WO2023088212A1 - 一种基于集成学习的在线机组负荷预测方法 - Google Patents

一种基于集成学习的在线机组负荷预测方法 Download PDF

Info

Publication number
WO2023088212A1
WO2023088212A1 PCT/CN2022/131741 CN2022131741W WO2023088212A1 WO 2023088212 A1 WO2023088212 A1 WO 2023088212A1 CN 2022131741 W CN2022131741 W CN 2022131741W WO 2023088212 A1 WO2023088212 A1 WO 2023088212A1
Authority
WO
WIPO (PCT)
Prior art keywords
unit load
data
method based
model
feature
Prior art date
Application number
PCT/CN2022/131741
Other languages
English (en)
French (fr)
Inventor
周东阳
曹军
万松森
王承文
郑小刚
刘爱君
安玉强
唐贝
张骁
王帆
宋志坚
蔡连成
Original Assignee
西安热工研究院有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 西安热工研究院有限公司 filed Critical 西安热工研究院有限公司
Publication of WO2023088212A1 publication Critical patent/WO2023088212A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06313—Resource planning in a project environment
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Definitions

  • the disclosure belongs to the field of thermal power station load forecasting, and relates to an online unit load forecasting method based on integrated learning.
  • the power demand of the power grid fluctuates greatly over time. According to the actual characteristics of power production and consumption, it is necessary to maintain a balanced relationship between the supply and demand sides.
  • the resulting changes in power generation on the power supply side are mainly regulated by thermal power units through unit load changes.
  • the power grid issues load planning curves to the power plants according to the actual conditions of the units and the corresponding power dispatching principles. There is a large deviation from the actual unit load command. Therefore, accurate unit actual load forecasting is helpful for the power plant to choose the adjustment method, respond to the load dispatch of the power grid in time, and keep the power grid running efficiently, stably, safely and economically.
  • the load forecasting methods are mainly used at present: (1) regression forecasting; (2) time series forecasting; (3) ) gray prediction; (4) artificial neural network; (5) support vector machine; (6) wavelet variation algorithm; (7) fuzzy regression model.
  • the non-linearity of the coefficients in the regression method reduces the accuracy of the model; the time prediction model is greatly affected by factors such as weather and climate; the prediction accuracy of the gray model is inversely proportional to the gray level, and when the dispersion of the data increases, the prediction accuracy decreases;
  • the design requirements of the neural network are relatively high, the number of hidden layers is difficult to judge, and the convergence speed is slow; the support vector machine is difficult to handle large-scale training samples, and cannot reflect the long-term change law of the unit load;
  • the data accuracy is high, and it is difficult to overcome the interference of complex factors; the fuzzy system lacks self-learning ability, and its fuzzy rules mainly rely on the expert system, and the scope of use has relatively large limitations.
  • the present disclosure provides an online load forecasting method based on ensemble learning, which determines a load regulation mode by predicting and judging a load change trend based on the result.
  • An online unit load forecasting method based on integrated learning comprising the following steps:
  • XD, LD1 and LD2 are spliced and used as the fully connected layer Dense2, and input into the linear regression model to obtain the final output result, that is, the final unit load forecast result.
  • a further improvement of the present disclosure lies in that the counting interval in the step (1) is 1 min.
  • a further improvement of the present disclosure lies in that the historical data in the step (1) is time-series historical data of unit load.
  • a further improvement of the present disclosure lies in that the data preprocessing process in the step (2) is to filter outliers in the original sample D.
  • a further improvement of the present disclosure is that in the step (2), according to the Raida criterion, the standard deviation is calculated and processed for the sample D, and exceeding the deviation interval is a gross error, and the load value at the previous moment is used to replace the error value.
  • the feature extraction in the step (3) includes discrete time features, autocorrelation features and partial autocorrelation features between the current value and its past values, and aggregation features of different time window average values, where time The spans are all 3, including t-1, t-2, and t-3.
  • a further improvement of the present disclosure lies in that the concatenation in the step (5) is matrix concatenation of TD and LD, rather than adding corresponding elements.
  • a further improvement of the present disclosure is that in the step (6), the linear regression model linearly combines the results of XD, LD1 and LD2 through the additive model, and constantly changes the weight during training to reduce the prediction residual to obtain a satisfactory training result.
  • the present disclosure has the following advantages:
  • the present disclosure is an algorithm-based online prediction method without additional hardware equipment.
  • the present disclosure is a unit load forecasting method based on operating data and integrating multiple algorithms for integrated learning.
  • the forecasting accuracy is high and the stability is strong, and the maximum forecasting error is small and stable over a long period of time.
  • the present disclosure establishes a block structure by sorting the models through the extreme gradient boosting (XGBoost) method during model training, reduces repeated data in iterative calculations, reduces computational complexity, and shortens computational time.
  • XGBoost extreme gradient boosting
  • This disclosure calculates the weights of different historical data through the lightweight and efficient gradient boosting (LGBM) algorithm, reduces the data with smaller weights in the modeling data, and improves the accuracy of the model.
  • LGBM lightweight and efficient gradient boosting
  • This disclosure strengthens the proportion of high-weight data through matrix splicing of multi-model output results, and saves low-weight data at the same time, improves the generalization ability of the model on the basis of ensuring the accuracy of the model, and ensures that the actual value in the embodiment is consistent with the model The root mean square error of the predicted value is reduced.
  • FIG. 1 is a flowchart of an online unit load forecasting method based on integrated learning in the present disclosure.
  • Fig. 2 is the time-series variation curve of unit load under the same working condition in the embodiment.
  • Fig. 3 is a comparison diagram between the time-series prediction result of the predicted unit load and the actual unit load in the embodiment.
  • the disclosed framework mainly consists of core steps such as historical data sampling, data preprocessing, feature extraction, XGBoost model training, LGBM model training, LSTM model training, integrated learning linear regression model training, and unit load forecasting.
  • core steps such as historical data sampling, data preprocessing, feature extraction, XGBoost model training, LGBM model training, LSTM model training, integrated learning linear regression model training, and unit load forecasting.
  • the detailed process is shown in Figure 1 . Taking a coal-fired unit as an example, the specific operation steps are as follows:
  • the training data is used to train the constructed prediction model, and the parameters of the model are shown in Table 1.
  • Figure 3 shows the comparison between the unit load forecast results and the actual measurement point data for a period of time (from 00:00 on May 9, 2020 to 00:00 on May 22, 2020).
  • RMSE root mean square error
  • MAAE absolute value of the relative percentage error
  • the measured load time series data of the unit has greater volatility, and the prediction results of the model can better identify the fluctuation law of the load, which fits well with the measured results, indicating that the unit load proposed in this disclosure
  • the prediction method can judge the change trend of the load in advance, and solve the problem that the load is difficult to respond to the command quickly during the operation of the unit.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Health & Medical Sciences (AREA)
  • Tourism & Hospitality (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Educational Administration (AREA)
  • Public Health (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Water Supply & Treatment (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本公开提出了一种基于集成学习的在线机组负荷预测方法，该方法包括：首先从数据库读取过去一个月的机组负荷历史数据，其次对该数据进行预处理，按照拉依达准则对粗大值进行检测，检测出来的粗大值使用向后填充的方式进行处理。进一步对预处理后数据进行特征提取，包括时间特征、相关性特征和聚合特征，构建特征数据集；通过该特征数据集训练极限梯度提升和轻量级高效梯度提升模型，并通过LGBM模型的预测结果和原始特征数据训练长短期记忆神经网络模型；最后将这三个模型输出结果作为三个输入来训练线性回归模型输出最后的负荷预测结果。在模型训练阶段，以预测平均误差最小为原则，调整网络参数，从而建立最优的机组负荷预测模型。

Description

一种基于集成学习的在线机组负荷预测方法 技术领域
本公开属于火电站负荷预测领域，涉及一种基于集成学习的在线机组负荷预测方法。
背景技术
受环境、季节等因素的影响，电网的用电需求随时间存在较大的波动，根据电力生产与消费的实际特点，需保持供需端的平衡关系。由此产生的供电侧发电量变化主要由火电机组通过机组负荷的变化调节，电网根据机组实际情况和相应的电力调度原则，向电厂下达负荷计划曲线，但因目前机组并网结构复杂，计划曲线与实际机组负荷指令存在较大偏差。因此准确的机组实际负荷预测有助于电厂选择调节方式，及时的响应电网的负荷调度，能够使电网保持高效、稳定、安全、经济地运行。
因负荷需求受各种条件因素的影响，具有不确定性、时间周期性和条件性等特点，目前主要采用以下几种负荷预测方法：(1)回归预测；(2)时间序列预测；(3)灰色预测；(4)人工神经网络；(5)支持向量机；(6)小波变化算法；(7)模糊回归模型。然而回归方法中系数的非线性，使模型的精确降低；时间预测模型受气象气候等因素影响大；灰色模型的预测精度与灰度成反比，当数据离散程度增大时，预测精度下降；人工神经网络的设计要求较高，隐含层个数难以判断，收敛速度慢；支持向量机难以处理大规模训练样本，无法反映机组负荷长时间的变化规律；小波变换算法过程较为复杂，对历史负荷数据准确性要求高，且难以克服复杂因素的干扰作用；模糊系统缺乏自学习能力， 其模糊规则主要依赖专家系统，使用范围有较大的局限性。
因此，为及时响应机组的负荷调度需求，当前急需一种适合发电机组负荷特性并且预测精度较高的负荷预测方法，通过有效的数据处理和挖掘工具，从复杂无序的负荷运行数据中获取稳定而准确的时序变化趋势。
发明内容
为了快速响应负荷调度指令，本公开提供一种基于集成学习的在线负荷预测方法，通过对结果预测判断负荷的变化趋势，决定负荷调节模式。
为解决上述技术问题，本公开采用如下技术方案：
一种基于集成学习的在线机组负荷预测方法，包括以下步骤：
(1)从机组SIS数据库中获取前一个月的负荷历史运行数据，得到原始数据样本D；
(2)对原始数据样本D进行数据预处理，根据拉依达准则去除粗大误差，处理后的样本记为CD；
(3)对样本CD进行特征提取，包括时间特征相关性特征和聚合特征，建立特征数据样本TD；
(4)将TD输入极限梯度提升和轻量级高效梯度提升模型，对应输出结果分别记为XD和LD1；
(5)将TD和LD1进行拼接并作为全连接层Dense1，输入长短期记忆神经网络，输出结果记为LD2；
(6)最后将XD、LD1和LD2进行拼接并作为全连接层Dense2，输入线性回归模型，得到最后的输出结果，即最终的机组负荷预测结果。
本公开进一步的改进在于，所述步骤(1)中取数间隔为1min。
本公开进一步的改进在于，所述步骤(1)中历史数据为机组负荷的时序历史数据。
本公开进一步的改进在于，所述步骤(2)中数据预处理过程为过滤原始样本D中的离群值。
本公开进一步的改进在于，所述步骤(2)中根据拉依达准则，对样本D计算处理得到标准偏差，超过偏差区间属于粗大误差，采用前一时刻的负荷值替代误差值。
本公开进一步的改进在于，所述步骤(3)中特征提取包括离散时间特征，当前值与其过去值之间的自相关特征和偏自相关特征以及不同时间窗平均值的聚合特征，其中，时间跨度均为3，即包括t-1、t-2、t-3。
本公开进一步的改进在于，所述步骤(5)中所述的拼接是将TD和LD进行矩阵拼接，并非对应元素相加。
本公开进一步的改进在于，所述步骤(6)中线性回归模型通过加法模型将XD、LD1和LD2结果进行线性组合，训练时不断改变权值来减小预测残差以获得满意的训练结果。
本公开相对于现有技术而言具备以下优点：
(1)本公开是基于算法的在线预测方法，无需额外的硬件设备。
(2)本公开是一种基于运行数据、融合多种算法进行集成学习的机组负荷预测方法，预测精度较高且稳定性强，在长时间段最大预测误差小且保持稳定。
(3)本公开通过极限梯度提升(XGBoost)法在训练模型时通过对模型的排序，建立block结构，降低迭代计算中的重复数据，降低计算复杂度，缩小计算时间。
(4)本公开通过轻量级高效梯度提升(LGBM)算法计算不同历史数据的 权重，减少建模数据中权重较小的数据，提高模型精确度。
(5)本公开通过多模型输出结果的矩阵拼接，加强高权重数据的的比例，同时保存低权重数据，在保证模型精确度的基础上提高模型泛化能力，确保实施例中实际值与模型预测值的均方根误差减小。
附图说明
图1为本公开一种基于集成学习的在线机组负荷预测方法的流程图。
图2为实施例中的同一工况下机组负荷时序变化曲线。
图3为实施例中的预测机组负荷时序预测结果与实际机组负荷的对比图。
具体实施方式
下面将参照附图更详细地描述本公开的示例性实施例。虽然附图中显示了本公开的示例性实施例，然而应当理解，可以以各种形式实现本公开而不应被这里阐述的实施例所限制。相反，提供这些实施例是为了能够更透彻地理解本公开，并且能够将本公开的范围完整的传达给本领域的技术人员。需要说明的是，在不冲突的情况下，本公开中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本公开。
本公开的框架主要由历史数据采样、数据预处理、特征提取、XGBoost模型训练、LGBM模型训练、LSTM模型训练、集成学习线性回归模型训练、机组负荷预测等核心步骤，详细流程如图1所示。以某燃煤机组为例，具体操作步骤如下：
从厂级监控信息系统(SIS)的数据库采集2020年5月1日00:00至2020年5月31日00:00的负荷历史数据值，取数间隔为1min，所得的负荷值序列如图2 所示。而后对数据进行预处理，根据拉依达准则计算标准偏差区间，清理粗大误差值。进一步地对数据进行特征提取，离散时间特征，提取当前值与其过去值之间的自相关特征和偏自相关特征以及不同时间窗平均值的聚合特征，其中，时间跨度均为3，包括t-1、t-2、t-3，并将特征数据按照4:1的比例划分数据集，前一部分作为训练数据，后一半部分作为测试数据。进一步地利用训练数据分别训练XGBoost，LGBM两个模型，学习器均采用了四折交叉验证进行训练，并使用了Sklearn类里面的GridSearch方法进行参数寻优。
将训练数据用于训练所构建的预测模型，模型的参数如表1所示。
表1：XGBoost模型参数
n_estimators 16000
min_child_weight 1
learning_rate 0.01
max_depth 6
subsample 0.6
colsample_bytree 0.4
colsample_bylevel 1
表2：LGBM模型参数
n_estimators 9000
min_child_samples 20
num_leaves 20
learning_rate 0.005
feature_fraction 0.8
subsample 0.4
LGBM和XGBoost中均设置了提前终止训练：当模型损失连续300次迭代均没有改进时停止训练。对于LSTM模型，采用单层LSTM，隐含层神经元数 目为200，迭代次数为200。
图3给出了一段时间内(2020年5月9日00:00到2020年5月22日00:00)机组负荷预测结果和实际测点数据的对比。对于整个测试集，模型预测值的均方根误差(RMSE)为3.2796，相对百分误差绝对值(MAPE)为2.147％。可以看到，模型预测误差很小，满足实际工程中的需要。同时，可以看到，实测的机组负荷时序数据具有较大的波动性，模型的预测结果能够较好的识别负荷的波动规律，与实测结果较好的贴合，表明本公开所提出的机组负荷预测方法能够提前判断负荷的变动趋势，解决机组运行过程中，负荷难以快速响应指令的问题。
以上所述仅是本公开的优选实施方式，应当指出：对于本技术领域的普通技术人员来说，在不脱离本公开原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本公开的保护范围。

Claims (8)

  1. 一种基于集成学习的在线机组负荷预测方法，其特征在于，包括以下步骤：
    (1)从机组SIS数据库中获取前一个月的负荷历史运行数据，得到原始数据样本D；
    (2)对原始数据样本D进行数据预处理，根据拉依达准则去除粗大误差，处理后的样本记为CD；
    (3)对样本CD进行特征提取，包括时间特征相关性特征和聚合特征，建立特征数据样本TD；
    (4)将TD输入极限梯度提升和轻量级高效梯度提升模型，对应输出结果分别记为XD和LD1；
    (5)将TD和LD1进行拼接并作为全连接层Dense1，输入长短期记忆神经网络，输出结果记为LD2；
    (6)最后将XD、LD1和LD2进行拼接并作为全连接层Dense2，输入线性回归模型，得到最后的输出结果，即最终的机组负荷预测结果。
  2. 根据权利要求1所述的一种基于集成学习的在线机组负荷预测方法，其特征在于，所述步骤(1)中取数间隔为1min。
  3. 根据权利要求1所述的一种基于集成学习的在线机组负荷预测方法，其特征在于，所述步骤(1)中历史数据为机组负荷的时序历史数据。
  4. 根据权利要求1所述的一种基于集成学习的在线机组负荷预测方法，其特征在于，所述步骤(2)中数据预处理过程为过滤原始样本D中的离群值。
  5. 根据权利要求4所述的一种基于集成学习的在线机组负荷预测方法，其特征在于，所述步骤(2)中根据拉依达准则，对样本D计算处理得到标准偏差，超过偏差区间属于粗大误差，采用前一时刻的负荷值替代误差值。
  6. 根据权利要求1所述的一种基于集成学习的在线机组负荷预测方法，其特征在于，所述步骤(3)中特征提取包括离散时间特征，当前值与其过去值之间的自相关特征和偏自相关特征以及不同时间窗平均值的聚合特征，其中，时间跨度均为3，即包括t-1、t-2、t-3。
  7. 根据权利要求1所述的一种基于集成学习的在线机组负荷预测方法，其特征在于，所述步骤(5)中所述的拼接是将TD和LD进行矩阵拼接，并非对应元素相加。
  8. 根据权利要求1所述的一种基于集成学习的在线机组负荷预测方法，其特征在于，所述步骤(6)中线性回归模型通过加法模型将XD、LD1和LD2结果进行线性组合，训练时不断改变权值来减小预测残差以获得满意的训练结果。
PCT/CN2022/131741 2021-11-16 2022-11-14 一种基于集成学习的在线机组负荷预测方法 WO2023088212A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111357431.X 2021-11-16
CN202111357431.XA CN114022021A (zh) 2021-11-16 2021-11-16 一种基于集成学习的在线机组负荷预测方法

Publications (1)

Publication Number Publication Date
WO2023088212A1 true WO2023088212A1 (zh) 2023-05-25

Family

ID=80064606

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/131741 WO2023088212A1 (zh) 2021-11-16 2022-11-14 一种基于集成学习的在线机组负荷预测方法

Country Status (2)

Country Link
CN (1) CN114022021A (zh)
WO (1) WO2023088212A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116544931A (zh) * 2023-06-27 2023-08-04 北京理工大学 基于集成片段变换和时间卷积网络的电力负荷分布预测方法
CN116632842A (zh) * 2023-07-26 2023-08-22 国网山东省电力公司信息通信公司 基于聚类特性的台区分布式光伏负荷概率预测方法及系统
CN117078047A (zh) * 2023-10-16 2023-11-17 华能济南黄台发电有限公司 一种基于lstm的热负荷预测与分配优化方法及系统
CN117239731A (zh) * 2023-09-21 2023-12-15 山东工商学院 基于混合模型的节假日短期电力负荷预测方法
CN117909928A (zh) * 2024-03-19 2024-04-19 国网四川省电力公司成都供电公司 基于大数据分析的空调负荷预测方法及系统

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114022021A (zh) * 2021-11-16 2022-02-08 西安热工研究院有限公司 一种基于集成学习的在线机组负荷预测方法
CN114444821A (zh) * 2022-04-12 2022-05-06 国网湖北省电力有限公司电力科学研究院 面向电力物联网的集成学习负荷预测方法、系统及介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150356213A1 (en) * 2014-06-06 2015-12-10 General Electric Company Method and system for generating electric load models
CN110084424A (zh) * 2019-04-25 2019-08-02 国网浙江省电力有限公司 一种基于lstm与lgbm的电力负荷预测方法
CN113205207A (zh) * 2021-04-19 2021-08-03 深圳供电局有限公司 一种基于XGBoost算法的用电短期负荷波动预测方法及系统
CN114022021A (zh) * 2021-11-16 2022-02-08 西安热工研究院有限公司 一种基于集成学习的在线机组负荷预测方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150356213A1 (en) * 2014-06-06 2015-12-10 General Electric Company Method and system for generating electric load models
CN110084424A (zh) * 2019-04-25 2019-08-02 国网浙江省电力有限公司 一种基于lstm与lgbm的电力负荷预测方法
CN113205207A (zh) * 2021-04-19 2021-08-03 深圳供电局有限公司 一种基于XGBoost算法的用电短期负荷波动预测方法及系统
CN114022021A (zh) * 2021-11-16 2022-02-08 西安热工研究院有限公司 一种基于集成学习的在线机组负荷预测方法

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116544931A (zh) * 2023-06-27 2023-08-04 北京理工大学 基于集成片段变换和时间卷积网络的电力负荷分布预测方法
CN116544931B (zh) * 2023-06-27 2023-12-01 北京理工大学 基于集成片段变换和时间卷积网络的电力负荷分布预测方法
CN116632842A (zh) * 2023-07-26 2023-08-22 国网山东省电力公司信息通信公司 基于聚类特性的台区分布式光伏负荷概率预测方法及系统
CN116632842B (zh) * 2023-07-26 2023-11-10 国网山东省电力公司信息通信公司 基于聚类特性的台区分布式光伏负荷概率预测方法及系统
CN117239731A (zh) * 2023-09-21 2023-12-15 山东工商学院 基于混合模型的节假日短期电力负荷预测方法
CN117239731B (zh) * 2023-09-21 2024-02-27 山东工商学院 基于混合模型的节假日短期电力负荷预测方法
CN117078047A (zh) * 2023-10-16 2023-11-17 华能济南黄台发电有限公司 一种基于lstm的热负荷预测与分配优化方法及系统
CN117078047B (zh) * 2023-10-16 2024-02-23 华能济南黄台发电有限公司 一种基于lstm的热负荷预测与分配优化方法及系统
CN117909928A (zh) * 2024-03-19 2024-04-19 国网四川省电力公司成都供电公司 基于大数据分析的空调负荷预测方法及系统
CN117909928B (zh) * 2024-03-19 2024-05-28 国网四川省电力公司成都供电公司 基于大数据分析的空调负荷预测方法及系统

Also Published As

Publication number Publication date
CN114022021A (zh) 2022-02-08

Similar Documents

Publication Publication Date Title
WO2023088212A1 (zh) 一种基于集成学习的在线机组负荷预测方法
CN110766212B (zh) 用于历史数据缺失电场的超短期光伏功率预测方法
CN108388962B (zh) 一种风电功率预测系统及方法
CN112734128B (zh) 一种基于优化rbf的7日电力负荷峰值预测方法
CN110119845A (zh) 一种轨道交通客流预测的应用方法
CN109615124B (zh) 一种基于深度学习的scada主站负荷预测方法
Oprea et al. Ultra-short-term forecasting for photovoltaic power plants and real-time key performance indicators analysis with big data solutions. Two case studies-PV Agigea and PV Giurgiu located in Romania
CN116167527B (zh) 纯数据驱动的电力系统静态安全运行风险在线评估方法
CN115130741A (zh) 基于多模型融合的多因素电力需求中短期预测方法
CN115587672A (zh) 一种配变负荷预测及重过载预警方法及系统
CN116167531A (zh) 一种基于数字孪生的光伏发电预测方法
CN113919545A (zh) 一种多数据模型融合的光伏发电功率预测方法和系统
CN114444660A (zh) 基于注意力机制和lstm的短期电力负荷预测方法
CN110991689B (zh) 基于LSTM-Morlet模型的分布式光伏发电系统短期预测方法
Wang et al. Probabilistic power curve estimation based on meteorological factors and density LSTM
CN113988421A (zh) 基于深度学习的风电机组功率智能预测方法
CN109447336A (zh) 一种上游水库与其反调节水库坝间水位优化控制方法
CN112821383A (zh) 一种基于深度学习的电力系统自然频率特性系数区间预测方法
CN117200352A (zh) 一种基于云边融合的光伏发电调控方法及系统
CN112232570A (zh) 一种正向有功总电量预测方法、装置及可读存储介质
CN111967660A (zh) 一种基于svr的超短期光伏预测残差修正方法
CN108345996B (zh) 一种降低风电功率考核电量的系统及方法
CN114936640A (zh) 一种新能源发电智能预测模型的在线训练方法
CN115759343A (zh) 一种基于e-lstm的用户电量预测方法和装置
CN116205123A (zh) 一种超短期风电场功率预测方法及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22894746

Country of ref document: EP

Kind code of ref document: A1