An intelligent micro-grid energy management system and method

Abstract

The application discloses an intelligent micro-grid energy management system and method, relates to the technical field of power grid regulation distribution, and solves the technical problem that the single and independent power consumption load prediction method is difficult to guarantee the accuracy of the prediction result in the existing micro-grid management, and further causes the rationality of the micro-grid management to decrease; the correlation of power consumption units is constructed according to power consumption data and a relationship network diagram; time period environment data corresponding to the power consumption units is acquired, an initial time sequence prediction sequence is generated according to the time period environment data, the initial time sequence prediction sequence is input into a power consumption load prediction model, and a preliminary estimated load is obtained; a time sequence prediction sequence for review is generated according to the correlation and the preliminary estimated load, the time sequence prediction sequence is input into the power consumption load prediction model, and a final estimated load is obtained; and finally, power dispatching is carried out according to the estimated power generation and the estimated load; the accuracy of the prediction result is guaranteed, and further the rationality of subsequent power dispatching according to the prediction result is guaranteed.

Description

Technical Field

[0001] This application belongs to the field of power grid regulation and distribution technology, specifically a smart microgrid energy management system and method. Background Technology

[0002] Smart microgrids are a new type of energy supply system that integrates distributed energy, intelligent control, data communication and optimized scheduling technologies. The core of it is to achieve efficient energy production, precise matching and flexible scheduling through "intelligent management". It can operate in an isolated grid mode or in a grid-connected mode that works in coordination with the public power grid. It is a key technology carrier for solving the problem of distributed energy consumption, improving power supply reliability and energy utilization efficiency.

[0003] Existing microgrids typically estimate power generation from generating equipment based on available environmental data, and combine this with historical electricity consumption data and current environmental data to predict user electricity demand. Power dispatch is then implemented based on the estimated power generation and demand to achieve effective microgrid management. However, traditional methods often rely solely on historical electricity consumption data for individual users or units. This singular, independent prediction approach fails to detect deviations in the prediction results for a particular user in a timely manner, thus affecting the rationality of subsequent microgrid management. Therefore, there is an urgent need to develop an intelligent microgrid energy management system and method. Summary of the Invention

[0004] This application provides an intelligent microgrid energy management system, method, and apparatus, which solves the technical problem that the existing microgrid management uses a single, independent power load forecasting method, which makes it difficult to guarantee the accuracy of the forecast results, thus reducing the rationality of microgrid management.

[0005] To achieve the above objectives, this application adopts the following technical solution:

[0006] Firstly, a smart microgrid energy management method is provided, including:

[0007] Obtain wind speed data for power generation equipment; based on the wind speed data of each power generation equipment, estimate the power generation corresponding to the power generation equipment to obtain the estimated power generation.

[0008] Obtain electricity consumption data for each electricity-consuming unit, as well as a network diagram showing the relationships between these units; construct the association relationships between the electricity-consuming units based on the electricity consumption data and the network diagram.

[0009] The process involves acquiring time-period environmental data corresponding to the electricity-consuming unit, generating an initial time-series prediction sequence based on the time-period environmental data, inputting the initial time-series prediction sequence into the electricity load prediction model to obtain a preliminary estimated load, generating a time-series prediction sequence for verification based on the correlation and the preliminary estimated load, inputting the time-series prediction sequence into the electricity load prediction model to obtain the final estimated load, wherein the time-series prediction sequence is a sequence composed of several electricity consumption quantities.

[0010] Power dispatch is based on estimated power generation and estimated load.

[0011] Based on the above technical solution, in the intelligent microgrid energy management system and method provided in this application, the following steps are taken: First, wind speed data of power generation equipment is acquired. Then, the power generation of each power generation equipment is estimated based on its wind speed data to obtain the estimated power generation. Next, power consumption data of each power-consuming unit and a network diagram showing the relationships between them are acquired. Then, the relationships between power-consuming units are constructed based on the power consumption data and the network diagram. Finally, time-series environmental data corresponding to each power-consuming unit is acquired, and an initial time-series prediction sequence is generated based on this data. This initial time-series prediction sequence is input into the power load prediction model to obtain a preliminary estimated load. A time-series prediction sequence for verification is generated based on the relationships and the preliminary estimated load. This time-series prediction sequence is input into the power load prediction model to obtain the final estimated load. Finally, power dispatch is performed based on the estimated power generation and estimated load. By constructing relationships between different power-consuming units and verifying and adjusting the preliminary estimated load based on these relationships, the accuracy of the prediction results is ensured, thereby guaranteeing the rationality of subsequent power dispatch based on the prediction results.

[0012] In conjunction with the first aspect above, in one possible implementation, the step of estimating the power generation corresponding to each power generation device based on wind speed data to obtain the estimated power generation includes:

[0013] Extract wind speed and direction at different altitudes from the wind speed data, and convert the wind speed into effective wind speed based on the wind direction and the orientation of the generator;

[0014] The effective wind speeds at different altitudes are integrated into a three-dimensional wind speed feature vector in ascending order of altitude. The three-dimensional wind speed feature vector is then input into the power generation prediction model to obtain the predicted power generation of the power generation equipment.

[0015] In conjunction with the first aspect above, in one possible implementation, a training method for the power generation prediction model includes:

[0016] Acquire some historical wind speed data and actual power generation, generate a three-dimensional wind speed feature vector based on the wind speed and wind direction at different heights in the historical wind speed data, and integrate the three-dimensional wind speed feature vector and actual power generation into some training data and test data.

[0017] The artificial intelligence model is trained using training data and tested using validation data. The final result is a power generation prediction model with a three-dimensional wind speed feature vector as input and an estimated power generation as output. The estimated power generation is the power generation corresponding to the three-dimensional wind speed feature vector.

[0018] In conjunction with the first aspect above, in one possible implementation, the construction of the association relationship of electricity-consuming units based on electricity consumption data and a relationship network diagram includes:

[0019] Select any electricity-consuming unit as the target electricity-consuming unit in the spatial matrix construction, extract the electricity-consuming units that are connected to the target electricity-consuming unit in the relationship network diagram, and mark them as actual associated electricity-consuming units; the relationship network diagram is the electricity-consuming relationship network between various electricity-consuming units. When there is an electricity-consuming relationship between two electricity-consuming units, there is a connection between the nodes corresponding to the two electricity-consuming units in the relationship network diagram.

[0020] Obtain the electricity consumption data of the target electricity user and the electricity consumption data of each actual associated electricity user. Based on the electricity consumption data of the target electricity user and the actual associated electricity users, determine the actual electricity consumption relationship between the target electricity user and the actual associated electricity users.

[0021] Obtain the electricity consumption data of the target electricity user and the electricity users in the relationship network diagram that are not connected to the target electricity user, and mark them as pseudo-related electricity users;

[0022] Based on the electricity consumption data of the target electricity user and the pseudo-related electricity user, the pseudo-electricity consumption relationship between the target electricity user and the pseudo-related electricity user is determined;

[0023] The target electricity user corresponding to the actual electricity consumption relationship is recorded as the verified electricity user, and the actual associated electricity user is recorded as the verified electricity user.

[0024] The target electricity user corresponding to the pseudo-electricity relationship is recorded as the verified electricity user, and the pseudo-related electricity user is recorded as the verified electricity user.

[0025] The association relationship includes the actual or pseudo-electricity consumption relationship between several verified electricity users and the verified electricity users.

[0026] In conjunction with the first aspect above, in one possible implementation, determining the actual electricity consumption relationship between the target electricity consumer and the actual associated electricity consumer based on their electricity consumption data includes:

[0027] Obtain the electricity consumption of the target electricity user in each time period from the electricity application data, and integrate the electricity consumption into a target electricity consumption matrix according to the chronological order of its corresponding time periods;

[0028] Obtain the electricity consumption of actual associated electricity users in each time period from the electricity application data, and integrate the electricity consumption into an associated electricity consumption matrix according to the chronological order of their corresponding time periods;

[0029] Substituting the target electricity consumption matrix and the associated electricity consumption matrix into the set associated time period difference calculation function, the corresponding associated time period difference is obtained; one expression form of the associated time period difference calculation function includes:

[0030] ;

[0031] in, For the time difference of association; To find the function corresponding to the independent variable that achieves the minimum value of the function, This is the mean function, used to calculate the mean. Let be the target electricity consumption matrix, which is a 1×n matrix; The associated electricity consumption matrix is ​​a 1×n matrix; for offset matrix, It is a positive integer, and , The upper limit of the set offset, It is a positive integer, and In this embodiment, each time period corresponds to one hour, and a day corresponds to 24 hours. It can be set to 24; When =1,

[0032] ;

[0033] A minimum deviation matrix is ​​generated based on the associated time difference and the associated electricity consumption matrix. A correlation coefficient matrix is ​​constructed based on the minimum deviation matrix and the target electricity consumption matrix. The average value of all non-zero elements in the correlation coefficient matrix is ​​calculated as the correlation coefficient between the target electricity consumption unit and the associated electricity consumption unit. The actual electricity consumption relationship includes the associated time difference and the correlation coefficient.

[0034] In conjunction with the first aspect above, in one possible implementation, determining the pseudo-electricity consumption relationship between the target electricity user and the pseudo-related electricity user based on their electricity consumption data includes:

[0035] Obtain the electricity consumption of the target electricity user in each time period from the electricity application data, calculate the difference between the electricity consumption of each time period and the electricity consumption of the previous time period, and record the element corresponding to the time period as 0 when the absolute value of the difference is less than or equal to a set difference threshold; record the element corresponding to the time period as 1 when the difference is greater than the set difference threshold; record the element corresponding to the time period as -1 when the difference is negative and the absolute value of the difference is greater than the set difference threshold; integrate the elements into a target element matrix according to the chronological order of their corresponding time periods;

[0036] Obtain the electricity consumption of each time period in the application electricity data of pseudo-correlated electricity users, calculate the difference between the electricity consumption of each time period and the electricity consumption of the previous time period, and record the element corresponding to the time period as 0 when the absolute value of the difference is less than or equal to a set difference threshold; record the element corresponding to the time period as 1 when the difference is greater than the set difference threshold; record the element corresponding to the time period as -1 when the difference is negative and the absolute value of the difference is greater than the set difference threshold; integrate the elements into a pseudo-correlated element matrix according to the chronological order of their corresponding time periods.

[0037] Substituting the target element matrix and the pseudo-correlated element matrix into a predefined correlation evaluation function yields the correlation degree between them; one expression of the correlation evaluation function includes:

[0038] ;

[0039] in, The degree of correlation between the target element matrix and the pseudo-correlation element matrix; Let be a random variable composed of the elements in the target element matrix; Let be a random variable composed of the elements in the pseudo-correlated element matrix; for One possible value, for One possible value; For random variables and random variables The joint probability distribution, for middle The probability distribution; for middle The probability distribution when This indicates that the target electricity user and the pseudo-related electricity user are independent of each other and have no correlation. The larger the value, the stronger the correlation between the target electricity user and the pseudo-correlated electricity user. ;

[0040] When the correlation degree is less than or equal to the set correlation degree threshold, the pseudo-correlated power consumption unit is recorded as an uncorrelated power consumption unit.

[0041] When the correlation degree is greater than the set correlation degree threshold, the electricity consumption of the target electricity user in each time period of the application electricity data is integrated into a target electricity consumption matrix according to the chronological order of the corresponding time periods; the electricity consumption of the pseudo-correlated electricity user in each time period of the application electricity data is integrated into a pseudo-correlated electricity consumption matrix according to the chronological order of the corresponding time periods; a correlation coefficient matrix is ​​constructed based on the pseudo-correlated electricity consumption matrix and the target electricity consumption matrix; the average value of all non-zero elements in the correlation coefficient matrix is ​​calculated as the correlation coefficient corresponding to the target electricity user and the pseudo-correlated electricity user.

[0042] In conjunction with the first aspect above, in one possible implementation, generating the initial time-series prediction sequence based on time-period environmental data includes:

[0043] Extract time period labels and several time period environmental parameters from the time period environmental data, and generate time period feature sequences based on the time period labels and several time period environmental parameters;

[0044] The system obtains a set sequence length, queries the database based on the time period feature sequence to obtain several corresponding time-series unit data, filters the time-series unit data according to the sequence length, and obtains several time-series unit data whose data storage time is closest to the current time; the number of time-series unit data obtained by the filtering is the same as the sequence length; the data storage time is the time when the time-series unit data is stored in the database; it can be understood that since each time period environmental data and electricity data corresponding to an electricity consumption unit are collected, they are processed and stored in the database, so the order of storage time is equivalent to the order of the corresponding time periods;

[0045] Extract the electricity consumption from each of the time-series data units, and integrate the electricity consumption into a time-series prediction sequence according to the storage time of the corresponding time-series data units.

[0046] In conjunction with the first aspect above, in one possible implementation, generating the time-series prediction sequence for review based on correlation and preliminary estimated load includes:

[0047] Extract the actual and pseudo electricity consumption relationships from the association relationships, as well as their corresponding verified electricity consumption units and the verified electricity consumption units;

[0048] When the correlation is a pseudo-electricity consumption relationship, obtain the correlation coefficient in the pseudo-electricity consumption relationship; obtain the preliminary estimated load corresponding to each verified electricity consumption unit and the verified electricity consumption unit; calculate the product of the preliminary estimated load corresponding to the verified electricity consumption unit and the correlation coefficient, and record it as the pseudo-verified electricity consumption load of the verified electricity consumption unit.

[0049] When the correlation is an actual power consumption relationship, obtain the correlation time difference and correlation coefficient in the actual power consumption relationship; obtain the power consumption of the verified power consumption unit before the time period corresponding to the preliminary estimated load, and the time period separated by the correlation time difference; calculate the product of the power consumption and the correlation coefficient as the actual verified power load of the verified power consumption unit.

[0050] Obtain several pseudo-verification power loads and actual verification power loads corresponding to the power user unit as the verified power user unit;

[0051] Obtain the preliminary estimated load corresponding to the electricity-consuming unit; calculate the deviation based on the preliminary estimated load, the pseudo-verification load, and the actual verification load to obtain the associated verification deviation;

[0052] Obtain the set correlation deviation adjustment threshold, substitute the correlation verification deviation and the correlation deviation adjustment threshold into the set sequence length adjustment function to obtain the adjusted sequence length;

[0053] Obtain the time-period characteristic sequence of the electricity consumption unit, query the database based on the time-period characteristic sequence to obtain several corresponding time-series unit data, filter the time-series unit data according to the sequence length, and obtain several time-series unit data whose data storage time is closest to the current time; the number of time-series unit data obtained by the filtering is the same as the sequence length.

[0054] Extract the electricity consumption from each of the time series data units, and integrate the electricity consumption into a time series prediction sequence for review according to the storage time of the corresponding time series data units.

[0055] In conjunction with the first aspect above, in one possible implementation, the power dispatching based on estimated power generation and estimated load includes:

[0056] Calculate the sum of all estimated power generation to obtain the total estimated power generation; and calculate the sum of the estimated loads of each power-consuming unit to obtain the total estimated load.

[0057] When the estimated total power generation is less than the estimated total load, the estimated total load and the estimated total power generation are calculated as the amount of electricity transferred in, which is the amount of electricity transferred from the main grid to the microgrid.

[0058] When the estimated total power generation is greater than the estimated total load, the estimated total power generation and the estimated total load are calculated as the power to be transferred out, and the power to be transferred in is the power transferred from the microgrid to the main grid.

[0059] Secondly, this application provides an intelligent microgrid energy management system, comprising: a data acquisition module, a data processing module, and a power dispatching module; wherein,

[0060] The data acquisition module is used to acquire wind speed data of power generation equipment, electricity consumption data of each electricity user, a network diagram showing the relationships between each electricity user, and time-period environmental data for each electricity user.

[0061] The data processing module is used to estimate the power generation of each power generation device based on the wind speed data of each power generation device, and obtain the estimated power generation.

[0062] The system constructs the relationships between electricity consumption units based on electricity consumption data and relationship diagrams; generates an initial time-series prediction sequence based on time-period environmental data, inputs the initial time-series prediction sequence into the electricity load prediction model to obtain a preliminary estimated load; and generates a time-series prediction sequence for verification based on the relationships and the preliminary estimated load, inputs the time-series prediction sequence into the electricity load prediction model to obtain the final estimated load.

[0063] The power dispatch module is used to perform power dispatch based on estimated power generation and estimated load.

[0064] This application provides a smart microgrid energy management method and system, which can acquire wind speed data of power generation equipment; estimate the power generation of each power generation equipment based on the wind speed data to obtain the estimated power generation; acquire the power consumption data of each power consumption unit and the relationship network diagram between the power consumption units; construct the correlation relationship of power consumption units based on the power consumption data and the relationship network diagram; acquire the time-period environmental data corresponding to the power consumption units, generate an initial time-series prediction sequence based on the time-period environmental data, input the initial time-series prediction sequence into the power load prediction model to obtain the preliminary estimated load; generate a time-series prediction sequence for verification based on the correlation relationship and the preliminary estimated load, input the time-series prediction sequence into the power load prediction model to obtain the final estimated load; and finally perform power dispatch based on the estimated power generation and estimated load. By constructing the correlation relationship between different power consumption units and verifying and adjusting the preliminary estimated load based on the correlation relationship, the accuracy of the prediction results is ensured, thereby ensuring the rationality of subsequent power dispatch based on the prediction results.

[0065] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a schematic diagram illustrating the steps of the smart microgrid energy management method in this application;

[0068] Figure 2 This is a schematic diagram of the module connections of the smart microgrid energy management system in this application. Detailed Implementation

[0069] The technical solutions of this application will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0070] Please see Figure 1 The first aspect of this application provides a smart microgrid energy management method, including:

[0071] The wind speed data of the power generation equipment is obtained. The wind speed data can be obtained from weather forecasts and includes wind speed and direction at different altitudes within a certain period of time. Alternatively, historical data can be collected from anemometers at different altitudes. The wind speed and direction at different altitudes within a certain period of time can be inferred by training a model based on the historical data. Based on the wind speed data of each power generation equipment, the power generation of the corresponding equipment is estimated to obtain the estimated power generation.

[0072] Obtain electricity consumption data for each electricity-consuming unit, as well as a network diagram showing the relationships between these units; construct the association relationships between the electricity-consuming units based on the electricity consumption data and the network diagram.

[0073] The process involves acquiring time-period environmental data for the electricity-consuming unit, which consists of environmental parameters for the area where the unit is located in the next time period. An initial time-series forecast sequence is generated based on this data and input into the electricity load forecasting model to obtain a preliminary estimated load. A further time-series forecast sequence for verification is generated based on the correlation and the preliminary estimated load and input into the electricity load forecasting model to obtain the final estimated load. The time-series forecast sequence is a sequence composed of several electricity consumption amounts.

[0074] Power dispatch is based on estimated power generation and estimated load.

[0075] Based on the above technical solution, the intelligent microgrid energy management system and method provided in this application can obtain wind speed data of power generation equipment; estimate the power generation of each power generation equipment based on the wind speed data of each power generation equipment to obtain the estimated power generation; obtain the power consumption data of each power consumption unit and the relationship network diagram of each power consumption unit; construct the association relationship of power consumption units based on the power consumption data and the relationship network diagram; obtain the time-period environmental data of the power consumption unit, generate an initial time-series prediction sequence based on the time-period environmental data, input the initial time-series prediction sequence into the power load prediction model to obtain the preliminary estimated load; generate a time-series prediction sequence for verification based on the association relationship and the preliminary estimated load, input the time-series prediction sequence into the power load prediction model to obtain the final estimated load; finally, perform power dispatch based on the estimated power generation and estimated load; construct the association relationship between different power consumption units, and verify and adjust the preliminary estimated load based on the association relationship to ensure the accuracy of the prediction results, thereby ensuring the rationality of subsequent power dispatch based on the prediction results.

[0076] In one possible implementation, the power generation of each power generation device is estimated based on the wind speed data of each device. This includes extracting wind speed and direction at different heights from the wind speed data, and converting the wind speed into an effective wind speed based on the wind direction and the orientation of the generator. Specifically, the wind speed is the average wind speed over a period of time. Based on the relationship between the wind direction and the orientation of the generator, the wind speed is decomposed into wind speeds that are favorable to the direction of generator power generation. These wind speeds are then considered effective wind speeds. The wind speed can be perpendicular to the direction of the blades. The specific process is a relatively existing technology and will not be elaborated upon here.

[0077] The effective wind speeds at different altitudes are integrated into a three-dimensional wind speed feature vector in ascending order of altitude. The three-dimensional wind speed feature vector is then input into the power generation prediction model to obtain the predicted power generation of the power generation equipment.

[0078] In one possible implementation, a training method for the power generation prediction model includes: acquiring historical wind speed data and actual power generation; generating a three-dimensional wind speed feature vector based on wind direction at different heights in the historical wind speed data; specifically, extracting wind speed and wind direction at different heights from the wind speed data; converting the wind speed into effective wind speed based on the wind direction and the orientation of the generator; integrating the effective wind speeds at different heights into a three-dimensional wind speed feature vector in ascending order of height; and integrating the three-dimensional wind speed feature vector and the actual power generation into several training and testing data sets.

[0079] The artificial intelligence model is trained using training data and tested using validation data. The final result is a power generation prediction model with a three-dimensional wind speed feature vector as input and an estimated power generation as output. The estimated power generation is the power generation corresponding to the three-dimensional wind speed feature vector.

[0080] In one possible implementation, the association relationship of electricity-consuming units is constructed based on electricity consumption data and a relationship network diagram, including: selecting any electricity-consuming unit as the target electricity-consuming unit for constructing a spatial matrix, extracting electricity-consuming units connected to the target electricity-consuming unit in the relationship network diagram, and marking them as actual associated electricity-consuming units; the relationship network diagram is a network of electricity consumption relationships between various electricity-consuming units, and when there is an electricity consumption relationship between two electricity-consuming units, there is a connection between the nodes corresponding to the two electricity-consuming units in the relationship network diagram;

[0081] Obtain the electricity consumption data of the target electricity user and the electricity consumption data of each actual associated electricity user. Based on the electricity consumption data of the target electricity user and the actual associated electricity users, determine the actual electricity consumption relationship between the target electricity user and the actual associated electricity users.

[0082] Obtain the electricity consumption data of the target electricity user and the electricity users in the relationship network diagram that are not connected to the target electricity user, and mark them as pseudo-related electricity users;

[0083] Based on the electricity consumption data of the target electricity user and the pseudo-related electricity user, the pseudo-electricity consumption relationship between the target electricity user and the pseudo-related electricity user is determined;

[0084] The target electricity user corresponding to the actual electricity consumption relationship is recorded as the verified electricity user, and the actual associated electricity user is recorded as the verified electricity user.

[0085] The target electricity user corresponding to the pseudo-electricity relationship is recorded as the verified electricity user, and the pseudo-related electricity user is recorded as the verified electricity user.

[0086] The relationships include the actual or pseudo-electricity consumption relationships between several verified electricity users and the verified electricity users.

[0087] In one possible implementation, the actual electricity consumption relationship between the target electricity consumption unit and the actual associated electricity consumption unit is determined based on the electricity consumption data of the target electricity consumption unit and the actual associated electricity consumption unit, including: obtaining the electricity consumption of the target electricity consumption unit in each time period in the electricity consumption data, and integrating the electricity consumption into a target electricity consumption matrix according to the chronological order of its corresponding time periods;

[0088] Obtain the electricity consumption of actual associated electricity users in each time period from the electricity application data, and integrate the electricity consumption into an associated electricity consumption matrix according to the chronological order of their corresponding time periods;

[0089] Substituting the target electricity consumption matrix and the associated electricity consumption matrix into the set associated time period difference calculation function yields the corresponding associated time period difference; one expression form of the associated time period difference calculation function generation function includes:

[0090] ;

[0091] in, For the time difference of association; To find the function corresponding to the independent variable that achieves the minimum value of the function, This is the mean function, used to calculate the mean. Let be the target electricity consumption matrix, which is a 1×n matrix; The associated electricity consumption matrix is ​​a 1×n matrix; for offset matrix, It is a positive integer, and , The upper limit of the set offset, It is a positive integer, and In this embodiment, each time period corresponds to one hour, and a day corresponds to 24 hours. It can be set to 24; When =1,

[0092] ;

[0093] A minimum deviation matrix is ​​generated based on the associated time difference and the associated electricity consumption matrix. A correlation coefficient matrix is ​​constructed based on the minimum deviation matrix and the target electricity consumption matrix. The average value of all non-zero elements in the correlation coefficient matrix is ​​calculated as the correlation coefficient between the target electricity consumption unit and the associated electricity consumption unit. The actual electricity consumption relationship includes the associated time difference and the correlation coefficient.

[0094] Specifically, one method for constructing a deviation matrix based on the associated time period difference and the associated electricity consumption matrix includes:

[0095] ;

[0096] in, It is the minimum deviation matrix;

[0097] One method for generating the correlation coefficient matrix based on the minimum deviation matrix and the target electricity consumption matrix includes: calculating the element at the corresponding position in the correlation coefficient matrix by using the corresponding elements in the minimum deviation matrix and the target electricity consumption matrix;

[0098] ;

[0099] in, This refers to the element in the first row and i-th column of the correlation coefficient matrix; This refers to the element in the first row and i-th column of the minimum deviation matrix; The element corresponding to the first row and i-th column of the target electricity consumption matrix is ​​defined; a correlation coefficient matrix is ​​constructed based on the elements at each position in the correlation coefficient matrix; and the average value of all non-zero elements in the correlation coefficient matrix is ​​calculated as the correlation coefficient between the target electricity consumption unit and the associated electricity consumption unit.

[0100] In one possible implementation, determining the pseudo-electricity consumption relationship between the target electricity user and the pseudo-related electricity user based on electricity consumption data of the target electricity user and the pseudo-related electricity user includes: obtaining the electricity consumption of the target electricity user for each time period in the electricity consumption data; calculating the difference between the electricity consumption of each time period and the electricity consumption of the previous time period; when the absolute value of the difference is less than or equal to a set difference threshold, recording the element corresponding to the time period as 0; when the difference is greater than the set difference threshold, recording the element corresponding to the time period as 1; when the difference is negative and the absolute value of the difference is greater than the set difference threshold, recording the element corresponding to the time period as -1; and integrating the elements into a target element matrix according to the chronological order of their corresponding time periods.

[0101] The method involves obtaining the electricity consumption data of pseudo-correlated electricity users for each time period, calculating the difference between the electricity consumption of each time period and the electricity consumption of the previous time period, and recording the element corresponding to the time period as 0 when the absolute value of the difference is less than or equal to a set difference threshold; recording the element corresponding to the time period as 1 when the difference is greater than the set difference threshold; and recording the element corresponding to the time period as -1 when the difference is negative and the absolute value of the difference is greater than the set difference threshold. The elements are then integrated into a pseudo-correlated element matrix according to the chronological order of their corresponding time periods. It is understood that the specific value of the difference threshold in this embodiment is set by experts based on experience. For example, if a residential building has 80 users, each time period is 1 hour, and each user experiences a fluctuation of 1 kWh per hour, the difference threshold can be set to 80 kWh. The difference threshold can also be other fixed values; this method can filter out small fluctuations in electricity consumption.

[0102] Substituting the target element matrix and the pseudo-correlated element matrix into a predefined correlation evaluation function yields the correlation degree between them; one expression of the correlation evaluation function includes:

[0103] ;

[0104] in, The degree of correlation between the target element matrix and the pseudo-correlation element matrix; Let be a random variable composed of the elements in the target element matrix; Let be a random variable composed of the elements in the pseudo-correlated element matrix; for One possible value, for One possible value; For random variables and random variables The joint probability distribution, for middle The probability distribution; for middle The probability distribution when This indicates that the target electricity user and the pseudo-related electricity user are independent of each other and have no correlation. The larger the value, the stronger the correlation between the target electricity user and the pseudo-correlated electricity user. ;

[0105] When the correlation degree is less than or equal to the set correlation degree threshold, the pseudo-correlated power consumption unit is recorded as an uncorrelated power consumption unit.

[0106] When the correlation degree is greater than the set correlation degree threshold, the electricity consumption of the target electricity user in each time period of the application electricity data is integrated into a target electricity consumption matrix according to the chronological order of the corresponding time periods; the electricity consumption of the pseudo-correlated electricity user in each time period of the application electricity data is integrated into a pseudo-correlated electricity consumption matrix according to the chronological order of the corresponding time periods; a correlation coefficient matrix is ​​constructed based on the pseudo-correlated electricity consumption matrix and the target electricity consumption matrix; the average value of all non-zero elements in the correlation coefficient matrix is ​​calculated as the correlation coefficient corresponding to the target electricity user and the pseudo-correlated electricity user.

[0107] Specifically, one method for constructing a correlation coefficient matrix based on the correlation electricity consumption matrix and the target electricity consumption matrix includes: calculating the element at the corresponding position in the correlation matrix from the corresponding elements in the correlation electricity consumption matrix and the target electricity consumption matrix;

[0108] ;

[0109] in, This refers to the element in the first row and i-th column of the correlation coefficient matrix; The element corresponding to the first row and i-th column in the pseudo-correlation electricity consumption matrix is ​​defined; a correlation coefficient matrix is ​​constructed based on the elements at each position in the correlation coefficient matrix; and the average value of all non-zero elements in the correlation coefficient matrix is ​​calculated as the correlation coefficient between the target electricity consumption unit and the pseudo-correlation electricity consumption unit.

[0110] This embodiment calculates the synchronous changes in the electricity consumption between the target electricity-consuming unit and pseudo-related electricity-consuming units. It can identify electricity-consuming units that are not actually related but whose electricity consumption changes synchronously due to external environmental influences, such as two different factories or different residential buildings. If the temperature rises and both turn on their air conditioners, the electricity consumption of the two factories or the two residential buildings will increase synchronously. If this situation occurs frequently in historical data, it indicates that there is a hidden relationship, i.e., a pseudo-related relationship, between the two electricity-consuming units. This embodiment finds this pseudo-related relationship and corrects the predicted electricity consumption based on it, thus ensuring the accuracy of electricity consumption prediction.

[0111] This embodiment analyzes the actual and pseudo-electricity consumption relationships between different electricity users, enabling the verification of electricity load using the interrelationships between electricity users. It determines whether the relationship between the estimated loads corresponding to each electricity user is the actual or pseudo-electricity consumption relationship, thereby judging the degree of deviation in the estimated load. If the deviation is large, the estimated load needs to be re-estimated to ensure the accuracy of the estimated load.

[0112] In one possible implementation, generating an initial time-series prediction sequence based on time-period environmental data includes: extracting time-period labels and several time-period environmental parameters from the time-period environmental data; generating a time-period feature sequence based on the time-period labels and several time-period environmental parameters; specifically, extracting the parameter values ​​corresponding to each time-period environmental parameter, querying the number of the range corresponding to the parameter value according to the parameter value, and integrating the number of the range corresponding to each time-period environmental parameter into a time-period feature sequence according to a set order. It is worth noting that in this embodiment, the range of each time-period environmental parameter is divided in a relatively detailed manner, which facilitates the subsequent final power load estimation and ensures the estimation accuracy; that is, the length of each range segment is very short. For example, when the time-period environmental parameter is temperature, a range is set every 0.5 degrees, i.e., 0-0.5 degrees is one range, 0.5 degrees to 1 degree is one range; the corresponding numbers can be numbered in ascending order of temperature. It should be noted that different ranges correspond to different numbers, i.e., the numbers are unique.

[0113] The system obtains a set sequence length, queries the database based on the time period feature sequence to obtain several corresponding time-series unit data, filters the time-series unit data according to the sequence length, and obtains several time-series unit data whose data storage time is closest to the current time; the number of time-series unit data obtained by the filtering is the same as the sequence length; the data storage time is the time when the time-series unit data is stored in the database; it can be understood that since each time period environmental data and electricity data corresponding to an electricity consumption unit are collected, they are processed and stored in the database, so the order of storage time is equivalent to the order of the corresponding time periods;

[0114] Extract the electricity consumption from each of the time-series data units, and integrate the electricity consumption into a time-series prediction sequence according to the storage time of the corresponding time-series data units.

[0115] The database is constructed by: acquiring time period labels and time period environmental parameters from historical time period environmental data corresponding to several time periods, and extracting electricity consumption from historical electricity consumption data corresponding to several time periods; the time period labels include "0" and "1", when the time period label is 0, it indicates that the time period belongs to the rest period, such as the time periods corresponding to holidays; when the time period label is 1, it indicates that the time period belongs to the working period, such as the working hours of a weekday; the time period environmental parameters are environmental factors such as temperature, humidity and light intensity of the area corresponding to the electricity consumption unit collected for the corresponding time period; generating time period feature sequences based on time period labels and environmental parameters of each time period; integrating the time period feature sequences and the corresponding electricity consumption into time series unit data and storing it in the database.

[0116] The electricity load prediction model is built based on an LSTM model. In this embodiment, the LSTM model structure includes: an input layer configuration where the input is set to the number of sequence features and the sequence itself. This means that when predicting electricity load, the input sequence length and the time-series prediction sequence are required, allowing the electricity load prediction model to support variable-length inputs. The hidden layer design includes: a first LSTM layer with 64-128 neurons, enabling the return sequence to capture long-term temporal dependencies, such as the delayed response logic of associated units; a Dropout layer with a ratio of 0.2-0.3 to prevent overfitting and avoid excessive reliance on a single associated feature; a second LSTM layer with 32-64 neurons, disabling the return sequence and integrating the temporal features extracted from the first layer; and a fully connected layer with 16-32 neurons, using ReLU activation to enhance the non-linear mapping capability of the features.

[0117] The output layer is designed with one neuron, no activation function, and performs a regression task, directly outputting the predicted electricity load of the target unit.

[0118] One training method for an electricity load forecasting model includes: grouping time-series unit data with the same time-period feature sequences in the database into the same data group, and sorting the time-series data in the data group according to the order of their corresponding time periods; selecting any number of time-series unit data in the data group; integrating the electricity consumption according to the time-series unit data in the order of their corresponding time periods into a data sequence for training and testing; obtaining the time period corresponding to the last electricity consumption in the data sequence, querying the time-period unit data corresponding to the next time period in the corresponding data group, recording the electricity consumption in the time-period unit data as the predicted load, recording the data sequence as the time-series prediction sequence, integrating several time-series prediction sequences and predicted loads into several training data and testing data, and using the training data and testing data to train the LSTM model to obtain the final electricity load forecasting model.

[0119] In one possible implementation, a time-series prediction sequence for verification is generated based on the correlation and preliminary estimated load, including: extracting the actual power consumption relationship and pseudo power consumption relationship in the correlation, as well as their corresponding verification power consumption unit and the verified power consumption unit;

[0120] When the correlation is a pseudo-electricity consumption relationship, obtain the correlation coefficient in the pseudo-electricity consumption relationship; obtain the preliminary estimated load corresponding to each verified electricity consumption unit and the verified electricity consumption unit; calculate the product of the preliminary estimated load corresponding to the verified electricity consumption unit and the correlation coefficient, and record it as the pseudo-verified electricity consumption load of the verified electricity consumption unit.

[0121] When the correlation is an actual power consumption relationship, obtain the correlation time difference and correlation coefficient in the actual power consumption relationship; obtain the power consumption of the verified power consumption unit before the time period corresponding to the preliminary estimated load, and the time period separated by the correlation time difference; calculate the product of the power consumption and the correlation coefficient as the actual verified power load of the verified power consumption unit.

[0122] Obtain several pseudo-verification power loads and actual verification power loads corresponding to the power user unit as the verified power user unit;

[0123] Obtain the preliminary estimated load corresponding to the electricity user; calculate the deviation based on the preliminary estimated load, the pseudo-verification load, and the actual verification load to obtain the associated verification deviation; specifically, the preliminary estimated load, the pseudo-verification load, and the actual verification load can be substituted into a set deviation calculation function; one expression of the deviation calculation function includes:

[0124] ;

[0125] in, To correlate with the verification bias, Number The corresponding actual verification power load, It is a positive integer, and ; When the relationship is an actual electricity consumption relationship, the total number of verified electricity consumption units corresponding to the electricity consumption unit when the electricity consumption unit is the verified electricity consumption unit; Number The corresponding pseudo-verification power load, It is a positive integer, and ; When the relationship is a pseudo-electricity consumption relationship, the total number of verified electricity consumption units corresponding to the electricity consumption unit when the electricity consumption unit is the verified electricity consumption unit; This is the preliminary estimated load corresponding to the electricity-consuming unit; The weighting coefficients corresponding to the deviations in the actual correlation relationships. The weighting coefficients corresponding to the spurious correlation Guaci bias are, and The specific numerical values ​​are set by experts based on experience. It can be understood that in this embodiment, the weight corresponding to the actual verification average power load is greater than the weight corresponding to the pseudo average power load. Specifically, the weight corresponding to the actual verification average power load is set to 0.75, and the weight corresponding to the pseudo average power load is set to 0.25.

[0126] Obtain the set correlation deviation adjustment threshold, substitute the correlation verification deviation and the correlation deviation adjustment threshold into the set sequence length adjustment function to obtain the adjusted sequence length; one expression of the sequence length adjustment function includes:

[0127] ;

[0128] in, The adjusted sequence length, To set the base sequence length, it can be the sequence length used when the first electricity load forecast was performed, or it can be longer than the sequence length used when the first electricity load forecast was performed. To correlate with the verification bias, The correlation deviation adjustment threshold is set to 25 degrees in this embodiment. The set adjustment ratio coefficient, with a specific value determined based on expert experience, is used to control the adjustment range of the sequence length. In this embodiment The value is set to 0.2; it is understandable that the above only provides one possible way to adjust the sequence length.

[0129] Obtain the time-period characteristic sequence of the electricity consumption unit, query the database based on the time-period characteristic sequence to obtain several corresponding time-series unit data, filter the time-series unit data according to the sequence length, and obtain several time-series unit data whose data storage time is closest to the current time; the number of time-series unit data obtained by the filtering is the same as the sequence length.

[0130] Extract the electricity consumption from each of the time series data units, and integrate the electricity consumption into a time series prediction sequence for review according to the storage time of the corresponding time series data units.

[0131] In another embodiment, lateral data augmentation is also performed by associating validation bias and associativity bias adjustment thresholds, including:

[0132] Substituting the correlation verification deviation and the correlation deviation adjustment threshold into the set lateral expansion adjustment function, the lateral adjustment magnitude corresponding to each environmental parameter is obtained; one expression of the lateral expansion adjustment function includes:

[0133] ;

[0134] in, For the number The horizontal adjustment range corresponding to the environmental parameters during the time period. To correlate with the verification bias, The correlation deviation adjustment threshold is set to 25 degrees in this embodiment. The set adjustment ratio coefficient, with a specific value determined based on expert experience, is used to control the adjustment range of the horizontal adjustment. In this embodiment Set to 0.1; it is understandable that the above only provides one possible way to achieve the horizontal adjustment range;

[0135] Several time-period environmental parameters are extracted from the time-period environmental data. Based on the corresponding horizontal adjustment range, the parameter values ​​are adjusted positively and negatively. Specifically, when the horizontal adjustment range is 5, intervals of 1x, 2x, 3x, 4x, and 5x are set respectively. For example, if the time-period environmental parameter is temperature, a range is set every 0.5 degrees, with the corresponding interval being 0.5 degrees. Positive adjustment involves adding the parameter value to the interval by 1x, 2x, 3x, 4x, and 5x respectively, while negative adjustment involves subtracting the interval from the parameter value by 1x, 2x, 3x, 4x, and 5x respectively, resulting in several positive and negative expansion values. The positive and negative expansion values ​​corresponding to each time-period environmental parameter are then obtained sequentially.

[0136] The parameter values, positive expansion values, and negative expansion values ​​corresponding to environmental parameters for each time period are randomly combined. Specifically, the parameter values, positive expansion values, and negative expansion values ​​are randomly selected as the parameter values ​​for the corresponding environmental parameters for each time period. The parameter values ​​of environmental parameters from different time periods are integrated into a single combination, resulting in several combinations of parameter values. Each combination of parameter values ​​contains one parameter value corresponding to the environmental parameter for each time period. The parameter value can be the original parameter value, or a positive expansion value or a negative expansion value. Based on this, different combinations can generate different time period feature sequences when generating time period feature sequences in the subsequent generation of time period feature sequences. Several time period feature sequences are generated based on the time period label and the combination of several time period environmental parameters.

[0137] Several time-period feature sequences of the electricity-consuming unit are obtained. Based on these time-period feature sequences, several corresponding time-series unit data are obtained by querying the database. The time-series unit data is then filtered according to the adjusted sequence length to obtain several time-series unit data whose data storage time is closest to the current time. The number of time-series unit data obtained by the filtering is the same as the sequence length.

[0138] Extract the electricity consumption from each of the time series data units, and integrate the electricity consumption into a time series prediction sequence for review according to the storage time of the corresponding time series data units.

[0139] Simply increasing the length of the time series prediction sequence used for prediction in the time direction may result in the addition of older data to the time series, which may reduce the accuracy of the prediction results. In this case, this embodiment expands the data in the non-time direction by horizontally expanding other data under similar environmental conditions, thus ensuring the validity of the data in the time series prediction sequence and thereby ensuring the accuracy of the prediction results.

[0140] In one possible implementation, power dispatch is based on estimated power generation and estimated load, including: calculating the sum of each estimated power generation to obtain the total estimated power generation; and the sum of the estimated loads corresponding to each power-consuming unit to obtain the total estimated load.

[0141] When the estimated total power generation is less than the estimated total load, the estimated total load and the estimated total power generation are calculated as the amount of electricity transferred in, which is the amount of electricity transferred from the main grid to the microgrid; the transfer process occurs before the predicted time period.

[0142] When the estimated total power generation is greater than the estimated total load, the estimated total power generation and the estimated total load are calculated as the power to be transferred out, and the power to be transferred in is the power transferred from the microgrid to the main grid.

[0143] Please see Figure 2 Secondly, this application provides an intelligent microgrid energy management system, comprising: a data acquisition module, a data processing module, and a power dispatching module; wherein,

[0144] The data acquisition module is used to acquire wind speed data of power generation equipment, electricity consumption data of each electricity user, a network diagram showing the relationships between each electricity user, and time-period environmental data for each electricity user.

[0145] The data processing module is used to estimate the power generation of each power generation device based on the wind speed data of each power generation device, and obtain the estimated power generation.

[0146] The system constructs the relationships between electricity consumption units based on electricity consumption data and relationship diagrams; generates an initial time-series prediction sequence based on time-period environmental data, inputs the initial time-series prediction sequence into the electricity load prediction model to obtain a preliminary estimated load; and generates a time-series prediction sequence for verification based on the relationships and the preliminary estimated load, inputs the time-series prediction sequence into the electricity load prediction model to obtain the final estimated load.

[0147] The power dispatch module is used to perform power dispatch based on estimated power generation and estimated load.

[0148] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0149] How this application works:

[0150] This system can acquire wind speed data from power generation equipment; estimate the power generation of each power generation equipment based on the wind speed data; acquire electricity consumption data for each electricity user and a network diagram showing the relationships between these users; construct the relationships between electricity users based on the electricity consumption data and the network diagram; acquire time-period environmental data for each electricity user and generate an initial time-series prediction sequence based on this data; input this initial time-series prediction sequence into the electricity load prediction model to obtain a preliminary estimated load; generate a time-series prediction sequence for verification based on the relationships and the preliminary estimated load; input this time-series prediction sequence into the electricity load prediction model to obtain the final estimated load; and finally, perform power dispatch based on the estimated power generation and estimated load. By constructing relationships between different electricity users and verifying and adjusting the preliminary estimated load based on these relationships, the system ensures the accuracy of the prediction results and, consequently, the rationality of subsequent power dispatch based on the prediction results.

[0151] The above embodiments are only used to illustrate the technical methods of this application and are not intended to limit it. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of this application without departing from the spirit and scope of the technical methods of this application.

Claims

1. A smart microgrid energy management method, characterized in that, include: Obtain wind speed data from the power generation equipment; Based on the wind speed data of each power generation device, the power generation of the corresponding power generation device is estimated to obtain the estimated power generation. Obtain electricity consumption data for each electricity-consuming unit, as well as a network diagram showing the relationships between these units; construct the association relationships between the electricity-consuming units based on the electricity consumption data and the network diagram. Obtain the time-period environmental data corresponding to the electricity-consuming unit, generate an initial time-series prediction sequence based on the time-period environmental data, input the initial time-series prediction sequence into the electricity load prediction model, and obtain a preliminary estimated load. Based on the correlation and preliminary estimated load, a time-series prediction sequence for verification is generated. The time-series prediction sequence is then input into the electricity load prediction model to obtain the final estimated load. The time-series prediction sequence is a sequence composed of several electricity consumption quantities. Power dispatch is based on estimated power generation and estimated load.

2. The intelligent microgrid energy management method according to claim 1, characterized in that, The construction of the relationship between electricity-consuming units based on electricity consumption data and a relational network diagram includes: Select any electricity-consuming unit as the target electricity-consuming unit in the spatial matrix construction, extract the electricity-consuming units that are connected to the target electricity-consuming unit in the relationship network diagram, and mark them as actual associated electricity-consuming units; Obtain the electricity consumption data of the target electricity user and the electricity consumption data of each actual associated electricity user. Based on the electricity consumption data of the target electricity user and the actual associated electricity users, determine the actual electricity consumption relationship between the target electricity user and the actual associated electricity users. Obtain the electricity consumption data of the target electricity user and the electricity users in the relationship network diagram that are not connected to the target electricity user, and mark them as pseudo-related electricity users; Based on the electricity consumption data of the target electricity user and the pseudo-related electricity user, the pseudo-electricity consumption relationship between the target electricity user and the pseudo-related electricity user is determined; The target electricity user corresponding to the actual electricity consumption relationship is recorded as the verified electricity user, and the actual associated electricity user is recorded as the verified electricity user. The target electricity user corresponding to the pseudo-electricity relationship is recorded as the verified electricity user, and the pseudo-related electricity user is recorded as the verified electricity user. The association relationship includes the actual or pseudo-electricity consumption relationship between several verified electricity users and the verified electricity users.

3. The intelligent microgrid energy management method according to claim 2, characterized in that, The step of determining the actual electricity consumption relationship between the target electricity user and the actual associated electricity user based on their electricity consumption data includes: Obtain the electricity consumption of the target electricity user in each time period from the electricity application data, and integrate the electricity consumption into a target electricity consumption matrix according to the chronological order of its corresponding time periods; Obtain the electricity consumption of actual associated electricity users in each time period from the electricity application data, and integrate the electricity consumption into an associated electricity consumption matrix according to the chronological order of their corresponding time periods; Substitute the target electricity consumption matrix and the associated electricity consumption matrix into the set associated time period difference calculation function to obtain the corresponding associated time period difference; A minimum deviation matrix is ​​generated based on the associated time difference and the associated electricity consumption matrix. A correlation coefficient matrix is ​​constructed based on the minimum deviation matrix and the target electricity consumption matrix. The average value of all non-zero elements in the correlation coefficient matrix is ​​calculated as the correlation coefficient between the target electricity consumption unit and the associated electricity consumption unit. The actual electricity consumption relationship includes the associated time difference and the correlation coefficient.

4. The intelligent microgrid energy management method according to claim 2, characterized in that, The process of determining the pseudo-electricity consumption relationship between the target electricity user and the pseudo-related electricity user based on electricity consumption data of the target electricity user and the pseudo-related electricity user includes: Obtain the electricity consumption of the target electricity user in each time period from the electricity application data, calculate the difference between the electricity consumption of each time period and the electricity consumption of the previous time period, and record the element corresponding to the time period as 0 when the absolute value of the difference is less than or equal to a set difference threshold; record the element corresponding to the time period as 1 when the difference is greater than the set difference threshold; record the element corresponding to the time period as -1 when the difference is negative and the absolute value of the difference is greater than the set difference threshold; integrate the elements into a target element matrix according to the chronological order of their corresponding time periods; Obtain the electricity consumption of each time period in the application electricity data of pseudo-correlated electricity users, calculate the difference between the electricity consumption of each time period and the electricity consumption of the previous time period, and record the element corresponding to the time period as 0 when the absolute value of the difference is less than or equal to a set difference threshold; record the element corresponding to the time period as 1 when the difference is greater than the set difference threshold; record the element corresponding to the time period as -1 when the difference is negative and the absolute value of the difference is greater than the set difference threshold; integrate the elements into a pseudo-correlated element matrix according to the chronological order of their corresponding time periods. Substitute the target element matrix and the pseudo-correlated element matrix into the set correlation degree evaluation function to obtain the correlation degree between the target element matrix and the pseudo-correlated element matrix; When the correlation degree is less than or equal to the set correlation degree threshold, the pseudo-correlated power consumption unit is recorded as an uncorrelated power consumption unit. When the correlation degree is greater than the set correlation degree threshold, the electricity consumption of the target electricity user in each time period of the application electricity data is integrated into a target electricity consumption matrix according to the chronological order of the corresponding time periods; the electricity consumption of the pseudo-correlated electricity user in each time period of the application electricity data is integrated into a pseudo-correlated electricity consumption matrix according to the chronological order of the corresponding time periods; a correlation coefficient matrix is ​​constructed based on the pseudo-correlated electricity consumption matrix and the target electricity consumption matrix; the average value of all non-zero elements in the correlation coefficient matrix is ​​calculated as the correlation coefficient corresponding to the target electricity user and the pseudo-correlated electricity user.

5. The intelligent microgrid energy management method according to claim 1, characterized in that, The generation of the initial time-series prediction sequence based on the time-period environmental data includes: Extract time period labels and several time period environmental parameters from the time period environmental data, and generate time period feature sequences based on the time period labels and several time period environmental parameters; The set sequence length is obtained, and several corresponding time-series unit data are obtained by querying the database based on the time period feature sequence. The time-series unit data is then filtered according to the sequence length to obtain several time-series unit data whose data storage time is closest to the current time. The number of time-series unit data obtained by the filtering is the same as the sequence length. Extract the electricity consumption from each of the time-series data units, and integrate the electricity consumption into a time-series prediction sequence according to the storage time of the corresponding time-series data units.

6. The intelligent microgrid energy management method according to claim 1, characterized in that, The generation of time-series forecast sequences for review based on correlation relationships and preliminary estimated loads includes: Extract the actual and pseudo electricity consumption relationships from the association relationships, as well as their corresponding verified electricity consumption units and the verified electricity consumption units; When the correlation is a pseudo-electricity consumption relationship, obtain the correlation coefficient in the pseudo-electricity consumption relationship; obtain the preliminary estimated load corresponding to each verified electricity consumption unit and the verified electricity consumption unit; calculate the product of the preliminary estimated load corresponding to the verified electricity consumption unit and the correlation coefficient, and record it as the pseudo-verified electricity consumption load of the verified electricity consumption unit. When the correlation is an actual power consumption relationship, obtain the correlation time difference and correlation coefficient in the actual power consumption relationship; obtain the power consumption of the verified power consumption unit before the time period corresponding to the preliminary estimated load, and the time period separated by the correlation time difference; calculate the product of the power consumption and the correlation coefficient as the actual verified power load of the verified power consumption unit. Obtain several pseudo-verification power loads and actual verification power loads corresponding to the power user unit as the verified power user unit; Obtain the preliminary estimated load corresponding to the electricity-consuming unit; calculate the deviation based on the preliminary estimated load, the pseudo-verification load, and the actual verification load to obtain the associated verification deviation; Obtain the set correlation deviation adjustment threshold, substitute the correlation verification deviation and the correlation deviation adjustment threshold into the set sequence length adjustment function to obtain the adjusted sequence length; The time-period characteristic sequence of the electricity consumption unit is obtained. Based on the time-period characteristic sequence, several corresponding time-series unit data are obtained by querying the database. The time-series unit data is filtered according to the sequence length to obtain several time-series unit data whose data storage time is closest to the current time. The number of time-series unit data obtained by the filtering is the same as the sequence length. Extract the electricity consumption from each of the time series data units, and integrate the electricity consumption into a time series prediction sequence for review according to the storage time of the corresponding time series data units.

7. The intelligent microgrid energy management method according to claim 1, characterized in that, The process of estimating the power generation of each power generation device based on wind speed data to obtain the estimated power generation includes: Extract wind speed and direction at different altitudes from the wind speed data, and convert the wind speed into effective wind speed based on the wind direction and the orientation of the generator; The effective wind speeds at different altitudes are integrated into a three-dimensional wind speed feature vector in ascending order of altitude. The three-dimensional wind speed feature vector is then input into the power generation prediction model to obtain the predicted power generation of the power generation equipment.

8. The intelligent microgrid energy management method according to claim 7, characterized in that, One training method for the power generation prediction model includes: Acquire some historical wind speed data and actual power generation, generate a three-dimensional wind speed feature vector based on the wind speed and direction at different heights in the historical wind speed data, and integrate the three-dimensional wind speed feature vector and actual power generation into some training data and test data. The artificial intelligence model is trained using training data and tested using validation data. The final result is a power generation prediction model with a three-dimensional wind speed feature vector as input and an estimated power generation as output. The estimated power generation is the power generation corresponding to the three-dimensional wind speed feature vector.

9. The intelligent microgrid energy management method according to claim 1, characterized in that, The power dispatching based on estimated power generation and estimated load includes: Calculate the sum of all estimated power generation to obtain the total estimated power generation; and calculate the sum of the estimated loads of each power-consuming unit to obtain the total estimated load. When the estimated total power generation is less than the estimated total load, the estimated total load and the estimated total power generation are calculated as the amount of electricity transferred in, which is the amount of electricity transferred from the main grid to the microgrid. When the estimated total power generation is greater than the estimated total load, the estimated total power generation and the estimated total load are calculated as the power to be transferred out, and the power to be transferred in is the power transferred from the microgrid to the main grid.

10. An intelligent microgrid energy management system, based on the application of the intelligent microgrid energy management method according to any one of claims 1-9, characterized in that, include: The system comprises a data acquisition module, a data processing module, and a power dispatching module; among which, The data acquisition module is used to acquire wind speed data of power generation equipment, electricity consumption data of each electricity user and a network diagram of relationships between each electricity user; and environmental data for each electricity user during different time periods. The data processing module is used to estimate the power generation of each power generation device based on the wind speed data of each power generation device, and obtain the estimated power generation. The system constructs the relationships between electricity consumption units based on electricity consumption data and relationship diagrams; generates an initial time-series prediction sequence based on time-period environmental data, inputs the initial time-series prediction sequence into the electricity load prediction model to obtain a preliminary estimated load; and generates a time-series prediction sequence for verification based on the relationships and the preliminary estimated load, inputs the time-series prediction sequence into the electricity load prediction model to obtain the final estimated load. The power dispatch module is used to perform power dispatch based on estimated power generation and estimated load.

Images