An energy scheduling system and method

Abstract

The application belongs to the technical field of power dispatching, and discloses an energy dispatching system and method, which comprises the following steps: collecting total power consumption data in a unit time in a collection area; analyzing the reading of each smart meter in a period in the collection area to obtain the basic power consumption of the minimum power consumption unit corresponding to each smart meter, wherein the period comprises m groups of unit time; collecting influence data in the unit time, wherein the influence data comprises timestamp data, environment data, holiday data and seasonal data; preprocessing the timestamp data to eliminate the year, month and date in the timestamp data to obtain preprocessed timestamp data; and preprocessing the environment data to obtain preprocessed environment data; and the robustness and stability of the consumption prediction model are improved, and the prediction accuracy is further improved.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology, and more specifically, to an energy dispatching system and method. Background Technology

[0002] Chinese Patent CN117410981B discloses a method and system for intelligent multi-regional power energy dispatching based on big data. The method includes acquiring historical data of the regional power system, predicting regional electricity consumption, acquiring initial regional electricity consumption prediction data, and obtaining regional electricity consumption prediction data based on the initial regional electricity consumption prediction data, regional construction electricity consumption data, and the regional electricity consumption environmental impact index. This invention improves the accuracy of regional electricity consumption prediction by supplementing and analyzing historical regional electricity consumption data, calculating the impact of weather and temperature on regional electricity consumption, and determining the power supply required for regional planning and construction. Furthermore, it judges the regional power supply status based on regional power supply data and dispatches power energy across multiple regions based on the regional power supply situation and the distance of dispatch routes between regions, thereby improving the efficiency of multi-regional power energy dispatching.

[0003] The existing technology still has the following problems:

[0004] 1. In the process of power energy dispatching, if the prediction of total power consumption in the data collection area is not accurate enough, it will affect the efficiency of power energy dispatching. When predicting the total power consumption data of the data collection area, the existing technology does not consider the impact of the basic power consumption of the smallest power consumption unit in the area on the prediction of the total power consumption of the area, which in turn affects the normal power demand per unit time in the future.

[0005] 2. When obtaining the basic power consumption of the smallest power consumption unit within a period, the situation of long-term power outages in the region during the collection period was not considered, which reduced the accuracy of the basic power consumption. Predicting the total power consumption per unit time in the future based solely on time series data is not accurate enough, thus affecting the accuracy of the prediction of the total regional power consumption.

[0006] In view of this, the present invention proposes an energy dispatching system and method to solve the above problems. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution:

[0008] An energy dispatching method, comprising:

[0009] Acquire the total power consumption data of the data collection area per unit time;

[0010] The readings of each smart meter in the data collection area are analyzed within one cycle to obtain the basic power consumption of the smallest power consumption unit corresponding to each smart meter. The cycle includes m sets of unit time.

[0011] Collect impact data within a unit of time period, including timestamp data, environmental data, holiday data, and seasonal data;

[0012] The timestamp data is preprocessed by removing the year, month, and date from the timestamp data to obtain preprocessed timestamp data;

[0013] Environmental data is preprocessed to obtain preprocessed environmental data;

[0014] Input the total power consumption data, the sum of the basic power consumption of the smallest power consumption unit, the preprocessed timestamp data, the preprocessed environmental data, the holiday data, and the seasonal data into the pre-trained power consumption prediction model, and output the predicted value of the total power consumption data per unit time in the future.

[0015] Obtain the total power supply data of the current unit of time in the data collection area;

[0016] Based on the predicted total power consumption data and the total power supply data of the current unit time in the collection area, additional power data for the future unit time is obtained, and based on the additional power data, it is determined whether to generate a dispatch instruction.

[0017] Based on dispatch instructions and additional power data, power dispatch is carried out for future units of time in the data collection area.

[0018] Furthermore, the total power consumption data is the sum of the first power consumption within the collection area and the second power consumption in the public area;

[0019] The method for collecting the first power consumption is as follows:

[0020] The first power consumption is the sum of all smart meter readings at the end of the unit time minus the sum of all smart meter readings at the end of the previous unit time.

[0021] The second power consumption is the sum of the power consumption of electrical equipment operating in the public area.

[0022] Furthermore, the method for obtaining the basic power consumption of the smallest power consumption unit corresponding to each smart meter includes:

[0023] Step 1: Mark the smallest electricity consumption units within the data collection area according to their geographical location, and mark them as follows: ; This represents the smallest electricity-consuming unit in the nth region, where n is an integer greater than or equal to 1;

[0024] Step 2: Collect the readings of m sets of smart meters for the smallest electricity consumption unit, that is, collect the data once at the end of each unit of time, and mark the smart meter readings in chronological order. , , Until ;

[0025] Step 3: Perform differential calculations on the smart meter readings to obtain the power consumption of the smallest power-consuming unit in group m-1 within a unit of time. The power consumption is the current smart meter reading minus the previous smart meter reading. The power consumption of group m-1 includes... , Until ,in This represents the electricity hour in the first unit of time. This represents the electricity hour in the second unit of time. This represents the electricity hour in the (m-1)th unit of time;

[0026] Step 4: Take the minimum value (excluding 0) of the power consumption in group m-1 as the basic power consumption of the smallest power consumption unit corresponding to the smart meter, and proceed to step 5; if all the power consumption values ​​in group m-1 are 0, then proceed to step 6.

[0027] Step 5: Repeat steps 2-4 according to the smallest power consumption unit marking order in step 1 until the smallest power consumption unit is obtained. The corresponding basic power consumption ends;

[0028] Step 6: Select a historical period before the current period that has the same holiday data and seasonal data as the current period, and whose environmental coefficient is within the preset error range. Use the basic power consumption of the smallest power consumption unit in the historical period as the basic power consumption of the smallest power consumption unit corresponding to the current period.

[0029] Furthermore, the method for generating the environmental coefficient includes:

[0030] Calculate the average temperature and average humidity values ​​for a given period, and then sum the average temperature and average humidity values ​​by weight to obtain the environmental coefficient.

[0031] Furthermore, the timestamp data is the time when the smart meter reading is collected, including the year, month, date, and time; the environmental data includes temperature and humidity values; the holiday data corresponds to whether each timestamp data is on holiday, including yes and no; the seasonal data is the seasonal information corresponding to each timestamp data, including spring, summer, autumn, and winter; and the preprocessed environmental data includes average temperature and average humidity.

[0032] Furthermore, the training method for the consumption prediction model includes:

[0033] The collected historical data is arranged in a sorted manner; the historical data includes total power consumption data of the smallest power consumption unit, the sum of basic power consumption of the smallest power consumption unit, preprocessed timestamp data, preprocessed environmental data, holiday data, and seasonal data;

[0034] The prediction time step, sliding step, and sliding window length are preset; the sliding window method is used to transform historical data into multiple training samples; the historical data is converted into labels corresponding to the historical data according to the sliding step, one training sample corresponds to one label, and they form a set of training data. Multiple sets of training data constitute the training set.

[0035] The training samples are used as input to the consumption prediction model, and the predicted total power consumption data per unit time in the future is used as the output of the consumption prediction model. The actual total power consumption data per unit time in the future is used as the prediction target, and the training objective is to minimize the sum of the prediction accuracies of all training data. The consumption prediction model is trained until the sum of prediction accuracies converges and training stops. The consumption prediction model is a time series prediction model, and the time series prediction model is a SARIMA model.

[0036] Furthermore, the additional power data is obtained by subtracting the total power supply data of the current unit time in the data collection area from the predicted total power consumption data for future units of time, as expressed below:

[0037] ;

[0038] In the formula, Additional power data required per unit of time in the future; This is a predicted value of total electricity consumption per unit time in the future. This is to collect the total power supply data for the current unit of time in the area.

[0039] Furthermore, methods for determining whether to generate a scheduling instruction include:

[0040] If the additional power data is greater than 0, a dispatch instruction is generated; if the additional power data is less than or equal to 0, a dispatch instruction is not generated.

[0041] Furthermore, it also includes:

[0042] Input additional power data, future unit time start timestamps, future unit time end timestamps, environmental data, holiday data, and seasonal data into the strategy prediction model, and output the compensation power supply for each node timestamp in the future unit time.

[0043] An energy dispatching system, implementing the energy dispatching method, includes:

[0044] The first acquisition module is used to acquire the total power consumption data of the collection area per unit time.

[0045] The second acquisition module analyzes the readings of each smart meter within a period in the acquisition area to obtain the basic power consumption of the smallest power consumption unit corresponding to each smart meter. The period includes m sets of unit time.

[0046] The third acquisition module is used to collect impact data within a unit of time. The impact data includes timestamp data, environmental data, holiday data, and seasonal data.

[0047] The first processing module is used to preprocess the timestamp data, removing the year, month and date from the timestamp data to obtain preprocessed timestamp data;

[0048] The second processing module is used to preprocess environmental data to obtain preprocessed environmental data.

[0049] The first prediction module inputs the total power consumption data, the sum of the basic power consumption of the smallest power consumption unit, the preprocessed timestamp data, the preprocessed environmental data, the holiday data, and the seasonal data into the pre-trained power consumption prediction model, and outputs the predicted value of the total power consumption data per unit time in the future.

[0050] The supply acquisition module is used to acquire the total power supply data of the current unit of time in the data collection area;

[0051] The third processing module obtains additional power data for future units of time based on the predicted total power consumption data and the total power supply data of the current unit time in the collection area, and determines whether to generate a dispatch instruction based on the additional power data.

[0052] The power grid dispatching module performs power dispatching for the future unit time of the data collection area based on dispatching instructions and additional power data.

[0053] The technical effects and advantages of the energy dispatching system and method of the present invention are as follows:

[0054] In the process of dispatching power energy, when predicting the total power consumption data of the collection area, the impact of the sum of the basic power consumption of the smallest power consumption unit in the collection area on the prediction of the total power consumption data of the area in the future per unit time is considered, which improves the robustness and stability of the consumption prediction model and further improves the prediction accuracy.

[0055] Meanwhile, when obtaining the basic power consumption of the smallest power consumption unit, considering the case where the collected power consumption of the smallest power consumption unit is all 0, the basic power consumption of the previous period for each smallest power consumption unit with the same holiday data and seasonal data, and with the environmental coefficient within the preset error range, is selected as the basic power consumption of the current smallest power consumption unit, which effectively improves the accuracy of the output of the consumption prediction model.

[0056] Secondly, by inputting total power consumption data, basic total power consumption data of the smallest power consumption unit, preprocessed timestamp data, preprocessed environmental data, holiday data, and seasonal data into the pre-trained consumption prediction model, the comprehensive influence of multiple factors is considered, further improving the accuracy of the total power consumption prediction value.

[0057] Finally, by combining the predicted total power consumption data for future units of time with the current total power supply data for the region for the current units of time, additional power data for future units of time can be obtained. This helps to schedule energy in the region in advance and ensure normal power consumption in the region for future units of time. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of an energy dispatching system according to Embodiment 1 of the present invention;

[0059] Figure 2 This is a schematic diagram of an energy dispatching system according to Embodiment 2 of the present invention;

[0060] Figure 3 This is a schematic diagram of an energy dispatching method according to Embodiment 3 of the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Example 1

[0063] Please see Figure 1 As shown, the energy dispatching system described in this embodiment includes a first acquisition module, a second acquisition module, a third acquisition module, a first processing module, a second processing module, a first forecasting module, a supply acquisition module, a third processing module, and a power grid dispatching module. Each module is connected via wired and / or wireless means.

[0064] The first acquisition module is used to acquire the total power consumption data of the collection area per unit time. The total power consumption data is the sum of the first power consumption in the collection area and the second power consumption in the public area. The collection area includes commercial areas, residential areas and industrial areas. The unit time is 1 hour, 2 hours or other.

[0065] The method for collecting the first power consumption is as follows:

[0066] The first power consumption is the sum of all smart meter readings at the end of the unit time minus the sum of all smart meter readings at the end of the previous unit time.

[0067] The second power consumption is the sum of the power consumption of electrical equipment operating in the public area.

[0068] By analyzing the total power consumption data of the collected area per unit time into the sum of the first power consumption and the second power consumption, differentiating the different sources of the total power consumption data, the predicted value of the first power consumption per unit time of the smallest power-consuming unit can be predicted more accurately. Based on the predicted value of the first power consumption per unit time of the future, the accuracy of predicting the total power consumption data of the future unit time can be further improved. The reason is that the power equipment in public areas, such as street lights and monitoring probes, have fixed usage time and relatively stable energy consumption, and have little impact on the predicted total power consumption data of the area in the future unit time.

[0069] The second acquisition module analyzes the readings of each smart meter within a period in the acquisition area to obtain the basic power consumption of the smallest power consumption unit corresponding to each smart meter. The period includes m sets of unit time.

[0070] Using the sum of basic power consumption of the smallest power consumption unit as one of the input features of the subsequent consumption prediction model reduces the impact of data noise caused by outliers or sudden changes in the consumption prediction model, improves the robustness and stability of the consumption prediction model, and further improves the prediction accuracy.

[0071] Methods for obtaining the basic power consumption of the smallest power consumption unit corresponding to each smart meter include:

[0072] Step 1: Mark the smallest electricity consumption units within the data collection area according to their geographical location, and mark them as follows: ; This represents the smallest electricity-consuming unit in the nth region, where n is an integer greater than or equal to 1. Proceed to step 2.

[0073] Step 2: Collect the readings of m sets of smart meters for the smallest electricity consumption unit, that is, collect the data once at the end of each unit of time, and mark the smart meter readings in chronological order. , , Until To ensure the accuracy of the basic power consumption data of the smallest power consumption unit per unit time, in this embodiment, m is an integer greater than or equal to 100, and step 3 is executed;

[0074] Step 3: Perform differential calculations on the smart meter readings to obtain the power consumption of the smallest power-consuming unit per unit time in group m-1. The power consumption of group m-1 includes... , Until ,in This represents the electricity hour in the first unit of time. This represents the electricity hour in the second unit of time. Represents the power hours of the (m-1)th unit of time; proceed to step 4.

[0075] Step 4: Take the minimum value of the power consumption in group m-1, excluding 0, as the basic power consumption of the smallest power consumption unit corresponding to the smart meter; proceed to step 5; if all the power consumption in group m-1 is 0, proceed to step 6. This indicates that there is a high probability of a long-term power outage in the area. If the total power consumption data per unit time in the future is predicted based on 0, the prediction will be significantly inaccurate, affecting subsequent power dispatch.

[0076] Step 5: Repeat steps 2-4 according to the smallest power consumption unit marking order in step 1 until the smallest power consumption unit is obtained. The corresponding basic power consumption ends;

[0077] Step 6: Select a historical period (reflecting the impact of environmental conditions on power consumption) that has the same holiday and seasonal data as the current period and whose environmental coefficient is within a preset error range. Use the basic power consumption of the smallest power consumption unit in the historical period as the basic power consumption of the smallest power consumption unit in the current period. That is, select the basic power consumption of the previous period for each smallest power consumption unit that has the same holiday and seasonal data and whose environmental coefficient is within a preset error range as the basic power consumption of the current smallest power consumption unit.

[0078] The method for generating the environmental coefficient includes:

[0079] Calculate the average temperature and average humidity values ​​for a given period, and then sum them using a weighted average to obtain the environmental coefficient. The specific method is as follows:

[0080] .

[0081] in, For environmental factors, This refers to the number of temperature or humidity values ​​corresponding to environmental data within a single period. Preset weights for temperature values, The preset weight for humidity values ​​reflects the different impacts of humidity and temperature on power consumption. This preset weight can be flexibly set according to seasonal or regional characteristics; for example, the weight of temperature can be increased in summer. In autumn, the focus is on humidity. To adapt to the needs of different scenarios, This is the p-th temperature value in the set of temperature values. This is the p-th humidity value in the group. Calculate the average value to eliminate the influence of differences in the number of sampling points and ensure the horizontal comparability of environmental coefficients. When the historical period and the current period's seasonal and holiday data are consistent, and... When the value is within the error range, it indicates that the environmental conditions of the two are similar. In this case, the basic power consumption of the lowest historical power consumption unit can be directly reused, avoiding double calculation and improving energy efficiency management.

[0082] It should be noted that when a localized power outage or other power failure occurs in the area, the power consumption data of the smallest power consumption unit per unit time is 0 according to the smart meter readings. However, the actual basic power consumption of the smallest power consumption unit is not 0. By removing the 0 values ​​from the m-1 groups of power consumption, the basic power consumption data of the smallest power consumption unit is obtained, which improves the accuracy of the power consumption data of the smallest power consumption unit and thus helps to improve the accuracy of the output of the subsequent consumption prediction model. However, there may be cases where the power supply has not been restored in the current statistical period, resulting in all m-1 groups of power consumption data being 0 values. Selecting the basic power consumption of each smallest power consumption unit in the previous period, where the holiday data and seasonal data are the same and the environmental coefficient is within the preset error range, as the basic power consumption of the current smallest power consumption unit, also helps to improve the accuracy of the output of the subsequent consumption prediction model.

[0083] The third acquisition module is used to collect impact data within a unit of time. The impact data includes timestamp data, environmental data, holiday data, and seasonal data. The timestamp data is the time when the smart meter reading is collected, including the year, month, date, and time. The environmental data includes temperature and humidity values. The holiday data corresponds to whether each timestamp data is on holiday, and the holiday data includes yes and no. The seasonal data is the seasonal information corresponding to each timestamp data, and the seasonal data includes spring, summer, autumn, and winter.

[0084] Timestamp data, environmental data, holiday data, and seasonal data are all key factors influencing primary electricity consumption. Timestamp data not only marks holiday and seasonal data, but the time in the timestamp data also affects the operating status of different electrical devices. For example, at 8 pm, many electric water heaters may be running, thus increasing the total electricity consumption of the smallest electricity unit in the area, or streetlights in public areas may be on at 8 pm and off at 8 am. Environmental and seasonal data affect the electricity consumption of air conditioning and heating equipment, thus affecting the primary electricity consumption in the data collection area. Holiday data leads to changes in the number of electricity users in the data collection area, which also affects primary electricity consumption. Inputting these influencing data into the consumption prediction model can more accurately predict the total electricity consumption per unit time in the future.

[0085] The timestamp data is obtained through the power system. When collecting smart meter readings, the power system marks the collection time and collects the data once every unit of time.

[0086] Temperature and humidity values ​​are obtained by relevant sensors installed in the data collection area.

[0087] The holiday data and seasonal data are obtained by matching pre-marked timestamp data.

[0088] The first processing module is used to preprocess the timestamp data, removing the year, month and date from the timestamp data to obtain preprocessed timestamp data, which only contains the time.

[0089] Since holiday and seasonal data have already been obtained through timestamp data, preprocessing the timestamp data to retain only the time points simplifies the input to the subsequent consumption prediction model and improves the training efficiency of the consumption prediction model.

[0090] The second processing module is used to preprocess the environmental data to obtain preprocessed environmental data, which includes average temperature and average humidity.

[0091] Similarly, by obtaining the average temperature and average humidity, the input to the subsequent consumption prediction model is simplified, while retaining the characteristics of temperature and humidity, thus improving the training efficiency of the consumption prediction model.

[0092] The first prediction module inputs total power consumption data, the sum of basic power consumption of the smallest power consumption unit, preprocessed timestamp data, preprocessed environmental data, holiday data, and seasonal data into a pre-trained power consumption prediction model, and outputs the predicted value of total power consumption per unit time in the future.

[0093] The training method for the consumption prediction model includes:

[0094] The collected historical data are arranged in a sorted manner; the historical data includes total power consumption data of the smallest power consumption unit, the sum of basic power consumption of the smallest power consumption unit, preprocessed timestamp data, preprocessed environmental data, holiday data, and seasonal data.

[0095] Preset prediction time step Sliding step size Sliding window length .

[0096] For example, =3, in hours; =1, in hours; =24, in hours;

[0097] Using a sliding window method, historical data is transformed into multiple training samples; based on the sliding step size... Historical data is converted into labels corresponding to the historical data. One training sample corresponds to one label and forms a set of training data. Multiple sets of training data constitute the training set.

[0098] The training samples are used as input to the consumption prediction model, and the predicted total power consumption data per unit time in the future is used as the output of the consumption prediction model. The actual total power consumption data per unit time in the future is used as the prediction target, and the training objective is to minimize the sum of the prediction accuracies of all training data. The consumption prediction model is trained until the sum of prediction accuracies converges and training stops. The consumption prediction model is a time series prediction model, which includes the SARIMA model or other suitable models.

[0099] The historical data collection methods include: collecting total power consumption data, the sum of basic power consumption of the smallest power consumption unit, preprocessed timestamp data, preprocessed environmental data, holiday data, and seasonal data; when labeling holiday data, "yes" is recorded as 1 and "no" as 0; when labeling seasonal data, "spring" is recorded as 1, "summer" as 2, "autumn" as 3, and "winter" as 4; and numerical labeling is performed on holiday data and seasonal data.

[0100] The supply acquisition module is used to acquire the total power supply data of the collection area within the current unit time; the total power supply data is the total amount of power supplied to the collection area within the current unit time.

[0101] The total power supply data is obtained from the power supplier in the data collection area, and is collected once per unit time.

[0102] The third processing module obtains the additional power data required for future unit time based on the predicted total power consumption data for future unit time and the total power supply data for the current unit time in the collection area. Based on the additional power data, it determines whether to generate a dispatch instruction. If a dispatch instruction is generated, it sends the dispatch instruction and the additional power data to the power grid dispatch module.

[0103] Methods for obtaining additional power data include:

[0104] ;

[0105] In the formula, Additional power data required per unit of time in the future; This is a predicted value of total electricity consumption per unit time in the future. This is to collect the total power supply data for the current unit of time in the area.

[0106] Methods for determining whether to generate a scheduling instruction include:

[0107] If the additional power data is greater than 0, a dispatch instruction is generated; if the additional power data is less than or equal to 0, a dispatch instruction is not generated.

[0108] The power grid dispatch module, based on dispatch instructions and additional power data, performs power dispatch for the data collection area in the future unit of time to ensure normal power consumption in the data collection area in the future unit of time.

[0109] Example 2

[0110] Please see Figure 2 As shown in the figure, the parts not described in detail in this embodiment are described in Embodiment 1. In order to further improve the grid stability and reduce the dispatching cost during power dispatching, a dispatching strategy module is also included.

[0111] The scheduling strategy module takes additional power data, the start timestamp of the future unit time, the end timestamp of the future unit time, environmental data, holiday data, and seasonal data into the strategy prediction model, and outputs the compensation power supply for each node timestamp in the future unit time.

[0112] The power grid dispatching module performs power dispatching for the data collection area based on the compensation power supply for each node timestamp within a future unit of time.

[0113] It should be noted that the nodes within a unit of time in the future will be designed by the staff according to the actual situation. Node timestamps can be set at 5-minute intervals, but this is not required here.

[0114] The training method for the policy prediction model includes:

[0115] Obtain k sets of scheduling data, where k is a positive integer greater than 1. The scheduling data includes additional power data, the start timestamp of the future unit time, the end timestamp of the future unit time, environmental data, holiday data, seasonal data, and the corresponding compensation power supply for each node timestamp within the future unit time. Use the scheduling data as a sample set, divide the sample set into a training set and a test set, and construct a classifier. Use the additional power data, the start timestamp of the future unit time, the end timestamp of the future unit time, the environmental data, the holiday data, and the seasonal data in the training set as input data, and use the compensation power supply for each node timestamp within the future unit time in the training set as output data to train the classifier and obtain an initial classifier. Use the test set to test the initial classifier, and output a classifier that meets the preset accuracy as the policy prediction model. The classifier is either a Naive Bayes model or a Support Vector Machine model.

[0116] Power dispatching by obtaining the compensation power supply for each node timestamp within a future unit of time not only avoids voltage fluctuations or large frequency deviations caused by arbitrary dispatching, which could lead to grid collapse or damage to grid power equipment and reduce grid operation stability, but also reduces resource waste caused by excess power supply at node timestamps, effectively reducing power dispatching costs.

[0117] Example 3

[0118] Please see Figure 3 An energy dispatching method, comprising:

[0119] Acquire the total power consumption data of the data collection area per unit time;

[0120] The readings of each smart meter in the data collection area are analyzed within one cycle to obtain the basic power consumption of the smallest power consumption unit corresponding to each smart meter. The cycle includes m sets of unit time.

[0121] Collect impact data within a unit of time period, including timestamp data, environmental data, holiday data, and seasonal data;

[0122] The timestamp data is preprocessed by removing the year, month, and date from the timestamp data to obtain preprocessed timestamp data;

[0123] Environmental data is preprocessed to obtain preprocessed environmental data;

[0124] Input the total power consumption data, the sum of the basic power consumption of the smallest power consumption unit, the preprocessed timestamp data, the preprocessed environmental data, the holiday data, and the seasonal data into the pre-trained power consumption prediction model, and output the predicted value of the total power consumption data per unit time in the future.

[0125] Obtain the total power supply data of the current unit of time in the data collection area;

[0126] Based on the predicted total power consumption data and the total power supply data of the current unit time in the collection area, obtain the additional power data required in the future unit time, and determine whether to generate a dispatch instruction based on the additional power data.

[0127] Based on dispatch instructions and additional power data, power dispatch is carried out for future units of time in the data collection area.

[0128] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0129] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An energy dispatching method, characterized in that, The application relates to a method for predicting and scheduling power consumption of a collection area. The method comprises the following steps: acquiring total power consumption data of the collection area in a unit time; analyzing each smart meter reading in a period of the collection area to acquire basic power consumption of each smart meter corresponding to a minimum power consumption unit, wherein the period comprises m groups of unit time; the method for acquiring the basic power consumption of each smart meter corresponding to the minimum power consumption unit comprises the following steps: step 1, marking the minimum power consumption units in the collection area according to geographical positions in sequence; step 2, collecting m groups of smart meter readings of the minimum power consumption units, i.e. collecting once at the end of each unit time; step 3, performing differential operation on the smart meter readings to obtain the power consumption of the minimum power consumption units in m-1 groups of unit time, wherein the power consumption is the current smart meter reading minus the previous smart meter reading; Step 5, in accordance with the minimum power consumption unit marking order of Step 1, Steps 2-4 are repeatedly executed until the minimum power consumption unit is acquired corresponding base power consumption time ends; step 4, taking the minimum value of m-1 groups of power consumption except 0 as the basic power consumption of the minimum power consumption unit corresponding to the smart meter, and executing step 5; if the m-1 groups of power consumption are all 0, executing step 6; step 6, selecting a historical period corresponding to the same holiday data and seasonal data and the same environment coefficient within a preset error range before the current period, and taking the basic power consumption of the minimum power consumption unit of the historical period as the basic power consumption of the minimum power consumption unit corresponding to the current period; acquiring influence data in a unit time, wherein the influence data comprises timestamp data, environment data, holiday data and seasonal data; preprocessing the timestamp data to eliminate the year, month and date in the timestamp data, and obtaining preprocessed timestamp data; preprocessing the environment data to obtain preprocessed environment data; inputting the total power consumption data, the sum of the basic power consumption of the minimum power consumption unit, the preprocessed timestamp data, the preprocessed environment data, the holiday data and the seasonal data into a pre-trained consumption prediction model, and outputting a predicted value of total power consumption data in a future unit time; acquiring total power supply data of the collection area in a current unit time; acquiring extra power data in a future unit time according to the predicted value of the total power consumption data and the total power supply data of the collection area in the current unit time, and judging whether to generate a scheduling instruction according to the extra power data; 2. The energy dispatching method of claim 1, wherein, performing power scheduling on the collection area in a future unit time according to the scheduling instruction and the extra power data. The total power consumption data is the sum of first power consumption in the collection area and second power consumption in a public area; the collection method of the first power consumption is as follows: the sum of all smart meter readings at the end of a unit time minus the sum of all smart meter readings at the end of the previous unit time, i.e. the first power consumption; 3. The energy dispatching method of claim 1, wherein, the second power consumption is the sum of power consumption of power equipment in the public area. The method for generating the environment coefficient comprises the following steps: respectively calculating the average value of temperature and the average value of humidity in a period, and performing weighted summation on the average value of temperature and the average value of humidity to obtain the environment coefficient.

4. The energy dispatching method of claim 3, wherein, The timestamp data is the time when the smart meter reading is collected, including year, month, date and time; the environmental data includes temperature value and humidity value; the holiday data is whether each timestamp data corresponds to a holiday, and the holiday data includes yes and no; the season data is the season information corresponding to each timestamp data, and the season data includes spring, summer, autumn and winter; the preprocessed environmental data includes temperature average value and humidity average value.

5. The energy dispatching method of claim 2, wherein, The training method of the consumption prediction model comprises: arranging the collected historical data; the historical data includes minimum power consumption data of the minimum power consumption unit, sum of the minimum power consumption unit, preprocessed timestamp data, preprocessed environmental data, holiday data and season data; presetting a prediction time step, a sliding step and a sliding window length; using a sliding window method, the historical data is converted into a plurality of training samples; according to the sliding step, the historical data is converted into labels corresponding to the historical data, one training sample corresponds to one label, and a group of training data is formed, and a plurality of groups of training data form a training set; using the training sample as the input of the consumption prediction model, and using the future unit time total power consumption data prediction value as the output of the consumption prediction model, taking the future unit time actual total power consumption data as the prediction target, and minimizing the sum of the prediction accuracy of all training data as the training target, the consumption prediction model is trained until the sum of the prediction accuracy reaches convergence, and the training is stopped; the consumption prediction model is a time series prediction model, and the time series prediction model is a SARIMA model.

6. The energy dispatching method of claim 1, wherein, The additional power data is the future unit time total power consumption data prediction value minus the total power supply data in the collection area in the current unit time.

7. The energy dispatching method of claim 1, wherein, The method for determining whether to generate a scheduling instruction comprises: if the additional power data is greater than 0, a scheduling instruction is generated, and if the additional power data is less than or equal to 0, no scheduling instruction is generated.

8. The energy dispatching method of claim 1, wherein, Further comprising: inputting the additional power data, the future unit time start timestamp, the future unit time end timestamp, the environmental data, the holiday data and the season data into the strategy prediction model, and outputting the compensation power supply of each node timestamp in the future unit time.

9. An energy scheduling system implementing the energy scheduling method of any one of claims 1-8. Comprise: a first acquisition module for acquiring total power consumption data in a unit time in a collection area; a second acquisition module for analyzing each smart meter reading in a period in the collection area to acquire the basic power consumption of each smart meter corresponding to the minimum power consumption unit, and the period comprises m groups of unit time; a third acquisition module for collecting influence data in a unit time, the influence data including timestamp data, environmental data, holiday data and season data; a first processing module for preprocessing the timestamp data, eliminating the year, month and date in the timestamp data, and obtaining preprocessed timestamp data; a second processing module for preprocessing the environmental data and obtaining preprocessed environmental data; The first prediction module inputs the total power consumption data, the sum of the basic power consumption of the minimum power consumption unit, preprocessed timestamp data, preprocessed environment data, holiday data and seasonal data into a pre-trained consumption prediction model, and outputs a predicted value of the total power consumption data in a future unit time; The supply acquisition module is configured to acquire total power supply data in a current unit time in the collection area; The third processing module acquires extra power data in the future unit time according to the predicted value of the total power consumption data and the total power supply data in the current unit time in the collection area, and determines whether to generate a dispatching instruction according to the extra power data; The power grid dispatching module performs power dispatching on the collection area in the future unit time according to the dispatching instruction and the extra power data.

Images