Machine learning-based distributed energy station ai intelligent regulation method and system

Abstract

The application discloses a distributed energy station AI intelligent regulation and control method and system based on machine learning, wherein the method comprises the steps of data acquisition, analysis, calculation, regulation and control, judgment, energy supply and feedback. By establishing a preset time sequence energy scheduling model trained by historical power consumption load data, the application can effectively capture the periodicity and burstiness characteristics of user power consumption behavior, effectively reflect the complementary characteristics between distributed energy stations, alleviate the local supply-demand imbalance in a specific period, timely optimize the scheduling strategy, improve the stability of energy load supply, and thus improve the accuracy of energy regulation and control.

Description

Technical Field

[0001] This invention relates to the technical field of energy regulation, specifically to an AI-based intelligent regulation method and system for distributed energy stations based on machine learning. Background Technology

[0002] After the distributed energy stations are built, it is necessary to schedule the energy supply between the main energy station and each energy station, analyze the usage demand of users under each energy station, and adjust the energy reserves of each distributed energy station in a timely manner to achieve efficient energy allocation and utilization and reduce the probability of energy waste or insufficient energy supply.

[0003] Existing energy dispatch technologies typically require monitoring of various energy-consuming devices used by users, such as heating, cooling, and electrical equipment, to meet users' energy needs and energy quality requirements. This involves collecting user data, analyzing energy dispatch data, and proposing optimization suggestions to obtain reasonable energy allocation and dispatch strategies, responding to economic and carbon emission policy requirements, and improving the efficiency, safety, and reliability of energy use.

[0004] Currently, Chinese patent CN119535986A proposes an intelligent building resource scheduling system and method under multiple weather scenarios. It determines and obtains the total operating data of all equipment in the target area through an energy analysis module, divides the total operating data into nonlinear data and linear data, calculates the proportion of nonlinear data and linear data, and obtains the scheduling allocation result through an energy scheduling module based on the proportion of nonlinear data and linear data. The energy supply module controls the energy supply of solar energy, wind energy, geothermal energy or electric energy to the target area based on the scheduling allocation result.

[0005] Regarding the above technical solutions, this method only divides the scheduling results by the ratio of linear and nonlinear data. The scheduling strategy is fixed, lacks environmental adaptability, makes it difficult to optimize the scheduling strategy in a timely manner, and reduces the accuracy of energy regulation. Summary of the Invention

[0006] The purpose of this invention is to provide an AI-based intelligent control method and system for distributed energy stations based on machine learning, in order to solve the problems mentioned in the background art.

[0007] Firstly, this invention provides an AI-based intelligent control method for distributed energy stations based on machine learning, employing the following technical solution to achieve the invention's objective:

[0008] The AI-based intelligent control method for distributed energy stations, based on machine learning, includes the following steps:

[0009] Data Acquisition: The central energy station connects to N distributed energy stations and their respective M groups of users, acquiring historical electricity load data of users, historical electricity load data of energy stations, and basic energy storage regulation. Net amount of electricity exchanged;

[0010] Analysis: A preset time-series energy dispatch model is established based on the user's historical electricity load, and the predicted electricity load of the m-th group of users at the n-th energy station within time period t is output. ;

[0011] Calculate: The predicted electricity load of the nth energy station during time period t is: ,in The priority weight for the m-th group of users at the n-th energy station;

[0012] Control: The predicted electricity load of the main energy station during time period t is: Where η is the loss coefficient, Let be the energy storage regulation amount of the nth energy station within time period t. The calculation model is as follows: ,in Let be the average net electricity exchange volume of the nth energy station;

[0013] Energy supply: During time period t, according to The energy supply of the central energy station is dispatched according to... Adjust the energy storage regulation amount of the nth energy station according to Allocate the energy supply to the nth energy station.

[0014] By adopting the above technical solutions and establishing a preset time-series energy dispatch model trained on historical electricity load data, the periodic and sudden characteristics of user electricity consumption behavior can be effectively captured. Compared with traditional proportional division methods or prediction methods based on fixed growth rates, machine learning models can also adaptively learn the impact of external factors such as seasons, weather, holidays, and energy prices on electricity load, reducing prediction bias caused by empirical assumptions. By aggregating the load demand at the user level to the distributed energy station level and then further statistically analyzing it to the energy central station level, the computational complexity can be reduced to a certain extent. Differentiated energy allocation is achieved through priority weights, improving the supply stability of key energy loads. The calculation of energy storage regulation incorporates the average of historical net electricity exchange, which can effectively reflect the complementary characteristics between distributed energy stations, alleviate local supply and demand imbalances in specific periods, improve the supply stability of energy loads, and thus improve the accuracy of energy regulation.

[0015] Optionally, the control steps also include... The calculation model was revised, and after revision The computational model is updated as follows: ,in The dynamic adjustment weight for the nth energy station. , The value is determined based on the user's historical electricity load. The static adjustment weight for the nth energy station. .

[0016] By adopting the above technical solution and setting dynamic and static adjustment weights, different energy stations can adjust their energy storage strategies according to their own historical load characteristics. The value of the dynamic adjustment weight is determined based on the user's historical load data, enabling the model to adapt to long-term load change trends. The static adjustment weight, as a fixed coefficient, can reduce the impact of short-term fluctuations on the overall energy dispatch strategy. For example, for energy stations with frequent power exchange, the static adjustment weight is increased to amplify the impact of net power exchange. When extreme weather causes a surge in temporary net power exchange at multiple energy stations, the static adjustment weight can play a constraining role, reducing the violent fluctuations in energy storage regulation, improving the stability of energy load supply, and thus improving the accuracy of energy regulation.

[0017] Optionally, in the data acquisition step, the average user electricity load of the nth energy station is also obtained. ;

[0018] In the adjustment process, according to Dynamic adjustment weights for the nth energy station The value is determined accordingly. The specific calculation model for the value is as follows: .

[0019] By adopting the above technical solution, the weight is dynamically adjusted by quantifying the difference between the user's real-time load and the average value during a specified period. This enables the energy storage regulation to respond to load fluctuations in a timely manner. For example, for energy stations with large load fluctuations, the dynamic adjustment weight can be increased to enhance the elasticity of the basic regulation, thereby reserving more energy storage buffer space, reducing the probability of dispatch failure due to demand surges, improving the stability of energy load supply, and thus improving the accuracy of energy regulation.

[0020] Optionally, a judgment step and a feedback step are also set after the control step;

[0021] Determine: The change in energy storage regulation at the nth energy station during time period t is... The average change in energy storage regulation during time period t is The warning error for time period t is Set the warning value to E. If the regulation is normal, the energy supply procedure will be executed. If the adjustment is abnormal, a feedback step will be executed.

[0022] Feedback: Generate an alert for abnormal energy supply regulation during time period t and upload it to the management terminal.

[0023] By adopting the above technical solution, anomaly monitoring is performed by calculating the warning error of energy storage regulation changes. The warning error is used to measure the dispersion of energy storage regulation changes at each energy station, which can effectively identify abnormal situations where the decision-making behavior of the energy station group deviates. Based on the comparison results of the warning error and the warning value, an anomaly early warning mechanism is established to reduce the probability of the spread of erroneous scheduling strategies, improve the accuracy of inter-site coordination and joint decision-making, improve the overall energy efficiency of distributed energy stations, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0024] Optionally, in the decision step, let For the first period, This is the second time period. Obtain the warning error for the first time period. Warning error in the second period ,according to and The value of the warning value E is determined, and the specific calculation model for the value of E is as follows: .

[0025] By adopting the above technical solution, the warning value is dynamically adjusted based on the warning error in the first and second time periods. This allows the warning value to automatically adjust in accordance with periodic or seasonal load changes. Using the average of the warning errors in similar time periods at certain intervals as the warning value reduces the possibility of sudden changes in the warning value due to abnormal data in a single time period. This reduces reliance on experience-based presets and lowers the probability of repeated manual intervention in adjusting the strategy. For example, the warning value increases during peak electricity consumption seasons to tolerate greater fluctuations in energy storage regulation. The system can adaptively update the warning value, reducing operation and maintenance costs, improving the stability of energy load supply, and thus improving the accuracy of energy regulation.

[0026] Secondly, this invention provides an AI-based intelligent control system for distributed energy stations based on machine learning, which achieves its objectives using the following technical solutions:

[0027] The distributed energy station AI intelligent control system based on machine learning includes the following modules:

[0028] Acquisition Module: The output end connects to the input end of the analysis module. The energy central station connects to N distributed energy stations and their respective M groups of users, and is used to acquire the historical electricity load of users and the historical electricity load of energy stations, as well as the basic energy storage regulation. Net amount of electricity exchanged;

[0029] Analysis module: The input end is connected to the output end of the acquisition module, and the output end is connected to the input end of the calculation module. It is used to establish a preset time-series energy dispatch model based on the user's historical electricity load, and output the predicted electricity load of the m-th group of users at the n-th energy station within time period t. ;

[0030] The calculation module's input terminal connects to the output terminal of the analysis module, and its output terminal connects to the input terminal of the control module. It is used to calculate based on... and the priority weight of the m-th group of users at the n-th energy station Calculate the predicted electricity load of the nth energy station during time period t. ;

[0031] Control module: Its input terminal is connected to the output terminal of the calculation module, and its output terminal is connected to the input terminal of the judgment module. It is used to control the input terminal based on the input terminal of the calculation module. The average net amount of electricity exchanged with the nth energy station Calculate the energy storage regulation of the nth energy station during time period t. and according to , The predicted electricity load of the main energy station during the time period t is calculated using the loss factor η. ;

[0032] Energy supply module: The input terminal is connected to the output terminal of the judgment module, and is used to determine the energy supply within time period t based on... The energy supply of the central energy station is dispatched according to... Adjust the energy storage regulation amount of the nth energy station according to Allocate the energy supply to the nth energy station.

[0033] By adopting the above technical solutions and establishing a preset time-series energy dispatch model trained on historical electricity load data, the periodic and sudden characteristics of user electricity consumption behavior can be effectively captured. Compared with traditional proportional division methods or prediction methods based on fixed growth rates, machine learning models can also adaptively learn the impact of external factors such as seasons, weather, holidays, and energy prices on electricity load, reducing prediction bias caused by empirical assumptions. By aggregating the load demand at the user level to the distributed energy station level and then further statistically analyzing it to the energy central station level, the computational complexity can be reduced to a certain extent. Differentiated energy allocation is achieved through priority weights, improving the supply stability of key energy loads. The calculation of energy storage regulation incorporates the average of historical net electricity exchange, which can effectively reflect the complementary characteristics between distributed energy stations, alleviate local supply and demand imbalances in specific periods, improve the supply stability of energy loads, and thus improve the accuracy of energy regulation.

[0034] Optionally, in the control module, the weights are dynamically adjusted based on the nth energy station. and statically adjusted weights right The calculation model was modified, and the modified model was then... The computational model is updated, in which , .

[0035] By adopting the above technical solution and setting dynamic and static adjustment weights, different energy stations can adjust their energy storage strategies according to their own historical load characteristics. The value of the dynamic adjustment weight is determined based on the user's historical load data, enabling the model to adapt to long-term load change trends. The static adjustment weight, as a fixed coefficient, can reduce the impact of short-term fluctuations on the overall energy dispatch strategy. For example, for energy stations with frequent power exchange, the static adjustment weight is increased to amplify the impact of net power exchange. When extreme weather causes a surge in temporary net power exchange at multiple energy stations, the static adjustment weight can play a constraining role, reducing the violent fluctuations in energy storage regulation, improving the stability of energy load supply, and thus improving the accuracy of energy regulation.

[0036] Optionally, the acquisition module also acquires the average user electricity load of the nth energy station. ;

[0037] In the control module, according to Dynamic adjustment weights for the nth energy station The value is determined accordingly. The specific calculation model for the value is as follows: .

[0038] By adopting the above technical solution, the weight is dynamically adjusted by quantifying the difference between the user's real-time load and the average value during a specified period. This enables the energy storage regulation to respond to load fluctuations in a timely manner. For example, for energy stations with large load fluctuations, the dynamic adjustment weight can be increased to enhance the elasticity of the basic regulation, thereby reserving more energy storage buffer space, reducing the probability of dispatch failure due to demand surges, improving the stability of energy load supply, and thus improving the accuracy of energy regulation.

[0039] Optionally, it may also include a judgment module and a feedback module;

[0040] Judgment Module: Its input terminal is connected to the output terminal of the control module, and its output terminal is connected to the input terminals of the energy supply module and the feedback module, respectively. It is used to calculate the change in energy storage regulation at the nth energy station during time period t. ,according to Calculate the average change in energy storage regulation during time period t. The warning error for time period t is calculated as follows: Set the warning value to E. If the adjustment is normal, the energy supply module will be executed. If the adjustment is abnormal, the feedback module will be executed.

[0041] Feedback module: The input end is connected to the output end of the judgment module, and it is used to generate an early warning of abnormal energy supply regulation during time period t and upload it to the management end.

[0042] By adopting the above technical solution, anomaly monitoring is performed by calculating the warning error of energy storage regulation changes. The warning error is used to measure the dispersion of energy storage regulation changes at each energy station, which can effectively identify abnormal situations where the decision-making behavior of the energy station group deviates. Based on the comparison results of the warning error and the warning value, an anomaly early warning mechanism is established to reduce the probability of the spread of erroneous scheduling strategies, improve the accuracy of inter-site coordination and joint decision-making, improve the overall energy efficiency of distributed energy stations, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0043] Optionally, in the judgment module, let For the first period, For the second time period, obtain the warning error of the first time period. Warning error in the second period ,according to and The value of the warning value E is determined, and the specific calculation model for the value of E is as follows: .

[0044] By adopting the above technical solution, the warning value is dynamically adjusted based on the warning error in the first and second time periods. This allows the warning value to automatically adjust in accordance with periodic or seasonal load changes. Using the average of the warning errors in similar time periods at certain intervals as the warning value reduces the possibility of sudden changes in the warning value due to abnormal data in a single time period. This reduces reliance on experience-based presets and lowers the probability of repeated manual intervention in adjusting the strategy. For example, the warning value increases during peak electricity consumption seasons to tolerate greater fluctuations in energy storage regulation. The system can adaptively update the warning value, reducing operation and maintenance costs, improving the stability of energy load supply, and thus improving the accuracy of energy regulation.

[0045] Compared with the prior art, the beneficial effects of the present invention are:

[0046] 1. By establishing a pre-set time-series energy dispatch model trained on historical electricity load data, the periodic and sudden characteristics of user electricity consumption behavior can be effectively captured. Compared with traditional proportional division methods or prediction methods based on fixed growth rates, machine learning models can also adaptively learn the impact of external factors such as seasons, weather, holidays, and energy prices on electricity load, reducing prediction bias caused by empirical assumptions. By aggregating the load demand at the user level to the distributed energy station level, and then further statistically analyzing it to the energy central station level, the computational complexity can be reduced to a certain extent. Differentiated energy allocation is achieved through priority weights, improving the supply stability of key energy loads. The calculation of energy storage regulation incorporates the average of historical net electricity exchange, which can effectively reflect the complementary characteristics between distributed energy stations, alleviate local supply and demand imbalances in specific periods, improve the supply stability of energy loads, and thus improve the accuracy of energy regulation.

[0047] 2. By setting dynamic and static adjustment weights, different energy stations can adjust their energy storage strategies according to their own historical load characteristics. The value of the dynamic adjustment weight is determined based on the user's historical load data, enabling the model to adapt to long-term load change trends. The static adjustment weight, as a fixed coefficient, can reduce the impact of short-term fluctuations on the overall energy dispatch strategy. For example, for energy stations with frequent power exchange, the static adjustment weight is increased to amplify the impact of net power exchange. When extreme weather causes a surge in temporary net power exchange at multiple energy stations, the static adjustment weight can play a constraining role, reducing the violent fluctuations in energy storage regulation, improving the stability of energy load supply, and thus improving the accuracy of energy regulation.

[0048] 3. By quantifying and dynamically adjusting the weights based on the difference between the user's real-time load and the average load during a specified period, the energy storage regulation can respond promptly to load fluctuations. For example, for energy stations with large load fluctuations, the dynamic adjustment weights can be increased to enhance the elasticity of the basic regulation, thereby reserving more energy storage buffer space, reducing the probability of dispatch failure due to demand surges, improving the stability of energy load supply, and thus improving the accuracy of energy regulation.

[0049] 4. By calculating the warning error of energy storage regulation changes, anomaly monitoring is carried out. The warning error is used to measure the dispersion of energy storage regulation changes at each energy station, which can effectively identify abnormal situations where the decision-making behavior of the energy station group deviates. Based on the comparison results of the warning error and the warning value, an anomaly early warning mechanism is established to reduce the probability of the spread of erroneous scheduling strategies, improve the accuracy of inter-site coordination and joint decision-making, improve the overall energy efficiency of distributed energy stations, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0050] 5. By dynamically adjusting the warning value calculation based on the warning error in the first and second time periods, the warning value can be automatically adjusted to follow periodic or seasonal load changes. Using the average of the warning errors in similar time periods at certain intervals as the warning value can reduce the possibility of sudden changes in the warning value due to abnormal data in a single time period, reduce reliance on experience presets, and reduce the probability of repeated manual intervention in adjustment strategies. For example, the warning value can be increased during peak electricity consumption seasons to tolerate greater fluctuations in energy storage regulation. The system can adaptively update the warning value, reduce operation and maintenance costs, improve the stability of energy load supply, and thus improve the accuracy of energy regulation. Attached Figure Description

[0051] Figure 1 This is a flowchart of the AI-based intelligent control method for distributed energy stations based on machine learning, as described in this invention.

[0052] Figure 2 This is a block diagram of the AI-powered intelligent control system for distributed energy stations based on machine learning, as described in this invention. Detailed Implementation

[0053] The following will be based on embodiments of the present invention. Figure 1 and Figure 2 The technical solutions in the embodiments of the present invention are clearly and completely described herein. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0054] Example 1: This example discloses an AI-based intelligent control method for distributed energy stations based on machine learning, referring to... Figure 1 This includes the following steps:

[0055] S1. Data Acquisition Steps: The central energy station connects to N distributed energy stations and their respective M groups of users, acquiring the users' historical electricity load, the energy station's historical electricity load, and the basic energy storage regulation. The net electricity exchanged yields the average user electricity load of the nth energy station. ,in .

[0056] S2. Analysis Steps: Establish a preset time-series energy dispatch model based on the user's historical electricity load, and output the predicted electricity load of the m-th group of users at the n-th energy station within time period t. ,in .

[0057] S3. Calculation steps: The predicted electricity load of the nth energy station during time period t is: ,in This represents the priority weight of the m-th group of users at the n-th energy station.

[0058] S4. Control Steps: The predicted electricity load of the main energy station during time period t is: Where η is the loss coefficient, Let be the energy storage regulation amount of the nth energy station within time period t. The calculation model is as follows: ,in The dynamic adjustment weight for the nth energy station. The specific calculation model for the value is as follows: , The static adjustment weight for the nth energy station. , Let be the average net electricity exchange of the nth energy station.

[0059] S5. Judgment Step: The change in energy storage regulation at the nth energy station during time period t is... The average change in energy storage regulation during time period t is The warning error for time period t is Set the alert value to E, and let For the first period, This is the second time period. Obtain the warning error for the first time period. Warning error in the second period ,according to and The value of the warning value E is determined, and the specific calculation model for the value of E is as follows: ,like If the adjustment is normal, proceed to step S6 of the energy supply procedure. If the adjustment is abnormal, then feedback step S7 is executed.

[0060] S6, Energy Supply Steps: During time period t, according to The energy supply of the central energy station is dispatched according to... Adjust the energy storage regulation amount of the nth energy station according to Allocate the energy supply to the nth energy station.

[0061] S7. Feedback Steps: Generate an early warning of abnormal energy supply regulation for time period t and upload it to the management terminal.

[0062] The implementation principle of the AI-based intelligent control method for distributed energy stations based on machine learning in this embodiment is as follows:

[0063] The central energy station connects to N distributed energy stations, each of which connects to M groups of users. It obtains historical electricity load data from users, historical electricity load data from energy stations, and basic energy storage regulation. The net electricity exchanged yields the average user electricity load of the nth energy station. A preset time-series energy dispatch model is established based on the user's historical electricity load, and the predicted electricity load of the m-th group of users at the n-th energy station during time period t is output. .

[0064] By establishing a pre-set time-series energy dispatch model trained on historical electricity load data using machine learning techniques, it is possible to effectively capture the periodic and sudden characteristics of user electricity consumption behavior. Machine learning techniques include, but are not limited to, LSTM time-series prediction models and neural network techniques. Compared with traditional proportional division methods or prediction methods based on fixed growth rates, machine learning models can also adaptively learn the impact of external factors such as seasons, weather, holidays, and energy prices on electricity load, reducing prediction bias caused by empirical assumptions. For example, in the summer or winter seasons in northern regions, the model can automatically analyze and process the relationship between increased air conditioning load and temperature changes, accurately predict peak demand, and optimize energy dispatch strategies in a timely manner, thereby improving the accuracy of energy regulation.

[0065] The predicted electricity load of the nth energy station during time period t is calculated as follows: ,in The priority weight of the m-th group of users at the n-th energy station. For example, industrial or institutional users can be assigned higher weights to improve the supply stability of energy load for critical energy-consuming equipment. The predicted electricity load of the main power station during time period t is: Where η is the energy transmission loss coefficient between the main energy station and the distributed energy stations. Let be the energy storage regulation amount of the nth energy station within time period t. The calculation model is as follows: ,in The dynamic adjustment weight for the nth energy station. The specific calculation model for the value is as follows: , The static adjustment weight for the nth energy station. , Let be the average net electricity exchange of the nth energy station.

[0066] By aggregating user-level load demands to the distributed energy station level and then further summarizing them to the central energy station level, the computational complexity can be reduced to some extent. Differentiated energy allocation can be achieved through priority weights, improving the supply stability of key energy loads. The calculation of energy storage regulation incorporates the average of historical net electricity exchange, which can effectively reflect the complementary characteristics between distributed energy stations and alleviate local supply-demand imbalances during specific periods. For example, during a period, energy stations with redundant resources can transfer energy to energy stations experiencing insufficient energy storage or real-time electricity price changes by utilizing their remaining energy storage regulation capacity, thereby improving the supply stability of energy loads and thus enhancing the accuracy of energy regulation.

[0067] By setting dynamic and static adjustment weights, different energy stations can adjust their energy storage strategies according to their own historical load characteristics. The value of the dynamic adjustment weight is determined based on the user's historical load data, enabling the model to adapt to long-term load change trends. The dynamic adjustment weight is quantified by the difference between the user's real-time load and the average load during a specified period, so that the energy storage regulation can respond to load fluctuations in a timely manner. For example, for energy stations with large load fluctuations, the dynamic adjustment weight can be increased to enhance the elasticity of the basic regulation, thereby reserving more energy storage buffer space and reducing the probability of dispatch failure due to demand surges.

[0068] As a fixed coefficient, the static adjustment weight can reduce the impact of short-term fluctuations on the overall energy dispatch strategy. For example, for energy stations with frequent power exchange, the static adjustment weight can be increased to amplify the impact of net power exchange. When extreme weather causes a surge in temporary net power exchange at multiple energy stations, the static adjustment weight can play a constraining role, reduce the violent fluctuations in energy storage regulation, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0069] Calculate the change in energy storage regulation at the nth energy station during time period t. The average change in energy storage regulation during time period t is The warning error for time period t is Set the alert value to E, and let For the first period, This is the second time period. Obtain the warning error for the first time period. Warning error in the second period ,according to and The value of the warning value E is determined, and the specific calculation model for the value of E is as follows: ,like If so, the adjustment is considered normal. If so, then the regulation is judged to be abnormal.

[0070] When an abnormal regulation is detected, an energy supply regulation abnormality warning for time period t is generated and uploaded to the management terminal. When the regulation is detected as normal, within time period t, according to... The energy supply of the central energy station is dispatched according to... Adjust the energy storage regulation amount of the nth energy station according to Allocate the energy supply to the nth energy station.

[0071] Anomaly monitoring is conducted by calculating the warning error of changes in energy storage regulation. The warning error is used to measure the dispersion of changes in energy storage regulation at each energy station, which can effectively identify abnormal situations where the decision-making behavior of the energy station group deviates. Based on the comparison results of the warning error and the warning value, an anomaly early warning mechanism is established to reduce the probability of the spread of erroneous scheduling strategies, improve the accuracy of inter-site coordination and joint decision-making, improve the overall energy efficiency of distributed energy stations, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0072] By dynamically adjusting the warning value calculation based on the warning error in the first and second time periods, the warning value can be automatically adjusted to follow periodic or seasonal load changes. Using the average of the warning errors in similar time periods at certain intervals as the warning value can reduce the possibility of sudden changes in the warning value due to abnormal data in a single time period, reduce reliance on experience presets, and reduce the probability of repeated manual intervention in adjustment strategies. For example, the warning value can be increased during peak electricity consumption seasons to tolerate greater fluctuations in energy storage regulation. The system can adaptively update the warning value, reduce operation and maintenance costs, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0073] Example 2: This example discloses an AI-powered intelligent control system for distributed energy stations based on machine learning, referring to... Figure 2 It includes the following modules:

[0074] Acquisition Module: The output end connects to the input end of the analysis module. The energy central station connects to N distributed energy stations and their respective M groups of users, and is used to acquire the historical electricity load of users and the historical electricity load of energy stations, as well as the basic energy storage regulation. The net electricity exchanged yields the average user electricity load of the nth energy station. .

[0075] Analysis module: The input end is connected to the output end of the acquisition module, and the output end is connected to the input end of the calculation module. It is used to establish a preset time-series energy dispatch model based on the user's historical electricity load, and output the predicted electricity load of the m-th group of users at the n-th energy station within time period t. .

[0076] The calculation module's input terminal connects to the output terminal of the analysis module, and its output terminal connects to the input terminal of the control module. It is used to calculate based on... and the priority weight of the m-th group of users at the n-th energy station Calculate the predicted electricity load of the nth energy station during time period t. .

[0077] Control module: Its input terminal is connected to the output terminal of the calculation module, and its output terminal is connected to the input terminal of the judgment module. It is used to control the input terminal based on the input terminal of the calculation module. and the average net electricity exchange of the nth energy station Dynamically adjust weights and statically adjusted weights Calculate the energy storage regulation of the nth energy station during time period t. , The specific calculation model for the value is as follows: , and according to , The predicted electricity load of the main energy station during the time period t is calculated using the loss factor η. .

[0078] Judgment Module: Its input terminal is connected to the output terminal of the control module, and its output terminal is connected to the input terminals of the energy supply module and the feedback module, respectively. It is used to calculate the change in energy storage regulation at the nth energy station during time period t. ,according to Calculate the average change in energy storage regulation during time period t. The warning error for time period t is calculated as follows: Set the alert value to E, and let For the first period, For the second time period, obtain the warning error of the first time period. Warning error in the second period ,according to and The value of the warning value E is determined, and the specific calculation model for the value of E is as follows: ,like If the adjustment is normal, the energy supply module will be executed. If the adjustment is abnormal, the feedback module will be executed.

[0079] Energy supply module: The input terminal is connected to the output terminal of the judgment module, and is used to determine the energy supply within time period t based on... The energy supply of the central energy station is dispatched according to... Adjust the energy storage regulation amount of the nth energy station according to Allocate the energy supply to the nth energy station.

[0080] Feedback module: The input end is connected to the output end of the judgment module, and it is used to generate an early warning of abnormal energy supply regulation during time period t and upload it to the management end.

[0081] The implementation principle of the distributed energy station AI intelligent control system based on machine learning in this embodiment is as follows:

[0082] The central energy station connects to N distributed energy stations, each of which connects to M groups of users. The acquisition module obtains the historical electricity load of the users, the historical electricity load of the energy stations, and the basic energy storage regulation. The net electricity exchanged yields the average user electricity load of the nth energy station. The analysis module establishes a preset time-series energy dispatch model based on the user's historical electricity load and outputs the predicted electricity load of the nth energy station and m groups of users within time period t. .

[0083] By establishing a pre-set time-series energy dispatch model trained on historical electricity load data using machine learning technology, it is possible to effectively capture the periodic and sudden characteristics of users' electricity consumption behavior. At the same time, it can adaptively learn the impact of external factors such as seasons, weather, holidays and energy prices on electricity load, reduce prediction bias caused by empirical assumptions, accurately predict peak demand, and optimize energy dispatch strategies in a timely manner, thereby improving the accuracy of energy regulation.

[0084] The calculation module is based on and the priority weight of the m-th group of users at the n-th energy station Calculate the predicted electricity load of the nth energy station during time period t. , If it is a positive number, the control module will adjust accordingly. and the average net electricity exchange of the nth energy station Dynamically adjust weights and statically adjusted weights Calculate the energy storage regulation of the nth energy station during time period t. , The specific calculation model for the value is as follows: , and according to , The predicted electricity load of the main energy station during the time period t is calculated using the loss factor η. .

[0085] By aggregating user-level load demands to the distributed energy station level and then further statistically analyzing them to the central energy station level, the computational complexity can be reduced to some extent. Differentiated energy allocation can be achieved through priority weights, improving the supply stability of key energy loads. The calculation of energy storage regulation incorporates the average of historical net electricity exchange, which can effectively reflect the complementary characteristics between distributed energy stations, alleviate local supply and demand imbalances in specific time periods, improve the supply stability of energy loads, and thus enhance the accuracy of energy regulation.

[0086] By setting dynamic and static adjustment weights, different energy stations can adjust their energy storage strategies according to their own historical load characteristics. The value of the dynamic adjustment weight is determined based on the user's historical load data, enabling the model to adapt to long-term load change trends. The dynamic adjustment weight is quantified by the difference between the user's real-time load and the average load during a specified period, allowing the energy storage regulation to respond to load fluctuations in a timely manner and reducing the probability of dispatch failure due to demand surges. The static adjustment weight, as a fixed coefficient, can reduce the impact of short-term fluctuations on the overall energy dispatch strategy, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0087] The judgment module calculates the change in energy storage regulation of the nth energy station during time period t. The average change in energy storage regulation during time period t is calculated as follows: The warning error for time period t is Set the alert value to E, and let For the first period, This is the second time period. Obtain the warning error for the first time period. Warning error in the second period ,according to and The value of the warning value E is determined, and the specific calculation model for the value of E is as follows: ,like If so, the adjustment is considered normal. If so, then the regulation is judged to be abnormal.

[0088] When an abnormal adjustment is detected, the feedback module generates an energy supply adjustment anomaly warning for time period t and uploads it to the management terminal. When the adjustment is detected as normal, the energy supply module, within time period t, according to... The energy supply of the central energy station is dispatched according to... Adjust the energy storage regulation amount of the nth energy station according to Allocate the energy supply to the nth energy station.

[0089] Anomaly monitoring is conducted by calculating the warning error of changes in energy storage regulation. The warning error is used to measure the dispersion of changes in energy storage regulation at each energy station, which can effectively identify abnormal situations where the decision-making behavior of the energy station group deviates. Based on the comparison results of the warning error and the warning value, an anomaly early warning mechanism is established to reduce the probability of the spread of erroneous scheduling strategies, improve the accuracy of inter-site coordination and joint decision-making, improve the overall energy efficiency of distributed energy stations, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0090] By dynamically adjusting the warning value calculation based on the warning error in the first and second time periods, the warning value can be automatically adjusted to follow periodic or seasonal load changes. Using the average of the warning errors in similar time periods at certain intervals as the warning value can reduce the possibility of sudden changes in the warning value due to abnormal data in a single time period, reduce reliance on experience presets, and reduce the probability of repeated manual intervention in adjustment strategies. For example, the warning value can be increased during peak electricity consumption seasons to tolerate greater fluctuations in energy storage regulation. The system can adaptively update the warning value, reduce operation and maintenance costs, improve the stability of energy load supply, and thus improve the accuracy of energy regulation.

[0091] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A machine learning-based AI-powered intelligent control method for distributed energy stations, characterized in that: Includes the following steps: Data Acquisition: The central energy station connects to N distributed energy stations and their respective M groups of users, acquiring historical electricity load data of users, historical electricity load data of energy stations, and basic energy storage regulation. The net electricity exchanged yields the average user electricity load of the nth energy station. ; Analysis: A preset time-series energy dispatch model is established based on the user's historical electricity load, and the predicted electricity load of the m-th group of users at the n-th energy station within time period t is output. ; Calculate: The predicted electricity load of the nth energy station during time period t is: ,in The priority weight for the m-th group of users at the n-th energy station; Control: The predicted electricity load of the main energy station during time period t is: Where η is the loss coefficient, Let be the energy storage regulation amount of the nth energy station within time period t. The calculation model is as follows: ,in, Let be the average net electricity exchange volume of the nth energy station; right The calculation model was revised, and after revision The computational model is updated as follows: ,in The dynamic adjustment weight for the nth energy station. , The value is determined based on the user's historical electricity load. The specific calculation model for the value is as follows: , The static adjustment weight for the nth energy station. ; Determine: The change in energy storage regulation at the nth energy station during time period t is... The average change in energy storage regulation during time period t is The warning error for time period t is Set the warning value to E. If the regulation is normal, the energy supply procedure will be executed. If the adjustment is abnormal, a feedback step will be executed. Feedback: Generate an alert for abnormal energy supply regulation during time period t and upload it to the management terminal; Energy supply: During time period t, according to The energy supply of the central energy station is dispatched according to... Adjust the energy storage regulation amount of the nth energy station according to Allocate the energy supply to the nth energy station.

2. The AI-based intelligent control method for distributed energy stations based on machine learning according to claim 1, characterized in that: In the decision step, let For the first period, For the second time period, obtain the warning error of the first time period. Warning error in the second period ,according to and The value of the warning value E is determined, and the specific calculation model for the value of E is as follows: .

3. A distributed energy station AI intelligent control system based on machine learning, characterized in that: Includes the following modules: Acquisition Module: The output end connects to the input end of the analysis module. The energy central station connects to N distributed energy stations and their respective M groups of users, and is used to acquire the historical electricity load of users and the historical electricity load of energy stations, as well as the basic energy storage regulation. The net electricity exchanged yields the average user electricity load of the nth energy station. ; Analysis module: The input end is connected to the output end of the acquisition module, and the output end is connected to the input end of the calculation module. It is used to establish a preset time-series energy dispatch model based on the user's historical electricity load, and output the predicted electricity load of the m-th group of users at the n-th energy station within time period t. ; The calculation module's input terminal connects to the output terminal of the analysis module, and its output terminal connects to the input terminal of the control module. It is used to calculate based on... and the priority weight of the m-th group of users at the n-th energy station Calculate the predicted electricity load of the nth energy station during time period t. ; Control module: Its input terminal is connected to the output terminal of the calculation module, and its output terminal is connected to the input terminal of the judgment module. It is used to control the input terminal based on the input terminal of the calculation module. The average net amount of electricity exchanged with the nth energy station Calculate the energy storage regulation capacity of the nth energy station within time period t. ; The weights are dynamically adjusted based on the nth energy station. and statically adjusted weights right The calculation model was modified according to... Dynamic adjustment weights for the nth energy station The value is determined accordingly. The specific calculation model for the value is as follows: And the revised The calculation model was updated and corrected. The computational model is updated as follows: ,in , and according to , The predicted electricity load of the main energy station during the time period t is calculated using the loss factor η. ; Judgment Module: Its input terminal is connected to the output terminal of the control module, and its output terminal is connected to the input terminals of the energy supply module and the feedback module, respectively. It is used to calculate the change in energy storage regulation at the nth energy station during time period t. ,according to The average change in energy storage regulation during time period t is calculated as follows: The warning error for time period t is calculated as follows: Set the warning value to E. If the adjustment is normal, the energy supply module will be executed. If the adjustment is abnormal, the feedback module will be executed. Feedback module: The input end is connected to the output end of the judgment module, and it is used to generate an early warning of abnormal energy supply regulation during time period t and upload it to the management end; Energy supply module: The input terminal is connected to the output terminal of the judgment module, and is used to determine the energy supply within time period t based on... The energy supply of the central energy station is dispatched according to... Adjust the energy storage regulation amount of the nth energy station according to Allocate the energy supply to the nth energy station.

4. The AI-powered intelligent control system for distributed energy stations based on machine learning according to claim 3, characterized in that: In the judgment module, let For the first period, For the second time period, obtain the warning error of the first time period. Warning error in the second period ,according to and The value of the warning value E is determined, and the specific calculation model for the value of E is as follows: .

Images