CN115513951A - Power load prediction method and system based on concept drift detection - Google Patents
Power load prediction method and system based on concept drift detection Download PDFInfo
- Publication number
- CN115513951A CN115513951A CN202211463933.5A CN202211463933A CN115513951A CN 115513951 A CN115513951 A CN 115513951A CN 202211463933 A CN202211463933 A CN 202211463933A CN 115513951 A CN115513951 A CN 115513951A
- Authority
- CN
- China
- Prior art keywords
- power load
- data
- load prediction
- concept drift
- prediction model
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 66
- 238000001514 detection method Methods 0.000 title claims abstract description 51
- 230000008569 process Effects 0.000 claims description 34
- 238000012545 processing Methods 0.000 claims description 11
- 230000015654 memory Effects 0.000 claims description 10
- 238000012549 training Methods 0.000 claims description 10
- 238000012360 testing method Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000007787 long-term memory Effects 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 238000001276 Kolmogorov–Smirnov test Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000006403 short-term memory Effects 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 description 9
- 238000013507 mapping Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 230000005611 electricity Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000009193 crawling Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
-
- 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
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Power Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application belongs to the technical field of power system load prediction, and particularly relates to a power load prediction method and system based on concept drift detection, which comprises the following steps: acquiring historical data of the power load; constructing a power load prediction model by adopting a sliding time window according to the acquired power load historical data; acquiring a power load predicted value at the next moment of the sliding time window based on the constructed power load prediction model; capturing the original data input during the prediction based on the power load prediction model, and judging whether the input data based on the new sliding time window and the input original data belong to the same distribution or not to realize the prediction of the power load; the method and the device aim at the problem that better accuracy cannot be obtained due to the fact that concept drift is not considered in load prediction, and improve the accuracy and the reasonability of power load prediction by detecting whether the concept drift phenomenon exists in known data or not.
Description
Technical Field
The application belongs to the technical field of power system load prediction, and particularly relates to a power load prediction method and system based on concept drift detection.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The power load prediction is to predict the future value of the power load according to the known data of the power load, so as to provide reliable basis for operation decision makers such as power grid enterprises, power grid users and the like; therefore, there is a high demand for the accuracy of the power load prediction.
To the knowledge of the inventor, most of the existing power load prediction processes are trained once to predict future values, and actually, the power load data exists in the form of streaming data, and the mapping relationship between the power load and its influencing factors fluctuates with the use habits of users and the changes of the states of the power equipment, for example: when the habit of the electricity consumer changes or the electricity equipment fails, the accuracy of the prediction model is reduced; when the power load prediction model changes over time due to data changes of the power load over time, this phenomenon is called concept drift, which is a problem in the power load prediction process.
Disclosure of Invention
In order to solve the problems, the power load prediction method and the power load prediction system based on concept drift detection are provided for solving the problem that better accuracy cannot be obtained due to the fact that the phenomenon of concept drift cannot be considered in load prediction in the prior art, and the prediction result of the power load is more accurate and reasonable by detecting whether the concept drift phenomenon exists in known data or not.
According to some embodiments, a first aspect of the present application provides a power load prediction method based on concept drift detection, which adopts the following technical solutions:
a power load prediction method based on concept drift detection comprises the following steps:
acquiring historical data of the power load;
constructing a power load prediction model by adopting a sliding time window according to the acquired power load historical data;
acquiring a power load predicted value at the next moment of the sliding time window based on the constructed power load prediction model;
capturing original data input during prediction based on the power load prediction model, and realizing concept drift detection in the power load prediction process;
judging whether input data based on a new sliding time window and input original data in a base window belong to the same distribution or not in the process of detecting concept drift, if so, not generating the concept drift in the process of predicting the power load, and continuously using the constructed power load prediction model to predict the power load; if not, the concept drift occurs in the process of predicting the power load, and the power load prediction model needs to be reconstructed to predict the power load.
As a further technical limitation, after acquiring historical data of the power load, performing data preprocessing on the acquired historical data;
the data preprocessing comprises missing data completion processing and normalization processing;
the missing data completion processing is to fill the default position by using average adjacent loads;
the normalization process uses Min-Max method to normalize the data set after missing data completion processing, that is, the data set is normalized
Wherein, in the process,
in order to normalize the processed data, the data is normalized,
as the original data, it is the original data,
in order to be the largest of the original data,
is the smallest raw data.
As a further technical limitation, a power load historical data set is obtained based on the obtained power load historical data, a sliding time window is adopted on the power load historical data set, power load values of the current time and the historical time in the time window are used as an input sequence of the constructed power load prediction model, a power load value of the next time of the sliding time window is used as a prediction object, iterative optimization of the constructed power load prediction model is performed through inputting a training set in the power load historical data set, and training of the constructed power load prediction model is completed.
Further, the power load prediction model adopts a Long Short-Term Memory network (LSTM); and predicting the test set in the power load historical data set by using the trained LSTM neural network prediction model to obtain a power load prediction value at the next moment of the sliding time window.
As a further technical limitation, in the process of judging whether the input data based on the new sliding time window and the input original data in the base window belong to the same distribution, a K-S test is adopted to test the statistic D n Comprises the following steps:
wherein, in the process,
is the supremum limit of the function value,
is the current time of the sliding time window,
for the input data sequence of the new sliding time window,
is the original data sequence in the input base window.
Further, suppose
Calculating the absolute difference of the accumulated frequency between the input data based on the new sliding time window and the input original data, making the maximum absolute difference be
(ii) a Based on sample capacity
And level of significance
Obtaining a critical value
(ii) a If it is
Sending out early warning of concept drift; if it is
If the assumption is not true, the input data based on the new sliding time window is not distributed in the same way as the input original data, and a warning of concept drift is issued.
Further, when the early warning of concept drift is received, starting a temporary sliding time window, receiving data which sends out early warning information and data which follow, and training a temporary power load prediction model by using the temporary sliding time window; and when the concept drift alarm is received, replacing the original power load prediction model with the temporary power load prediction model, emptying the temporary sliding time window, and moving the base window to the next moment of the concept drift point.
According to some embodiments, a second aspect of the present application provides a power load prediction system based on concept drift detection, which adopts the following technical solutions:
a power load prediction system based on concept drift detection, comprising:
an acquisition module configured to acquire power load history data;
a modeling module configured to construct a power load prediction model using a sliding time window according to the acquired power load historical data;
a prediction module configured to obtain a power load prediction value at a time next to the time at which the sliding time window is located based on the constructed power load prediction model;
a concept drift detection module configured to capture raw data input in prediction based on the power load prediction model, and realize concept drift detection in a power load prediction process;
judging whether input data based on a new sliding time window and input original data in a base window belong to the same distribution or not in a concept drift detection module, if so, not generating concept drift in the process of predicting the power load, and continuously using the constructed power load prediction model to predict the power load; if not, the concept drift occurs in the process of predicting the power load, and the power load prediction model needs to be reconstructed to predict the power load.
According to some embodiments, a third aspect of the present application provides a computer-readable storage medium, which adopts the following technical solutions:
a computer readable storage medium, having stored thereon a program which, when executed by a processor, carries out the steps in the power load prediction method based on concept drift detection as described in the first aspect of the present application.
According to some embodiments, a fourth aspect of the present application provides an electronic device, which adopts the following technical solutions:
an electronic device comprising a memory, a processor and a program stored on the memory and executable on the processor, the processor implementing the steps of the power load prediction method based on concept drift detection according to the first aspect of the present application when executing the program.
Compared with the prior art, the beneficial effect of this application is:
by introducing the LSTM neural network, the long-term memory of the LSTM network to the power load data is fully utilized, the contribution of the current prediction of the historical information can be automatically adjusted according to the current state, the defect of poor long-term memory of the load data in the prior art is overcome, and the accuracy of the power load prediction is improved; due to the existence of the concept drift phenomenon, the data distribution is gradually changed in a data number value change mode and does not belong to the same distribution with the old data, a concept drift detection mechanism is introduced, the original data used for predicting by capturing the LSTM neural network is used, whether the original data and the old data of a new entering window belong to the same distribution or not is judged, if the original data and the old data belong to the same distribution, the mapping relation between the current data and the influence factors of the current data is not changed, the current prediction model is still effective, if the original prediction model does not belong to the same distribution, the mapping relation between the data and the influence factors of the data is changed, the original prediction model is not established, at the moment, an alarm is sent out, the prediction model under the new mapping relation is retrained by the system to replace the original model for performing prediction tasks, the prediction model is made to adapt to the changed new data in time, and the accuracy and the rationality of power load prediction are improved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, are included to provide a further understanding of the application, and the description of the exemplary embodiments and illustrations of the application are intended to explain the application and are not intended to limit the application.
Fig. 1 is a flowchart of a power load prediction method based on concept drift detection in an embodiment one of the present application;
FIG. 2 is a flowchart illustrating specific steps of a power load prediction method based on concept drift detection according to an embodiment of the present application;
FIG. 3 is a schematic structural diagram of a sliding window of power load data according to an embodiment of the present application;
FIG. 4 is a schematic diagram of a structure of a circulation unit of the LSTM neural network according to the first embodiment of the present application;
FIG. 5 is a flow chart of a conceptual drift detection mechanism in an embodiment one of the present application;
FIG. 6 is a schematic diagram of a sliding window structure of a conceptual drift detection mechanism according to an embodiment of the present application;
fig. 7 is a block diagram of a power load prediction system based on concept drift detection according to a second embodiment of the present application.
Detailed Description
The present application will be further described with reference to the following drawings and examples.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an", and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Example one
The embodiment of the application introduces a power load prediction method based on concept drift detection.
A power load prediction method based on concept drift detection as shown in fig. 1 and 2 includes:
acquiring historical data of the power load;
constructing a power load prediction model by adopting a sliding time window according to the acquired power load historical data;
acquiring a power load predicted value at a time next to the time of the sliding time window based on the constructed power load prediction model;
capturing original data input during prediction based on the power load prediction model, and realizing concept drift detection in the power load prediction process;
judging whether input data based on a new sliding time window and input original data in a base window belong to the same distribution or not in the process of detecting concept drift, if so, not generating the concept drift in the process of predicting the power load, and continuously using the constructed power load prediction model to predict the power load; if not, the concept drift occurs in the process of predicting the power load, and the power load prediction model needs to be reconstructed to predict the power load.
As one or more embodiments, in the process of acquiring historical data of the power load, taking the current time as t, and crawling 5000 time series power load data in the power monitoring equipment from the current time t to the front through Python script as historical data of the power load and inputting the power load into a power load prediction model; the power load data includes date, time, weather conditions, highest temperature, lowest temperature, active power, reactive power, real-time electricity price, and the like.
Specifically, after acquiring historical data of the power load, performing data preprocessing on the acquired historical data of the power load to obtain a data set as a sample, and dividing the data set into a training set and a test set;
the data preprocessing comprises missing data completion processing and normalization processing;
filling the default position by using the average adjacent load;
the normalization process uses Min-Max method to normalize the data set after missing data completion processing, that is, the data set is normalized
Wherein, in the process,
in order to normalize the processed data, the data is normalized,
as a result of the original data, it is,
in order to be the largest of the original data,
is the smallest raw data.
Obtaining a power load historical data set based on the obtained power load historical data, adopting a sliding time window on the power load historical data set, and as shown in fig. 3, taking the power load values of the current time and the historical time in the time window as an input sequence of the constructed power load prediction model, taking the power load value of the next time of the sliding time window as a prediction object, performing iterative optimization on the constructed power load prediction model by inputting a training set in the power load historical data set, and completing training of the constructed power load prediction model.
In the embodiment, the power load prediction model adopts an LSTM neural network; and predicting the test set in the power load historical data set by using the trained LSTM neural network prediction model to obtain a power load predicted value at the next moment of the sliding time window.
The structure of the circulation unit in the LSTM neural network, as shown in FIG. 4, includes the state of the cellsC t Forgetting doorf t Input gatei t And an output gateo t Defining and maintaining an internal memory cell state, i.e. the cell state, throughout the cycleC t (ii) a Passing forgetting doorf t Input gatei t And output gateo t Three gates update cell state and output information to external stateh t At each moment of timetInternal state ofC t The history information up to the current time is recorded, and the calculation method of different gates at each time in the loop unit structure is as follows:
wherein the content of the first and second substances,i t 、f t 、o t an input gate, a forgetting gate and an output gate,
the memory cell of the last time is the memory cell,
for a candidate state obtained by a non-linear function,W i 、W f 、W o 、W C 、b i 、b f 、b o 、b C respectively representing the corresponding weight coefficient matrix and bias term,
is an input for the current time of day,
is the external state at the last moment in time,
andtanhare respectively asLogisticFunctions and hyperbolic tangent activation functions.
In the present embodiment, selection of the sliding time Window length Window _ Size: setting the length of a time Window of the sliding interception load data Window _ Size, and ensuring that the Window _ Size is smaller than the lengths of a training set and a test set; the sliding time Window length Window Size ranges from 100 to 300.
As shown in fig. 5, the concept drift detection in the power load prediction process specifically includes the following steps:
(1) Connecting the new sliding time window with the base window, sliding on the data set, and starting to perform concept drift detection after the base window and the new window are filled with data, as shown in fig. 6;
(2) Adopting K-S test to judge whether the input data based on the new sliding time window and the input original data in the base window belong to the same distribution; test statistic D n Comprises the following steps:
wherein, in the step (A),
is the supremum limit of the function value,
is the current time instant of the sliding time window,
for the input data sequence of the new sliding time window,
inputting an original data sequence in the base window;
suppose that
Calculating the input data based on the new sliding time window and the original input dataAbsolute difference of accumulated frequency between data, let maximum absolute difference be
(ii) a Based on sample capacity
And level of significance
Obtaining a critical value
(ii) a If it is
Sending out early warning of concept drift; if it is
If the assumption is false, the input data based on the new sliding time window and the input original data do not belong to the same distribution, and an alarm of concept drift is sent out;
the determined conceptual drift nodes are displayed in a data list in a highlighted form; the concept drift node can be displayed in a highlight mode to play a warning role.
(3) When early warning of concept drift is received, starting a temporary sliding time window, receiving data which sends out early warning information and data which follow, and training a temporary power load prediction model by using the temporary sliding time window; and when a concept drift alarm is received, replacing the original power load prediction model with the temporary power load prediction model, emptying the temporary sliding time window, and moving the base window to the next moment of the concept drift point.
By introducing the LSTM neural network, the method makes full use of the long-term memory of the LSTM network to the power load data, can automatically adjust the contribution of the current prediction of the historical information according to the current state, avoids the defect of poor long-term memory of the load data in the prior art, and improves the precision of the power load prediction; due to the existence of the concept drift phenomenon, the data distribution is gradually changed in the form of the change of the data number value and does not belong to the same distribution as the old data, the original data used for predicting by capturing the LSTM neural network is introduced through a concept drift detection mechanism, whether the original data and the old data of a new entering window belong to the same distribution or not is judged, if the original data and the old data belong to the same distribution through detection, the mapping relation between the current data and the influence factors thereof is not changed, the current prediction model is still effective, if the original data and the influence factors thereof do not belong to the same distribution, the mapping relation between the data and the influence factors thereof is changed, the original prediction model is not established, at the moment, an alarm is sent out, the system retrains the prediction model under the new mapping relation to replace the original model to perform a prediction task, the prediction model is made to adapt to the changed new data in time, and the accuracy and the rationality of power load prediction are improved.
Example two
The second embodiment of the present application introduces a power load prediction system based on concept drift detection.
A power load prediction system based on concept drift detection as shown in fig. 7, comprising:
an acquisition module configured to acquire power load history data;
a modeling module configured to construct a power load prediction model using a sliding time window according to the acquired power load historical data;
a prediction module configured to obtain a power load prediction value at a time next to the time at which the sliding time window is located based on the constructed power load prediction model;
a concept drift detection module configured to capture raw data input in prediction based on the power load prediction model, and realize concept drift detection in a power load prediction process;
judging whether input data based on a new sliding time window and input original data in a base window belong to the same distribution or not in a concept drift detection module, if so, not generating concept drift in the process of predicting the power load, and continuously using the constructed power load prediction model to predict the power load; if not, the concept drift occurs in the process of predicting the power load, and the power load prediction model needs to be reconstructed to predict the power load.
The detailed steps are the same as those of the power load prediction method based on concept drift detection provided in the first embodiment, and are not described herein again.
EXAMPLE III
The third embodiment of the application provides a computer-readable storage medium.
A computer-readable storage medium, on which a program is stored, which when executed by a processor implements the steps in the power load prediction method based on concept drift detection as described in the first embodiment of the present application.
The detailed steps are the same as those of the power load prediction method based on concept drift detection provided in the first embodiment, and are not described herein again.
Example four
The fourth embodiment of the application provides electronic equipment.
An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the power load prediction method based on concept drift detection according to the first embodiment of the present application.
The detailed steps are the same as those of the power load prediction method based on concept drift detection provided in the first embodiment, and are not described herein again.
The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (10)
1. A power load prediction method based on concept drift detection is characterized by comprising the following steps:
acquiring historical data of the power load;
constructing a power load prediction model by adopting a sliding time window according to the acquired power load historical data;
acquiring a power load predicted value at the next moment of the sliding time window based on the constructed power load prediction model;
capturing original data input during prediction based on the power load prediction model, and realizing concept drift detection in the power load prediction process;
judging whether input data based on a new sliding time window and input original data in a base window belong to the same distribution or not in the process of detecting concept drift, if so, not generating the concept drift in the process of predicting the power load, and continuously using the constructed power load prediction model to predict the power load; if not, the concept drift occurs in the process of predicting the power load, and the power load prediction model needs to be reconstructed to predict the power load.
2. A power load prediction method based on concept drift detection as claimed in claim 1, characterized in that after obtaining historical data of the power load, the obtained historical data is subjected to data preprocessing;
the data preprocessing comprises missing data completion processing and normalization processing;
the missing data completion processing is to fill the default position by using average adjacent loads;
the normalization process uses a Min-Max method to normalize the data set after missing data completion processing, namely
Wherein, in the step (A),
in order to normalize the processed data, the data is normalized,
as a result of the original data, it is,
in order to be the largest of the original data,
is the smallest raw data.
3. The method as claimed in claim 1, wherein a power load prediction model based on concept drift detection is obtained based on the obtained power load historical data, a sliding time window is adopted on the power load historical data set, the power load values of the current time and the historical time in the time window are used as an input sequence of the constructed power load prediction model, the power load value of the time next to the sliding time window is used as a prediction object, and the training of the constructed power load prediction model is completed by inputting a training set in the power load historical data set to perform iterative optimization of the constructed power load prediction model.
4. A method for predicting a power load based on concept drift detection as claimed in claim 3, wherein said power load prediction model employs a long and short term memory network; and predicting the test set in the historical data set of the power load by using the trained long-short term memory network prediction model to obtain the predicted value of the power load at the next moment of the sliding time window.
5. A method for power load prediction based on concept drift detection as claimed in claim 1, characterized in that in the process of determining whether the input data based on the new sliding time window and the original data in the input base window belong to the same distribution, the K-S test is used to test the statistic D n Comprises the following steps:
wherein, in the process,
is the supremum limit of the function value,
is the current time of the sliding time window,
for the input data sequence of the new sliding time window,
is the original data sequence in the input base window.
6. A method for power load prediction based on concept drift detection as claimed in claim 5, characterized in that it is assumed that
Calculating the absolute difference of the accumulated frequency between the input data based on the new sliding time window and the input original data, making the maximum absolute difference be
(ii) a Based on sample capacity
And level of significance
Obtaining a critical value
(ii) a If it is
Sending out early warning of concept drift; if it is
If the assumption is not true, the input data based on the new sliding time window is not distributed in the same way as the input original data, and a warning of concept drift is issued.
7. The method as claimed in claim 6, wherein when the pre-warning of the concept drift is received, a temporary sliding time window is started, data which sends out pre-warning information and data which follow are received, and a temporary power load prediction model is trained by using the temporary sliding time window; and when a concept drift alarm is received, replacing the original power load prediction model with the temporary power load prediction model, emptying the temporary sliding time window, and moving the base window to the next moment of the concept drift point.
8. A power load prediction system based on concept drift detection, comprising:
an acquisition module configured to acquire power load history data;
a modeling module configured to construct a power load prediction model using a sliding time window according to the acquired power load historical data;
a prediction module configured to obtain a power load prediction value at a time next to the time at which the sliding time window is located based on the constructed power load prediction model;
a concept drift detection module configured to capture raw data input in prediction based on the power load prediction model, so as to realize concept drift detection in a power load prediction process;
judging whether input data based on a new sliding time window and input original data in a base window belong to the same distribution or not in a concept drift detection module, if so, not generating concept drift in the process of power load prediction, and continuously predicting the power load by using a constructed power load prediction model; if not, the concept drift occurs in the process of predicting the power load, and the power load prediction model needs to be reconstructed to predict the power load.
9. A computer-readable storage medium, on which a program is stored, which program, when being executed by a processor, carries out the steps of the method for power load prediction based on concept drift detection according to any one of claims 1 to 7.
10. An electronic device comprising a memory, a processor and a program stored on the memory and executable on the processor, wherein the processor implements the steps of the method for power load prediction based on concept drift detection according to any of claims 1-7 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211463933.5A CN115513951B (en) | 2022-11-17 | 2022-11-17 | Power load prediction method and system based on concept drift detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211463933.5A CN115513951B (en) | 2022-11-17 | 2022-11-17 | Power load prediction method and system based on concept drift detection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115513951A true CN115513951A (en) | 2022-12-23 |
| CN115513951B CN115513951B (en) | 2023-04-18 |
Family
ID=84513832
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211463933.5A Active CN115513951B (en) | 2022-11-17 | 2022-11-17 | Power load prediction method and system based on concept drift detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115513951B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116049757A (en) * | 2023-04-03 | 2023-05-02 | 湖南工商大学 | Flow abnormal behavior detection method based on concept drift |
| CN116227543A (en) * | 2023-05-09 | 2023-06-06 | 南方电网科学研究院有限责任公司 | Detection method and detection device for abnormal electricity consumption, electronic device and electronic equipment |
| CN116308304A (en) * | 2023-05-24 | 2023-06-23 | 山东建筑大学 | New energy intelligent operation and maintenance method and system based on meta learning concept drift detection |
| CN117114195A (en) * | 2023-08-31 | 2023-11-24 | 国网浙江电动汽车服务有限公司 | Multi-type electric vehicle charging demand real-time prediction method based on concept drift |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109034504A (en) * | 2018-09-14 | 2018-12-18 | 云南电网有限责任公司 | A kind of Short-term Load Forecasting Model method for building up and device |
| CN109978253A (en) * | 2019-03-22 | 2019-07-05 | 电子科技大学 | A kind of short-term load forecasting method based on incremental learning |
| CN110298501A (en) * | 2019-06-21 | 2019-10-01 | 河海大学常州校区 | Electric load prediction technique based on long Memory Neural Networks in short-term |
| CN111861039A (en) * | 2020-07-31 | 2020-10-30 | 广东电网有限责任公司广州供电局 | Power load prediction method, system, equipment and storage medium based on LSTM and generalized predictive control algorithm |
| CA3133280A1 (en) * | 2020-10-06 | 2022-04-06 | Bank Of Montreal | Systems and methods for predicting operational events |
| CN114330845A (en) * | 2021-12-15 | 2022-04-12 | 北京工业大学 | MSWI process dioxin emission prediction method based on multi-window concept drift detection |
| EP4027277A1 (en) * | 2021-01-11 | 2022-07-13 | Fundación Tecnalia Research & Innovation | Method, system and computer program product for drift detection in a data stream |
-
2022
- 2022-11-17 CN CN202211463933.5A patent/CN115513951B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109034504A (en) * | 2018-09-14 | 2018-12-18 | 云南电网有限责任公司 | A kind of Short-term Load Forecasting Model method for building up and device |
| CN109978253A (en) * | 2019-03-22 | 2019-07-05 | 电子科技大学 | A kind of short-term load forecasting method based on incremental learning |
| CN110298501A (en) * | 2019-06-21 | 2019-10-01 | 河海大学常州校区 | Electric load prediction technique based on long Memory Neural Networks in short-term |
| CN111861039A (en) * | 2020-07-31 | 2020-10-30 | 广东电网有限责任公司广州供电局 | Power load prediction method, system, equipment and storage medium based on LSTM and generalized predictive control algorithm |
| CA3133280A1 (en) * | 2020-10-06 | 2022-04-06 | Bank Of Montreal | Systems and methods for predicting operational events |
| US20220108238A1 (en) * | 2020-10-06 | 2022-04-07 | Bank Of Montreal | Systems and methods for predicting operational events |
| EP4027277A1 (en) * | 2021-01-11 | 2022-07-13 | Fundación Tecnalia Research & Innovation | Method, system and computer program product for drift detection in a data stream |
| CN114330845A (en) * | 2021-12-15 | 2022-04-12 | 北京工业大学 | MSWI process dioxin emission prediction method based on multi-window concept drift detection |
Non-Patent Citations (2)
| Title |
|---|
| MAREK MOLEDA: "Concept Drift and Avoiding its Negative Effects in Predictive Modeling of Failures of Electricity Production Units in Power Plants", 《2020 28TH INTERNATIONAL SYMPOSIUM ON MODELING, ANALYSIS, AND SIMULATION OF COMPUTER AND TELECOMMUNICATION SYSTEMS (MASCOTS)》 * |
| 计远帆: "园区级微网功率动态预测方法研究", 《中国知网硕士电子期刊》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116049757A (en) * | 2023-04-03 | 2023-05-02 | 湖南工商大学 | Flow abnormal behavior detection method based on concept drift |
| CN116227543A (en) * | 2023-05-09 | 2023-06-06 | 南方电网科学研究院有限责任公司 | Detection method and detection device for abnormal electricity consumption, electronic device and electronic equipment |
| CN116227543B (en) * | 2023-05-09 | 2023-08-04 | 南方电网科学研究院有限责任公司 | Detection method and detection device for abnormal electricity consumption, electronic device and electronic equipment |
| CN116308304A (en) * | 2023-05-24 | 2023-06-23 | 山东建筑大学 | New energy intelligent operation and maintenance method and system based on meta learning concept drift detection |
| CN116308304B (en) * | 2023-05-24 | 2023-08-25 | 山东建筑大学 | New energy intelligent operation and maintenance method and system based on meta learning concept drift detection |
| CN117114195A (en) * | 2023-08-31 | 2023-11-24 | 国网浙江电动汽车服务有限公司 | Multi-type electric vehicle charging demand real-time prediction method based on concept drift |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115513951B (en) | 2023-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115513951B (en) | Power load prediction method and system based on concept drift detection | |
| CN111210024A (en) | Model training method and device, computer equipment and storage medium | |
| CN112118143B (en) | Traffic prediction model training method, traffic prediction method, device, equipment and medium | |
| CN112990556A (en) | User power consumption prediction method based on Prophet-LSTM model | |
| CN114925623B (en) | Oil and gas reservoir yield prediction method and system | |
| CN111881023B (en) | Software aging prediction method and device based on multi-model comparison | |
| CN107886160B (en) | BP neural network interval water demand prediction method | |
| CN116307215A (en) | Load prediction method, device, equipment and storage medium of power system | |
| CN111044926A (en) | Method for predicting service life of proton exchange membrane fuel cell | |
| CN112836920A (en) | Coal electric unit energy efficiency state evaluation method and device and coal electric unit system | |
| CN112990587A (en) | Method, system, equipment and medium for accurately predicting power consumption of transformer area | |
| CN117175588A (en) | Space-time correlation-based electricity load prediction method and device | |
| CN116562514A (en) | Method and system for immediately analyzing production conditions of enterprises based on neural network | |
| CN111143777A (en) | Data processing method and device, intelligent terminal and storage medium | |
| CN108459991B (en) | Method for obtaining equipment reliability data | |
| CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
| CN117332898A (en) | New energy small time scale power time sequence rolling prediction method based on machine learning | |
| CN113159395A (en) | Deep learning-based sewage treatment plant water inflow prediction method and system | |
| CN116610416A (en) | Load prediction type elastic expansion system and method based on Kubernetes | |
| CN116381510A (en) | Method, system and storage medium for predicting residual life of lithium battery | |
| US20220243347A1 (en) | Determination method and determination apparatus for conversion efficiency of hydrogen production by wind-solar hybrid electrolysis of water | |
| CN113868957B (en) | Residual life prediction and uncertainty quantitative calibration method under Bayes deep learning | |
| CN115310355A (en) | Multi-energy coupling-considered multi-load prediction method and system for comprehensive energy system | |
| CN115459982A (en) | Power network false data injection attack detection method | |
| CN115640542A (en) | Intelligent electric energy meter state evaluation method and evaluation device based on Bayesian theory |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 250000 No. 1000, Fengming Road, Lingang Development Zone, Licheng District, Jinan City, Shandong Province Patentee after: SHANDONG JIANZHU University Patentee after: Shuifa Digital Industry (Shanghai) Co.,Ltd. Address before: 250000 No. 1000, Fengming Road, Lingang Development Zone, Licheng District, Jinan City, Shandong Province Patentee before: SHANDONG JIANZHU University Patentee before: Shuifa Digital Industry (Shandong) Co.,Ltd. |
|
| CP01 | Change in the name or title of a patent holder |