Virtual power plant baseline calculation method introducing user confidence

Abstract

The invention relates to a virtual power plant baseline calculation method introducing user confidence, and the method comprises the following steps: carrying out the analysis based on a user load prediction curve, and obtaining a historical prediction load value before an event release day and a user load prediction value of the event release day; calculating a user confidence coefficient by using the historical predicted load value and the corresponding actual load value; screening out a plurality of typical days before the demand response day; calculating a historical baseline load value according to the actual load value of the typical day; and calculating a virtual power plant load baseline based on the user confidence, the historical baseline load value and the user load prediction value. According to the method, the load baseline of the virtual power plant can be optimized, the influence of the load prediction deviation on the baseline is reduced, and users are better encouraged to adjust energy consumption behaviors.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual power plant baseline calculation, and in particular to a virtual power plant baseline calculation method introducing user confidence. Background Art

[0002] A baseline is a crucial basis for evaluating regulatory markets. It typically refers to the expected electricity consumption patterns or load levels of electricity users before load management measures are implemented and ancillary markets are activated. The baseline load measures changes in user load profiles after implementing measures like demand response. It serves as a benchmark for calculating the regulatory volume of new operators, such as virtual power plants, when participating in peak-shaving ancillary services and demand response transactions, and is a key basis for calculating their compensation. Existing technologies often use the average load value over a specified number of days before a transaction, without executing orders. However, due to factors such as weather and production schedules that influence user energy consumption, actual user plans can deviate significantly from the calculated baseline, leading to situations where users earn revenue despite responding to uncalculated or unactual responses. Furthermore, if there are a large number of demand responses, the time corresponding to the data used in baseline calculation may deviate significantly from the current time, potentially leading to a phenomenon where more responses lead to a more equal baseline. Of course, there are also cases where users maliciously claim revenue. Therefore, it is necessary to more scientifically consider the relationship between historical data, user forecast curves, and user credibility to optimize baseline calculation methods.

[0003] Document CN119518740A discloses a load baseline prediction method, which mainly considers the calculation of the baseline based on historical data. However, as the number of demand responses increases, this method has the contradiction of long data forward time and large deviation from the user's operation plan. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a virtual power plant baseline calculation method that introduces user confidence, which can more scientifically consider the relationship between historical data, user prediction curve and user credibility, and optimize the baseline calculation method.

[0005] The technical solution adopted by the present invention to solve the technical problem is to provide a virtual power plant baseline calculation method that introduces user confidence, including the following steps:

[0006] Based on the user load forecast curve, the historical forecast load value before the event release date and the user load forecast value on the event release date are analyzed;

[0007] Calculating user confidence using the historical predicted load value and its corresponding actual load value;

[0008] Filter out several typical days before the demand response day;

[0009] Calculating a historical baseline load value based on the actual load value of the typical day;

[0010] Based on the user confidence, the historical baseline load value and the user load forecast value, a virtual power plant load baseline is calculated.

[0011] Furthermore, the user confidence is expressed as

[0012] ,

[0013] in, is the data collection point, represents the user confidence, represents the historical forecast load value, represents the actual load value corresponding to the historical predicted load value, Indicates the total number of data collection times.

[0014] Furthermore, the user confidence is calculated based on data from several normal days before the event is released, and the normal days do not include demand response days, emergency peak-avoidance days, and power outage maintenance days.

[0015] Furthermore, the historical baseline load value is a weighted average of the actual load values ​​of each typical day.

[0016] Furthermore, the virtual power plant load baseline is expressed as

[0017] ,

[0018] in, Indicates the The virtual power plant load baseline for the day, represents the user confidence, Indicates the The historical forecast load value of the day, Indicates the The historical baseline load value of the day.

[0019] Furthermore, it also includes the step of introducing a first adjustment factor to correct the load baseline of the virtual power plant, wherein the first adjustment factor is the ratio of the load average of a first set number of time periods before the event occurs to the historical load average, and the historical load average is the average value of the total load of all the same-name time periods of the typical days relative to the total number of time periods.

[0020] Furthermore, the virtual power plant load baseline after correction by introducing the first adjustment factor is the product of the virtual power plant load baseline and the first adjustment factor.

[0021] Further, the step of introducing a second adjustment factor to correct the virtual power plant load baseline is included, wherein the second adjustment factor is a percentage value less than 1.

[0022] Further, the corrected virtual power plant load baseline is represented as

[0023] ,

[0024] wherein, represents the virtual power plant load baseline on the th day, represents the virtual power plant load baseline on the th day after being corrected by the second adjustment factor, represents the second adjustment factor, represents the actual load value on the th day when the event occurs.

[0025] Further, the second adjustment factor is 90%.

[0026] Further, the multiple typical days before the demand response day are selected, including:

[0027] If the demand response day is a weekday, the first set number of first typical days before the demand response day are selected, wherein the first typical days satisfy a first set condition.

[0028] If the demand response day is a non-weekday, the second set number of second typical days before the demand response day are selected, wherein the second typical days satisfy a second set condition.

[0029] Further, the first set condition is that the first typical days do not include non-weekdays, maintenance days, demand response days and auxiliary service days, and the daily load fluctuation rate is not more than a threshold value.

[0030] Further, the second set condition is that the second typical days do not include weekdays, maintenance days, demand response days and auxiliary service days, and the daily load fluctuation rate is not more than a threshold value.

[0031] Beneficial effects

[0032] Compared with the prior art, the present application has the following advantages and positive effects: the present application can better optimize the load baseline and reduce the influence of load prediction deviation by combining and correcting the predicted load reported by the virtual power plant and other subjects according to the accuracy of user historical load prediction; further, the influence of weather change can be reduced by introducing the first correction factor to further correct the optimized load baseline; for the user with strong load volatility, the second correction factor is introduced to make post-correction according to the load volatility correction result, on the one hand, the situation that the user adjusts the plan but cannot get the subsidy is avoided as much as possible, and on the other hand, the adverse influence of user speculation declaration is avoided as much as possible. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 is a flowchart of the embodiment of the present application;

[0034] Figure 2 is a flowchart of the preferred embodiment in the embodiment of the present application. DETAILED DESCRIPTION

[0035] The present application will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present application and not used to limit the scope of the present application. In addition, it should be understood that after reading the content taught by the present application, those skilled in the art can make various modifications or changes to the present application, and these equivalent forms also fall within the scope defined by the appended claims of the present application.

[0036] The professional terms involved in the present embodiment include:

[0037] The embodiment of the present application relates to a virtual power plant baseline calculation method introducing user confidence, as shown in Figure 1 , comprising the following steps:

[0038] S0 obtains a user load prediction curve;

[0039] S1 calculates the historical load prediction accuracy as the user confidence according to the historical predicted load value before the event release date and the corresponding actual load value;

[0040] S2 selects a plurality of typical days before the demand response day;

[0041] S3 calculates the historical baseline load value according to the actual load value of the typical day;

[0042] S4 calculates the virtual power plant load baseline based on the historical load prediction accuracy, the historical baseline load value and the user load prediction value.

[0043] In some embodiments, the load prediction result involved in step S0 can be autonomously confirmed according to the self operation plan, or concluded according to the load prediction method, and the load prediction can be performed by using an LSTM algorithm or the like.

[0044] More specifically, the historical load prediction accuracy involved in step S1 is the difference between the predicted load and the actual load data of the user in the event period corresponding to the M normal days before the event (M is a reference value of 5, without special situations such as demand response, emergency peak avoidance, power grid maintenance, and the like), and the average value is taken to obtain the historical load prediction accuracy, denoted as The algorithm is designed as follows:

[0045] ,

[0046] Among them, represents the historical predicted load value of the data collection point of the M normal days before the event, represents the historical actual load value of the data collection point of the M normal days, the frequency of data collection can be 15 minutes, represents the total number of data collection, and the collected data can be user reported or real-time load data.

[0047] The typical day selection involved in step S2 can be divided into two cases of weekdays and non-weekdays, and needs to meet the corresponding set conditions. The following method can be used:

[0048] 1) If the demand response and other regulatory market execution demand occur on weekdays, select days (which can be 5, 7, or 10) before the execution day, among which non-weekdays, maintenance days, and days of user participation in demand response, auxiliary service, and load fluctuation rate exceeding the threshold value are excluded. After exclusion, if there are less than days, the days are sequentially selected forward, and days should be supplemented. From the above days, the two days with the largest maximum and minimum daily maximum load of power users are excluded, and the remaining -2) days are called typical days.

[0049] 2) If the demand response and other regulatory market execution demand occurs on non-weekdays, the nearest 3 non-weekdays before the execution day are selected as typical days, among which maintenance days, days of user participation in demand response, auxiliary service, and load fluctuation rate exceeding the threshold value are excluded. After exclusion, if there are less than 3 days, the days are sequentially selected forward, and 3 days should be supplemented.

[0050] The baseline calculation involved in step S3 takes the weighted average load curve in the response period corresponding to the typical day as the historical baseline load, and the formula is as follows:​

[0051]

[0052] In the formula:

[0053] is the baseline load value of the jth day;

[0054] is the load value of the jth day d days ago;

[0055] is the total number of typical days;

[0056] is the weight coefficient of each typical day, which can be set according to the local load characteristics. For example, if the typical day is 5 days, the weight coefficient A can be set to 50%, 25%, 12.5%, 6.25%, and 6.25% in the positive order from the first day to the fifth day before the typical day.

[0057] Step S4 multiplies the predicted load value reported by the user subject at each time point by the historical load prediction accuracy, and adds (1-historical load prediction accuracy) multiplied by the baseline load to obtain a new baseline load value at each time point. The higher the load prediction accuracy of the user, the closer the baseline load will be to the user prediction curve; otherwise, it will be closer to the baseline load calculated by the conventional method. The algorithm is designed as follows:

[0058] ,

[0059] In the formula:

[0060] is the baseline load value of the jth day; is the new baseline load after correction;

[0061] is the baseline load value of the jth day;

[0062] is the historical load prediction accuracy.

[0063] In some preferred embodiments, when extreme weather or other weather changes are encountered, a meteorological adjustment factor can be introduced to correct the virtual power plant load baseline obtained in step S4.

[0064] Specifically, the ratio of the average load of several hours before the event to the total average load of the selected typical day in the corresponding period can be selected as the weather adjustment factor. Specifically, the average load of 2 hours before the event can be selected, which is closest to the load at the time of the event and has less possibility of user speculation. The user-adjusted baseline load should be obtained by multiplying the calculated uncorrected baseline load by the weather adjustment factor.

[0065] For users with strong load fluctuation, the load fluctuation correction result can also be used for post-correction. Specifically, a percentage coefficient X can be introduced as an adjustment factor, X is used as the weight of the baseline load, (1-X) is used as the weight of the load value at the time of the event, and the weighted sum of the two is calculated as the new baseline load. Among them, the initial basic load should account for a large proportion to curb the user's speculative behavior.

[0066] It should be noted that the calculation of the load baseline can be performed for different time granularities such as minutes, hours, and days, and those skilled in the art can make corresponding adjustments according to actual needs.

[0067] One preferred embodiment of the present embodiment is as follows: Figure 2 As shown in the figure, it includes:

[0068] 1. The user load curve prediction result is reported by the virtual power plant or generated by the load prediction algorithm;

[0069] 2. Calculate the historical load prediction accuracy of the virtual power plant;

[0070] 3. Select a typical day;

[0071] 4. Calculate the baseline based on the baseline load calculation method based on the historical data of the typical day;

[0072] 5. Determine the optimized virtual power plant baseline based on the load curve prediction result and the conventional baseline calculation result;

[0073] 6. Revise as needed according to the weather;

[0074] 7. Revise as needed according to the load fluctuation rate;

[0075] 8. Publish the baseline load.

[0076] Among them, the historical load prediction accuracy is to select the predicted load and actual load data of the user in M normal days before the event (M reference value is 5, without demand response, emergency peak avoidance, cooperation with power grid power outage maintenance and other special circumstances), calculate the difference between the two and take the average to get the historical load prediction accuracy, denoted as The algorithm design is as follows:

[0077] ,

[0078] wherein, represents the historical predicted load value of the data collection point of the M normal days before the event publishing, represents the historical actual load value of the data collection point of the M normal days, The frequency of data collection can be 15 minutes, represents the total number of data collection, the predicted data can be the load prediction value reported by the user or obtained based on the prediction algorithm, and the actual load value is the real-time load data collected.

[0079] The selection of typical days can be divided into two cases of working days and non-working days, which need to meet the corresponding set conditions. The following method can be used:

[0080] 1) If the demand response and other regulatory market execution demand occurs on a working day, the execution day is selected day( 5, 7 or 10 can be taken), among which non-working days, maintenance days, user participation in demand response, auxiliary service days, and load fluctuation rate exceeding the threshold value are excluded, and the part less than day is sequentially selected forward, and day should be made up, and the two days with the largest maximum and minimum daily maximum load of the above day are excluded again, and the remaining -2) days are called typical days;

[0081] 2) If the demand response and other regulatory market execution demand occurs on a non-working day, the last 3 non-working days before the execution day are selected as the typical days, among which maintenance days, user participation in demand response, auxiliary service days, and load fluctuation rate exceeding the threshold value are excluded, and the part less than 3 days is sequentially selected forward, and 3 days should be made up.

[0082] The baseline calculation takes the weighted average load curve in the response event period corresponding to the typical day as the historical baseline load, and the response event period can be specified according to the actual demand, and the formula is as follows:

[0083]

[0084] In the formula,

[0085] is the baseline load value of the event period corresponding to the jth day;

[0086] is the average load value of the event period corresponding to the dth day before the jth day;

[0087] ​​​ is the total number of typical days;

[0088] The weight coefficient for each typical day can be set according to the local load characteristics. For example, if there are 5 typical days, the weight coefficient A can be set in a positive sequence from 1 day before the typical day to 5 days before the typical day, with values ​​of 50%, 25%, 12.5%, 6.25%, and 6.25% respectively.

[0089] The predicted load value at each time point reported by the user is multiplied by the historical load prediction accuracy, and then (1-historical load prediction accuracy) is added to the baseline load to obtain the new baseline load value at each time point. The higher the user's load prediction accuracy, the closer the baseline load will be to the user's prediction curve; conversely, the closer it will be to the baseline load calculated by conventional methods. The algorithm is designed as follows:

[0090] ,

[0091] Where:

[0092] is the historical load forecast accuracy for day j Corrected corresponding event period baseline load;

[0093] is the event period corresponding to day j Baseline load value;

[0094] is the accuracy of historical load forecast.

[0095] The selection of correction factors is optional. For example, when encountering extreme weather or other situations where weather changes need to be considered, meteorological adjustment factors should be introduced for correction.

[0096] Step 6: Select the ratio of the average load K hours before the event to the average total load for the selected typical day corresponding to the time period as the meteorological adjustment factor. The K hours before the event should be selected considering that the load at this time is closest to the load at the time of the event and that users are less likely to speculate. The reference value for K is 2. The user-adjusted baseline load should be calculated by multiplying the calculated unadjusted baseline load by the meteorological adjustment factor. The algorithm is designed as follows:

[0097]

[0098] The jth day's meteorological adjustment factor is further adjusted to correspond to the event period baseline load;

[0099] is the historical load forecast accuracy for day j the baseline load value of the event period corresponding to the jth day after the correction;

[0100] the load value of the event period corresponding to the jth day before d days (i.e., the selected typical day); the historical load average value K hours before the event occurred, which can be calculated by the following formula:

[0101] the baseline load value of the event period corresponding to the jth day before d days (i.e., the selected typical day) after the correction;

[0102] .

[0103] Step 7: For users with strong load fluctuations, the baseline load can be corrected after the event, and the percentage coefficient X of the baseline load fluctuation correction can be added to the percentage coefficient (1-X) of the load value at the time of the event as the new baseline load. The initial baseline load should account for a large proportion to curb the speculative behavior of users, such as setting it to 90%. The algorithm design is as follows:

[0104]

[0105] the new baseline load of the event period corresponding to the jth day after the correction of load fluctuation;

[0106] the baseline load value of the event period corresponding to the jth day after the correction of historical load prediction accuracy

[0107] the load value of the event period corresponding to the jth day after the correction of historical load prediction accuracy

[0108] X is the correction coefficient, which is a percentage value.

[0109] The software carrying carrier of the embodiment is a corresponding system module, and the system is composed of a baseline publisher and a user declarer. The baseline publisher is the main system, which is connected to the user declarer system through an interface. The user declarer can use mobile terminals, web pages, etc. to carry.

[0110] The system of the baseline publisher has functions of data collection, user load prediction data acquisition, prediction accuracy evaluation, typical day data selection, baseline calculation based on historical data, optimized virtual power plant baseline calculation, meteorological factor correction, fluctuation correction, baseline release, etc.

[0111] The system of the user declarer has functions of load prediction data declaration, baseline viewing, etc.​​​​​

Claims

1. A virtual power plant baseline calculation method incorporating user confidence, characterized by, The method comprises the following steps: Based on the user load prediction curve, the historical predicted load value before the event release date and the user load prediction value on the event release date are analyzed; The user confidence is calculated by using the historical predicted load value and the corresponding actual load value; A number of typical days before the demand response day are screened out; The historical baseline load value is calculated according to the actual load value of the typical days; The virtual power plant load baseline is calculated based on the user confidence, the historical baseline load value and the user load prediction value.

2. The method of claim 1, wherein, The user confidence is represented as ， wherein, is a data collection point, represents the user confidence, represents the historical predicted load value, represents the actual load value corresponding to the historical predicted load value, represents the total number of data collection.

3. The method of claim 2, wherein, The user confidence is calculated according to the data of a number of normal days before the event release date, and the normal days do not include the demand response day, the emergency peak avoidance day and the power outage maintenance day.

4. The method of claim 1, wherein, The historical baseline load value is the weighted average of the actual load values of each typical day.

5. The method of claim 1, wherein, The virtual power plant load baseline is represented as ， in, Indicates the The virtual power plant load baseline for the day, represents the user confidence, Indicates the The historical forecast load value of the day, Indicates the The historical baseline load value of the day.

6. The method of claim 1, wherein, The step of introducing a first adjustment factor to modify the virtual power plant load baseline is further included, the first adjustment factor is the ratio of the load average value of the first set number of time periods before the event to the historical load average value, and the historical load average value is the average value of the total load of the same time period of all the typical days relative to the total number of time periods.

7. The method of claim 6, wherein, The virtual power plant load baseline modified by the first adjustment factor is the product of the virtual power plant load baseline and the first adjustment factor.

8. The method of claim 1, wherein, The step of introducing a second adjustment factor to modify the virtual power plant load baseline is further included, and the second adjustment factor is a set percentage value less than 1.

9. The method of claim 8, wherein, The modified virtual power plant load baseline is represented as ， wherein, represents the virtual power plant load baseline of the day, represents the virtual power plant load baseline of the day, represents the virtual power plant load baseline of the day, represents the virtual power plant load baseline of the day, represents the second adjustment factor, represents the virtual power plant load baseline of the day, represents the actual load value at the time of the day event.

10. The method of claim 9, wherein, The second adjustment factor is 90%.

Images