CN114021848A - Generating capacity demand prediction method based on LSTM deep learning - Google Patents

Generating capacity demand prediction method based on LSTM deep learning Download PDF

Info

Publication number
CN114021848A
CN114021848A CN202111406645.1A CN202111406645A CN114021848A CN 114021848 A CN114021848 A CN 114021848A CN 202111406645 A CN202111406645 A CN 202111406645A CN 114021848 A CN114021848 A CN 114021848A
Authority
CN
China
Prior art keywords
data
model
lstm
network
power generation
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
CN202111406645.1A
Other languages
Chinese (zh)
Inventor
谢云明
李�杰
王垚
赵子龙
陈建亮
徐创学
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian Thermal Power Research Institute Co Ltd
Dezhou Power Plant of Huaneng International Power Co Ltd
Huaneng Shandong Power Generation Co Ltd
Original Assignee
Xian Thermal Power Research Institute Co Ltd
Dezhou Power Plant of Huaneng International Power Co Ltd
Huaneng Shandong Power Generation Co Ltd
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 Xian Thermal Power Research Institute Co Ltd, Dezhou Power Plant of Huaneng International Power Co Ltd, Huaneng Shandong Power Generation Co Ltd filed Critical Xian Thermal Power Research Institute Co Ltd
Priority to CN202111406645.1A priority Critical patent/CN114021848A/en
Publication of CN114021848A publication Critical patent/CN114021848A/en
Pending legal-status Critical Current

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
    • G06Q10/04—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
    • 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
    • G06Q10/06—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
    • G06Q10/06315—Needs-based resource requirements planning or analysis
    • 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Abstract

A generating capacity demand prediction method based on LSTM deep learning comprises the steps of firstly, acquiring historical generating capacity and main influence factor data to obtain a sequence data set, and performing normalization processing; dividing the prepared data into a training set and a test set according to a proportion; establishing a generating capacity LSTM network model, and adjusting internal parameters of the network; completing LSTM model fitting by using training set data; combining the predictions with test data for overfitting evaluation of the model; after model evaluation and verification, inputting the current generated data sampled on line into a verified LSTM neural network to predict the generated energy value at the future time; the generating capacity demand forecasting method based on LSTM deep learning can record required data for a long time and forecast on line, has long forecasting period and high forecasting generating capacity precision, and meets the requirement of generating enterprises on generating capacity demand forecasting.

Description

Generating capacity demand prediction method based on LSTM deep learning
Technical Field
The invention relates to the technical field of decision making of operation requirements of a power supply side, in particular to a method for predicting a power generation demand based on LSTM deep learning.
Technical Field
Electrical energy has its particularity in the production and consumption of electricity. Under the current technical conditions, the electric energy storage technology cannot store and release abundant electric quantity on a large scale. The built energy storage is small in scale and low in economy. Therefore, ensuring the balance of power generation and power utilization is still the only feasible means for improving the power utilization quality and improving the economical efficiency. Meanwhile, in order to realize the safety of the power system, accurate prediction of the power generation (utilization) demand must be realized.
The contradiction between supply and demand exists in the production and consumption links of the electric energy. On one hand, the energy supply is insufficient, the generated electric energy cannot meet the requirements of users, and the gate is usually pulled to limit the electricity, so that the requirements of normal work and life of enterprises, public institutions and residents are influenced; on the other hand, the power generated by the supply side exceeds the power demand of the user, which causes system safety and energy waste. According to the data display of the national energy statistics bureau, China can waste electric energy by billions every year. Therefore, an accurate electric energy load prediction method is sought, the demand change trend of the electric energy is accurately predicted, the adverse effect of the electric energy on the environment can be reduced, and meanwhile, the electric energy is saved. The generated energy demand prediction takes historical time sequence data as a data source, and a generated energy demand prediction mathematical model is established by utilizing technologies such as data mining, deep learning and the like, so that the generated (used) energy demand in a future time period is predicted, a power generation enterprise can conveniently and effectively manage the electric energy production, and the electric energy waste is reduced.
For power generation enterprises, the power generation demand forecast is generally divided into medium-term and short-term forecasts. The general primary purpose of medium and long term forecasting is to provide guidance for fuel procurement and storage and equipment servicing of the power plant. The short-term prediction generally mainly provides guidance for the output arrangement of the unit and the formulation of a dispatching plan.
Due to the difference of economic structures, development levels and climates of various regions, the power generation demand prediction does not have a universally applicable mode or prediction model and can be universal. In addition, the power consumption in the same area is affected by various factors, such as seasons, holidays, weekend effects and the like, and the power consumption presents complex variation characteristics, so that the accurate prediction of the power consumption is difficult. The existing generating capacity load forecasting methods mostly carry out linear forecasting according to the single or combined factors, the forecasting time is short, the precision is low, some methods need complex modeling, the occupied resources are high, and engineering implementation and application popularization for enterprises on the generating side are difficult to carry out.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a power generation demand prediction method based on LSTM deep learning, which predicts the power generation demand of the next time period according to the power generation and the main influence factor data of the previous time period; the accuracy and the practicability of the generated energy demand prediction are improved by introducing a plurality of time series prediction LSTM models of main influence factor data, and a power generation enterprise can effectively organize coal reserve supply and reasonably arrange equipment maintenance time and output plans according to the predicted generated energy demand, so that stable generated energy production supply is achieved, and the economic benefit and the social benefit of the power generation enterprise are improved.
In order to achieve the purpose, the technical scheme of the invention is as follows:
a method for predicting a power generation demand based on LSTM deep learning is characterized by comprising the following steps:
step 1: collecting power generation capacity and historical data of main influence factors, arranging the historical data according to time to obtain a sequence sample data set, then carrying out normalization processing on the data to enable the numerical value of the data to be between [0 and 1], and obtaining a data set as a sample for supervised learning, wherein the historical data of the main influence factors comprises: weather data, date attribute data (season, week, holidays), power generation equipment output and reliability data and power grid scheduling data;
step 2: dividing the data into a training set and a test set; the prepared sequence data are processed according to the following steps of 2: 1, dividing the data into a training set and a test set in proportion, and ensuring that the sampling period of each data set can represent a characteristic change sample in the same time period;
and step 3: defining and establishing an LSTM network model of generating capacity demand, setting data input of a plurality of main influence factors, and adjusting internal parameters of the network; after format conversion of an input layer, transmitting the data collected by the LSTM network model to the LSTM layer for LSTM network training, and outputting a classification result by a classification output layer after sequentially passing through a full connection layer and a SoftMax layer;
step 3, setting a plurality of main influence factor data inputs, and decomposing the generated energy into basic electric quantity and weather sensitive electric quantity; the basic electric quantity takes seasonal changes, holiday effects and weekend effects into consideration; weather sensitive electric quantity is correlated with weather data, and the weather data comprises: air temperature, humidity, wind speed;
adjusting the internal parameters of the network, adjusting the learning rate and the iteration times according to a set value, acquiring input weight, cycle weight and deviation, and adjusting the parameters of a corresponding input gate, a forgetting gate, a selected gate and an output gate by using a gate activation function and a state activation function respectively;
and 4, step 4: training and fitting the model; training an LSTM network model with the generated energy requirement, inputting the training set data subjected to the normalization processing in the step (1) into the LSTM network model with the generated energy requirement for training until the network is converged;
in the process, an average absolute error (MAE) loss function and an Adam algorithm with a random gradient descending are used, on one hand, the learning rate of each parameter is dynamically modified, and on the other hand, a momentum method is introduced, so that the parameter updating has more chances to jump out of local optimum, and the network convergence is accelerated and optimized;
and 5: and (3) performing model overfitting evaluation, after the model is fitted, predicting the test data set obtained in the step (2) by using the LSTM model fitted in the step (4), adjusting the scale of the test data set by using an expected power generation amount value and an operation period index, and performing prediction and test on the data according to the ratio of 1: 1, combining the two models, inputting the two models into a fitted LSTM model, displaying the model loss of a training set and the model loss of a testing set in a graph, and judging whether the model has an overfitting phenomenon;
step 6: model prediction; after model evaluation and verification, inputting the currently generated data sampled on line into the LSTM neural network verified in the step 5 to predict the power generation amount value at the future time.
The invention has the beneficial effects that:
the invention sets the model supervised learning problem as: and predicting the power generation demand of the next time period according to the power generation of the time period and the data of the main influence factors. Firstly, generating capacity and historical data of main influence factors are collected. Acquiring data related to power generation from historical data of an enterprise production monitoring system, wherein the data comprises power generation data and main influence factor data, arranging the data according to time to obtain a sequence sample data set, and performing normalization processing; in order to facilitate model training, dividing prepared data into a training set and a test set in proportion; establishing an LSTM network model for generating capacity demand prediction, inputting a plurality of main influence factor data into the model, decomposing the generating capacity into basic electric quantity and meteorological sensitive electric quantity, and adjusting network internal parameters; inputting the normalized training set data into the generated energy LSTM network model for model training, introducing a momentum method, dynamically modifying the learning rate of each parameter, and completing LSTM network model fitting; combining the prediction with the test data, adjusting the scale of the test data set by using the expected power generation amount, inputting the adjusted scale to a fitted LSTM model, comparing and displaying model training and test loss data, and judging whether the model has an overfitting phenomenon; after model evaluation and verification, inputting the current generated data sampled on line into a verified LSTM neural network to predict the power generation amount value at the future time.
Drawings
Fig. 1 is an annual power generation timing diagram for establishing a power generation demand forecast.
FIG. 2 is a schematic diagram of steps of a method for predicting a power generation demand based on LSTM deep learning.
Fig. 3 is a diagram of an LSTM network model for power generation demand provided by the present invention.
Fig. 4 is a flow chart inside the power generation demand prediction LSTM network provided by the present invention.
FIG. 5 is a graph of the present invention for determining loss of training and testing for overfitting of the LSTM model.
Detailed Description
The invention is further illustrated by the following detailed description in conjunction with the accompanying drawings.
A method for predicting a power generation demand based on LSTM deep learning is characterized by comprising the following steps:
step 1: generating capacity and main influence factor historical data are collected and arranged according to time to obtain a sequence sample data set, then the data are normalized to enable the numerical value to be between [0 and 1], and the data set is obtained and used as a sample for supervised learning. This step prepares the dataset for the LSTM model. Data related to the power generation capacity, including power generation capacity data and main influence factor data, are collected from historical data of the enterprise production monitoring system. The main influence factor data comprises: weather data, date attribute data (season, week, holidays), power generation equipment output and reliability data, power grid scheduling data and the like.
First, the power generation capacity and the historical data of the main influence factors are collected. These data are typically obtained from a supervisory control and data acquisition (SCADA/SIS/ERP) system via an API/SDK interface, and FIG. 1 is a sample set of a year-round (2018.1.1-2018.12.31) power generation sequence in a certain area. We choose the weather temperature as the correlation with the main impact factor of the power generationData, collecting the highest weather temperature and the lowest weather temperature of the same region in the same period and generating capacity to form an original data set, arranging the original data set according to time to obtain a sequence sample data set, and carrying out normalization processing to obtain an original collection sequence St＝{Xt(1),Xt(2),Xt(3),...Xt(u)},XtE to { U, I }, and carrying out normalization processing on the original collected data, wherein the calculation formula is as follows:
Xi＝(Xi-Xmin)/(Xmax-Xmin)
wherein: xi-a normalized value of the ith value in the collected sample
Xi-the ith value in the collected sample
Xmin-minimum value in the collected sample
Xmax- -maximum value in the collected sample
Step 2: the data is divided into a training set and a test set. To facilitate model training, we will fit the model using only 2 years of data and then evaluate the remaining 1 year of data, i.e., training set and test set data size 2: 1, ensuring that each data set sampling period can represent a characteristic change sample in the same time interval;
and step 3: defining and establishing an LSTM network model of generating capacity demand, setting data input of a plurality of main influence factors, and adjusting internal parameters of the network; after format conversion of an input layer, transmitting the data collected by the LSTM network model to the LSTM layer for LSTM network training, and outputting a classification result by a classification output layer after sequentially passing through a full connection layer and a SoftMax layer;
the model inputs a plurality of main influence factor data and decomposes the generated energy into basic electric quantity and weather sensitive electric quantity. The basic electric quantity takes seasonal changes, holiday effects and weekend effects into consideration; weather sensitive electric quantity is correlated with weather data, and the weather data comprises: air temperature, humidity, wind speed, etc.
And adjusting the internal parameters of the network, adjusting the learning rate and the iteration times according to a set value, acquiring the input weight, the cycle weight and the deviation, and adjusting the parameters of the corresponding input gate, the forgetting gate, the selected gate and the output gate by using a gate activation function and a state activation function respectively.
Establishing an electric energy production demand LSTM network model, and establishing the electric energy production demand LSTM network model shown in figure 3, wherein the LSTM model comprises the following steps: the system comprises a data sampling layer a, a data preprocessing layer b, an input layer c, an LSTM layer d, a dropout layer e, a full connection layer f, a SoftMax layer g and a classification output layer h. Wherein: the layer a is used for collecting generating capacity and weather data, the layer b is used for carrying out normalization processing on the layer a data, the layer c is used for data diversity and format conversion, the layer d is used for LSTM network training, the layer e is used for monitoring network overfitting, and finally the layers f, g and h are used for outputting and optimizing results,
the model inputs a plurality of main influence factor data and is characterized in that the generated energy is decomposed into basic electric quantity and meteorological sensitive electric quantity, the basic electric quantity takes seasonal changes, holiday effects and weekend effects into consideration, and the meteorological sensitive electric quantity has the best correlation with the highest air temperature and the lowest air temperature.
The input data of the LSTM model is reconstructed into the LSTM expected 3D format (samples, time steps, input dim), where samples represents the number of samples, time steps represents the time step, and input dim represents the dimension at each time step. And predicting the load value var (t + n) of the future time (t +1, t + n) as an output sequence Y by taking the load values var (t-n) and var (t) of the current time t and the past time (t-n, t-1) in the sequence data set as an input sequence X of the LSTM neural network prediction model, wherein the prediction step n is a positive integer.
The established model adjusts the learning rate and the iteration times according to a set value, obtains an input weight W, a loop weight R and a deviation b, and adjusts corresponding hyper-parameters of an input gate i, a forgetting gate f, an alternative gate g and an output gate o by respectively utilizing a gate activation function and a state activation function, which is specifically shown in fig. 4. The LSTM network internal learning weights are:
W＝[Wi,Wf,Wg,Wo]T
R＝[Ri,Rf,Rg,Ro]T
b＝[bi,bf,bg,bo]T
the one-time iteration process of the generated energy demand prediction LSTM network model comprises the following steps:
Figure BDA0003372493840000081
wherein σgFor the gate activation function, a sigmoid function is used, and σ (X) ═ 1+ e-x)-1,σcFor the state activation function, use is made of the tanh function, Ct,htThe output value of the network unit after one calculation.
And 4, step 4: training and fitting the model; and (3) training the LSTM network model of the power generation requirement, and inputting the training set data subjected to the normalization processing in the step (1) into the LSTM network model of the power generation requirement for training until the network is converged. In the process, an average absolute error (MAE) loss function and an efficient Adam algorithm with a descending random gradient are used, on one hand, the learning rate of each parameter is dynamically modified, and on the other hand, a momentum method is introduced, so that the parameter updating has more chances to jump out of local optimum, and the network convergence is accelerated and optimized.
And 5: and (5) evaluating model overfitting. After fitting, we used the model to predict the test data set obtained in step 2. At this time, the scale of the test data set is adjusted by using the expected generated electricity quantity value and the operation period index, and the prediction and test data are calculated according to the following ratio of 1: 1, inputting the model loss of the training set and the model loss of the testing set into a fitted LSTM model, displaying the model loss of the training set and the model loss of the testing set in a graph, and judging whether the model has an overfitting phenomenon.
The error fraction of the model is calculated using the Root Mean Square Error (RMSE) of the unit errors that are the same as the variables from the initial predicted and actual values. The loss of training and testing is output at the end of each training epoch. Finally, the final RMSE of the model to the test data set is output, as shown in FIG. 5, and we can see that the model achieves better RMSE test results.
Step 6: and (5) model prediction. After model evaluation and verification, inputting the currently generated data sampled on line into the LSTM neural network verified in the step 5 to predict the power generation amount value at the future time.

Claims (4)

1. A method for predicting a power generation demand based on LSTM deep learning is characterized by comprising the following steps:
step 1: collecting power generation capacity and historical data of main influence factors, arranging the historical data according to time to obtain a sequence sample data set, then carrying out normalization processing on the data to enable the numerical value of the data to be between [0 and 1], and obtaining a data set as a sample for supervised learning, wherein the historical data of the main influence factors comprises: weather data, date attribute data, power generation equipment output and reliability data and power grid dispatching data.
Step 2: dividing the data into a training set and a test set; the prepared sequence data are processed according to the following steps of 2: 1, dividing the data into a training set and a test set in proportion, and ensuring that the sampling period of each data set can represent a characteristic change sample in the same time period;
and step 3: defining and establishing an LSTM network model of generating capacity demand, setting data input of a plurality of main influence factors, and adjusting internal parameters of the network; after format conversion of an input layer, transmitting the data collected by the LSTM network model to the LSTM layer for LSTM network training, and outputting a classification result by a classification output layer after sequentially passing through a full connection layer and a SoftMax layer;
and 4, step 4: training and fitting the model; training an LSTM network model with the generated energy requirement, inputting the training set data subjected to the normalization processing in the step (1) into the LSTM network model with the generated energy requirement for training until the network is converged;
and 5: and (3) performing model overfitting evaluation, after the model is fitted, predicting the test data set obtained in the step (2) by using the LSTM model fitted in the step (4), adjusting the scale of the test data set by using an expected power generation amount value and an operation period index, and performing prediction and test on the data according to the ratio of 1: 1, combining the two models, inputting the two models into a fitted LSTM model, displaying the model loss of a training set and the model loss of a testing set in a graph, and judging whether the model has an overfitting phenomenon;
step 6: model prediction; after model evaluation and verification, inputting the currently generated data sampled on line into the LSTM neural network verified in the step 5 to predict the power generation amount value at the future time.
2. The method according to claim 1, wherein step 3 sets a plurality of data inputs of main influence factors to decompose the power generation into basic power and weather sensitive power; the basic electric quantity takes seasonal changes, holiday effects and weekend effects into consideration; weather sensitive electric quantity is correlated with weather data, and the weather data comprises: air temperature, humidity, wind speed.
3. The method of claim 1, wherein the step 3 adjusts the internal parameters of the network, adjusts the learning rate and the iteration times according to the set values, obtains the input weight, the loop weight and the deviation, and adjusts the parameters of the corresponding input gate, the forgetting gate, the selected gate and the output gate by using the gate activation function and the state activation function respectively.
4. The method of claim 1, wherein in the step 4 network convergence process, an average absolute error (MAE) loss function and an Adam algorithm with a random gradient descent are used, so that on one hand, the learning rate of each parameter is dynamically modified, and on the other hand, a momentum method is introduced, so that the parameter update has more opportunities to jump out of local optimum, and the network convergence is accelerated and optimized.
CN202111406645.1A 2021-11-24 2021-11-24 Generating capacity demand prediction method based on LSTM deep learning Pending CN114021848A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111406645.1A CN114021848A (en) 2021-11-24 2021-11-24 Generating capacity demand prediction method based on LSTM deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111406645.1A CN114021848A (en) 2021-11-24 2021-11-24 Generating capacity demand prediction method based on LSTM deep learning

Publications (1)

Publication Number Publication Date
CN114021848A true CN114021848A (en) 2022-02-08

Family

ID=80066141

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111406645.1A Pending CN114021848A (en) 2021-11-24 2021-11-24 Generating capacity demand prediction method based on LSTM deep learning

Country Status (1)

Country Link
CN (1) CN114021848A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114529071A (en) * 2022-02-11 2022-05-24 杭州致成电子科技有限公司 Method for predicting power consumption of transformer area
CN115249090A (en) * 2022-07-04 2022-10-28 重庆大学 Electric quantity prediction method and system based on homomorphic encryption
CN116565846A (en) * 2023-05-11 2023-08-08 国网上海市电力公司 Virtual power plant demand prediction method, system and readable storage medium

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114529071A (en) * 2022-02-11 2022-05-24 杭州致成电子科技有限公司 Method for predicting power consumption of transformer area
CN115249090A (en) * 2022-07-04 2022-10-28 重庆大学 Electric quantity prediction method and system based on homomorphic encryption
CN116565846A (en) * 2023-05-11 2023-08-08 国网上海市电力公司 Virtual power plant demand prediction method, system and readable storage medium

Similar Documents

Publication Publication Date Title
Lin et al. An improved moth-flame optimization algorithm for support vector machine prediction of photovoltaic power generation
CN109711620B (en) Short-term power load prediction method based on GRU neural network and transfer learning
CN113962364B (en) Multi-factor power load prediction method based on deep learning
CN110766212B (en) Ultra-short-term photovoltaic power prediction method for historical data missing electric field
CN114021848A (en) Generating capacity demand prediction method based on LSTM deep learning
CN108921339B (en) Quantile regression-based photovoltaic power interval prediction method for genetic support vector machine
Wang et al. LSTM-based short-term load forecasting for building electricity consumption
CN112116130B (en) Short-term scheduling rule extraction method based on variable structure deep learning framework
CN113496311A (en) Photovoltaic power station generated power prediction method and system
CN110969306B (en) Power distribution low-voltage transformer area load prediction method and device based on deep learning
CN113554466A (en) Short-term power consumption prediction model construction method, prediction method and device
CN106446440A (en) Short-term photovoltaic generation power prediction method based on online sequential extreme learning machine
CN115688993A (en) Short-term power load prediction method suitable for power distribution station area
CN115965132A (en) Distributed photovoltaic digital twin system power prediction method based on GA-BP neural network
CN115481918A (en) Active sensing and predictive analysis system for unit state based on source network load storage
CN114611757A (en) Electric power system short-term load prediction method based on genetic algorithm and improved depth residual error network
CN114444805A (en) Control method for smooth output of multi-photovoltaic power station shared energy storage system
Alharbi et al. Short-term wind speed and temperature forecasting model based on gated recurrent unit neural networks
CN111177278A (en) Grid user short-term load prediction real-time processing tool
CN116128211A (en) Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene
CN108345996A (en) A kind of system and method reducing wind power checking energy
Huang et al. Power Prediction Method of Distributed Photovoltaic Digital Twin System Based on GA-BP
CN112836876A (en) Power distribution network line load prediction method based on deep learning
Liu et al. Short-Term Wind Speed and Wind Power Prediction Based on Meteorological Model Modification
Zhao et al. Short‐term wind power prediction based on combined long short‐term memory

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination