A method for load forecasting of an energy storage system

Abstract

The application relates to an energy storage system load prediction method, which comprises the following steps: firstly, generating a sample time-power curve based on sample running time information and sample running power information in sample running data; secondly, marking the sample time-power curve with sample running weather information to obtain a first-marked sample time-power curve; thirdly, marking the first-marked sample time-power curve with sample running temperature information to obtain a second-marked sample time-power curve; and finally, marking the second-marked sample time-power curve with sample running humidity information to obtain a third-marked sample time-power curve. The influence of different running information on the sample time-power curve is combined to obtain the influence of different running information on power supply in sample data, and then a power demand prediction model is constructed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy storage, in particular to a load prediction method for an energy storage system. BACKGROUND

[0002] Under the background of the current power system transformation, with the widespread popularity of renewable energy, the operation characteristics of the power system have changed significantly, especially the intensification of peak-valley price, which provides strong impetus for the development of industrial and commercial energy storage field.

[0003] However, this development has also brought new challenges. Under the background of two-part electricity price, industrial and commercial energy storage systems may face the problem of imbalance between supply and demand in the process of realizing peak clipping and valley filling, because of the uncertainty and volatility of load, which may trigger the maximum demand. SUMMARY

[0004] Therefore, it is necessary to provide a load prediction method for an energy storage system to solve the problem that the supply and demand cannot be balanced in the discharging process of the traditional energy storage system.

[0005] The present application provides a load prediction method for an energy storage system, comprising:

[0006] Obtaining sample data of a plurality of samples, the sample data comprising sample operation data;

[0007] Analyzing the sample operation data to obtain sample operation time information, sample operation power information, sample operation weather information, sample operation temperature information, and sample operation humidity information in the sample operation data;

[0008] Generating a sample time-power curve for each sample data based on the sample operation time information and the sample operation power information of each sample data;

[0009] Labeling the sample time-power curve of each sample data based on the sample operation weather information of each sample data to obtain a first labeled sample time-power curve of each sample data;

[0010] Labeling the first labeled sample time-power curve of each sample data based on the sample operation temperature information of each sample data to obtain a second labeled sample time-power curve of each sample data;

[0011] Labeling the second labeled sample time-power curve of each sample data based on the sample operation humidity information of each sample data to obtain a third labeled sample time-power curve of each sample data;

[0012] constructing an electricity demand prediction model based on the per-sample running time information, the per-sample running power information, the per-sample running weather information, the per-sample running temperature information, the per-sample running humidity information, the per-first-time-labeled sample time-power curve, the per-second-time-labeled sample time-power curve, and the per-third-time-labeled sample time-power curve;

[0013] acquiring a sample data to be tested;

[0014] parsing the sample data to be tested to obtain sample running time information, sample running power information, sample running weather information, sample running temperature information, and sample running humidity information in the sample data to be tested;

[0015] inputting the obtained sample running time information, sample running power information, sample running weather information, sample running temperature information, and sample running humidity information into the electricity demand prediction model, starting the electricity demand prediction model, and obtaining predicted demand electricity output by the electricity demand prediction model.

[0016] The present application relates to a kind of energy storage system load prediction method, by first based on sample running data in sample running time information and sample running power information Generation sample time-power curve, then with sample running weather information sample time-power curve is marked to obtain the first time labeled sample time-power curve, again with sample running temperature information to the first time labeled sample time-power curve mark to obtain the second time labeled sample time-power curve, finally with sample running humidity information to the second time labeled sample time-power curve mark to obtain the third time labeled sample time-power curve, the influence of different running information to sample time-power curve is combined to obtain the influence of different running information to the power supply in sample data, to construct electricity demand prediction model, and a large amount of data training is carried out to electricity demand prediction model to make that electricity demand prediction model has the function of predicting demand electricity. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The flowchart of the energy storage system load prediction method provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application is further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0019] As Figure 1As shown, in an embodiment of the present application, the energy storage system load prediction method comprises the following S001 to S010:

[0020] S001, sample data of a plurality of samples is acquired, and the sample data comprises sample operation data.

[0021] Specifically, the sample operation data refers to data of the energy storage system in a historical discharge process.

[0022] S002, sample operation time information, sample operation power information, sample operation weather information, sample operation temperature information, and sample operation humidity information in the sample operation data are obtained by analyzing the sample operation data.

[0023] S003, a sample time-power curve of each sample data is generated based on the sample operation time information and the sample operation power information of each sample data.

[0024] Specifically, the sample time-power curve can refer to a curve of the power of the energy storage system in a unit time, and the unit time can be in units of hours, days, weeks, etc. The area enclosed by the curve and the time axis in the sample time-power curve is the power output of the energy storage system.

[0025] S004, the sample time-power curve of each sample data is marked based on the sample operation weather information of each sample data, to obtain a first marked sample time-power curve of each sample data.

[0026] S005, the first marked sample time-power curve of each sample data is marked based on the sample operation temperature information of each sample data, to obtain a second marked sample time-power curve of each sample data.

[0027] S006, the second marked sample time-power curve of each sample data is marked based on the sample operation humidity information of each sample data, to obtain a third marked sample time-power curve of each sample data.

[0028] S007, an electricity demand prediction model is constructed based on each sample operation time information, each sample operation power information, each sample operation weather information, each sample operation temperature information, each sample operation humidity information, each first marked sample time-power curve, each second marked sample time-power curve, and each third marked sample time-power curve.

[0029] S008, a to-be-tested sample data is acquired.

[0030] S009, analyze the sample data to be tested to obtain the sample running time information, sample running power information, sample running weather information, sample running temperature information, and sample running humidity information in the sample data to be tested.

[0031] S010, input the obtained sample running time information, sample running power information, sample running weather information, sample running temperature information, and sample running humidity information into the power demand prediction model, start the power demand prediction model, and obtain the predicted demand power output by the power demand prediction model.

[0032] In this embodiment, the sample time-power curve is first generated based on the sample running time information and the sample running power information in the sample running data, then the sample time-power curve is marked with the sample running weather information to obtain the first marked sample time-power curve, the first marked sample time-power curve is marked with the sample running temperature information to obtain the second marked sample time-power curve, and finally the second marked sample time-power curve is marked with the sample running humidity information to obtain the third marked sample time-power curve. The influence of different running information on the sample time-power curve is combined to obtain the influence of different running information on the power supply in the sample data, and the power demand prediction model is constructed, and a large amount of data training is performed on the power demand prediction model to make the power demand prediction model have the function of predicting the demand power.

[0033] In an embodiment of the present application, the sample time-power curve of each sample data is generated based on the sample running time information and the sample running power information of each sample data, including the following S003a to S003e:

[0034] S003a, selecting one sample running data.

[0035] S003b, obtaining the sample running time information and the sample running power information in the sample running data.

[0036] S003c, creating a two-dimensional coordinate system.

[0037] S003d, based on the sample running time information, the sample running power information is included in the two-dimensional coordinate system to obtain the sample time-power curve.

[0038] S003e, returning to the selected one sample running data until each sample running data has been selected once.

[0039] Specifically, in the two-dimensional coordinate system, time is taken as a unit of one coordinate axis and power is taken as a unit of another coordinate axis.

[0040] In the embodiment, the power information corresponding to different time in the sample running data is incorporated into a two-dimensional coordinate system, and then a sample time-power curve is obtained. In the sample time-power curve, the area of the figure enclosed by the curve and the time coordinate axis is the power value in the time period or the short time period.

[0041] In an embodiment of the present application, the sample running weather information based on each sample data marks the sample time-power curve of each sample data, and obtains the first marked sample time-power curve of each sample data, including the following S004a to S004e:

[0042] S004a, selecting one sample running data.

[0043] S004b, obtaining the sample time-power curve in the sample running data.

[0044] S004c, obtaining the sample running weather information in the sample running data, the sample running weather information including a plurality of weather types, and the weather types including sunny, cloudy, overcast, light rain, moderate rain and heavy rain.

[0045] S004d, marking the sample time-power curve based on each weather type and the time period corresponding to each weather type, and obtaining the first marked sample time-power curve.

[0046] S004e, returning to the selected one sample running data until each sample running data is selected once.

[0047] Specifically, marking the sample time-power curve based on each weather type and the time period corresponding to each weather type, and obtaining the first marked sample time-power curve includes:

[0048] selecting one weather type;

[0049] analyzing the weather type to obtain all time intervals corresponding to the weather type;

[0050] marking the curve segment corresponding to all time intervals in the sample time-power curve as the weather type;

[0051] returning to the selected one weather type until each weather type is selected once.

[0052] In the embodiment, the time intervals of different weather types are obtained, and the corresponding time intervals are mapped to the sample time-power curve of the corresponding time, and then the multi-segment sample time-power curve of the weather type is obtained, so as to distinguish different weather types to increase the accuracy of the prediction of the discharge amount.

[0053] In an embodiment of the present application, the sample running weather information based on each sample data marks the sample time-power curve of each sample data to obtain the first marked sample time-power curve of each sample data, and then further includes the following S004f to S004l:

[0054] S004f, selecting one weather type.

[0055] S004g, analyzing the weather type to obtain at least one first marked sample time-power curve in the weather type.

[0056] S004h, selecting one first marked sample time-power curve.

[0057] S004i, analyzing the first marked sample time-power curve to obtain the actual power supply value in the first marked sample time-power curve.

[0058] S004j, returning to the selected first marked sample time-power curve until each first marked sample time-power curve is selected once to obtain a plurality of actual power supply values.

[0059] S004k, taking the maximum actual power supply value in the obtained plurality of actual power supply values as the first prediction range maximum boundary value, and taking the minimum actual power supply value in the obtained plurality of actual power supply values as the first prediction range minimum boundary value to obtain the first prediction range.

[0060] S004l, returning to selecting one weather type until each weather type is selected once.

[0061] Specifically, the actual power supply value in the first marked sample time-power curve refers to at least one unit time power value contained in the multi-segment curve of one weather type in the first marked sample time-power curve, and each segment curve contains at least one unit time.

[0062] In the embodiment, the actual power supply amount in a unit time under the same weather type is counted, and then the range of the actual power supply amount in a unit time under the weather type is obtained, which is used as the bottom logic to calculate the range of the actual power supply amount of different weather types, and then the actual power supply amount in a unit time corresponding to different weather types is obtained.

[0063] In an embodiment of the present application, the sample running temperature information of each sample data is used to mark the first marked sample time-power curve of each sample data, and the second marked sample time-power curve of each sample data is obtained, including the following S005a to S005e.

[0064] S005a, selecting a sample running data.

[0065] S005b, obtaining the first marked sample time-power curve in the sample running data.

[0066] S005c, obtaining the sample running temperature information in the sample running data, wherein the sample running temperature information includes at least one temperature value.

[0067] S005d, marking the first marked sample time-power curve based on all temperature values in the sample running temperature information to obtain the second marked sample time-power curve.

[0068] S005e, returning to selecting a sample running data until each sample running data is selected once.

[0069] Specifically, marking the first marked sample time-power curve based on all temperature values in the sample running temperature information to obtain the second marked sample time-power curve includes:

[0070] selecting a temperature value;

[0071] analyzing the temperature value to obtain all time intervals corresponding to the temperature value;

[0072] marking the curve segment corresponding to all time intervals in the sample time-power curve as the temperature value;

[0073] returning to selecting a temperature value until each temperature value is selected once.

[0074] In the embodiment, the time intervals of different temperature values are obtained, and the corresponding time intervals are mapped to the sample time-power curve of the corresponding time, and then the multi-segment sample time-power curve of the temperature value is obtained, so as to distinguish different temperature values to further increase the accuracy of the prediction of the discharge amount.

[0075] In an embodiment of the present application, the sample running temperature information based on each sample data marks the sample time-power curve after the first marking of each sample data, to obtain the sample time-power curve after the second marking of each sample data, and then further includes S005f to S005l as follows:

[0076] S005f, select a temperature value.

[0077] S005g, analyze the temperature value to obtain at least one sample time-power curve after the second marking in the temperature value.

[0078] S005h, select a sample time-power curve after the second marking.

[0079] S005i, analyze the sample time-power curve after the second marking to obtain the actual power supply value in the sample time-power curve after the second marking.

[0080] S005j, return to select a sample time-power curve after the second marking until each sample time-power curve after the second marking is selected once, to obtain a plurality of actual power supply values.

[0081] S005k, take the maximum actual power supply value in the plurality of actual power supply values as the maximum boundary value of the second prediction range, and take the minimum actual power supply value in the plurality of actual power supply values as the minimum boundary value of the second prediction range, to obtain the second prediction range.

[0082] S005l, return to select a temperature value until each temperature value is selected once.

[0083] Specifically, the actual power supply value in the sample time-power curve after the second marking refers to at least one unit time power value contained in the multi-segment curve of a temperature value in the sample time-power curve after the second marking, and each segment of the curve contains at least one unit time.

[0084] In the embodiment, the actual power supply amount of the same temperature value in a unit time is counted, and then the range of the actual power supply amount value appearing in a unit time under the condition of the temperature value is obtained, which is used as the bottom logic to calculate the range of the actual power supply amount value of different temperature values, and then the actual power supply amount value in a unit time corresponding to different temperature values is obtained.

[0085] In an embodiment of the present application, the sample running humidity information based on each sample data marks the secondly marked sample time-power curve of each sample data to obtain the thirdly marked sample time-power curve of each sample data, including the following S006a to S006e:

[0086] S006a, selecting a sample running data;

[0087] S006b, obtaining the secondly marked sample time-power curve in the sample running data;

[0088] S006c, obtaining the sample running humidity information in the sample running data, wherein the sample running humidity information includes at least one humidity value;

[0089] S006d, marking the secondly marked sample time-power curve based on all humidity values in the sample running humidity information to obtain the thirdly marked sample time-power curve;

[0090] S006e, returning to the selected sample running data until each sample running data is selected once.

[0091] Specifically, marking the secondly marked sample time-power curve based on all humidity values in the sample running humidity information to obtain the thirdly marked sample time-power curve includes:

[0092] selecting a humidity value;

[0093] analyzing the humidity value to obtain all time intervals corresponding to the humidity value;

[0094] marking the curve segment corresponding to all time intervals in the sample time-power curve as the humidity value;

[0095] returning to the selected humidity value until each humidity value is selected once.

[0096] In the embodiment, the time intervals of different humidity values are obtained, and the corresponding time intervals are mapped to the sample time-power curve of the corresponding time, and then the multiple sample time-power curves of the humidity values are obtained, so as to distinguish different humidity values to further increase the accuracy of the prediction of the discharge amount.

[0097] In an embodiment of the present application, the sample running humidity information based on each sample data marks the secondly marked sample time-power curve of each sample data to obtain the thirdly marked sample time-power curve of each sample data, and then further includes the following S006f to S006l:

[0098] S006f, selecting a humidity value.

[0099] S006g, analyzing the humidity value to obtain at least one thirdly marked sample time-power curve in the humidity value.

[0100] S006h, selecting a thirdly marked sample time-power curve.

[0101] S006i, analyzing the thirdly marked sample time-power curve to obtain the actual power supply value in the thirdly marked sample time-power curve.

[0102] S006j, returning to the selected thirdly marked sample time-power curve until each thirdly marked sample time-power curve is selected once to obtain multiple actual power supply values.

[0103] S006k, taking the maximum actual power supply value in the obtained multiple actual power supply values as the third prediction range maximum boundary value, and taking the minimum actual power supply value in the obtained multiple actual power supply values as the third prediction range minimum boundary value to obtain the third prediction range.

[0104] S006l, returning to select a humidity value until each humidity value is selected once.

[0105] Specifically, the actual power supply value in the secondly marked sample time-power curve refers to at least one unit time power value contained in the multiple curves of a humidity value in the secondly marked sample time-power curve, and each curve contains at least one unit time.

[0106] In the embodiment, the actual power supply amount of the same humidity value in a unit time is counted, and then the range of the actual power supply amount value in a unit time under the condition of the humidity value is obtained, which is used as the bottom logic to calculate the range of the actual power supply amount value of different humidity values, and then the actual power supply amount value in a unit time corresponding to different humidity values is obtained.

[0107] In an embodiment of the present application, the sample running humidity information based on each sample data marks the sample time-power curve after the second marking of each sample data, and obtains the sample time-power curve after the third marking of each sample data, and then further comprises the following S016 to S116:

[0108] S016, judges whether the first prediction range and the second prediction range have intersection.

[0109] S026, if the first prediction range and the second prediction range have intersection, judges whether the intersection of the first prediction range and the second prediction range has intersection with the third prediction range.

[0110] S036, if the intersection of the first prediction range and the second prediction range has intersection with the third prediction range, calculates the average value of the intersection of the first prediction range and the second prediction range and the intersection of the third prediction range, and takes the obtained average value as the target demand power.

[0111] S046, if the intersection of the first prediction range and the second prediction range has no intersection with the third prediction range, calculates the average value between the average value of the intersection of the first prediction range and the second prediction range and the average value of the third prediction range, and takes the obtained average value as the target demand power.

[0112] S056, if the first prediction range and the second prediction range have no intersection, judges whether the first prediction range and the third prediction range have intersection.

[0113] S066, if the first prediction range and the third prediction range have intersection, judges whether the second prediction range and the third prediction range have intersection.

[0114] S076, if the second prediction range and the third prediction range have intersection, calculates the average value of the average value of the intersection of the first prediction range and the third prediction range and the average value of the intersection of the second prediction range and the third prediction range, and takes the obtained average value as the target demand power.

[0115] S086, if the second prediction range and the third prediction range have no intersection, calculates the average value of the average value of the intersection of the first prediction range and the third prediction range and the average value of the third prediction range, and takes the obtained average value as the target demand power.

[0116] S096, if the first prediction range and the third prediction range do not have intersection, then judging whether the second prediction range and the third prediction range have intersection.

[0117] S106, if the second prediction range and the third prediction range have intersection, then calculating the average value of the first prediction range and the average value of the intersection of the second prediction range and the third prediction range, and taking the obtained average value as the target demand power.

[0118] S116, if the second prediction range and the third prediction range do not have intersection, then calculating the average value of the first prediction range, the average value of the second prediction range and the average value of the third prediction range, and taking the obtained average value as the target demand power.

[0119] Specifically, in addition to the existing sample running time information, sample running power information, sample running weather information, sample running temperature information, sample running humidity information and other conditions, multiple limiting conditions such as power consumption peak period and power consumption valley period can also be added.

[0120] In the embodiment, by establishing the multiple power consumption ranges under the sample running time information, sample running power information, sample running weather information, sample running temperature information, sample running humidity information, and by narrowing the different conditions, the corresponding matching power consumption range is obtained, and the intersection calculation of the power consumption range and the intersection calculation between the intersections of the power consumption range are further carried out to calculate the target demand power under the narrowing of different conditions.

[0121] In an embodiment of the present application, the sample running humidity information based on each sample data marks the secondly marked sample time-power curve of each sample data to obtain the thirdly marked sample time-power curve of each sample data, and then further includes the following S126 to SS146:

[0122] S126, obtaining the target demand power.

[0123] S136, calculating the sum of the target demand power and the target demand power multiplied by N%, and defining the obtained sum as the compensated target demand power.

[0124] S146, outputting the compensated target demand power.

[0125] Specifically, the sum of the target demand power and the target demand power multiplied by N% is calculated, and the obtained sum is defined as the compensated target demand power. The compensated target demand power refers to the target demand power after a certain amount of power is added or subtracted. For example, if the target demand power is in the peak period of power consumption, a certain amount of power is added to the target demand power; if the target demand power is in the valley period of power consumption, a certain amount of power is subtracted from the target demand power; and further, the direction and amount of compensation are adjusted in different seasons.

[0126] In the embodiment, by adding or subtracting a certain percentage of the target demand power from the target demand power, the compensated target demand power is obtained, thereby increasing the fault tolerance of the emergency.

[0127] The technical features of the above embodiments can be combined in any manner, and the execution order of the method steps is not limited. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not contradict, they should be considered as the scope of the present application.

[0128] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for load forecasting of an energy storage system, characterized in that, The energy storage system load forecasting method includes: Acquire sample data for multiple samples, including sample execution data; The sample operation data is analyzed to obtain sample operation time information, sample operation power information, sample operation weather information, sample operation temperature information, and sample operation humidity information. Based on the sample running time information and sample running power information of each sample data, a sample time-power curve is generated for each sample data. Based on the sample operation weather information of each sample data, the sample time-power curve of each sample data is marked to obtain the sample time-power curve of each sample data after the first marking. Based on the sample operating temperature information of each sample data, the sample time-power curve of each sample data after the first labeling is labeled to obtain the sample time-power curve of each sample data after the second labeling. Based on the sample operating humidity information of each sample data, the sample time-power curve of each sample data after the second labeling is labeled to obtain the sample time-power curve of each sample data after the third labeling. A power demand prediction model is constructed based on the running time information, running power information, running weather information, running temperature information, running humidity information, the time-power curve of each sample after the first labeling, the time-power curve of each sample after the second labeling, and the time-power curve of each sample after the third labeling. The first, second, and third prediction ranges of power supply are obtained based on the time-power curves of each marked sample. The target power demand is then calculated based on the intersection of the three prediction ranges. Obtain a sample of data to be tested; The sample data to be tested is analyzed to obtain the sample running time information, sample running power information, sample running weather information, sample running temperature information, and sample running humidity information. The obtained data on the running time, power, weather, temperature, and humidity of the sample to be tested are input into the power demand prediction model. The power demand prediction model is then started to obtain the predicted power demand output by the power demand prediction model.

2. The energy storage system load forecasting method according to claim 1, characterized in that, The process of generating a sample time-power curve for each sample data point based on its sample runtime and sample power information includes: Select a sample of running data; Obtain the sample runtime information and sample power information from the sample's runtime data; Create a two-dimensional coordinate system; Based on the sample running time information, the sample running power information is incorporated into a two-dimensional coordinate system to obtain the sample time-power curve; Return to the selected sample data, and continue until each sample data has been selected once.

3. The energy storage system load forecasting method according to claim 2, characterized in that, The sample time-power curve of each sample data is marked based on the sample operation weather information of each sample data, resulting in the first marked sample time-power curve of each sample data, including: Select a sample of running data; Obtain the sample time-power curve from the sample's runtime data; Obtain the sample operation weather information from the sample operation data. The sample operation weather information includes multiple weather types, including sunny, cloudy, overcast, light rain, moderate rain, and heavy rain. The sample time-power curve is marked based on each weather type and the time period corresponding to each weather type, resulting in the sample time-power curve after the first marking. Return to the selected sample data, and continue until each sample data has been selected once.

4. The energy storage system load forecasting method according to claim 3, characterized in that, The sample time-power curve of each sample data point is marked based on the sample operation weather information of each sample data point, resulting in the first marked sample time-power curve of each sample data point. This is followed by: Select a weather type; Analyze the weather type to obtain at least one sample time-power curve after the first labeling for that weather type; Select a time-power curve for a sample after the first labeling; Analyze the time-power curve of the sample after the first marking to obtain the actual power supply value in the time-power curve of the sample after the first marking. Return to the selected sample time-power curve after the first marking, until each sample time-power curve after the first marking has been selected once, to obtain multiple actual power supply values; The largest actual power supply value among the multiple actual power supply values ​​is taken as the maximum boundary value of the first prediction range; the smallest actual power supply value among the multiple actual power supply values ​​is taken as the minimum boundary value of the first prediction range, thus obtaining the first prediction range. Return to the selected weather type, until each weather type has been selected at least once.

5. The energy storage system load forecasting method according to claim 4, characterized in that, The process of marking the sample time-power curve of each sample data after the first marking based on the sample operating temperature information of each sample data to obtain the sample time-power curve of each sample data after the second marking includes: Select a sample of running data; Obtain the time-power curve of the sample after the first labeling in the sample's runtime data; Obtain the sample running temperature information from the sample running data, wherein the sample running temperature information includes at least one temperature value; Based on all temperature values ​​in the sample operating temperature information, the sample time-power curve after the first labeling is labeled to obtain the sample time-power curve after the second labeling. Return to the selected sample data, and continue until each sample data has been selected once.

6. The load forecasting method for energy storage systems according to claim 5, characterized in that, The sample time-power curve of each sample data point after the first labeling is marked based on the sample operating temperature information of each sample data point, resulting in the sample time-power curve of each sample data point after the second labeling. This process further includes: Select a temperature value; Analyze the temperature value to obtain at least one sample time-power curve after the second labeling of the temperature value; Select a time-power curve for a sample after the second labeling; Analyze the sample time-power curve after the second marking to obtain the actual power supply value in the sample time-power curve after the second marking. Return to the selected sample time-power curve after the second mark, until each sample time-power curve after the second mark has been selected once, to obtain multiple actual power supply values; The largest actual power supply value among the multiple actual power supply values ​​is taken as the maximum boundary value of the second prediction range; the smallest actual power supply value among the multiple actual power supply values ​​is taken as the minimum boundary value of the second prediction range, thus obtaining the second prediction range. Return to the selected temperature value, and continue until each temperature value has been selected once.

7. The energy storage system load forecasting method according to claim 6, characterized in that, The sample time-power curve of each sample data after the second labeling is marked based on the sample operating humidity information of each sample data, resulting in the sample time-power curve of each sample data after the third labeling, including: Select a sample of running data; Obtain the time-power curve of the sample after the second labeling in the sample's runtime data; Obtain the sample operation humidity information from the sample operation data, wherein the sample operation humidity information includes at least one humidity value; Based on all humidity values ​​in the sample operation humidity information, the sample time-power curve after the second labeling is labeled to obtain the sample time-power curve after the third labeling; Return to the selected sample data, and continue until each sample data has been selected once.

8. The energy storage system load forecasting method according to claim 7, characterized in that, The sample time-power curve of each sample data point after the second labeling is marked based on the sample operating humidity information of each sample data point, resulting in the sample time-power curve of each sample data point after the third labeling. This process further includes: Select a humidity value; Analyze the humidity value to obtain at least one sample time-power curve after the third labeling of the humidity value; Select a time-power curve for a sample after the third labeling; Analyze the sample time-power curve after the third mark to obtain the actual power supply value in the sample time-power curve after the third mark; Return to the selected sample time-power curve after the third mark, until each sample time-power curve after the third mark has been selected once, to obtain multiple actual power supply values; The largest actual power supply value among the multiple actual power supply values ​​is taken as the maximum boundary value of the third prediction range; the smallest actual power supply value among the multiple actual power supply values ​​is taken as the minimum boundary value of the third prediction range, thus obtaining the third prediction range. Return to select a humidity value, until each humidity value has been selected once.

9. The energy storage system load forecasting method according to claim 8, characterized in that, The sample time-power curve of each sample data point after the second labeling is marked based on the sample operating humidity information of each sample data point, resulting in the sample time-power curve of each sample data point after the third labeling. This process further includes: Determine whether there is any overlap between the first prediction range and the second prediction range; If the first prediction range intersects with the second prediction range, then determine whether the intersection of the first prediction range and the second prediction range intersects with the third prediction range. If the intersection of the first forecast range and the second forecast range intersects with the third forecast range, then the average value of the intersection of the first forecast range and the second forecast range and the third forecast range is calculated, and the obtained average value is used as the target electricity demand. If the intersection of the first forecast range and the second forecast range does not intersect with the third forecast range, then the average value between the average value of the intersection of the first forecast range and the second forecast range and the average value of the third forecast range is calculated, and the obtained average value is taken as the target electricity demand. If the first prediction range and the second prediction range do not intersect, then determine whether the first prediction range and the third prediction range intersect. If the first prediction range and the third prediction range intersect, then determine whether the second prediction range and the third prediction range intersect. If the second forecast range and the third forecast range intersect, then the average of the average of the intersection of the first forecast range and the third forecast range and the average of the intersection of the second forecast range and the third forecast range is calculated, and the resulting average is taken as the target demand electricity. If the second forecast range and the third forecast range do not intersect, then the average of the intersection of the first forecast range and the third forecast range and the average of the third forecast range are calculated, and the resulting average is taken as the target electricity demand. If the first prediction range and the third prediction range do not intersect, then determine whether the second prediction range and the third prediction range intersect. If the second forecast range and the third forecast range intersect, then calculate the average of the average of the first forecast range and the average of the intersection of the second forecast range and the third forecast range, and use the resulting average as the target electricity demand. If the second and third forecast ranges do not overlap, the average of the first, second, and third forecast ranges is calculated, and the resulting average is taken as the target electricity demand.

10. The energy storage system load forecasting method according to claim 9, characterized in that, The sample time-power curve of each sample data point after the second labeling is marked based on the sample operating humidity information of each sample data point, resulting in the sample time-power curve of each sample data point after the third labeling. This process further includes: Obtain the target power requirement; Calculate the sum of the target electricity demand and the target electricity demand multiplied by N%, and define the sum as the compensated target electricity demand. Output the target power demand after compensation.

Images