CN114970997A - Short-term prediction method for regional power load - Google Patents
Short-term prediction method for regional power load Download PDFInfo
- Publication number
- CN114970997A CN114970997A CN202210540765.9A CN202210540765A CN114970997A CN 114970997 A CN114970997 A CN 114970997A CN 202210540765 A CN202210540765 A CN 202210540765A CN 114970997 A CN114970997 A CN 114970997A
- Authority
- CN
- China
- Prior art keywords
- data
- load
- training
- prediction
- short
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 230000006870 function Effects 0.000 claims abstract description 49
- 230000005284 excitation Effects 0.000 claims abstract description 18
- 238000013528 artificial neural network Methods 0.000 claims abstract description 16
- 238000004458 analytical method Methods 0.000 claims abstract description 13
- 238000012360 testing method Methods 0.000 claims abstract description 11
- 238000010606 normalization Methods 0.000 claims abstract description 10
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 29
- 210000002569 neuron Anatomy 0.000 claims description 22
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims description 18
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims description 18
- 238000003062 neural network model Methods 0.000 claims description 4
- 238000012897 Levenberg–Marquardt algorithm Methods 0.000 claims description 3
- 230000009191 jumping Effects 0.000 claims description 3
- 238000007689 inspection Methods 0.000 claims description 2
- 238000007418 data mining Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 abstract description 2
- 239000000446 fuel Substances 0.000 abstract description 2
- 238000010219 correlation analysis Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 230000001932 seasonal effect Effects 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G06N3/045—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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Tourism & Hospitality (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- General Business, Economics & Management (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Marketing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a short-term prediction method of regional power load, for input data, firstly, adopting a statistical function model to carry out preliminary analysis on the correlation between relevant influence factors and load data, and secondly, adopting a Z-score standardization algorithm to carry out further normalization processing on the basis of the first analysis result to obtain a final input result; thirdly, processing input data by using BP neural networks with 4 different excitation functions to obtain predicted data; and fourthly, predicting the test sample through the established prediction model. Compared with the prior art, the method has the advantages of strong nonlinear fitting capability in experimental data, higher prediction precision, lower average absolute error percentage, suitability for engineering practicality, capability of serving as a basic technology of power plant power load prediction data, auxiliary application to the field work of power plant spot market transaction, fuel scheduling, output analysis, data mining and the like.
Description
Technical Field
The invention relates to the technical field of power load prediction, in particular to a short-term prediction method for regional power load.
Background
The regional power load has the characteristics of strong uncertainty and large time correlation, and the short-term prediction of the regional power load is the basic work of planning and scheduling of a power distribution network system and becomes one of the hot spots of energy research. The currently common data analysis and prediction methods can be mainly classified into a class-extrapolation method, a function method, a qualitative prediction method, a time sequence prediction method, a causal relationship prediction method and the like.
The regional power load is suitable for a data analysis and prediction method based on a time sequence, is a regression prediction method, belongs to quantitative prediction, acknowledges the continuity of the development of objects, performs statistical analysis by using past time sequence data, and predicts the development trend of the objects; however, the method has the problems that the analysis data needs to be specified by human experience, the prediction precision is low, the selection of the load prediction method has large influence on the prediction result, the method has the defect of sensitivity to an abnormal value, the universality is poor, and the method is not suitable for management and other works.
Disclosure of Invention
The present invention is directed to a method for short-term prediction of regional power load, which overcomes the above-mentioned drawbacks of the prior art.
The purpose of the invention can be realized by the following technical scheme:
a method of short term prediction of regional power loads, the method comprising the steps of:
step 1: regional power load data (training data and test data sources) are collected, and a statistical function model is adopted to perform preliminary analysis on the correlation between relevant influence factors and the load data. The method comprises the following specific steps:
step 1.1: selecting hourly load data of data N days except holidays in a certain area and meteorological data such as dry bulb temperature, dew point temperature, humidity and the like;
step 1.2: analyzing the correlation between the relevant influence factors and the load data by adopting a CORREL statistical function model; the expression of the CORREL statistical function model is as follows:
in the formula: corel (X, Y) is the correlation coefficient; x and y are numerical values of all variables;is the average value of the variables. Different values of the correlation coefficient Correl (X, Y) represent different characteristic standards, and when the value of the correlation coefficient Correl (X, Y) is 0-0.3, the characteristic standards are micro-correlation; when the value of the correlation coefficient Correl (X, Y) is 0.3-0.5, the characteristic standard is real correlation; when the value of the correlation coefficient Correl (X, Y) is 0.5-0.8, the characteristic standard is significant correlation; the characteristic criterion is highly correlated when the correlation coefficient Correl (X, Y) has a value of 0.8 to 1.0.
Step 1.3: analyzing the correlation between each variable and actual load data according to the analysis result in the step 1.2;
step 1.4: repeating the steps 1.2 and 1.3 until all relevant influence factors complete the relevance analysis;
relevant influencing factors include dry bulb temperature, dew point, relative humidity, wind speed level, average load in the previous 24 hours, load at the same time of the previous day, load at the same time of the previous week, etc.
Step 1.5: taking out data X from data set (data set composed of all analyzed related influence factors) according to correlation characteristic standard i For data X i Performing normalization treatment, and giving X i The mean and standard deviation of the data (mean) were normalized. The processed data are in accordance with a standard normal distribution, i.e. mean value of0, standard deviation of 1; the normalization process is performed using the following formula:
in the formula: z is the variable after normalization, and μ and σ are the variables X, respectively i Sample mean and sample standard deviation.
Step 2: and (3) training a BP neural network algorithm model for the normalized data set obtained in the step (1). Specifically, the method comprises the following steps:
step 2.1: selecting the number N of hidden layer neurons of the BP neural network, setting excitation functions of the hidden layer neurons and the output layer neurons as a tansig function and a purelin function respectively, randomly initializing a network weight and a threshold, and taking the number of iterations as 5000.
Step 2.2: repeating training on the training sample for 10 times by adopting a trainlm learning function and a Levenberg-Marquardt algorithm, and calculating the average absolute error percentage MAPE average value of 10 training results; the number of network epoch iterations is 5000, and the expected error target is 0.00000001.
Step 2.3: if the average absolute error percentage MAPE is the minimum value, ending; otherwise, jumping to the step 2.1 to continuously reselect the hidden layer number N;
step 2.4: and (3) randomly initializing a network weight and a threshold for the neural network with the number of the hidden layer neurons being N, and taking the iteration number as 30000 times. Training was repeated using the trainglm learning function.
And step 3: and (3) carrying out precision inspection, selection and training on the BP neural network model obtained in the step (2), and predicting the regional power load by using a prediction model. The method comprises the following specific steps:
step 3.1: setting the iteration number to be 100000 times, and performing deep training on the neural network in which the number of the selected hidden layer is N, the number of the hidden layer neurons is k, and excitation functions of the hidden layer neurons and the output layer neurons are respectively a tansig function and a purelin function.
Step 3.2: and comparing the MAPE values, selecting an excitation function model, and performing iterative training.
Step 3.3: and ending the iterative training when the target error is reached.
Step 3.4: in order to test the generalization capability of the model, the established prediction model is used for predicting the regional power load test sample.
Compared with the prior art, the short-term prediction method of the regional power load provided by the invention at least comprises the following beneficial effects:
1) according to the method, the test sample is predicted through the established prediction model, the method shows strong nonlinear fitting capability in experimental data, the prediction precision is higher, the average absolute error percentage (MAPE) is lower, and the method is suitable for engineering application;
2) the algorithm of the invention gets rid of the limitation of a single hidden layer function on the prediction data by comparing the prediction results under the conditions of different hidden layer numbers, neuron numbers and excitation functions, and tests the fitted prediction data, so that the prediction precision is higher, the data is more visual, the universality is higher, and the algorithm is suitable for management work;
3) the method can be used as a basic technology of power plant power load prediction data, and can be used for assisting the work in the fields of power plant spot market transaction, fuel scheduling, output analysis, data mining and the like.
Drawings
FIG. 1 is a flow chart illustrating a short-term prediction method for regional power loads according to an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating a short-term prediction method of regional power loads according to an embodiment of the present invention;
FIG. 3 is a flow chart of a BP neural network algorithm in an embodiment;
FIG. 4 is a purelin-purelin excitation function model MAPE curve in an example;
FIG. 5 is a graph of a purelin-tansig excitation function model MAPE in an embodiment;
FIG. 6 is a tan sig-tan sig stimulus function model MAPE curve in an embodiment;
FIG. 7 is a tan sig-purelin excitation function model MAPE curve in an embodiment;
FIG. 8 is a graph of a single implicit layer error versus a mean absolute error distribution for an embodiment;
FIG. 9 is a graph of the prediction error statistics divided by hour in the example;
FIG. 10 is a graph of the prediction error statistics by the division of the week in the example;
FIG. 11 illustrates an embodiment of a single hidden layer test for predicted load and actual load;
FIG. 12 is a single implicit layer mean absolute error percentage MAPE in an example.
Detailed Description
The invention is described in detail below with reference to the figures and the specific embodiments. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, shall fall within the scope of protection of the present invention.
Examples
At present, statistical analysis is carried out on seasonal trend of load and exogenous variable data such as temperature, humidity and wind speed of an area, and good correlation is proved to exist between the seasonal trend of load and the exogenous variable data; considering the temperature similarity and the date proximity, dividing the data of the power load data of a certain area into smaller areas according to the data of the power load data of the certain area; the method is characterized in that a multi-purpose curve method, an index method and a comparison method are used for analyzing the load characteristics of years, months, weeks, days and various industries and is used for researching a correlation analysis method of special load and daily load characteristics.
The invention relates to a short-term prediction method of regional power load, which is a regional power load prediction method based on Tansig-Purelin, Purelin-Tansig, Tansig-Tansig and Purelin-Purelin excitation function BP neural network, and the output load is determined by the precision of MAPE value. Specifically, as shown in fig. 1 and 2, the method includes the following steps:
acquiring regional power load data, and performing primary analysis on the correlation between relevant influence factors and the load data by adopting a statistical function model; specifically, the method comprises the following steps:
1.1, the hourly load data of 1771 days of data in three years in a certain area except 54 days of holidays and festivals are selected, and meteorological data such as dry bulb temperature, dew point temperature, humidity and the like are selected.
1.2, analyzing the correlation between the relevant influence factors and the load data by using a CORREL statistical function model;
the expression of the CORREL statistical function model is as follows:
in the formula: corel (X, Y) is the correlation coefficient; x and y are numerical values of all variables;are the variable means. The different values of the correlation coefficient Correl (X, Y) represent different characteristic criteria, as shown in table 1.
TABLE 1 correlation coefficient and characteristic standards
And 1.3, analyzing the correlation of each variable and the actual load data according to the analysis program in the step 1.2.
1.4, repeating the steps 1.2 and 1.3 until all relevant influence factors complete correlation analysis; all relevant influencing factors include dry bulb temperature, dew point, relative humidity, wind speed rating, average load over the first 24 hours, load at the same time of the day before, load at the same time of the week before, etc.
The correlation between the load data and the variables analyzed in this embodiment is shown in table 2:
TABLE 2 correlation of load data to variables
Wherein, V, C, Dry, Dew, R, WS, AL, HL, WL represent variables, correlation coefficients, Dry bulb temperature, Dew point, relative humidity, wind speed level, average load of the previous 24 hours, load at the same time of the previous day, and load at the same time of the previous week, respectively.
1.5, forming a data set by all the analyzed related influence factors, and taking out data X from the data set according to the correlation characteristic standard i For data X i Performing normalization treatment, and giving X i The mean and standard deviation of the data (mean) were normalized. The processed data are in accordance with standard normal distribution, namely the mean value is 0 and the standard deviation is 1; the normalization process is performed using the following formula:
in the formula: z is the variable after normalization, and μ and σ are the sample mean and sample standard deviation, respectively, of the variable V.
Step two, training a BP neural network algorithm model for the normalized data set obtained in the step one; the BP neural network algorithm model is shown in fig. 3. The method specifically comprises the following steps:
2.1, selecting the number N of hidden layer neurons of the BP neural network, and setting excitation functions of the hidden layer neurons and the output layer neurons as a tansig function and a purelin function respectively, as shown in Table 3. And randomly initializing a network weight and a threshold, and taking 5000 iterations.
TABLE 3 different excitation functions
2.2, repeatedly training 10 times on the training sample by adopting a trainlm learning function and a Levenberg-Marquardt algorithm, and calculating the average absolute error percentage MAPE average value of 10 training results; the number of network epoch iterations is 5000, and the expected error target is 0.00000001.
2.3, if the average absolute error percentage MAPE is the minimum value, ending; otherwise, jumping to 2.1 to continuously reselect the hidden layer number N;
2.4, randomly initializing a network weight and a threshold for the neural network with the hidden layer neuron number of 19, and taking 30000 iterations. The training is repeated using the trainglm learning function. MAPE curves obtained by different excitation functions are shown in FIGS. 4-7, and specific MAPE values obtained by each excitation function are shown in Table 4.
TABLE 4 different excitation functions
Step three, training and precision testing are carried out on the BP neural network model obtained in the step two; specifically, the method comprises the following steps:
3.1, setting the iteration number to be 100000 times, selecting the number of hidden layers to be 1, setting the number of hidden layer neurons to be 19, and enabling the activation functions of the hidden layer neurons and the output layer neurons to be respectively a tansig function and a purelin function. And (5) carrying out deep training on the neural network to achieve the target error, and ending the iterative training. The resulting single hidden layer error versus average absolute error is shown in fig. 8.
3.2, the statistical information of prediction errors divided by hour is shown in FIG. 9, and the statistical information of prediction errors divided by week is shown in FIG. 10. The simulation curve of the training sample is shown in fig. 11, wherein the dark black curve is actual load prediction, the light gray curve is predicted load, it can be known that the fitting trend of the predicted load is consistent with the actual sample trend, and further, the average absolute error percentage MAPE of the predicted load and the actual load is shown in fig. 12, which further illustrates that the BP neural network model has better fitting capability on the training sample for the short-term power load data of the a area.
While the invention has been described with reference to specific embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (6)
1. A short-term prediction method for regional power load is characterized by comprising the following steps:
1) acquiring power load data of an area to be predicted as a source of training data and test data, and analyzing the correlation between the load data and relevant influence factors by adopting a statistical function model to obtain an analysis result;
2) training a BP neural network algorithm model based on the analysis result of the step 1);
3) carrying out precision inspection, excitation function selection and prediction model training on the BP neural network model trained in the step 2); and predicting the power load test samples of the area to be predicted based on the prediction model.
2. The method of short-term prediction of regional power loads according to claim 1, wherein the relevant influencing factors include, but are not limited to, dry bulb temperature, dew point, relative humidity, wind speed rating, average load over the previous 24 hours, load at the same time of the previous day and load at the same time of the previous week.
3. The method for short-term prediction of regional power loads according to claim 2, wherein the specific steps of step 1) comprise:
101) selecting hourly load data of data N days except holidays of an area to be predicted and related influence factors;
102) analyzing the correlation between the relevant influence factors and the load data by using a CORREL statistical function model;
103) analyzing the correlation between each relevant influence factor and the actual load data according to the analysis result in the step 102);
104) repeating steps 102) and 103) until all relevant influencing factors are subjected to relevance analysis;
105) and (4) forming a data set by all the analyzed related influence factors, taking out data from the data set according to the correlation characteristic standard, and carrying out normalization processing on the data.
4. The method for short-term prediction of regional power loads according to claim 3, wherein the normalization process is performed using the following equation:
in the formula: x i For data taken from the data set according to the correlation property criteria, z is X i The variables, mu and sigma, after normalization are respectively the sample mean and the sample standard deviation of the variable V corresponding to the data X i 。
5. The method for short-term prediction of regional power loads according to claim 1, wherein the specific steps of step 2) comprise:
201) selecting the number N of hidden layer neurons of a BP neural network, setting excitation functions of the hidden layer neurons and the output layer neurons as a tansig function and a purelin function respectively, randomly initializing a network weight and a threshold, and taking the number of iterations as 5000;
202) repeating training on the training sample for 10 times by adopting a trainlm learning function and a Levenberg-Marquardt algorithm, and calculating the average absolute error percentage MAPE average value of 10 training results; setting the iteration times of the network epoch to be 5000 times, and setting the expected error target to be 0.00000001 time;
203) comparing the average absolute error percentage in real time, and finishing iteration if the average absolute error percentage MAPE reaches the minimum value; otherwise, jumping to step 201) to continuously reselect the hidden layer number N;
204) and (3) for the neural network with the number of the neurons in the hidden layer being N, randomly initializing a network weight and a threshold, taking the number of iterations as 30000, and adopting a thinglm learning function to train repeatedly.
6. The method for short-term prediction of regional power loads according to claim 5, wherein the specific steps of step 3) comprise:
301) setting the iteration number to be 100000 times, and carrying out deep training on the neural network in which the number of the selected hidden layer is N, the number of the hidden layer neurons is k, and excitation functions of the hidden layer neurons and the output layer neurons are respectively a tansig function and a purelin function;
302) comparing the MAPE values, selecting an excitation function model, and performing iterative training;
303) judging whether the target error is reached in real time, and if so, ending the iterative training;
304) and predicting the power load test sample of the region to be predicted by using the established prediction model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210540765.9A CN114970997A (en) | 2022-05-17 | 2022-05-17 | Short-term prediction method for regional power load |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210540765.9A CN114970997A (en) | 2022-05-17 | 2022-05-17 | Short-term prediction method for regional power load |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114970997A true CN114970997A (en) | 2022-08-30 |
Family
ID=82983781
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210540765.9A Pending CN114970997A (en) | 2022-05-17 | 2022-05-17 | Short-term prediction method for regional power load |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114970997A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116204792A (en) * | 2023-04-28 | 2023-06-02 | 北京航空航天大学 | Training method for generating causal interpretation model |
-
2022
- 2022-05-17 CN CN202210540765.9A patent/CN114970997A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116204792A (en) * | 2023-04-28 | 2023-06-02 | 北京航空航天大学 | Training method for generating causal interpretation model |
CN116204792B (en) * | 2023-04-28 | 2023-07-14 | 北京航空航天大学 | Training method for generating causal interpretation model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Sun et al. | Using Bayesian deep learning to capture uncertainty for residential net load forecasting | |
CN110046743B (en) | Public building energy consumption prediction method and system based on GA-ANN | |
Liu et al. | A comparative study of the data-driven day-ahead hourly provincial load forecasting methods: From classical data mining to deep learning | |
CN112116144B (en) | Regional power distribution network short-term load prediction method | |
CN111860658A (en) | Transformer fault diagnosis method based on cost sensitivity and integrated learning | |
CN113011796B (en) | Edible oil safety early warning method based on' analytic hierarchy process-neural network | |
CN114358213B (en) | Error ablation processing method, system and medium for nonlinear time series data prediction | |
CN114970997A (en) | Short-term prediction method for regional power load | |
Fang et al. | DESA: a novel hybrid decomposing-ensemble and spatiotemporal attention model for PM2. 5 forecasting | |
CN115049019B (en) | Method and device for evaluating arsenic adsorption performance of metal organic framework and related equipment | |
CN113139605A (en) | Power load prediction method based on principal component analysis and LSTM neural network | |
CN114169639A (en) | Enterprise power consumption prediction method | |
Charkha | Stock price prediction and trend prediction using neural networks | |
Ahmed et al. | Recognizing a spatial extreme dependence structure: A deep learning approach | |
CN114581141A (en) | Short-term load prediction method based on feature selection and LSSVR | |
DeYoreo et al. | Bayesian mixture models with focused clustering for mixed ordinal and nominal data | |
Ticona et al. | Hybrid model based on genetic algorithms and neural networks to forecast tax collection: application using endogenous and exogenous variables | |
Gilljam et al. | The colour of environmental fluctuations associated with terrestrial animal population dynamics | |
Semenov et al. | CHARACTERIZATION OF UNCERTAINTIES IN THE ASSESSMENTS OF THE INTERGOVERNMENTAL PANEL ON CLIMATE CHANGE | |
CN115907131A (en) | Method and system for building electric heating load prediction model in northern area | |
CN116388150A (en) | Matching short-term load prediction method based on digital twin and neural network technology | |
CN115860797A (en) | Electric quantity demand prediction method suitable for new electricity price reform situation | |
CN114862032A (en) | XGboost-LSTM-based power grid load prediction method and device | |
Petruseva et al. | Model for predicting construction time by using general regression neural network | |
CN115456287A (en) | Long-and-short-term memory network-based multi-element load prediction method for comprehensive energy system |
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 |