Power balance scene generation method based on source-load prediction

Abstract

The invention discloses an electric quantity balance scene generation method based on source-load prediction, and the method comprises a feature collection layer, a data processing layer, a prediction layer and a scene generation layer, and is used for making an electricity purchase plan, solving a new energy output condition, calculating wind and light abandoning data, and making an energy storage charging and discharging plan. In order to overcome the defect that the calculation of an electricity purchasing scheme and the wind curtailment and light curtailment amount is inaccurate due to inaccurate short-term prediction of load power and new energy power generation power when an electric quantity balance scene is generated, the scheme remarkably reduces the errors of the wind curtailment and light curtailment amount and the electricity purchasing scheme when the electric quantity balance is calculated; related information such as the relation between meteorological data and load data, the law of new energy power generation and the like is utilized to the maximum extent, traditional defects can be overcome, and new energy output prediction can be accurately calculated on the basis of less available data. Therefore, the precision of calculating the power purchase scheme and the wind and light curtailment amount in the power balance scene generation process is improved.

Description

Technical Field

[0001] The present invention relates to the field of power system operation and control, and in particular to a method for generating an electricity balance scenario based on source-load prediction. Background Art

[0002] Wind power and photovoltaic power generation have randomness, volatility and uncertainty. Therefore, the growing scale of wind power and photovoltaic power generation will greatly increase the difficulty of balancing power supply and demand in space and time. In order to ensure the accuracy of the power balance scenario generation strategy, the impact of new energy must be fully considered.

[0003] The characteristics of wind and solar energy in nature determine that wind power and photovoltaic power have strong volatility, which will bring huge challenges to the balance of power supply and demand in space and time. Accurately predicting photovoltaic and wind power in advance can more accurately calculate power purchase plans and energy storage charging and discharging plans, thereby effectively improving the grid's ability to absorb new energy.

[0004] At the same time, in the operation of the power grid, it is necessary to adjust the power generation according to the changes in load to maintain the balance of the power grid. Generally speaking, the load shows obvious fluctuation characteristics during the day, usually with a larger load during the peak hours in the morning and evening, and a lower load during the early morning hours. Therefore, it is necessary to reasonably arrange the power generation plan according to the load situation to ensure the stable operation of the power grid. In addition, with the large-scale access of renewable energy such as wind power and solar energy to the power grid, the uncertainty of the load will also increase, which has a huge impact on the prediction of future power balance scenarios.

[0005] At present, the research on renewable energy power generation prediction is becoming more and more in-depth at home and abroad. Among various prediction methods, the most direct way is to use the power data collected at the most recent time point as the predicted value at the next time point. This method has a simple model, but its accuracy decreases rapidly as the time scale increases. Other common methods include time series prediction, linear regression, support vector machine, decision tree, CatBoost, etc., but their prediction accuracy needs to be improved. In addition, the current domestic load prediction is usually predicted by machine learning the relationship between meteorological data and load. However, these methods usually ignore the time lag of weather on load, so their prediction accuracy needs to be improved. Summary of the invention

[0006] In order to overcome the drawbacks of the above-mentioned prior art, the purpose of the present invention is to provide a method for generating a power balance scenario that takes into account the delayed characteristics of the impact of meteorology on the load and a new energy power prediction method that takes into account the regularity of new energy power generation itself, which can significantly reduce the errors in the amount of wind and solar power abandoned and the power purchase plan when calculating the power balance. This method makes maximum use of the relationship between meteorological data and load data, the laws of new energy power generation itself and other related information, which can not only make up for the drawback that the traditional load power prediction method based on historical meteorological data does not consider the continuity of the impact of weather factors on the load, but also can accurately calculate the new energy output forecast based on less available data, thereby improving the accuracy of calculating the power purchase plan and the amount of wind and solar power abandoned in the process of generating the power balance scenario.

[0007] The technical solution adopted by the present invention comprehensively considers the following factors: 1. Local meteorological, wind power, photovoltaic, and load power data; 2. The time lag of weather’s impact on load; Based on the above factors, the steps of the method for generating a power balance scenario based on source-load prediction include: S1: Feature collection; S2: Data processing; S3: prediction; S4: Scenario generation; The feature collection specifically includes: Collect historical active output power series, historical meteorological data time series, and historical load data time series; Standardize the data; Determine the time scale over which the power balance scenario needs to be predicted , get the corresponding prediction data time point ; Find the causal relationship between historical meteorological data and historical load data, and get Delay time ; The data processing specifically includes: in accordance with and Parameters, respectively, split the photovoltaic power generation data and wind power generation data into four different data matrices: photovoltaic power generation feature matrix, photovoltaic power generation mapping matrix, wind power generation feature matrix, photovoltaic power generation mapping matrix; in accordance with Delay time Convert historical meteorological data into historical temperature feature matrix and historical light intensity feature matrix; The forecast specifically includes: Three prediction models were established for Forecast the load, photovoltaic power generation and wind power generation at each time point; The scene generation is specifically as follows: Using predicted Based on the load, photovoltaic power generation, and wind power generation data at each time point, we can formulate power purchase plans, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate energy storage charging and discharging plans.

[0008] Furthermore, in said S1: and The relationship between them satisfies the following formula: ; The D-Learner method is used to find the causal relationship between historical meteorological data and historical load data, and all the All values ​​less than 0.05 Delay time ,in, .

[0009] Furthermore, in S2: the data segmentation method is specifically: Set the time point of the first meteorological data collected to The second time point is , and so on, then the time point of the first historical load data after the split , the time point of the second historical load data , and so on, satisfying the following relationship: ; And the total number of historical load data after segmentation Satisfies the following relationship: ; in, Indicates the total number of time points for collecting historical load data; From the time point of the first historical load data Reverse all time points that have an impact on the first historical load data. The reverse formula is as follows: ; In the formula, Indicates that the delay factor affected by weather is , the time point corresponding to the meteorological data that has an impact on the first historical load data; The historical temperature feature matrix and the historical light intensity feature matrix are formed as follows: ; ; in, Representative The temperature corresponding to each time point, in degrees Celsius; Representative The light intensity corresponding to each time point, in lux; Furthermore, the data matrix is ​​specifically: If we say that before a certain point in time The data collected at a time point and the data collected at that time point are a group of data, and each The data at each time point corresponds to a row of the photovoltaic power generation characteristic matrix or the wind power generation characteristic matrix. For a data set with data, the number of groups contained in it is the number of samples in the sample space of machine learning. for: ; The photovoltaic power generation characteristic matrix is ​​as follows: ; Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points Indicates photovoltaic power generation, unit ; The photovoltaic power generation mapping matrix is ​​shown below: ; Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points ; The wind power generation characteristic matrix is ​​as follows: ; Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points Indicates photovoltaic power generation, unit kW; The wind power generation mapping matrix is ​​shown below: ; Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points .

[0010] Furthermore, the S3 is specifically: Standardize the photovoltaic power generation characteristic matrix, photovoltaic power generation mapping matrix, wind power generation characteristic matrix, and photovoltaic power generation mapping matrix as a whole; Standardize the historical meteorological characteristic moments; Using machine learning algorithms, The data at each time point and the The relationship between data at different time points.

[0011] Furthermore, in S4, the power balance constraint and energy storage SOC constraint requirements are as follows: At each point in time: ; ; Among them, the superscript Indicates the corresponding time point It represents the load predicted by machine learning in kW. represents the photovoltaic power generation predicted by machine learning, in kW, represents the wind power generation predicted by machine learning, in kW, Indicates the energy storage charge and discharge capacity, positive for discharge, negative for charge, unit is kW, Indicates the amount of electricity purchased, in kW. Indicates the total energy storage capacity in kWh. Indicates time point Energy storage capacity, in kWh; On this basis, a new variable, new energy differential load, is introduced Satisfies the following relationship: .

[0012] Furthermore, in S4, the strategy for generating a power balance scenario is as follows: When the new energy difference load is greater than 0 at this time point, energy storage charging is given priority to consume new energy to save new energy resources: When the energy storage capacity is less than 90% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than or equal to 90% of the time: ; Among them, the superscript Indicates the corresponding time point Indicates the charge and discharge efficiency; When the energy storage capacity is less than 90% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is greater than 90% of the time: ; Among them, the superscript Indicates the corresponding time point Represents the amount of wind and solar power curtailment; When the energy storage capacity is equal to 90% of the time: ; Among them, the superscript Indicates the corresponding time point .

[0013] When the new energy difference load is less than 0 at this time point, energy storage discharge is given priority to provide power to save costs: When the energy storage capacity is less than 10% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than or equal to 10% of: ; Among them, the superscript Indicates the corresponding time point ; When the energy storage capacity is greater than 10% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than 10% of: ; Among them, the superscript Indicates the corresponding time point ; When the energy storage capacity is equal to 10% of: ; Among them, the superscript Indicates the corresponding time point .

[0014] Furthermore, the S4 also includes minimizing the objective function: Using predicted The load, photovoltaic power generation, and wind power generation data at each time point are used to generate the power balance scenario at each time point with economy as the objective function, formulate the power purchase plan, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate the objective function of the energy storage charging and discharging plan. as follows: ; Among them, the superscript Indicates the corresponding time point represents the electricity purchase price, represents the cost of photovoltaic power generation, represents the cost of wind power, Represents the penalty cost for curtailing wind and solar power.

[0015] In a second aspect, the present invention further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of any one of the methods described in the above-mentioned schemes when executing the computer program.

[0016] In a third aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any one of the methods described in the above-mentioned scheme.

[0017] In a fourth aspect, the present invention provides a power balance scenario generation device based on source-load prediction, comprising: a feature acquisition module, a data processing module, a prediction module, and a scenario generation module; The feature acquisition module is used for: Collect historical active output power series, historical meteorological data time series, and historical load data time series; Standardize the data; Determine the time scale over which the power balance scenario needs to be predicted , get the corresponding prediction data time point ; Find the causal relationship between historical meteorological data and historical load data, and get Delay time ; The data processing module is used for: in accordance with and Parameters, respectively, split the photovoltaic power generation data and wind power generation data into four different data matrices: photovoltaic power generation feature matrix, photovoltaic power generation mapping matrix, wind power generation feature matrix, photovoltaic power generation mapping matrix; in accordance with Delay time Convert historical meteorological data into historical temperature feature matrix and historical light intensity feature matrix; The prediction module is used to: Three prediction models were established for Forecast the load, photovoltaic power generation and wind power generation at each time point; The scene generation module is used for: Using predicted Based on the load, photovoltaic power generation, and wind power generation data at each time point, we can formulate power purchase plans, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate energy storage charging and discharging plans.

[0018] Furthermore, in the data processing module: the data segmentation method is specifically: Set the time point of the first meteorological data collected to The second time point is , and so on, then the time point of the first historical load data after the split , the time point of the second historical load data , and so on, satisfying the following relationship: ; And the total number of historical load data after segmentation Satisfies the following relationship: ; in, Indicates the total number of time points for collecting historical load data; From the time point of the first historical load data Reverse all time points that have an impact on the first historical load data. The reverse formula is as follows: ; In the formula, Indicates that the delay factor affected by weather is , the time point corresponding to the meteorological data that has an impact on the first historical load data; The historical temperature feature matrix and the historical light intensity feature matrix are formed as follows: ; ; in, Representative The temperature corresponding to each time point, in degrees Celsius; Representative The light intensity corresponding to each time point, in lux; Furthermore, the data matrix is ​​specifically: If we say that before a certain point in time The data collected at a time point and the data collected at that time point are a group of data, and each The data at each time point corresponds to a row of the photovoltaic power generation characteristic matrix or the wind power generation characteristic matrix. For a data set with data, the number of groups contained in it is the number of samples in the sample space of machine learning. for: ; The photovoltaic power generation characteristic matrix is ​​as follows: ; Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points Indicates photovoltaic power generation, unit kW; The photovoltaic power generation mapping matrix is ​​shown below: ; Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points ; The wind power generation characteristic matrix is ​​as follows: ; Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points Indicates photovoltaic power generation, unit kW; The wind power generation mapping matrix is ​​shown below: ; Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points .

[0019] Furthermore, the prediction module specifically implements: Standardize the photovoltaic power generation characteristic matrix, photovoltaic power generation mapping matrix, wind power generation characteristic matrix, and photovoltaic power generation mapping matrix as a whole; Standardize the historical meteorological characteristic moments; Using machine learning algorithms, The data at each time point and the The relationship between data at different time points.

[0020] Furthermore, in S4, the power balance constraint and energy storage SOC constraint requirements are as follows: At each point in time: ; ; Among them, the superscript Indicates the corresponding time point It represents the load predicted by machine learning in kW. represents the photovoltaic power generation predicted by machine learning, in kW, represents the wind power generation predicted by machine learning, in kW, Indicates the energy storage charge and discharge capacity, positive for discharge, negative for charge, unit is kW, Indicates the amount of electricity purchased, in kW. Indicates the total energy storage capacity in kWh. Indicates time point Energy storage capacity, in kWh; On this basis, a new variable, new energy differential load, is introduced Satisfies the following relationship: .

[0021] Furthermore, in the scenario generation module, the strategy for generating a power balance scenario is as follows: When the new energy difference load is greater than 0 at this time point, energy storage charging is given priority to consume new energy to save new energy resources: When the energy storage capacity is less than 90% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than or equal to 90% of the time: ; Among them, the superscript Indicates the corresponding time point Indicates the charge and discharge efficiency; When the energy storage capacity is less than 90% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is greater than 90% of the time: ; Among them, the superscript Indicates the corresponding time point Represents the amount of wind and solar power curtailment; When the energy storage capacity is equal to 90% of the time: ; Among them, the superscript Indicates the corresponding time point .

[0022] When the new energy difference load is less than 0 at this time point, energy storage discharge is given priority to provide power to save costs: When the energy storage capacity is less than 10% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than or equal to 10% of: ; Among them, the superscript Indicates the corresponding time point ; When the energy storage capacity is greater than 10% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than 10% of: ; Among them, the superscript Indicates the corresponding time point ; When the energy storage capacity is equal to 10% of: ; Among them, the superscript Indicates the corresponding time point .

[0023] Furthermore, the scene generation module further includes making the objective function reach a minimum value: Using predicted The load, photovoltaic power generation, and wind power generation data at each time point are used to generate the power balance scenario at each time point with economy as the objective function, formulate the power purchase plan, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate the objective function of the energy storage charging and discharging plan. as follows: ; Among them, the superscript Indicates the corresponding time point , represents the electricity purchase price, represents the cost of photovoltaic power generation, represents the cost of wind power, Represents the penalty cost for curtailing wind and solar power.

[0024] Beneficial effects of the present invention: The present invention overcomes the drawbacks of inaccurate calculation of power purchase plans and wind and solar power abandonment due to inaccurate short-term predictions of load power and new energy power generation when generating power balance scenarios, and provides a power balance scenario generation method that considers the delayed characteristics of meteorological impact on load and a new energy power prediction method that considers the regularity of new energy power generation itself, which can significantly reduce the errors in the wind and solar power abandonment and power purchase plan when calculating power balance. The method makes maximum use of the relationship between meteorological data and load data, the regularity of new energy power generation itself and other related information, which can not only make up for the drawback that the traditional load power prediction method based on historical meteorological data does not consider the continuity of the impact of weather factors on load, but also can accurately calculate the new energy output prediction based on less available data, thereby improving the accuracy of calculating power purchase plans and wind and solar power abandonment in the process of generating power balance scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The present invention has the following accompanying drawings: Figure 1 Schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0026] In order to make the objects, advantages and features of the present invention more obvious, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] Consider the following factors: 1. Local meteorological, wind power, photovoltaic, and load power data; 2. The time lag of weather’s impact on load; Based on the above factors, the steps of the method for generating a power balance scenario based on source-load prediction include: A. Feature collection layer: Collect the historical active output power series of photovoltaic power plants and wind farms, as well as the historical meteorological data (hereinafter referred to as historical meteorological data) time series and historical load data time series including temperature and light intensity data, standardize some of the collected data, and determine the time scale for predicting the power balance scenario, that is, the total number of time points that need to be predicted. (According to the practice of power system, the length between two time points is 15 minutes) and then use the total number of time points to be predicted Get the corresponding data time point for prediction At the same time, the D-Learner method is used to find the causal relationship between historical meteorological data and historical load data, and all the All values ​​less than 0.05 Delay time ; B. Data processing layer: based on and Parameters, according to the continuity of photovoltaic power generation and wind power generation, the photovoltaic power generation data and wind power generation data are divided into four different data matrices: photovoltaic power generation feature matrix, photovoltaic power generation mapping matrix, wind power generation feature matrix, photovoltaic power generation mapping matrix. At the same time, according to Delay time Convert historical meteorological data into historical temperature feature matrix and historical light intensity feature matrix; C. Prediction layer: Three prediction models are established respectively. The algorithms of the three models are integrated with XGBoost and CatBoost in the Voting method. Forecast the load, photovoltaic power generation and wind power generation at each time point; D. Scene generation layer: using the predicted load, photovoltaic power generation, and wind power generation data at each time point, formulate power purchase plans, solve the output of new energy, calculate wind and solar power abandonment data, and formulate energy storage charging and discharging plans; Based on the above scheme, in step A, the process of preprocessing the collected historical meteorological data (including temperature, light intensity, humidity data, hereinafter referred to as historical meteorological data) and historical load data is as follows: collecting the historical active output power data of photovoltaic power field, the historical active output power data of wind power field, and the historical load data and historical meteorological data of the same time period according to the granularity of 15 minutes, and standardizing the historical meteorological data and historical load data. Based on the above scheme, in step A, according to the continuity of photovoltaic power generation and wind power generation, it can be considered that the new energy output data at a time point before the time period to be predicted is of reference significance for predicting the new energy output during this period. Unit: pcs. According to the practice of power system, the length between two time points represents 15 minutes on the time axis. Then, the total number of time points to be predicted is used. Get the corresponding data time point for prediction , the relationship between them satisfies the following formula: ; Based on the above scheme, in step A, the D-Learner method is used to process the historical meteorological data and the historical load data to obtain all the All values ​​less than 0.05 Weather delay factor Indicates The number of delayed time points of meteorological data that have a delay on historical load data, that is, for the load at a certain time point, the previous number The meteorological data at a certain time point will affect the load at that time point, and for the load at the same time point, it is affected by The impact of meteorological data at different time points.

[0028] It should be noted that the D-Learner method is a mature causal testing algorithm well known to professionals in the field, so the present invention only describes how to use the data obtained by the D-Learner method without introducing it in depth. Based on the above scheme, in step B, according to Weather delay factor The historical meteorological data is converted into the historical temperature feature matrix and the historical light intensity feature matrix. Because the historical meteorological data and historical load data of the same time interval are selected during sampling, when considering the subsequent sample space, the historical meteorological data at any time point is not the main influencing factor of the load data at that point due to the continuity of the impact of weather on load. Therefore, the original data set needs to be segmented, and the time point of the first meteorological data collected is set to (The second time point is , and so on), then the time point of the first historical load data after segmentation (The second time point of historical load data , and so on) satisfy the following relationship: ; And the total number of historical load data after segmentation (also the total number of data in each row of the historical temperature feature matrix and the historical light intensity feature matrix) The following relationship should be satisfied: ; in, Indicates the total number of time points for collecting historical load data (also the total number of time points for collecting historical meteorological data). Due to the weather delay coefficient obtained by the D-Learner method are consecutive integers, indicating the The number of delayed time points of meteorological data that has a delay on historical load data, so the time point of the first historical load data can be Reverse all time points that have an impact on the first historical load data. The reverse formula is as follows: ; In the formula, Indicates that the delay factor affected by weather is When the time point corresponding to the meteorological data that has an impact on the first historical load data is , the above formula can be used to deduce that if the weather affects the delay coefficient The corresponding time point of the first meteorological data is , then the weather affects the delay coefficient The corresponding time point of the first meteorological data is .

[0029] Finally, considering that temperature, light, and load are all continuous, the historical temperature feature matrix and the historical light intensity feature matrix are as follows: ; ; in: Representative The temperature corresponding to each time point, in degrees Celsius; Representative The light intensity corresponding to each time point, in lux; On the basis of the above scheme, in step B, according to the continuity of photovoltaic power generation and wind power generation, the photovoltaic power generation data and wind power generation data are converted into four different data matrices: photovoltaic power generation feature matrix, photovoltaic power generation mapping matrix, wind power generation feature matrix, photovoltaic power generation mapping matrix. No matter it is photovoltaic data or wind power data, if it is called a certain time point before The data collected at a time point and the data collected at that time point are a group of data, and each The data at each time point corresponds to a row of the photovoltaic power generation characteristic matrix or the wind power generation characteristic matrix. For a data set with data, the number of groups contained in it is the number of samples in the sample space of machine learning. (It is also the number of rows of the photovoltaic power generation characteristic matrix or the wind power generation characteristic matrix and the number of rows of the photovoltaic power generation mapping matrix or the wind power generation characteristic matrix) is: ; Taking photovoltaic data as an example, let the time point of the first photovoltaic data collection be (The second time point for collecting photovoltaic data is , and so on), then it and the subsequent common The data at each time point has a corresponding relationship in the subsequent machine learning. The first time point in (The first time point of collecting photovoltaic data and the subsequent The data at each time point has a corresponding relationship in the subsequent machine learning. The second time point is The second time point of collecting PV data is the same as the subsequent The data at each time point has a corresponding relationship in the subsequent machine learning. The first time point of the time points is , and so on) satisfy the following relationship: ; For photovoltaic power generation, the photovoltaic power generation characteristic matrix is ​​as follows: ; Among them: superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points Indicates photovoltaic power generation, unit kW; The photovoltaic power generation mapping matrix is ​​shown below: ; Among them: superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points ; Similarly, let the first time point of collecting wind power data be (The second time point for collecting photovoltaic data is , and so on), then it and the subsequent common The data at each time point has a corresponding relationship in the subsequent machine learning. The first time point in (The first time point of collecting photovoltaic data and the subsequent The data at each time point has a corresponding relationship in the subsequent machine learning. The second time point is The second time point of collecting PV data is the same as the subsequent The data at each time point has a corresponding relationship in the subsequent machine learning. The first time point of the time points is , and so on) satisfy the following relationship: ; The wind power generation characteristic matrix is ​​as follows: ; Among them: superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points Indicates photovoltaic power generation, unit kW; The wind power generation mapping matrix is ​​shown below: ; Among them: superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points ; Based on the above scheme, in step C, for the prediction of photovoltaic and wind power generation, the essence of using the feature matrix to predict the mapping matrix is ​​to use the machine learning algorithm to learn the previous The data at each time point and the The relationship between the data at each time point, that is, the relationship between a row of data in the photovoltaic power generation feature matrix or the wind power generation feature matrix and a row of data in the corresponding photovoltaic power generation mapping matrix or the wind power generation mapping matrix, before which the photovoltaic power generation feature matrix, the photovoltaic power generation mapping matrix, the wind power generation feature matrix, and the photovoltaic power generation mapping matrix need to be standardized as a whole; Based on the above scheme, in step C, the load prediction is to use machine learning algorithm to learn the relationship between the historical meteorological characteristic matrix and the historical load. Before this, the historical meteorological characteristic matrix needs to be standardized; Based on the above scheme, in step D, using the predicted The load, photovoltaic power generation, and wind power generation data at each time point are used to generate power balance scenarios with economy as the objective function, formulate power purchase plans, solve the output of new energy, calculate wind and solar power abandonment data, and formulate power balance constraints and energy storage SOC constraints for energy storage charging and discharging plans as follows: At each point in time: ; ; Among them: superscript Indicates the corresponding time point It represents the load predicted by machine learning in kW. represents the photovoltaic power generation predicted by machine learning, in kW, represents the wind power generation predicted by machine learning, in kW, Indicates the energy storage charge and discharge capacity, positive for discharge, negative for charge, unit is kW, Indicates the amount of electricity purchased, in kW. Indicates the total energy storage capacity in kWh. Indicates time point Energy storage capacity, in kWh; On this basis, a new variable, new energy differential load, is introduced Satisfies the following relationship: ; Based on the above solution, the present invention proposes the following strategy for generating a power balance scenario: (1) When the new energy differential load is greater than 0 at this time point, energy storage charging is given priority to consume new energy to save new energy resources: When the energy storage capacity is less than 90% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than or equal to 90% of the time: ; Among them: superscript Indicates the corresponding time point Indicates the charge and discharge efficiency; When the energy storage capacity is less than 90% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is greater than 90% of the time: ; Among them: superscript Indicates the corresponding time point Represents the amount of wind and solar power curtailment; When the energy storage capacity is equal to 90% of the time: ; Among them: superscript Indicates the corresponding time point .

[0030] (2) When the new energy difference load is less than 0 at this time point, energy storage discharge is given priority to provide electricity to save costs: When the energy storage capacity is less than 10% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than or equal to 10% of: ; Among them: superscript Indicates the corresponding time point ; When the energy storage capacity is greater than 10% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than 10% of: ; Among them: superscript Indicates the corresponding time point ; When the energy storage capacity is equal to 10% of: ; Among them: superscript Indicates the corresponding time point .

[0031] Based on the above solution, in step D, it is necessary to achieve the minimum value of the objective function while completing the strategy in claim 9. The load, photovoltaic power generation, and wind power generation data at each time point are used to generate the power balance scenario at each time point with economy as the objective function, formulate the power purchase plan, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate the objective function of the energy storage charging and discharging plan. as follows: ; Among them: superscript Indicates the corresponding time point Represents the electricity purchase price, unit: Yuan, Represents the cost of photovoltaic power generation, unit: yuan, Represents the cost of wind power, unit: yuan, Represents the penalty cost for curtailing wind and solar power, unit: yuan.

[0032] Based on the same technical concept, another embodiment of the present invention further provides a power balance scenario generation device based on source-load prediction, wherein each module is coupled in sequence, including: a feature acquisition module, a data processing module, a prediction module, and a scenario generation module; The feature acquisition module is used for: Collect historical active output power series, historical meteorological data time series, and historical load data time series; Standardize the data; Determine the time scale over which the power balance scenario needs to be predicted , get the corresponding prediction data time point ; Find the causal relationship between historical meteorological data and historical load data, and get Delay time ; The data processing module is used for: in accordance with and Parameters, respectively, split the photovoltaic power generation data and wind power generation data into four different data matrices: photovoltaic power generation feature matrix, photovoltaic power generation mapping matrix, wind power generation feature matrix, photovoltaic power generation mapping matrix; in accordance with Delay time Convert historical meteorological data into historical temperature feature matrix and historical light intensity feature matrix; The prediction module is used to: Three prediction models were established for Forecast the load, photovoltaic power generation and wind power generation at each time point; The scene generation module is used for: Using predicted Based on the load, photovoltaic power generation, and wind power generation data at each time point, we can formulate power purchase plans, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate energy storage charging and discharging plans.

[0033] Those skilled in the art can understand that the power balance scenario generation device based on source-load prediction and the power balance scenario generation method based on source-load prediction can refer to each other.

[0034] It should be noted that any process or method description in the embodiments may be understood as representing a module, fragment or portion of a code including one or more executable instructions for implementing steps of a specific logical function or process, and that the scope of the preferred embodiments of the present invention includes alternative implementations in which the functions may not be performed in the order shown or discussed, including performing the functions in a substantially simultaneous manner or in the reverse order according to the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0035] It should be noted that the logic and / or steps in the embodiments, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or equipment (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or equipment and execute instructions), or in combination with these instruction execution systems, devices or equipment. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or equipment, or in combination with these instruction execution systems, devices or equipment. More specific examples of computer-readable media (non-exhaustive list) include the following: an electrical connection with one or more wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0036] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0037] A person skilled in the art may understand that all or part of the steps of implementing the above-mentioned embodiment method may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one of the steps of the method embodiment or a combination thereof.

[0038] In addition, each functional module in the embodiment of the present invention may be integrated into a processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0039] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0040] The above embodiments describe the technical solutions of the present invention in detail. Obviously, the present invention is not limited to the described embodiments. Based on the embodiments of the present invention, people familiar with the technical field can also make various changes accordingly, but any changes that are equivalent or similar to the present invention belong to the scope of protection of the present invention.

[0041] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.

Claims

1. A method for generating a power balance scenario based on source-load prediction, characterized in that: The steps include: S1: Feature collection; S2: Data processing; S3: prediction; S4: Scenario generation; The S1 specifically includes: Collect historical active output power series, historical meteorological data time series, and historical load data time series; Standardize the data; Determine the time scale over which the power balance scenario needs to be predicted , get the corresponding prediction data time point ; Find the causal relationship between historical meteorological data and historical load data, and get Delay time ; The S2 specifically includes: in accordance with and Parameters, respectively, split the photovoltaic power generation data and wind power generation data into four different data matrices: photovoltaic power generation feature matrix, photovoltaic power generation mapping matrix, wind power generation feature matrix, photovoltaic power generation mapping matrix; in accordance with Delay time Convert historical meteorological data into historical temperature feature matrix and historical light intensity feature matrix; The S3 specifically includes: Three prediction models were established for Forecast the load, photovoltaic power generation and wind power generation at each time point; The S4 specifically includes: Using predicted Based on the load, photovoltaic power generation, and wind power generation data at each time point, we can formulate power purchase plans, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate energy storage charging and discharging plans.

2. A method for generating a power balance scenario based on source-load prediction according to claim 1, characterized in that: In S1: and The relationship between them satisfies: ； The D-Learner method is used to find the causal relationship between historical meteorological data and historical load data, and all the All values ​​less than 0.05 Delay time ,in, .

3. The method for generating a power balance scenario based on source-load prediction according to claim 1, characterized in that: In S2: the data segmentation method is specifically as follows: Set the time point of the first meteorological data collected to The second time point is , and so on, then the time point of the first historical load data after the split , the time point of the second historical load data , and so on, satisfying: ； And the total number of historical load data after segmentation satisfy: ； in, Indicates the total number of time points for collecting historical load data; From the time point of the first historical load data Reverse all time points that have an impact on the first historical load data, and the reverse calculation is as follows: ; In the formula, Indicates that the delay factor affected by weather is , the time point corresponding to the meteorological data that has an impact on the first historical load data; The historical temperature feature matrix and the historical light intensity feature matrix are formed as follows: ； ； in, Representative The temperature corresponding to each time point, in degrees Celsius; Representative The light intensity corresponding to a time point, in lux.

4. The method for generating a power balance scenario based on source-load prediction according to claim 3, characterized in that: In S2, the data matrix is ​​specifically: If we say that before a certain point in time The data collected at a time point and the data collected at that time point are a group of data, and each The data at each time point corresponds to a row of the photovoltaic power generation characteristic matrix or the wind power generation characteristic matrix. For a data set with data, the number of groups contained in it is the number of samples in the sample space of machine learning. for: ； The photovoltaic power generation characteristic matrix is ​​as follows: ； Among them, the superscript Indicates the corresponding time point , superscript The corresponding number of time points Indicates photovoltaic power generation, unit kW; The photovoltaic power generation mapping matrix is ​​as follows: ； Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points ; The wind power generation characteristic matrix is ​​as follows: ； Among them, the superscript Indicates the corresponding time point , superscript The corresponding number of time points Indicates photovoltaic power generation, unit kW; The wind power generation mapping matrix is ​​as follows: ； Among them, the superscript Indicates the corresponding time point , superscript Indicates the number of corresponding time points .

5. The method for generating a power balance scenario based on source-load prediction according to claim 1, characterized in that: The S3 specifically includes: The photovoltaic power generation characteristic matrix, the photovoltaic power generation mapping matrix, the wind power generation characteristic matrix, and the photovoltaic power generation mapping matrix are standardized as a whole; Standardize the historical meteorological characteristic moments; Using machine learning algorithms, learning The data at each time point and the The relationship between data at different time points.

6. The method for generating a power balance scenario based on source-load prediction according to claim 1, characterized in that: In S4, the power balance constraint and energy storage SOC constraint requirements are as follows: At each point in time: ； ； Among them, the superscript Indicates the corresponding time point It represents the load predicted by machine learning in kW. represents the photovoltaic power generation predicted by machine learning, in kW, represents the wind power generation predicted by machine learning, in kW, Indicates the energy storage charge and discharge capacity, positive for discharge, negative for charge, unit is kW, Indicates the amount of electricity purchased, in kW. Represents the total energy storage capacity in units of , Indicates time point Energy storage capacity, in ; On this basis, a new variable, new energy differential load, is introduced satisfy: 。 7. The method for generating a power balance scenario based on source-load prediction according to claim 1, characterized in that: In S4, the strategy for generating a power balancing scenario is as follows: When the new energy difference load is greater than 0 at this time point, energy storage charging is given priority to consume new energy to save new energy resources: When the energy storage capacity is less than 90% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than or equal to 90% of the time: ； Among them, the superscript Indicates the corresponding time point Indicates the charge and discharge efficiency; When the energy storage capacity is less than 90% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is greater than 90% of the time: ； Among them, the superscript Indicates the corresponding time point Represents the amount of wind and solar power curtailment; When the energy storage capacity is equal to 90% of the time: ； Among them, the superscript Indicates the corresponding time point ; When the new energy difference load is less than 0 at this time point, energy storage discharge is given priority to provide power to save costs: When the energy storage capacity is less than 10% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than or equal to 10% of: ； Among them, the superscript Indicates the corresponding time point ; When the energy storage capacity is greater than 10% of the total energy storage capacity after charging, and the estimated energy storage capacity after charging is less than 10% of: ； Among them, the superscript Indicates the corresponding time point ; When the energy storage capacity is equal to 10% of: ； Among them, the superscript Indicates the corresponding time point .

8. The method for generating a power balance scenario based on source-load prediction according to claim 1, characterized in that: The S4 also includes minimizing the objective function: Using predicted The load, photovoltaic power generation, and wind power generation data at each time point are used to generate the power balance scenario at each time point with economy as the objective function, formulate the power purchase plan, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate the objective function of the energy storage charging and discharging plan. as follows: ； Among them, the superscript Indicates the corresponding time point , represents the electricity purchase price, represents the cost of photovoltaic power generation, represents the cost of wind power, Represents the penalty cost for curtailing wind and solar power.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A device for generating a power balance scenario based on source-load prediction, characterized in that: include: Feature collection module, data processing module, prediction module, and scene generation module; The feature acquisition module is used for: Collect historical active output power series, historical meteorological data time series, and historical load data time series; Standardize the data; Determine the time scale over which the power balance scenario needs to be predicted , get the corresponding prediction data time point ; Find the causal relationship between historical meteorological data and historical load data, and get Delay time ; The data processing module is used for: in accordance with and Parameters, respectively, split the photovoltaic power generation data and wind power generation data into four different data matrices: photovoltaic power generation feature matrix, photovoltaic power generation mapping matrix, wind power generation feature matrix, photovoltaic power generation mapping matrix; in accordance with Delay time Convert historical meteorological data into historical temperature feature matrix and historical light intensity feature matrix; The prediction module is used to: Three prediction models were established for Forecast the load, photovoltaic power generation and wind power generation at each time point; The scene generation module is used for: Using predicted Based on the load, photovoltaic power generation, and wind power generation data at each time point, we can formulate power purchase plans, solve the output of new energy, calculate the wind and solar power abandonment data, and formulate energy storage charging and discharging plans.

Images