Electric vehicle charging pile load baseline prediction method and system considering market type
The gradient lifter and random forest model respectively predict the load baseline of charging piles in recent markets and continuous markets, which solves the accuracy of load baseline calculation in the prior art, and improves the efficiency and optimization effect of charging facilities and power grid operation.
Patent Information
- Application Number
- CN202411142987.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-12-13
AI Technical Summary
It is difficult to accurately calculate the load baseline of electric vehicle charging piles in different markets, which affects the accuracy of grid load management and charging facility planning.
The gradient hoist model and the random forest model are used to predict the load baseline data of the charging piles in the recent market and the continuous market respectively. By acquiring and processing the data, the corresponding model is constructed to achieve accurate calculation of the load baseline value.
Accurate prediction of the load baseline of charging piles in different markets is achieved, charging facility efficiency and grid operation optimization are improved, and performance is better than Fourier series model, support vector machine model and traditional historical numerical moving average method.
Smart Images
Figure CN119130507A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power system load baseline measurement, and in particular to an electric vehicle charging pile load baseline considering market types. Background Art
[0002] By calculating the baseline values of charging pile loads in different markets, it can help operators rationally plan the construction of charging stations, set a reasonable charging pile baseline, and enable charging stations to operate safely and smoothly, thereby ensuring the safe and stable operation of the power grid.
[0003] Calculating the load baseline of charging piles can help the power grid to manage and peak load, optimize the load distribution of the power grid, ensure the safety and stability of the power grid, and guide the rational planning and construction of charging stations, improve the utilization rate and economic benefits of charging facilities, and maintain the safe and economic operation of charging stations. Load baseline calculation is crucial for the coordinated development of electric vehicles and power grids.
[0004] Calculating the charging pile load baseline in the day-ahead market can achieve advance planning and balance of the power system, optimize resource allocation, avoid market risks, and provide market players with clear price signals to guide their trading decisions; calculating the charging pile load baseline in the continuous market focuses on real-time response, which can provide flexibility for demand-side management, support the formation of real-time prices, stimulate demand response, and achieve instant balance of the power system. Summary of the invention
[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a method and system for predicting the load baseline of electric vehicle charging piles taking into account the market type.
[0006] The purpose of the present invention can be achieved by the following technical solutions:
[0007] As a first aspect of the present invention, a method for predicting a baseline load of an electric vehicle charging pile considering market types is provided, the steps comprising:
[0008] Obtain charging pile load baseline data and determine whether the charging pile load baseline data comes from the day-ahead market or the continuous market;
[0009] Based on the baseline data of charging pile load in the day-ahead market, the gradient boosting model is used to predict the baseline value of charging pile load in the day-ahead market.
[0010] Based on the baseline data of charging pile load in the continuous market, the random forest model is used for prediction to obtain the baseline value of the predicted charging pile load in the continuous market.
[0011] As a preferred technical solution, the mathematical expression of the gradient boosting model is as follows:
[0012]
[0013] In the formula, y is the true value; is the calculated value of the current model at step t; r t is the residual between the calculated value and the true value of the current model; v is the optimal contribution value; L is the loss function; h t (x) is the calculation of the t-th weak learner on the input x; n is the number of training samples; is the calculated value at step t+1 after the model is updated with the best contribution; T is the number of model iterations; F(x) is the calculation of the final model for the input x.
[0014] As a preferred technical solution, the random forest mathematical expression is as follows:
[0015]
[0016] In the formula, I is the sample set in the node; a is the feature; T is the number of categories of feature a; I t is the sample set of category t of a in I, and the information gain IG is used to select the most informative feature to split the decision tree;
[0017]
[0018] In the formula, m is the number of categories; p j is the proportion of samples belonging to the jth class; when further splitting cannot significantly reduce the node Gini impurity value, a sufficient number of decision trees have been constructed;
[0019]
[0020] In the formula, B is the total number of trees in the forest; is the calculation result of the i-th tree; the final load baseline calculation result of the random forest model is the average of the calculation results of all trees.
[0021] As a preferred technical solution, the training process of the gradient boosting machine model and the random forest model is as follows:
[0022] Select the baseline data of the electric vehicle charging pile load on the previous day and create a data set;
[0023] Determine whether the data comes from the day-ahead or continuous market;
[0024] A gradient boosting model is built for the baseline data of charging pile load in the day-ahead market, and a random forest model is built for the baseline data of charging pile load in the continuous market.
[0025] The model is trained based on the data in the constructed dataset.
[0026] As a preferred technical solution, when creating the data set, abnormal data in the data is cleaned, blank data in the data is filled, and the data in the data set is normalized.
[0027] As a preferred technical solution, the model training selects mean square error as the loss function of the model:
[0028]
[0029] In the formula, p i is the calculated value of the i-th load baseline; t i is the actual value of the ith load baseline; n is the number of samples.
[0030] As a second aspect of the present invention, a system for predicting the load baseline of an electric vehicle charging pile considering market types is provided, wherein the system executes the above-mentioned method for predicting the load baseline of an electric vehicle charging pile considering market types, including:
[0031] Data acquisition module: used to obtain the charging pile load baseline data and determine whether the charging pile load baseline data comes from the day-ahead market or the continuous market;
[0032] The charging pile load baseline data prediction module of the day-ahead market: The charging pile load baseline data of the day-ahead market is predicted based on the gradient boosting model to obtain the predicted charging pile load baseline value under the day-ahead market;
[0033] Continuous market charging pile load baseline data prediction module: For the charging pile load baseline data of the continuous market, the random forest model is used for prediction to obtain the predicted charging pile load baseline value in the continuous market.
[0034] As a preferred technical solution, in the previous day market charging pile load baseline data prediction module, the mathematical expression of the gradient boosting machine model is as follows:
[0035]
[0036] In the formula, y is the true value; is the calculated value of the current model at step t; r t is the residual between the calculated value and the true value of the current model; v is the optimal contribution value; L is the loss function; h t (x) is the calculation of the t-th weak learner on the input x; n is the number of training samples; is the calculated value at step t+1 after the model is updated with the best contribution; T is the number of model iterations; F(x) is the calculation of the final model for the input x.
[0037] As a preferred technical solution, in the continuous market charging pile load baseline data prediction module, the random forest mathematical expression is as follows:
[0038]
[0039] In the formula, I is the sample set in the node; a is the feature; T is the number of categories of feature a; I t is the sample set of category t of a in I, and the information gain IG is used to select the most informative feature to split the decision tree;
[0040]
[0041] In the formula, m is the number of categories; p j is the proportion of samples belonging to the jth class; when further splitting cannot significantly reduce the node Gini impurity value, a sufficient number of decision trees have been constructed;
[0042]
[0043] In the formula, B is the total number of trees in the forest; is the calculation result of the i-th tree; the final load baseline calculation result of the random forest model is the average of the calculation results of all trees.
[0044] As a preferred technical solution, the continuous market charging pile load baseline data prediction module and the model training of the continuous market charging pile load baseline data prediction module select mean square error as the loss function of the model:
[0045]
[0046] In the formula, p i is the calculated value of the i-th load baseline; t i is the actual value of the ith load baseline; n is the number of samples.
[0047] Compared with the prior art, the present invention has the following beneficial effects:
[0048] The present invention proposes a method for calculating the load baseline of electric vehicle charging piles taking into account the market type. For the charging pile load baselines in the day-ahead market and the continuous market, the gradient boosting machine model and the random forest model are used for predictive calculation. Practical comparison shows that the scheme proposed in the present application has good performance, and the prediction effect far exceeds the Fourier series model, the support vector machine model and the traditional historical numerical moving average method. It can realize the accurate calculation of the charging pile load baselines in different markets, which is of great significance to improving the efficiency of charging facilities in charging stations and optimizing the operation of power grids. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Figure 1A flow chart of a method for calculating a load baseline of an electric vehicle charging pile considering market types according to the present invention;
[0050] Figure 2 This is a graph showing the baseline results of charging pile loads calculated by the gradient boosting model in the day-ahead market in an embodiment of the present invention;
[0051] Figure 3 This is a graph showing the baseline results of charging pile loads calculated using the random forest model in a continuous market in an implementation example of the present invention. DETAILED DESCRIPTION
[0052] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.
[0053] Example 1
[0054] As one implementation of the present invention, this embodiment provides a method for predicting the load baseline of electric vehicle charging piles taking into account the market type, such as Figure 1 As shown, the specific steps of the present invention are as follows:
[0055] (1) Construct a dataset for calculating the baseline load of electric vehicle charging piles and normalize the data in the dataset.
[0056] The data normalization method is as follows:
[0057]
[0058] In the formula, x i is the i-th data in the data set; min(x i ) is the data with the smallest value in the data set; max(x i ) is the data with the largest value in the data set; y i For x i The new data value after normalization.
[0059] After the data is normalized according to the data normalization method, the data in the data set are mapped to [0,1]. Through data normalization, the problem of poor model prediction results caused by the difference in load baseline values between different dates can be eliminated.
[0060] (2) Define the loss function to evaluate the model performance.
[0061] Select Mean Square Error (MSE) as the loss function of the model:
[0062]
[0063] In the formula, p i is the calculated value of the i-th load baseline; t i is the actual value of the ith load baseline; n is the number of samples.
[0064] (3) Construct gradient boosting machine model and random forest model.
[0065] The mathematical expression of the constructed gradient boosting machine model is as follows:
[0066]
[0067] In the formula, y is the true value; is the calculated value of the current model at step t; r t is the residual between the calculated value and the true value of the current model; v is the optimal contribution value; L is the loss function; h t (x) is the calculation of the t-th weak learner on the input x; n is the number of training samples; is the calculated value at step t+1 after the model is updated with the best contribution; T is the number of model iterations; F(x) is the calculation of the final model for the input x.
[0068] The mathematical expression of the constructed random forest is as follows:
[0069]
[0070] In the formula, I is the sample set in the node; a is the feature; T is the number of categories of feature a; I t is the sample set of category t of a in I. Information gain IG can be used to select the most informative features to split the decision tree.
[0071]
[0072] In the formula, m is the number of categories; p j is the proportion of samples belonging to the jth class. When further splitting cannot significantly reduce the node Gini impurity value, it proves that a sufficient number of decision trees have been constructed.
[0073]
[0074] In the formula, B is the total number of trees in the forest; is the calculation result of the i-th tree. The final load baseline calculation result of the random forest model is the average of the calculation results of all trees.
[0075] (6) Select an appropriate model based on the data source and calculate the load baseline of the electric vehicle charging station according to the above steps.
[0076] Example 2
[0077] As one of the specific implementation examples of the present invention, in this embodiment, the load baseline data of various charging piles in the day-ahead market and the continuous market in the Beijing-Tianjin-Tangshan area are used to select a corresponding model to calculate the load baseline value.
[0078] The experiment selected the load baseline values of the five days before the load baseline calculation day to construct a data set and calculated the load baseline value on the calculation day. In order to ensure the accuracy and stability of the load baseline value calculation results of the constructed model, multiple training verifications were carried out. The model calculation results are as follows: Figure 2-3 As shown:
[0079] The results of the charging pile load baseline calculated by the gradient boosting model in the day-ahead market are very close to the actual load baseline results, with an average error rate of 4.2%; the results of the charging pile load baseline calculated by the random forest model in the continuous market are also very close to the actual load baseline results, with an average error rate of 6.82%. At the same time, in order to verify the performance of the model, the constructed model is compared with the load baseline calculation effects of the Fourier series model, the support vector machine model, and the historical numerical moving average method. The results are shown in the following table:
[0080] Table 1 Comparison of average error rates of load baseline calculations for various models under the day-ahead market
[0081]
[0082]
[0083] Table 2 Comparison of average error rates of load baseline calculations for each model under continuous market
[0084]
[0085] It can be seen that the gradient boosting machine model and random forest model used in the electric vehicle charging pile load baseline calculation method considering market types proposed in the present invention have good performance, and the prediction effect far exceeds the Fourier series model, support vector machine model and traditional historical numerical moving average method. It can realize the accurate calculation of the charging pile load baseline in different markets, which is of great significance to improving the efficiency of charging facilities in charging stations and optimizing power grid operation.
[0086] Example 3
[0087] As another embodiment of the present invention, this embodiment further provides a system for implementing the above-mentioned method for predicting the load baseline of electric vehicle charging piles considering market types, including:
[0088] Data acquisition module: used to obtain the charging pile load baseline data and determine whether the charging pile load baseline data comes from the day-ahead market or the continuous market;
[0089] The charging pile load baseline data prediction module of the day-ahead market: The charging pile load baseline data of the day-ahead market is predicted based on the gradient boosting model to obtain the predicted charging pile load baseline value under the day-ahead market;
[0090] Continuous market charging pile load baseline data prediction module: For the charging pile load baseline data of the continuous market, the random forest model is used for prediction to obtain the predicted charging pile load baseline value in the continuous market.
[0091] In the day-ahead market charging pile load baseline data prediction module, the mathematical expression of the gradient boosting model is as follows:
[0092]
[0093] In the formula, y is the true value; is the calculated value of the current model at step t; r t is the residual between the calculated value and the true value of the current model; v is the optimal contribution value; L is the loss function; h t (x) is the calculation of the t-th weak learner on the input x; n is the number of training samples; is the calculated value at step t+1 after the model is updated with the best contribution; T is the number of model iterations; F(x) is the calculation of the final model for the input x.
[0094] In the continuous market charging pile load baseline data prediction module, the random forest mathematical expression is as follows:
[0095]
[0096] In the formula, I is the sample set in the node; a is the feature; T is the number of categories of feature a; I t is the sample set of category t of a in I, and the information gain IG is used to select the most informative feature to split the decision tree;
[0097]
[0098] In the formula, m is the number of categories; p j is the proportion of samples belonging to the jth class; when further splitting cannot significantly reduce the node Gini impurity value, a sufficient number of decision trees have been constructed;
[0099]
[0100] In the formula, B is the total number of trees in the forest; is the calculation result of the i-th tree; the final load baseline calculation result of the random forest model is the average of the calculation results of all trees.
[0101] In the model training of the continuous market charging pile load baseline data prediction module and the continuous market charging pile load baseline data prediction module, the mean square error is selected as the loss function of the model:
[0102]
[0103] In the formula, p i is the calculated value of the i-th load baseline; t i is the actual value of the ith load baseline; n is the number of samples.
[0104] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.
Claims
1. A method for predicting the load baseline of electric vehicle charging piles considering market types, characterized in that the steps include: Obtain charging pile load baseline data and determine whether the charging pile load baseline data comes from the day-ahead market or the continuous market; Based on the baseline data of charging pile load in the day-ahead market, the gradient boosting model is used to predict the baseline value of charging pile load in the day-ahead market. Based on the baseline data of charging pile load in the continuous market, the random forest model is used for prediction to obtain the baseline value of the predicted charging pile load in the continuous market.
2. The method for predicting the load baseline of electric vehicle charging piles considering market types according to claim 1 is characterized in that: The mathematical expression of the gradient boosting machine model is as follows: In the formula, y is the true value; is the calculated value of the current model at step t; r t is the residual between the calculated value and the true value of the current model; v is the optimal contribution value; L is the loss function; h t (x) is the calculation of the t-th weak learner on the input x; n is the number of training samples; is the calculated value at step t+1 after the model is updated with the best contribution; T is the number of model iterations; F(x) is the calculation of the final model for the input x.
3. The method for predicting the load baseline of electric vehicle charging piles considering market types according to claim 1, characterized in that: The mathematical expression of random forest is as follows: In the formula, I is the sample set in the node; a is the feature; T is the number of categories of feature a; I t is the sample set of category t of a in I, and the information gain IG is used to select the most informative feature to split the decision tree; In the formula, m is the number of categories; p j is the proportion of samples belonging to the jth class; when further splitting cannot significantly reduce the node Gini impurity value, a sufficient number of decision trees have been constructed; In the formula, B is the total number of trees in the forest; is the calculation result of the i-th tree; the final load baseline calculation result of the random forest model is the average of the calculation results of all trees.
4. The method for predicting the load baseline of electric vehicle charging piles considering market types according to claim 1, characterized in that: The training process of the gradient boosting machine model and the random forest model is as follows: Select the baseline data of the electric vehicle charging pile load on the previous day and create a data set; Determine whether the data comes from the day-ahead or continuous market; A gradient boosting model is built for the baseline data of charging pile loads in the day-ahead market, and a random forest model is built for the baseline data of charging pile loads in the continuous market. The model is trained based on the data in the constructed dataset.
5. The method for predicting the load baseline of electric vehicle charging piles considering market types according to claim 4, characterized in that: When creating the data set, abnormal data in the data is cleaned, blank data in the data is filled, and the data in the data set is normalized.
6. The method for predicting the load baseline of electric vehicle charging piles considering market types according to claim 4, characterized in that: The model training described above selects mean square error as the loss function of the model: In the formula, p i is the calculated value of the i-th load baseline; t i is the actual value of the ith load baseline; n is the number of samples.
7. A baseline prediction system for electric vehicle charging pile load considering market types, characterized in that: The system executes the method for predicting the load baseline of an electric vehicle charging pile considering market types as described in any one of claims 1 to 6, including: Data acquisition module: used to obtain the charging pile load baseline data and determine whether the charging pile load baseline data comes from the day-ahead market or the continuous market; The charging pile load baseline data prediction module of the day-ahead market: The charging pile load baseline data of the day-ahead market is predicted based on the gradient boosting model to obtain the predicted charging pile load baseline value under the day-ahead market; Continuous market charging pile load baseline data prediction module: For the charging pile load baseline data of the continuous market, the random forest model is used for prediction to obtain the predicted charging pile load baseline value in the continuous market.
8. The electric vehicle charging pile load baseline prediction system considering market types according to claim 7, characterized in that: In the above-mentioned day-ahead market charging pile load baseline data prediction module, the mathematical expression of the gradient boosting machine model is as follows: In the formula, y is the true value; is the calculated value of the current model at step t; r t is the residual between the calculated value and the true value of the current model; v is the optimal contribution value; L is the loss function; h t (x) is the calculation of the t-th weak learner on the input x; n is the number of training samples; is the calculated value at step t+1 after the model is updated with the best contribution; T is the number of model iterations; F(x) is the calculation of the final model for the input x.
9. The electric vehicle charging pile load baseline prediction system considering market types according to claim 7, characterized in that: In the continuous market charging pile load baseline data prediction module, the random forest mathematical expression is as follows: In the formula, I is the sample set in the node; a is the feature; T is the number of categories of feature a; I t is the sample set of category t of a in I, and the information gain IG is used to select the most informative feature to split the decision tree; In the formula, m is the number of categories; p j is the proportion of samples belonging to the jth class; when further splitting cannot significantly reduce the node Gini impurity value, a sufficient number of decision trees have been constructed; In the formula, B is the total number of trees in the forest; is the calculation result of the i-th tree; the final load baseline calculation result of the random forest model is the average of the calculation results of all trees.
10. The electric vehicle charging pile load baseline prediction system considering market types according to claim 7, characterized in that: The continuous market charging pile load baseline data prediction module and the model training of the continuous market charging pile load baseline data prediction module select the mean square error as the loss function of the model: In the formula, p i is the calculated value of the i-th load baseline; t i is the actual value of the ith load baseline; n is the number of samples.