各模型的最优参数。将其用于青海省龙羊峡水库的中长期径流预报中,结果表明,通过Stacking融合算法建立的集成预测模型相较于单一模型,取得了更高的预测精度(R2值由0.71提升至0.82)。此方法可为提升流域中长期径流预测精度提供一定参考。

基于集群辨识和卷积神经网络-双向长短期记忆-时序模式注意力机制的区域级短期负荷预测

为了解决区域级短期电力负荷预测时输入特征过多和负荷时序性较强的问题,提出一种基于集群辨识和卷积神经网络(convolutional neural networks,CNN)-双向长短期记忆网络(bi-directional long short-term memory,BiLSTM)-时序模式注意力机制(temporal pattern attention,TPA)的预测方法。首先,将用电模式和天气作为影响因素,基于二阶聚类算法对区域内的负荷节点进行集群辨识,再从每个集群中挑选代表特征作为深度学习模型的输入,这样既能减少输入特征维度,降低计算复杂度,又能综合考虑预测区域的整体特征,提升预测精度。然后,针对区域电力负荷时序性的特点,用CNN-BiLSTM-TPA模型完成训练和预测,该模型能提取输入数据的双向信息生成隐状态矩阵,并对隐状态矩阵的重要特征加权,从多时间步上捕获双向时序信息用于预测。最后,在美国加利福尼亚州实例上分析验证了所提方法的有效性。

基于可解释机器学习的黄河源区径流分期组合预报

黄河源区是黄河流域重要的产流区和我国重要的清洁能源基地,提高黄河源区径流预报准确率可为流域水资源科学调配和水风光清洁能源高效利用提供重要支撑。以黄河源区唐乃亥和玛曲水文站为研究对象,基于不同月份径流组分的差异,考虑积雪覆盖率及融雪水当量变化,构建了中长期径流分期组合机器学习预报模型及其可解释性分析框架。研究结果表明:1)年内的径流预报时段可划分为融雪影响期(3—6月)和非融雪主导(以降雨和地下水补给为主)期(7月—次年2月);2)与传统不分期模型相比,唐乃亥站和玛曲站分期组合预报模型的纳什效率系数分别达0.897、0.835,确定系数(R2)分别达0.897、0.839,均方根误差分别降低了10%、17%,提高了径流预报准确率,通过分位数映射校正,唐乃亥站和玛曲站预报模型的R2分别进一步提升至0.926和0.850;3)基于SHAP机器学习可解释性分析框架,辨识了预报因子对径流预报结果的贡献程度,由高到低依次为降水、前一个月流量、蒸发、气温、相对湿度、融雪水当量等,发现了不同预报因子之间交互作用散点分布具有拖尾式或阶跃式的特征。

基于双重分解和双向长短时记忆网络的中长期负荷预测模型

针对中长期电力负荷序列噪声含量高、难以直接提取序列周期规律从而影响预测精度的问题,提出了一种基于完全自适应噪声集合经验模态分解(complete ensemble empirical mode decomposition with adaptive noise,CEEMDAN)和奇异谱分析(singular spectrum analysis,SSA)双重分解的双向长短时记忆网络(bidirectional long and short time memory,BiLSTM)预测模型。首先,采用CEEMDAN对历史负荷进行分解,以得到若干个周期规律更为清晰的子序列;再利用多尺度熵(multiscale entropy,MSE)计算所有子序列的复杂程度,根据不同时间尺度上的样本熵值将相似的子序列重构聚合;然后,利用SSA去噪的功能,对高度复杂的新序列进行二次分解,去除序列中的噪声并提取更为主要的规律,从而进一步提高中长序列预测精度;再将得到的最终一组子序列输入BiLSTM进行预测;最后,考虑到天气、节假日等外部因素对电力负荷的影响,提出了一种误差修正技术。选取了巴拿马某地区的用电负荷进行实验,实验结果表明,经过双重分解可以将均方根误差降低87.4%;预测未来一年的负荷序列时,采用的BiLSTM模型将拟合系数最高提高2.5%;所提出的误差修正技术可将均方根误差降低9.7%。

基于集合Kalman滤波的中长期径流预报

为降低中长期径流预报的不确定性,增加水电站水库的发电效益,针对现有方法侧重于提高单一预报模型确定性预报结果的准确性以降低径流预报不确定性的问题,提出一种基于集合Kalman滤波的入库径流确定性预报方法。以旬为预见期的锦西水库实例验证结果表明:相比传统的单一预报模型和传统的信息融合预报模型,基于集合Kalman滤波的中长期径流预报可使RMSE降低4.78 m^(3)/s,合格率可提高0.56%,且更有效地降低了汛期预报的不确定性,得到了更加准确、可靠的确定性径流预报结果,可为开展流域梯级水电站优化调度提供技术支持。

基于改进Q学习算法和组合模型的超短期电力负荷预测

单一模型在进行超短期负荷预测时会因负荷波动而导致预测精度变差,针对此问题,提出一种基于深度学习算法的组合预测模型｡首先,采用变分模态分解对原始负荷序列进行分解,得到一系列的子序列。其次,分别采用双向长短期记忆网络和优化后的深度极限学习机对每个子序列进行预测。然后,利用改进Q学习算法对双向长短期记忆网络的预测结果和深度极限学习机的预测结果进行加权组合,得到每个子序列的预测结果。最后,将各个子序列的预测结果进行求和,得到最终的负荷预测结果。以某地真实负荷数据进行预测实验,结果表明所提预测模型较其他模型在超短期负荷预测中表现更佳,预测精度达到98%以上。

融合CNN与BiLSTM模型的短期电能负荷预测

针对卷积神经网络(CNN)在捕捉预测序列间历史相关性方面的不足以及在变量复杂情况下出现的无法精准提取预测关键信息的问题,提出一种将双向长短期记忆网络(BiLSTM)与卷积神经网络结合的CNN-BiLSTM模型.首先,采用数据预处理方法保证数据的正确性和完整性,并对数据进行分析以探究多变量之间的相关性;其次,通过CNN与L1正则化对多维输入特征进行特征筛选,选取与预测相关的重要性特征向量;最后,使用BiLSTM对CNN输出的关键特征信息进行保存,形成向量与预测序列,并通过分析时序特征的潜在特点,提取用户的内在消费模式.实验比较了该模型与其他时序模型在不同时间分辨率下的预测效果,实验结果表明,CNN-BiLSTM模型在不同的回望时间间隔下表现出了最佳的预测性能,能够实现更好的短期负荷预测.

基于Stacking融合的LSTM-SA-RBF短期负荷预测

为了解决单个神经网络预测的局限性和时间序列的波动性,提出了一种奇异谱分析(singular spectrum analysis,SSA)和Stacking框架相结合的短期负荷预测方法。利用随机森林筛选出与历史负荷相关性强烈的特征因素,采用SSA为负荷数据降噪,简化模型计算过程;基于Stacking框架,结合长短期记忆(long and short-term memory,LSTM)-自注意力机制(self-attention mechanism,SA)、径向基(radial base functions,RBF)神经网络和线性回归方法集成新的组合模型,同时利用交叉验证方法避免模型过拟合;选取PJM和澳大利亚电力负荷数据集进行验证。仿真结果表明,与其他模型比较,所提模型预测精度高。

基于双模态分解的发电站母线短期负荷预测

母线负荷预测是电力系统运营和规划中至关重要的一项任务,针对电力负荷数据的非线性强以及影响因素多等问题,提出了一种基于双模态分解、深度学习和注意力机制的负荷预测模型。首先,对输入数据进行经验模态分解(Empirical mode decomposition,EMD),通过K-means聚类分析对复杂度相似的分量进行集合得到三个组合分量。其次,使用变分模态分解(Variational mode decomposition, VMD)对组合分量再次进行分解得到不同分量,使用麻雀搜索算法(Sparrow search algorithm,SSA)对变分模态分解的参数进行优化。再次,将变分模态分解得到的分量与影响因素连接并输入长短期记忆网络(Long short-term memory network, LSTM),通过注意力机制挖掘数据内部的相关性,并使用SSA对LSTM网络的参数进行优化。最后,采用宁夏某电站一年的负荷数据进行验证,经过与不同模型的对比分析,所提模型有更高的预测精度。

基于STE-TCN的中短期电力负荷预测

目的 针对传统电力负荷预测模型对长序列预测精度低的问题,提出一种结合跳级卷积连接与时间编码网络的新型时序卷积神经网络(TCN)模型——STE-TCN模型。方法 首先对TCN模型加入跨周期的膨胀卷积通道(Skip-convolution)提取电力数据周期信息;再进行特征融合得到Skip-TCN网络,使网络抓取周期规律,增加信息利用长度;最后设计日期编码网络(Time encoding network)捕捉生活周期和季节性特征,与Skip-TCN进行特征融合得到STE-TCN模型,实现对电力负荷数据长序列预测。结果 实验表明:在与TCN模型和传统时序网络的对比下,Skip-TCN的预测精度均有提升,在预测长度更长的测试上提升尤为明显。结论 实验结果验证了通过对更长跨度时序关系的捕捉,STE-TCN网络改进方法有效提升了对长序列电力负荷的预测精度。

基于CNN-LSTM电力消耗预测模型及系统开发

有效预测电能负荷,对提高电力负荷时间序列测量准确度及合理制定用电能管理措施具有重要意义。针对传统预测模型在电能负荷预测中无法充分挖掘时间序列数据中隐藏特征的问题,基于电能数据时间序列的趋势,融合数值信息提出一种卷积神经网络(convolutional neuralnetwork,CNN)与长期短期记忆循环神经网络(long short-term memory network,LSTM)相结合的混合多隐层CNN-LSTM电力能耗预测模型。首先,通过设定最小目标函数作为优化目标,Adam优化算法更新神经网络的权重,并对网络层和批大小进行自适应调优以确定最佳层数和批大小。其次,构建混合多隐层模型并进行隐层组合优化与讨论,确定最佳时间维度的参数,进行时间维度的特征学习进而预测下一时间序列的耗电量。然后以某公司的电力负荷数据为例进行验证,并与LSTM、CNN、RNN等模型的预测结果分析比较。结果表明上述混合多隐层模型预测准确度达98.94%,平均绝对误差(MAE)达到0.0066,均优于其他相关模型,证明以上混合预测模型在电力负荷预测精度方面具有更好的性能。基于上述理论,开发了能耗监控决策系统,实现设备状态实时监控和能耗智能预测功能,为解决传统制造业能耗需求不精确和能源库存浪费问题提供参考和指导。

基于经验模态分解和优化BiLSTM的短期负荷预测

针对电力负荷数据的非线性和不稳定性问题,提出了一种基于经验模态分解改进麻雀搜索算法双向长短期记忆神经网络相结合的EMD ISSA BiLSTM预测模型。首先采用EMD处理非线性负荷数据,将原始负荷数据分解为多个不同尺度的本征模态函数(IMF),引入反向学习策略和Levy飞行策略分别改进麻雀搜索算法(SSA)的收敛速度慢和容易陷入局部最优问题,利用改进麻雀搜索算法(ISSA)对BiLSTM神经网络进行参数寻优。然后再利用优化后的BiLSTM模型对每个分量进行预测,并将各预测结果叠加组合,得到整个负荷序列的预测结果。最后通过实际算例分析,证明该方法相对于传统的预测方法具有更好的预测精度和稳定性,可作为一种有效的短期负荷预测方法。