CN115455799A - Transformer area electric vehicle load prediction method and system based on feature engineering - Google Patents
Transformer area electric vehicle load prediction method and system based on feature engineering Download PDFInfo
- Publication number
- CN115455799A CN115455799A CN202110639309.5A CN202110639309A CN115455799A CN 115455799 A CN115455799 A CN 115455799A CN 202110639309 A CN202110639309 A CN 202110639309A CN 115455799 A CN115455799 A CN 115455799A
- Authority
- CN
- China
- Prior art keywords
- feature
- load
- prediction
- electric vehicle
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Strategic Management (AREA)
- Software Systems (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Engineering & Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Geometry (AREA)
- Game Theory and Decision Science (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a transformer district electric vehicle load prediction method and system based on feature engineering, wherein the method comprises the steps of obtaining a preset prediction model, and the prediction model is obtained by converting an original transformer district electric vehicle load into a feature set for training and inputting the feature set to a neural network built based on an LSTM algorithm for feature learning; and predicting the load of the electric automobile in the transformer area based on the prediction model. According to the invention, through systematic feature engineering, the target to be predicted is subjected to accurate feature modeling, and the most suitable prediction algorithm is adopted according to the energy consumption characteristics, so that the prediction accuracy is improved.
Description
Technical Field
The invention belongs to the technical field of electric vehicle load prediction, and particularly relates to a transformer area electric vehicle load prediction method and system based on characteristic engineering.
Background
The Electric Vehicle (EV) is used as an important adjustable and controllable resource in a low-voltage distribution transformer area, the 'peak-to-peak' is easily caused without additional control, the peak-to-valley difference and the network loss of a power distribution network can be reduced by effectively regulating and controlling the EV, the economy and the stability of distribution transformer operation in the transformer area are improved, and the capacity-increasing transformation investment cost of the transformer area is saved. Each low-voltage distribution transformer area is small in coverage area, and the number of contained private charging piles is small. Therefore, accurate and reliable electric vehicle load prediction has important significance for energy utilization optimization and demand response of the low-voltage distribution transformer area.
For the research on the aspect of electric vehicle load prediction, at present, the prediction of a large-scale urban charging station is mainly focused, and the prediction modeling needs to include various vehicle types corresponding to various vehicle types such as taxies, buses and private cars, and the number of contained vehicles is also large. On the low-voltage distribution station side, the electric vehicle load prediction needs to be accurate for each vehicle and only includes one type of vehicle for the private residents. On the other hand, the complex vehicle composition causes the difficulty of data feature extraction, effective features behind data cannot be accurately selected in the process of building a neural network, feature selection determines the upper limit of a prediction result, and finally the prediction result is limited.
In general, there are two main problems in the research on the load prediction of the electric vehicle in the low-voltage distribution transformer area: on one hand, a prediction method specially aiming at the load of a private electric vehicle in a low-voltage distribution transformer area is lacked; on the other hand, the user behavior characteristics are not extracted sufficiently, and the load prediction of the electric automobile in the low-voltage distribution transformer area is not facilitated.
Disclosure of Invention
Aiming at the problems, the invention provides a platform area electric vehicle load prediction method and system based on characteristic engineering.
In order to achieve the technical purpose and achieve the technical effects, the invention is realized by the following technical scheme:
in a first aspect, the invention provides a platform area electric vehicle load prediction method based on feature engineering, which comprises the following steps:
acquiring a preset prediction model, wherein the prediction model is obtained by converting the electric automobile load of an original distribution area into a feature set for training and inputting the feature set into a neural network built based on an LSTM algorithm for feature learning;
and predicting the load of the electric automobile in the transformer area based on the prediction model.
Optionally, the method for acquiring a feature set for training includes:
preprocessing the electric automobile load in the original distribution area;
and carrying out characteristic engineering processing on the data obtained after preprocessing to obtain a characteristic set.
Optionally, the preprocessing includes identifying anomalous data, processing missing values, and/or dimensionless.
Optionally, the method for acquiring the feature set includes:
obtaining user behavior characteristics, time characteristics and weather characteristics from the data obtained after preprocessing, and adding the user behavior characteristics, the time characteristics and the weather characteristics into the candidate characteristic set;
and selecting features by using the correlation coefficient, scoring each feature in the candidate feature set according to the divergence or the correlation, setting the number of thresholds or thresholds to be selected, and screening out an optimal feature subset as a final feature set.
Optionally, the obtaining method of the prediction model includes:
segmenting a feature set used for training, and selecting a training set, a test set and a verification set;
and respectively inputting the training set, the test set and the verification set to a neural network built based on an LSTM algorithm, and continuously correcting parameters of the neural network until satisfactory prediction precision is achieved, so as to obtain a final prediction model.
In a second aspect, the present invention provides a platform region electric vehicle load prediction system based on feature engineering, including:
the system comprises an acquisition module, a prediction module and a prediction module, wherein the acquisition module is used for acquiring a preset prediction model, and the prediction model is obtained by converting the load of the electric automobile in an original distribution area into a feature set for training and inputting the feature set to a neural network built based on an LSTM algorithm for feature learning;
and the prediction module is used for predicting the load of the electric automobile in the transformer area based on the prediction model.
Optionally, the method for acquiring a feature set for training includes:
preprocessing the electric automobile load in the original distribution area;
and carrying out characteristic engineering processing on the data obtained after the preprocessing to obtain a characteristic set.
Optionally, the preprocessing comprises identifying anomalous data, processing missing values, and/or non-dimensionalizing.
Optionally, the method for acquiring the feature set includes:
obtaining user behavior characteristics, time characteristics and weather characteristics from the data obtained after preprocessing, and adding the user behavior characteristics, the time characteristics and the weather characteristics into the candidate characteristic set;
and selecting features by using the correlation coefficient, scoring each feature in the candidate feature set according to the divergence or the correlation, setting the number of thresholds or thresholds to be selected, and screening out an optimal feature subset as a final feature set.
Optionally, the obtaining method of the prediction model includes:
segmenting a feature set used for training, and selecting a training set, a test set and a verification set;
and respectively inputting the training set, the test set and the verification set to a neural network built based on an LSTM algorithm, and continuously correcting parameters of the neural network until satisfactory prediction precision is achieved, so as to obtain a final prediction model.
Compared with the prior art, the invention has the following beneficial effects:
the invention provides a station area electric vehicle load prediction method and system based on characteristic engineering.
Drawings
In order that the manner in which the present invention is more fully understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings, wherein:
fig. 1 is a schematic diagram of a platform electric vehicle load prediction method based on feature engineering according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of the invention.
The following detailed description of the principles of the invention is provided in connection with the accompanying drawings.
Example 1
The embodiment of the invention provides a platform area electric vehicle load prediction method based on characteristic engineering, which comprises the following steps of:
acquiring a preset prediction model, wherein the prediction model is obtained by converting the electric automobile load of an original distribution area into a feature set for training and inputting the feature set into a neural network built based on an LSTM algorithm for feature learning;
and predicting the load of the electric automobile in the transformer area based on the prediction model.
In a specific implementation manner of the embodiment of the present invention, the method for acquiring a feature set used for training includes:
preprocessing the electric automobile load in the original distribution area; in particular, the preprocessing includes identifying anomalous data, processing missing values, and/or dimensionless;
and carrying out characteristic engineering processing on the data obtained after the preprocessing to obtain a characteristic set.
In a specific implementation manner of the embodiment of the present invention, the method for acquiring a feature set includes:
obtaining user behavior characteristics, time characteristics and weather characteristics from the data obtained after preprocessing, and adding the user behavior characteristics, the time characteristics and the weather characteristics into the candidate characteristic set;
and selecting features by using the correlation coefficient, scoring each feature in the candidate feature set according to the divergence or the correlation, setting the number of thresholds or thresholds to be selected, and screening out an optimal feature subset as a final feature set.
In a specific implementation manner of the embodiment of the present invention, the method for obtaining the prediction model includes:
segmenting a feature set used for training, and selecting a training set, a test set and a verification set;
and respectively inputting the training set, the test set and the verification set into a neural network built based on an LSTM algorithm, and continuously correcting parameters of the neural network until satisfactory prediction precision is achieved, so as to obtain a final prediction model.
The following describes the modeling process of the prediction model in the embodiment of the present invention in detail with reference to specific embodiments.
Step (1): based on data from an open-power-system-data (OPSD) of a power system modeling open platform, electric vehicle charging load actual measurement data of 11 families are selected. And preprocessing the original observation data, which mainly comprises identifying abnormal data, processing missing values, dimensionless data and the like.
Step (2): and (3) combining the operation principle of the electric automobile and the statistical knowledge, performing characteristic engineering processing on the original data, and constructing a characteristic set for a next prediction model. The method mainly analyzes and designs characteristics from three aspects of reflecting data distribution characteristics, reflecting time sequence and variation trend and reflecting multivariate coupling relation, and adds the obtained user behavior characteristics, time characteristics and weather characteristics as new derivative characteristics into a candidate characteristic set to wait for subsequent screening. Specifically, the method comprises the following steps:
the user behavior characteristics, that is, the statistical indexes in the traditional statistics are used to perform statistical description on the user behavior data, so that the distribution characteristics of the data, including a concentration trend, a deviation trend, a distribution form and the like, can be well reflected. And carrying out statistical analysis on the data of the electric automobile such as the charging time period, the charging frequency and the charging type. Including average indicators such as arithmetic mean, mode, median, etc. that reflect the general level or central tendency of the population; and the variation indexes are used for reflecting variation conditions or dispersion degrees of the overall distribution, such as range, variance and standard deviation, dispersion coefficients and the like. Accordingly, the characteristics of the user type, the charging frequency, the charging type and the like are obtained.
In the time characteristic, although the load of the electric automobile in the low-voltage distribution transformer area has a certain sudden change, the load still generally shows a certain regular periodic change, and shows an obvious fluctuation trend along with the time. Therefore, the idea of a time series model can be used for feature construction. On one hand, the time is disassembled, and information such as year, month, day, time, season and the like is extracted. On the other hand, considering that the data does not have mutation, the rate of change can be introduced as a new feature. The characteristics of date type, season and the like are obtained.
And weather characteristics, namely acquiring historical weather data of local load data in the same period through a public weather website, wherein the historical weather data comprises a plurality of weather indexes such as wind speed, wind direction, precipitation type, precipitation amount and illumination intensity, and screening in the next step.
And (3): the method comprises the steps of selecting features by using correlation coefficients, grading each feature according to divergence or correlation, setting the number of thresholds or thresholds to be selected, screening an optimal feature subset from a candidate feature set for model construction, calculating the Pearson coefficient between each index and a charging load by taking weather features as an example, and selecting the first three items with the largest numerical value as a final weather type according to an obtained result.
And (4): and (3) integrating the data and considering the characteristics of each characteristic item, constructing a corresponding neural network model by using an LSTM algorithm for characteristic learning, constructing a prediction model aiming at the target to be measured, segmenting the original data, selecting a training set, a test set and a verification set, and continuously correcting the prediction model until satisfactory prediction accuracy is achieved.
Example 2
The embodiment of the invention provides a platform area electric vehicle load prediction system based on characteristic engineering, which comprises:
the system comprises an acquisition module, a prediction module and a prediction module, wherein the acquisition module is used for acquiring a preset prediction model, and the prediction model is obtained by converting the load of the electric automobile in an original distribution area into a feature set for training and inputting the feature set to a neural network built based on an LSTM algorithm for feature learning;
and the prediction module is used for predicting the load of the electric automobile in the transformer area based on the prediction model.
In a specific implementation manner of the embodiment of the present invention, the method for acquiring a feature set for training includes:
preprocessing the electric automobile load in the original distribution area; in particular, the preprocessing comprises identifying anomalous data, processing missing values, and/or non-dimensionalizing;
and carrying out characteristic engineering processing on the data obtained after preprocessing to obtain a characteristic set.
In a specific implementation manner of the embodiment of the present invention, the method for acquiring a feature set includes:
obtaining user behavior characteristics, time characteristics and weather characteristics from the data obtained after preprocessing, and adding the user behavior characteristics, the time characteristics and the weather characteristics into the candidate characteristic set;
and selecting features by using the correlation coefficient, scoring each feature in the candidate feature set according to the divergence or the correlation, setting the number of thresholds or thresholds to be selected, and screening out an optimal feature subset as a final feature set.
In a specific implementation manner of the embodiment of the present invention, the method for obtaining the prediction model includes:
segmenting a feature set used for training, and selecting a training set, a test set and a verification set;
and respectively inputting the training set, the test set and the verification set to a neural network built based on an LSTM algorithm, and continuously correcting parameters of the neural network until satisfactory prediction precision is achieved, so as to obtain a final prediction model.
The foregoing shows and describes the general principles and features of the present invention, together with the advantages thereof. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are given by way of illustration of the principles of the present invention, but that various changes and modifications may be made without departing from the spirit and scope of the invention, and such changes and modifications are within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (10)
1. A method for predicting the load of an electric vehicle in a transformer area based on feature engineering is characterized by comprising the following steps:
acquiring a preset prediction model, wherein the prediction model is obtained by converting the load of the electric automobile in the original distribution area into a feature set for training and inputting the feature set to a neural network built based on an LSTM algorithm for feature learning;
and predicting the load of the electric automobile in the transformer area based on the prediction model.
2. The feature engineering based platform electric vehicle load prediction method according to claim 1, characterized in that: the method for acquiring the feature set for training comprises the following steps:
preprocessing the electric automobile load in the original distribution area;
and carrying out characteristic engineering processing on the data obtained after preprocessing to obtain a characteristic set.
3. The method for predicting the load of the electric vehicle in the transformer area based on the feature engineering as claimed in claim 2, characterized in that: the preprocessing includes identifying anomalous data, processing missing values, and/or non-dimensionalizing.
4. The feature engineering-based platform electric vehicle load prediction method according to claim 2, wherein the feature set acquisition method comprises:
obtaining user behavior characteristics, time characteristics and weather characteristics from the data obtained after preprocessing, and adding the user behavior characteristics, the time characteristics and the weather characteristics into the candidate characteristic set;
and selecting features by using the correlation coefficient, scoring each feature in the candidate feature set according to the divergence or the correlation, setting the number of thresholds or thresholds to be selected, and screening out an optimal feature subset as a final feature set.
5. The method for predicting the load of the electric vehicle in the transformer area based on the feature engineering as claimed in claim 1, wherein the method for obtaining the prediction model comprises the following steps:
segmenting a feature set used for training, and selecting a training set, a test set and a verification set;
and respectively inputting the training set, the test set and the verification set to a neural network built based on an LSTM algorithm, and continuously correcting parameters of the neural network until satisfactory prediction precision is achieved, so as to obtain a final prediction model.
6. A platform district electric automobile load prediction system based on feature engineering is characterized by comprising:
the system comprises an acquisition module, a prediction module and a prediction module, wherein the acquisition module is used for acquiring a preset prediction model, and the prediction model is obtained by converting the load of the electric automobile in an original distribution area into a feature set for training and inputting the feature set to a neural network built based on an LSTM algorithm for feature learning; and the prediction module is used for predicting the load of the electric automobile in the transformer area based on the prediction model.
7. The system for predicting the load of the electric vehicle in the transformer district based on the feature engineering as claimed in claim 6, wherein the method for acquiring the feature set for training comprises:
preprocessing the electric automobile load in the original distribution area;
and carrying out characteristic engineering processing on the data obtained after preprocessing to obtain a characteristic set.
8. The feature engineering based platform electric vehicle load prediction system according to claim 7, wherein the preprocessing comprises identifying abnormal data, processing missing values and/or non-dimensionalization.
9. The system for predicting the load of the electric vehicle in the transformer area based on the feature engineering as claimed in claim 7, wherein the method for acquiring the feature set comprises:
obtaining user behavior characteristics, time characteristics and weather characteristics from the data obtained after preprocessing, and adding the user behavior characteristics, the time characteristics and the weather characteristics into the candidate characteristic set;
and selecting features by using the correlation coefficient, scoring each feature in the candidate feature set according to the divergence or the correlation, setting the number of thresholds or thresholds to be selected, and screening out an optimal feature subset as a final feature set.
10. The feature engineering based platform electric vehicle load prediction system according to claim 6, wherein the obtaining method of the prediction model comprises:
segmenting a feature set used for training, and selecting a training set, a test set and a verification set;
and respectively inputting the training set, the test set and the verification set into a neural network built based on an LSTM algorithm, and continuously correcting parameters of the neural network until satisfactory prediction precision is achieved, so as to obtain a final prediction model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110639309.5A CN115455799A (en) | 2021-06-08 | 2021-06-08 | Transformer area electric vehicle load prediction method and system based on feature engineering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110639309.5A CN115455799A (en) | 2021-06-08 | 2021-06-08 | Transformer area electric vehicle load prediction method and system based on feature engineering |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115455799A true CN115455799A (en) | 2022-12-09 |
Family
ID=84295316
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110639309.5A Pending CN115455799A (en) | 2021-06-08 | 2021-06-08 | Transformer area electric vehicle load prediction method and system based on feature engineering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115455799A (en) |
-
2021
- 2021-06-08 CN CN202110639309.5A patent/CN115455799A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113962364B (en) | Multi-factor power load prediction method based on deep learning | |
Wu et al. | An intuitionistic fuzzy multi-criteria framework for large-scale rooftop PV project portfolio selection: Case study in Zhejiang, China | |
CN111091143B (en) | Distribution transformer weight overload early warning method based on deep belief network and K-means clustering | |
CN102183621B (en) | Aquaculture dissolved oxygen concentration online forecasting method and system | |
CN108921339B (en) | Quantile regression-based photovoltaic power interval prediction method for genetic support vector machine | |
CN108985965A (en) | A kind of photovoltaic power interval prediction method of combination neural network and parameter Estimation | |
CN104573879A (en) | Photovoltaic power station output predicting method based on optimal similar day set | |
CN105069525A (en) | All-weather 96-point daily load curve prediction and optimization correction system | |
CN111754037B (en) | Long-term load hybrid prediction method for regional terminal integrated energy supply system | |
CN107292534A (en) | The yardstick competition evaluation method and device of urban power distribution network long term dynamics investment | |
CN112016755A (en) | Construction method of universal design cost standardization technology module of power transmission and transformation project construction drawing | |
CN115829105A (en) | Photovoltaic power prediction method based on historical data feature search | |
CN110163444A (en) | A kind of water demand prediction method based on GASA-SVR | |
CN110909958A (en) | Short-term load prediction method considering photovoltaic grid-connected power | |
CN117977568A (en) | Power load prediction method based on nested LSTM and quantile calculation | |
CN116187835A (en) | Data-driven-based method and system for estimating theoretical line loss interval of transformer area | |
CN106600055A (en) | Wind speed prediction method the basis of self excitation threshold autoregression model | |
CN115829145A (en) | Photovoltaic power generation capacity prediction system and method | |
CN116796168A (en) | CNN-BiLSTM high-altitude multi-factor power transmission line audible noise prediction method based on multi-head attention mechanism | |
Dang et al. | A novel multivariate grey model for forecasting periodic oscillation time series | |
Fan | Research on deep learning energy consumption prediction based on generating confrontation network | |
CN111461921A (en) | Load modeling typical user database updating method based on machine learning | |
CN116826745B (en) | Layered and partitioned short-term load prediction method and system in power system background | |
CN113570414A (en) | Electricity price prediction method for optimizing deep neural network based on improved Adam algorithm | |
CN116611785A (en) | Power transmission and transformation project cost model construction method, system, equipment and medium based on big data |
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 |