Urban energy intelligent management method and system based on deep learning
By dynamically dividing virtual partitioning modes using deep learning technology and combining real-time monitoring and prediction data, the problem of control mismatch caused by distributed power fluctuations in virtual partitioning management has been solved, thus achieving safe, stable operation and efficient control of the urban energy system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-07
- Publication Date
- 2026-04-14
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing virtual partition management methods are unable to cope with boundary drift and control authority confusion caused by sudden changes in wind and solar power output when faced with a high proportion of distributed power access, leading to risks such as line overload and voltage abnormalities.
A deep learning-based intelligent urban energy management method is adopted. By monitoring and predicting data in real time, the system dynamically divides the city into a high-capacity autonomous mode and a low-capacity auxiliary mode. It utilizes a central coordinator and local controllers to coordinate and regulate the city, enabling cross-regional support and ensuring that the regulation logic matches the actual power flow.
It effectively reduced the control mismatch caused by boundary drift, realized the transformation from passive response to active defense, improved the forward-looking early warning and preventive control capabilities of the urban energy management system, and ensured the safe and stable operation of the system.
Smart Images
Figure CN121860581A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent energy management technology, specifically to an intelligent urban energy management method and system based on deep learning. Background Technology
[0002] With the acceleration of urbanization and the transformation of energy structure, the penetration rate of distributed renewable energy, energy storage systems and flexible loads in urban power grids continues to increase. This has led to the evolution of urban energy system operation from the traditional centralized control based on source-load dynamics to a distributed collaborative model with multi-level interaction between source, grid, load and storage. To address this change, virtual partitioning technology has emerged. Its core idea is to divide the physically continuous urban power grid into multiple logical virtual partitions based on the relative consistency of topology and load distribution. Each virtual partition acts as a relatively independent autonomous control unit, which can achieve rapid self-balancing of power generation and load within itself. At the same time, global optimization is achieved through coordination between partitions. Existing technical solutions mostly focus on how to statically divide these virtual partitions based on historical data or scenarios, and preset the control strategies and boundaries of each partition in order to improve management efficiency and operational economy under normal operating conditions.
[0003] However, the aforementioned virtual zone management method based on static presets reveals its inherent limitations when facing the strong uncertainties brought about by the high proportion of distributed power sources. Distributed power sources, such as wind power and photovoltaic clusters, have intermittent and fluctuating outputs. When these fluctuations are large and rapid, they can instantly change the power distribution and power flow direction within the urban power grid. This change may cause the actual power balance range of a virtual zone to far exceed its preset electrical boundary, i.e., boundary drift. For example, when a wind-rich virtual zone in the eastern part of the city experiences a surge in output due to a sudden increase in wind speed, its surplus power may flow into the adjacent western zone. If the management system still makes decisions based on preset static boundaries, the eastern zone... The controller may mistakenly attempt to reduce power generation to match the load within its original boundary, while the western controller, lacking authority to manage additional power, is unable to proactively increase acceptance. This severe disconnect between control authority and actual physical power flow caused by boundary drift not only renders zonal autonomy ineffective but also directly triggers immediate operational risks such as interconnection line overload and local voltage anomalies. Ultimately, this may trigger protection devices, causing unplanned power outages, completely deviating from the original intention of achieving safety and optimized management through virtual zoning. Therefore, how to enable virtual zonal boundaries to have dynamic adaptive capabilities and ensure that control authority tracks and matches the actual energy flow distribution in real time has become a core technical challenge that urgently needs to be addressed in current urban energy intelligent management. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a deep learning-based intelligent urban energy management method and system, which can effectively solve the problems mentioned in the background technology.
[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: a deep learning-based urban energy intelligent management method, comprising dividing the urban energy intelligent management power grid into virtual partitions based on the physical and electrical structure parameters of the power grid, deploying data acquisition units in each virtual partition, monitoring and acquiring the operating status parameters of each virtual partition, and analyzing and obtaining the operating characteristic evaluation value of each virtual partition.
[0006] Predictive data from each virtual power grid zone is collected synchronously, and the predictive data feature values of each virtual power grid zone are obtained through processing. Based on the operational characteristic evaluation values of each virtual power grid zone and the predictive data feature values of each virtual power grid zone, the dynamic carrying capacity index values of each virtual power grid zone are obtained through comprehensive processing.
[0007] Based on the dynamic carrying capacity index value of each power grid virtual partition, the operating mode of each power grid virtual partition is dynamically determined. The operating modes of each power grid virtual partition include a high carrying capacity autonomous mode and a low carrying capacity auxiliary mode.
[0008] Grant high autonomy to virtual power grid partitions marked as high-capacity autonomous mode, automatically trigger global auxiliary mode for virtual power grid partitions marked as low-capacity auxiliary mode, and generate cross-partition support commands to adjust the operational balance of virtual power grid partitions.
[0009] Furthermore, the physical and electrical structure parameters of the smart urban energy management power grid are divided, specifically as follows: the physical and electrical structure parameters include the connection relationship of each line, the distribution relationship of substations, and the direction of power transmission.
[0010] Based on the connection relationships of various lines, the distribution relationships of substations, and the direction of power transmission in the urban energy smart management power grid, the urban energy smart management power grid is divided into multiple regions, resulting in virtual partitions of each power grid.
[0011] Furthermore, the analysis yields the operational characteristic evaluation values of each virtual power grid partition. The specific analysis process involves monitoring and acquiring the operational status parameters of each virtual power grid partition, which include the total active power, total reactive power, average voltage amplitude, line load rate, and reserve capacity of each virtual power grid partition.
[0012] The operating status parameters of each power grid virtual partition are analyzed to obtain the operating characteristic evaluation value of each power grid virtual partition. The operating characteristic evaluation value of each power grid virtual partition represents the quantitative result of the operating status parameters of each power grid virtual partition on the current operating health status of the power grid virtual partition.
[0013] Furthermore, the processing obtains the predicted data feature values of each power grid virtual partition. The specific processing procedure is as follows: the predicted data of each power grid virtual partition is collected synchronously, and the predicted data of each power grid virtual partition includes the load prediction value and the power generation prediction value of each power grid virtual partition.
[0014] Based on the prediction data of each power grid virtual partition, the prediction data feature values of each power grid virtual partition are obtained. The prediction data feature values of each power grid virtual partition represent the quantitative result of the prediction data of each power grid virtual partition on the operational risk status of the power grid virtual partition.
[0015] Furthermore, the comprehensive processing yields the dynamic carrying capacity index value of each power grid virtual partition. The specific processing procedure is as follows: based on the operational characteristic evaluation value and the predicted data feature value of each power grid virtual partition, the comprehensive processing yields the dynamic carrying capacity index value of each power grid virtual partition. The dynamic carrying capacity index value of each power grid virtual partition represents the quantitative result of the operational characteristic evaluation value and the predicted data feature value of each power grid virtual partition on maintaining autonomous operation capability.
[0016] Furthermore, the dynamic determination of the operating mode of each power grid virtual partition is specifically as follows: the dynamic carrying capacity index value of each power grid virtual partition is compared with the set dynamic carrying capacity index threshold. If the dynamic carrying capacity index value of a certain power grid virtual partition is higher than the set dynamic carrying capacity index threshold, the operating mode of the power grid virtual partition is marked as a high carrying capacity autonomous mode, and the power grid virtual partition is recorded as a high carrying capacity power grid virtual partition. Otherwise, the operating mode of the power grid virtual partition is marked as a low carrying capacity auxiliary mode, and the power grid virtual partition is recorded as a low carrying capacity power grid virtual partition.
[0017] Furthermore, the specific process of granting high autonomy permissions to the power grid virtual partition marked as high-capacity autonomous mode is as follows: the central coordinator sends an autonomy authorization instruction to the local controller of the power grid virtual partition marked as high-capacity autonomous mode.
[0018] After receiving the autonomous authorization command, the local controller of the high-capacity power grid virtual partition executes the charging and discharging scheduling of the energy storage system within the high-capacity power grid virtual partition and reports the key results to the central coordinator.
[0019] Furthermore, the specific process of generating cross-regional support instructions is as follows: when a certain power grid virtual region is marked as low-capacity auxiliary mode, the central coordinator automatically triggers the global auxiliary mode.
[0020] After automatically triggering the global auxiliary mode, cross-zone support instructions are generated, including active power support instructions and reactive power support instructions.
[0021] Furthermore, the specific process of adjusting the operational balance of the power grid virtual partition is as follows: the central coordinator will issue cross-partition support instructions to the local controller of the high-capacity power grid virtual partition and the local controller of the high-capacity power grid virtual partition that is electrically adjacent to the low-capacity power grid virtual partition.
[0022] Each local controller that receives the cross-regional support command drives the distributed power source and reactive power compensation equipment to perform the corresponding power output adjustment. For the active power support command, the preset active power support amount generated by the high-capacity grid virtual region is injected into the low-capacity grid virtual region through the designated electrical connection path.
[0023] Simultaneously, the reactive power support command causes the preset reactive power support amount provided by the electrically adjacent high-capacity grid virtual partition to be injected into the low-capacity grid virtual partition.
[0024] The second aspect of the present invention provides a deep learning-based urban energy intelligent management system, comprising: an operation evaluation module, used to divide the urban energy intelligent management power grid into virtual partitions based on the physical and electrical structure parameters of the power grid, deploy data acquisition units in each virtual partition, monitor and acquire the operation status parameters of each virtual partition, and analyze and obtain the operation characteristic evaluation value of each virtual partition.
[0025] The load-bearing capacity analysis module is used to synchronously collect the prediction data of each power grid virtual partition, process it to obtain the prediction data feature value of each power grid virtual partition, and comprehensively process the operation characteristic evaluation value and the prediction data feature value of each power grid virtual partition to obtain the dynamic load-bearing capacity index value of each power grid virtual partition.
[0026] The mode determination module is used to dynamically determine the operating mode of each power grid virtual partition based on the dynamic carrying capacity index value of each power grid virtual partition. The operating modes of each power grid virtual partition include a high carrying capacity autonomous mode and a low carrying capacity auxiliary mode.
[0027] The operation balancing module is used to grant high autonomy to virtual power grid partitions marked as high-capacity autonomous mode, automatically trigger global auxiliary mode for virtual power grid partitions marked as low-capacity auxiliary mode, and generate cross-partition support instructions to adjust the operation balance of virtual power grid partitions.
[0028] The present invention has the following beneficial effects: (1) This invention addresses the core problem of control mismatch caused by virtual partition boundary drift under high-proportion distributed power access. Existing technologies mostly rely on statically preset virtual partition boundaries and fixed control strategies, which cannot cope with the real-time power flow reconstruction caused by sudden changes in wind and solar power output, resulting in disordered permissions. This invention calculates dynamic carrying capacity index by integrating real-time operating status and ultra-short-term forecast data, and uses this as the sole criterion to dynamically divide the two operating modes into high-carrying-capacity autonomous and low-carrying-capacity auxiliary modes. This makes the control logic of the management system jump from static rules based on preset geographical boundaries to dynamic authorization based on real-time capability assessment. When the actual jurisdiction of a partition changes due to power fluctuations, i.e., when the boundary drifts, its operating mode and permissions can be adaptively adjusted accordingly, ensuring that the jurisdiction of the control command always matches the real physical power flow, thereby effectively preventing the chain safety risks such as line overload and voltage collapse caused by misalignment of rights and responsibilities.
[0029] (2) This invention constructs a two-layer resilience assessment system that combines quantitative assessment of the current state with forward-looking early warning of future risks, realizing a paradigm shift from passive response to active defense. Existing methods mostly focus on monitoring the historical or current state and lack the ability to predict and prepare for upcoming fluctuations. This invention not only accurately quantifies the current operating characteristics assessment value of the partition through multi-dimensional operating parameters, but also introduces and processes the predicted data of load and new energy, extracts the predictive data feature values that characterize the future, and integrates the two into a dynamic carrying capacity index. This enables the system to identify the attenuation trend of its carrying capacity in advance before the partition actually becomes unstable, and trigger different levels of control measures in advance, thereby improving the forward-looking early warning and preventive control capabilities of the urban energy intelligent management system.
[0030] (3) This invention proposes a hierarchical collaborative and source-load interactive elastic control strategy, which realizes the efficient unity of global optimization and local autonomy. For partitions with different carrying capacities, this invention designs a differentiated and refined collaborative control mechanism. For partitions with high carrying capacity, it grants them high autonomy, allowing them to use local energy storage and other resources to quickly smooth out internal fluctuations, giving full play to the rapid response advantage of distributed control and reducing the computational and communication burden of the central system. For partitions with low carrying capacity, it automatically triggers the global auxiliary mode, and the central coordinator coordinates and schedules the surplus resources of other partitions for cross-regional support. This reflects the global optimization advantage of centralized control. This hierarchical collaborative architecture with autonomy as the main focus and support as the auxiliary focus not only ensures the rapid on-site resolution of local problems, but also ensures the safety and stability of the global system when facing local failures, achieving the best balance between resource utilization efficiency and system operational resilience. Attached Figure Description
[0031] Figure 1 This is a schematic diagram of the method flow of the present invention.
[0032] Figure 2 This is a schematic diagram of the system module connections of the present invention. Detailed Implementation
[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0034] Please see Figure 1 As shown, the first aspect of the present invention provides a technical solution: a deep learning-based urban energy intelligent management method, which includes dividing the urban energy intelligent management power grid based on the physical and electrical structure parameters to obtain each power grid virtual partition, deploying data acquisition units in each power grid virtual partition, monitoring and acquiring the operating status parameters of each power grid virtual partition, and analyzing and obtaining the operating characteristic evaluation value of each power grid virtual partition.
[0035] Predictive data from each virtual power grid zone is collected synchronously, and the predictive data feature values of each virtual power grid zone are obtained through processing. Based on the operational characteristic evaluation values of each virtual power grid zone and the predictive data feature values of each virtual power grid zone, the dynamic carrying capacity index values of each virtual power grid zone are obtained through comprehensive processing.
[0036] Based on the dynamic carrying capacity index value of each power grid virtual partition, the operating mode of each power grid virtual partition is dynamically determined. The operating modes of each power grid virtual partition include a high carrying capacity autonomous mode and a low carrying capacity auxiliary mode.
[0037] Grant high autonomy to virtual power grid partitions marked as high-capacity autonomous mode, automatically trigger global auxiliary mode for virtual power grid partitions marked as low-capacity auxiliary mode, and generate cross-partition support commands to adjust the operational balance of virtual power grid partitions.
[0038] Specifically, the division is based on the physical and electrical structure parameters of the urban energy smart management power grid. The specific process is as follows: the physical and electrical structure parameters include the connection relationship of each line, the distribution relationship of substations, and the direction of power transmission.
[0039] Based on the connection relationships of various lines, the distribution relationships of substations, and the direction of power transmission in the urban energy smart management power grid, the urban energy smart management power grid is divided into multiple regions, resulting in virtual partitions of each power grid.
[0040] It should be added that, based on the connection relationships of the various lines, multiple connected subnetworks composed of electrically directly or indirectly connected lines and nodes are identified. Each connected subnetwork is used as an initial partitioned area. Then, based on the distribution relationship of the substations, load nodes and power supply nodes belonging to the same substation's power supply range and whose electrical distance is within a preset distance threshold are partitioned into the same initial partitioned area. Then, based on the power transmission direction, boundary verification and merging are performed on the initial partitioned areas with a clear main power supply direction. If, on the connecting lines between two adjacent initial partitioned areas, the active power direction continuously points from one area to another within a statistical period exceeding a preset proportion, then these two areas are merged. The domains are merged into a single virtual power grid partition. If an initially partitioned region simultaneously outputs power to multiple adjacent regions, and the power value of each output channel exceeds the preset ratio threshold of its line transmission capacity, then the initially partitioned region is independently divided into a single virtual power grid partition. Finally, all regions that have been merged or independently partitioned are output as the virtual power grid partitions. This partitioning method based on physical electrical structure parameters can fully consider factors such as the actual electrical connection of the power grid, the power supply range of substations, and the direction of power transmission, making the partitioning of virtual partitions more scientific and reasonable. It effectively avoids the problem of unreasonable partitioning caused by static preset boundaries, improves the stability and adaptability of partitions, and better copes with the uncertainties brought about by distributed power source access.
[0041] In this embodiment, by dividing the power grid into virtual partitions based on physical electrical structure parameters, the actual topology and load distribution of the urban power grid can be better matched, laying a data foundation for subsequent accurate assessment of the operating status of each partition.
[0042] Specifically, the operational characteristic evaluation values of each virtual power grid partition are obtained through analysis. The specific analysis process is as follows: monitoring and acquiring the operational status parameters of each virtual power grid partition, which include the total active power, total reactive power, average voltage amplitude, line load rate, and reserve capacity of each virtual power grid partition.
[0043] It should be noted that the total active power of each virtual power grid zone reflects the net balance between power generation and consumption in that zone. It represents the difference between the total active power generated by all power sources and the total active power consumed by all loads within the zone. This difference can be obtained by real-time monitoring and summing using smart meters and data acquisition units deployed at each grid connection point within the zone. The total reactive power of each virtual power grid zone reflects the balance between reactive power compensation and consumption within that zone. It represents the difference between the reactive power generated by all reactive power sources, such as capacitors and SVG, and the reactive power consumed by all inductive loads. This is also obtained by monitoring and summing the reactive power at each node. The average voltage amplitude of each virtual power grid zone reflects the overall voltage level of that zone. The arithmetic mean of the effective voltage values of all key monitoring nodes within the demonstration zone is obtained by reading the measured values of the voltage transformers at each node through the data acquisition unit and calculating the mean value. The line load rate of each virtual power grid zone reflects the utilization rate and safety margin of the key transmission channels within the zone. It represents the average ratio of the active power currently flowing through each key line to its thermal stability rated capacity, obtained by monitoring line power and querying equipment parameters. The reserve capacity of each virtual power grid zone reflects the zone's ability to adjust to power disturbances in real time. It represents the sum of the available capacity of all online generators or energy storage systems within the zone to adjust power upwards and downwards under the current output state, obtained through the urban energy intelligent management and monitoring system.
[0044] The operating status parameters of each power grid virtual partition are analyzed to obtain the operating characteristic evaluation value of each power grid virtual partition. The operating characteristic evaluation value of each power grid virtual partition represents the quantitative result of the operating status parameters of each power grid virtual partition on the current operating health status of the power grid virtual partition.
[0045] It should be added that the total active power, total reactive power, average voltage amplitude, line load rate, and reserve capacity of each virtual power grid partition are normalized. Specifically, this normalization process involves: dividing the total active power by the absolute value of the maximum total active power within a preset historical statistical period to obtain the normalized total active power value; dividing the total reactive power by the absolute value of the maximum total reactive power within a preset historical statistical period to obtain the normalized total reactive power value; subtracting a preset rated voltage value from the average voltage amplitude, taking the absolute value, and then dividing it by a preset maximum allowable voltage deviation value to obtain the normalized average voltage amplitude deviation value; directly using the line load rate as the normalized line load rate value between 0 and 1; and dividing the reserve capacity by the corresponding preset maximum adjustable capacity sum to obtain the normalized reserve capacity value.
[0046] In this embodiment, the operational characteristic evaluation values of each virtual power grid partition can be obtained through the following analysis method, with the specific analysis conditions as follows: ; In the formula, This represents the operational characteristic evaluation value of the i-th virtual power grid partition. This represents the normalized total active power value of the i-th virtual power grid partition. This represents the characteristic evaluation factor corresponding to the set total active power value. This represents the normalized total reactive power value of the i-th virtual power grid partition. This represents the characteristic evaluation factor corresponding to the set total reactive power value. This represents the normalized average voltage amplitude deviation value of the i-th virtual power grid partition. This represents the characteristic evaluation factor corresponding to the set average voltage amplitude deviation value. This represents the normalized line load rate value of the i-th virtual power grid partition. This represents the characteristic evaluation factor corresponding to the set line load rate value. This represents the normalized reserve capacity value of the i-th virtual power grid partition. This represents the characteristic evaluation factor corresponding to the set reserve capacity value, where i represents the number of each virtual power grid partition, i=1, 2, 3, ..., n, and n represents the total number of virtual power grid partitions.
[0047] It should be added that, in this embodiment, the characteristic evaluation factors corresponding to the preset total active power value, total reactive power value, average voltage amplitude deviation value, line load rate value and reserve capacity value are obtained from the urban energy management database.
[0048] It should be explained that these characteristic evaluation factors are used to adjust the importance of the data in the operating status parameters of each power grid virtual partition in the process of analyzing and obtaining the operating characteristic evaluation value. For example, a mapping relationship between the operating status parameters of each power grid virtual partition and the characteristic evaluation factors is set in the urban energy management database. Through the pre-set mapping relationship, the characteristic evaluation factors corresponding to the real-time operating status parameters of each power grid virtual partition can be matched. By matching the operating status parameters of each power grid virtual partition with the pre-set mapping relationship, the characteristic evaluation factors corresponding to the total active power value, total reactive power value, average voltage amplitude deviation value, line load rate value, and reserve capacity value are obtained.
[0049] In this implementation plan, the total active power, total reactive power, average voltage amplitude deviation, line load rate, and reserve capacity values of each virtual power grid zone are correlated and not independent. For example, an imbalance between the supply and demand of total active power directly affects the stability of the average voltage amplitude, while anomalies in the average voltage amplitude, in turn, affect the safe current-carrying capacity of the lines, i.e., the line load rate. Insufficient compensation for total reactive power leads to weak voltage support, exacerbates the deviation of the average voltage amplitude, and also affects the transmission efficiency and load rate of the lines. An excessively high line load rate reduces the voltage regulation margin and increases operational risks. Reserve capacity directly reflects the inherent buffering capacity of the virtual power grid zone in maintaining internal balance when dealing with the aforementioned power and voltage disturbances. By comprehensively analyzing the operational characteristic evaluation values of each virtual power grid zone, the ability of the virtual power grid zone to maintain stable and autonomous operation when facing internal fluctuations or external disturbances can be quantitatively assessed, thereby helping to improve the overall safety and operational efficiency of the urban energy intelligent management system.
[0050] Specifically, the predicted data feature values of each power grid virtual partition are obtained through processing. The specific processing procedure is as follows: the predicted data of each power grid virtual partition are collected synchronously. The predicted data of each power grid virtual partition includes the load prediction value and the power generation prediction value of each power grid virtual partition.
[0051] It should be noted that the load forecast values for each virtual power grid zone reflect the trend of active power demand changes in that virtual power zone over a specific future period. This data is calculated using machine deep learning algorithms based on historical load data, weather forecast information, date type, and real-time operating conditions, through a load forecasting system deployed within the zone. The power generation forecast values for each virtual power grid zone specifically refer to the expected active power output of distributed renewable energy sources within the zone, such as wind power and photovoltaics, over a specific future period. This data is calculated using a physical conversion algorithm based on numerical weather forecast data, such as irradiance, wind speed, and physical characteristic parameters of power generation units, through a renewable energy power generation forecasting system deployed within the zone.
[0052] Based on the prediction data of each power grid virtual partition, the prediction data feature values of each power grid virtual partition are obtained. The prediction data feature values of each power grid virtual partition represent the quantitative result of the prediction data of each power grid virtual partition on the operational risk status of the power grid virtual partition.
[0053] In this embodiment, the predicted data feature values of each power grid virtual partition can be obtained through the following analysis method, with the specific analysis conditions as follows: ; In the formula, This represents the predicted data feature value of the i-th virtual power grid partition. This represents the normalized load forecast value for the i-th virtual power grid partition. This represents the reference load forecast value for the i-th virtual power grid partition. This represents the weighting factor corresponding to the set load forecast value. This represents the normalized power generation prediction value for the i-th virtual power grid partition. This represents the reference power generation prediction value for the i-th virtual power grid partition. This represents the weighting factor corresponding to the set power generation forecast value, where i represents the number of each virtual power grid partition, i=1, 2, 3, ..., n, and n represents the total number of virtual power grid partitions.
[0054] It should be added that, in this embodiment, the weighting factors corresponding to the preset load forecast value and power generation forecast value are obtained from the urban energy management database. These weighting factors are used to adjust the importance of the data in the forecast data of each power grid virtual partition in the process of analyzing and obtaining the characteristic value of the forecast data.
[0055] In this implementation plan, the predictive data characteristic values of each virtual power grid partition are obtained through comprehensive analysis. This allows for a quantitative assessment of the operational risk level of the partition due to supply and demand mismatch in a specific future period, thereby helping to improve the forward-looking early warning capability and preventive control level of the urban energy intelligent management system.
[0056] Specifically, the dynamic carrying capacity index value of each power grid virtual partition is obtained through comprehensive processing. The specific processing process is as follows: based on the operation characteristic evaluation value and the predicted data feature value of each power grid virtual partition, the dynamic carrying capacity index value of each power grid virtual partition is obtained through comprehensive processing. The dynamic carrying capacity index value of each power grid virtual partition represents the quantitative result of the operation characteristic evaluation value and the predicted data feature value of each power grid virtual partition on maintaining autonomous operation capability.
[0057] In this embodiment, the dynamic carrying capacity index value of each power grid virtual zone can be obtained through the following analysis method, with the specific analysis conditions as follows: ; In the formula, This represents the dynamic carrying capacity index value of the i-th virtual power grid partition. This represents the operational characteristic evaluation value of the i-th virtual power grid partition. This represents the weighting coefficient corresponding to the set operational characteristic evaluation value. This represents the predicted data feature value of the i-th virtual power grid partition. The value represents the weighting coefficient corresponding to the set predictive data feature value, i represents the number of each power grid virtual partition, i=1,2,3,...,n, and n represents the total number of power grid virtual partitions.
[0058] It should be added that the weighting coefficients corresponding to the preset operation characteristic assessment values and predicted data feature values are obtained from the urban energy management database. These weighting coefficients are used to adjust the importance of the operation characteristic assessment values and predicted data feature values of each power grid virtual zone in the process of analyzing and obtaining the dynamic carrying capacity index value.
[0059] In this embodiment, the larger the operational characteristic evaluation value of each power grid virtual partition, the larger the dynamic carrying capacity index value obtained from the analysis; the smaller the predicted data feature value of each power grid virtual partition, the larger the dynamic carrying capacity index value obtained from the analysis. By synchronously collecting and processing the predicted data to obtain the corresponding predicted data feature value, and then combining it with the operational characteristic evaluation value to obtain the dynamic carrying capacity index value, it is possible to comprehensively consider the current operating status and future trends, more accurately quantify the ability of each power grid virtual partition to maintain autonomous operation, discover potential risks in advance, and provide data basis for subsequent decision-making.
[0060] Specifically, the operating mode of each power grid virtual partition is dynamically determined. The specific determination process is as follows: the dynamic carrying capacity index value of each power grid virtual partition is compared with the set dynamic carrying capacity index threshold. If the dynamic carrying capacity index value of a certain power grid virtual partition is higher than the set dynamic carrying capacity index threshold, the operating mode of the power grid virtual partition is marked as high carrying capacity autonomous mode, and the power grid virtual partition is recorded as a high carrying capacity power grid virtual partition. Otherwise, the operating mode of the power grid virtual partition is marked as low carrying capacity auxiliary mode, and the power grid virtual partition is recorded as a low carrying capacity power grid virtual partition.
[0061] In this embodiment, the operation mode is determined based on the dynamic carrying capacity index value, which can reflect the actual carrying capacity and operation status of each power grid virtual zone in real time, providing accurate data reference for subsequent targeted management measures and ensuring that different zones can operate in appropriate modes.
[0062] Specifically, granting high autonomy to virtual power grid zones marked as high-capacity autonomous mode involves the central coordinator sending an autonomy authorization command to the local controller of the virtual power grid zone marked as high-capacity autonomous mode.
[0063] It should be added that the central coordinator is a central control unit deployed in the city's intelligent energy management system, responsible for collecting global information, making calculations and decisions, and issuing coordination instructions. Its core function is to generate and issue coordination control instructions such as autonomous authorization or cross-regional support. The local controller is a distributed control unit deployed in each virtual power grid region, responsible for receiving and executing instructions from the central coordinator, while monitoring and regulating resources within the region. Its core function is to autonomously execute precise regulation of energy storage, distributed power sources, and other equipment within the region after obtaining authorization, so as to quickly eliminate local power deviations, maintain regional autonomous balance, and report the execution results.
[0064] After receiving the autonomous authorization command, the local controller of the high-capacity power grid virtual partition executes the charging and discharging scheduling of the energy storage system within the high-capacity power grid virtual partition and reports the key results to the central coordinator.
[0065] It needs to be explained that the specific process of executing the charging and discharging scheduling of the energy storage system within the high-capacity grid virtual zone is as follows: The total active power of the high-capacity grid virtual zone is extracted, and the difference between the total active power of the high-capacity grid virtual zone and a preset power balance target value, such as zero, is calculated and recorded as the total active power deviation value of the high-capacity grid virtual zone. When the absolute value of the total active power deviation value of the high-capacity grid virtual zone is higher than a preset power deviation action threshold, the internal scheduling of the energy storage system is activated. Based on the dynamic capacity index value of the high-capacity grid virtual zone, the adjustment coefficient of the high-capacity grid virtual zone is obtained. The product of the total active power deviation value of the high-capacity grid virtual zone and the adjustment coefficient is recorded as the demand power adjustment amount of the high-capacity grid virtual zone. The local controller performs charging and discharging scheduling based on the demand power adjustment amount of the high-capacity grid virtual zone.
[0066] It should be noted that the adjustment coefficient of the high-capacity power grid virtual zone is obtained by matching the dynamic carrying capacity index value of the high-capacity power grid virtual zone with the adjustment coefficients corresponding to each dynamic carrying capacity index value stored in the urban energy management database. The adjustment coefficient corresponding to the dynamic carrying capacity index value is then retrieved and recorded as the adjustment coefficient of the high-capacity power grid virtual zone. The dynamic carrying capacity index value directly quantifies the comprehensive ability of the zone to maintain stable and autonomous operation. The larger the index value, the healthier its current operation and the more abundant its internal adjustable resources. The more abundant the resources, the stronger the tolerance to future disturbances. Therefore, when such a partition detects internal power deviations and initiates local regulation, the system should allow it to use stronger regulation, i.e., a larger regulation coefficient, to make full use of its own good condition and abundant resources, quickly and thoroughly eliminate power imbalances, thereby consolidating its autonomy and preventing condition deterioration. Conversely, if the load-bearing capacity index value is low, a smaller regulation coefficient should be matched, and a more conservative and slow regulation strategy should be adopted to avoid causing excessive impact on its relatively fragile operating state. This correspondence ensures that the strength of the control strategy is accurately matched with the real-time capability of the partition itself.
[0067] It should be added that the local controller performs charging and discharging scheduling based on the demand power adjustment of the high-capacity grid virtual zone. Specifically, the local controller determines the sign of the demand power adjustment of the high-capacity grid virtual zone. If the demand power adjustment is negative, the local controller sends a charging command to the energy storage system within the high-capacity grid virtual zone. The charging power value included in the charging command is the smaller of the absolute value of the demand power adjustment and the maximum allowable charging power set by the energy storage system. If the demand power adjustment is positive, the local controller sends a discharging command to the energy storage system. The discharging power value included in the discharging command is the smaller of the absolute value of the demand power adjustment and the maximum allowable discharging power set by the energy storage system. Before executing the charging or discharging command, the local controller needs to obtain the current state of charge (SOC) value of the energy storage system. If the current SOC value is higher than or equal to a preset charging cutoff SOC threshold when executing the charging command, the charging command is canceled. If the current SOC value is lower than or equal to a preset discharging cutoff SOC threshold when executing the discharging command, the discharging command is canceled.
[0068] It should be noted that the current state of charge (SOC) of an energy storage system represents the percentage of the remaining usable energy of the system at the current moment relative to its nominal total energy capacity. This SOC is calculated as follows: First, the battery management system deployed inside the energy storage system monitors the remaining usable energy of the system in real time. Then, the remaining usable energy is divided by the nominal total energy capacity of the energy storage system, and the final value obtained is the current SOC of the energy storage system.
[0069] Specifically, the process of generating cross-regional support instructions is as follows: when a certain power grid virtual region is marked as low-capacity auxiliary mode, the central coordinator automatically triggers the global auxiliary mode.
[0070] After automatically triggering the global auxiliary mode, cross-zone support instructions are generated, including active power support instructions and reactive power support instructions.
[0071] Specifically, the process of adjusting the operational balance of the power grid virtual partitions is as follows: the central coordinator will issue cross-partition support instructions to the local controllers of the high-capacity power grid virtual partitions and the local controllers of the high-capacity power grid virtual partitions that are electrically adjacent to the low-capacity power grid virtual partitions.
[0072] Each local controller that receives the cross-regional support command drives the distributed power source and reactive power compensation equipment to perform the corresponding power output adjustment. For the active power support command, the preset active power support amount generated by the high-capacity grid virtual region is injected into the low-capacity grid virtual region through the designated electrical connection path.
[0073] It should be explained that the preset active power support amount generated by the high-capacity grid virtual zone is injected into the low-capacity grid virtual zone through a designated electrical connection path. Specifically, the local controller of the designated high-capacity grid virtual zone extracts the preset active power support amount, allocates the active power to the dispatchable distributed power sources within the high-capacity grid virtual zone, and issues a control command containing the specific active power support amount to the selected distributed power source to drive the distributed power source to increase its active power output. The active power support amount is then injected into the low-capacity grid virtual zone through the electrical connection path specified by the local controller.
[0074] Simultaneously, for reactive power support commands, a preset reactive power support amount provided by an electrically adjacent high-capacity grid virtual zone is injected into the low-capacity grid virtual zone. Specifically, the local controller of the high-capacity grid virtual zone, which is electrically adjacent to the low-capacity grid virtual zone, extracts the preset reactive power support amount and controls the reactive power compensation devices, such as static var generators or capacitor banks, on the boundary of the high-capacity grid virtual zone to adjust their output. The specific value of the adjusted output is the reactive power support amount. The local controller issues a control command containing the specific reactive power support amount to the designated reactive power compensation device, driving the reactive power compensation device to inject the corresponding reactive power into the grid, thereby injecting the reactive power support amount into the low-capacity grid virtual zone through the local controller.
[0075] In this embodiment, different measures are taken for partitions with different operating modes. Granting autonomous permissions to high-capacity partitions can give full play to their autonomous control capabilities. Low-capacity partitions trigger global auxiliary mode and generate cross-partition support instructions, which helps to solve the problem of partition power imbalance caused by factors such as distributed power fluctuations, avoid risks such as overload of interconnection lines and local voltage anomalies, and realize the global optimization and safe and stable operation of the urban energy system.
[0076] The second aspect of the present invention provides a deep learning-based urban energy intelligent management system, comprising: an operation evaluation module, used to divide the urban energy intelligent management power grid into virtual partitions based on the physical and electrical structure parameters of the power grid, deploy data acquisition units in each virtual partition, monitor and acquire the operation status parameters of each virtual partition, and analyze and obtain the operation characteristic evaluation value of each virtual partition.
[0077] The load-bearing capacity analysis module is used to synchronously collect the prediction data of each power grid virtual partition, process it to obtain the prediction data feature value of each power grid virtual partition, and comprehensively process the operation characteristic evaluation value and the prediction data feature value of each power grid virtual partition to obtain the dynamic load-bearing capacity index value of each power grid virtual partition.
[0078] The mode determination module is used to dynamically determine the operating mode of each power grid virtual partition based on the dynamic carrying capacity index value of each power grid virtual partition. The operating modes of each power grid virtual partition include a high carrying capacity autonomous mode and a low carrying capacity auxiliary mode.
[0079] The operation balancing module is used to grant high autonomy to virtual power grid partitions marked as high-capacity autonomous mode, automatically trigger global auxiliary mode for virtual power grid partitions marked as low-capacity auxiliary mode, and generate cross-partition support instructions to adjust the operation balance of virtual power grid partitions.
[0080] It should be noted that a deep learning-based urban energy intelligent management method and system also includes an urban energy management database, which stores a first parameter set, a second parameter set, a third parameter set, and a fourth parameter set obtained by analyzing historical data.
[0081] The first parameter set includes a preset distance threshold, a preset ratio, a preset ratio threshold for line transmission capacity, the absolute value of the maximum total active power within a historical statistical period, the absolute value of the maximum total reactive power within a historical statistical period, the rated voltage value, the maximum allowable voltage deviation value, the sum of the maximum adjustable capacities, the characteristic evaluation factor corresponding to the total active power value, the characteristic evaluation factor corresponding to the total reactive power value, the characteristic evaluation factor corresponding to the average voltage amplitude deviation value, the characteristic evaluation factor corresponding to the line load rate value, and the characteristic evaluation factor corresponding to the reserve capacity value.
[0082] The second parameter set includes the reference load forecast value for each virtual power grid partition, the reference power generation forecast value for each virtual power grid partition, the weight factor corresponding to the load forecast value, the weight factor corresponding to the power generation forecast value, the weight coefficient corresponding to the operating characteristic evaluation value, and the weight coefficient corresponding to the forecast data characteristic value.
[0083] The third parameter set includes dynamic bearing capacity index thresholds.
[0084] The fourth parameter set includes the power balance target value, power deviation action threshold, adjustment coefficients corresponding to each dynamic load capacity index value, the maximum allowable charging power set by the energy storage system, the maximum allowable discharging power set by the energy storage system, the charging cut-off state of charge threshold, the discharging cut-off state of charge threshold, active power support amount, and reactive power support amount.
[0085] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0086] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.
Claims
1. A deep learning-based intelligent urban energy management method, characterized in that, include: Based on the physical and electrical structure parameters of the urban energy smart management grid, each grid is divided into virtual zones. Data acquisition units are deployed in each virtual zone to monitor and acquire the operating status parameters of each virtual zone, and the operating characteristic evaluation values of each virtual zone are analyzed. Predictive data from each virtual power grid zone is collected synchronously, and the predictive data feature values of each virtual power grid zone are obtained through processing. Based on the operational characteristic evaluation values of each virtual power grid zone and the predictive data feature values of each virtual power grid zone, the dynamic carrying capacity index values of each virtual power grid zone are obtained through comprehensive processing. Based on the dynamic carrying capacity index value of each power grid virtual partition, the operating mode of each power grid virtual partition is dynamically determined. The operating modes of each power grid virtual partition include a high carrying capacity autonomous mode and a low carrying capacity auxiliary mode. Grant high autonomy to virtual power grid partitions marked as high-capacity autonomous mode, automatically trigger global auxiliary mode for virtual power grid partitions marked as low-capacity auxiliary mode, and generate cross-partition support commands to adjust the operational balance of virtual power grid partitions.
2. The urban energy intelligent management method based on deep learning according to claim 1, characterized in that: The physical and electrical structure parameters of the urban energy smart management power grid are divided, and the specific process is as follows: The physical and electrical structure parameters include the connection relationships of each line, the distribution relationships of substations, and the direction of power transmission; Based on the connection relationships of various lines, the distribution relationships of substations, and the direction of power transmission in the urban energy smart management power grid, the urban energy smart management power grid is divided into multiple regions, resulting in virtual partitions of each power grid.
3. The urban energy intelligent management method based on deep learning according to claim 1, characterized in that: The analysis yielded operational characteristic evaluation values for each virtual power grid partition. The specific analysis process is as follows: The operating status parameters of each virtual power grid partition are monitored and acquired. The operating status parameters of each virtual power grid partition include the total active power, total reactive power, average voltage amplitude, line load rate, and reserve capacity of each virtual power grid partition. The operating status parameters of each power grid virtual partition are analyzed to obtain the operating characteristic evaluation value of each power grid virtual partition. The operating characteristic evaluation value of each power grid virtual partition represents the quantitative result of the operating status parameters of each power grid virtual partition on the current operating health status of the power grid virtual partition.
4. The urban energy intelligent management method based on deep learning according to claim 1, characterized in that: The processing yields the predicted data feature values for each virtual power grid partition. The specific processing procedure is as follows: The prediction data of each virtual power grid zone is collected synchronously. The prediction data of each virtual power grid zone includes the load prediction value and the power generation prediction value of each virtual power grid zone. Based on the prediction data of each power grid virtual partition, the prediction data feature values of each power grid virtual partition are obtained. The prediction data feature values of each power grid virtual partition represent the quantitative result of the prediction data of each power grid virtual partition on the operational risk status of the power grid virtual partition.
5. The urban energy intelligent management method based on deep learning according to claim 1, characterized in that: The comprehensive processing yields the dynamic carrying capacity index values for each virtual power grid partition. The specific processing procedure is as follows: Based on the operational characteristic evaluation values and predicted data feature values of each virtual power grid partition, the dynamic carrying capacity index value of each virtual power grid partition is obtained through comprehensive processing. The dynamic carrying capacity index value of each virtual power grid partition represents the quantitative result of the operational characteristic evaluation values and predicted data feature values of each virtual power grid partition on maintaining autonomous operation capability.
6. The urban energy intelligent management method based on deep learning according to claim 5, characterized in that: The dynamic determination yields the operating mode of each power grid virtual partition. The specific determination process is as follows: The dynamic carrying capacity index value of each power grid virtual partition is compared with the set dynamic carrying capacity index threshold. If the dynamic carrying capacity index value of a certain power grid virtual partition is higher than the set dynamic carrying capacity index threshold, the operation mode of the power grid virtual partition is marked as high carrying capacity autonomous mode and the power grid virtual partition is recorded as high carrying capacity power grid virtual partition. Otherwise, the operation mode of the power grid virtual partition is marked as low carrying capacity auxiliary mode and the power grid virtual partition is recorded as low carrying capacity power grid virtual partition.
7. The urban energy intelligent management method based on deep learning according to claim 6, characterized in that: The specific process for granting high-level autonomy permissions to the virtual power grid partitions marked as high-capacity autonomous mode is as follows: The central coordinator sends autonomous authorization commands to the local controllers of the virtual power grid zones that are marked as high-capacity autonomous modes; After receiving the autonomous authorization command, the local controller of the high-capacity power grid virtual partition executes the charging and discharging scheduling of the energy storage system within the high-capacity power grid virtual partition and reports the key results to the central coordinator.
8. The urban energy intelligent management method based on deep learning according to claim 1, characterized in that: The specific process for generating cross-partition support instructions is as follows: When a virtual power grid partition is marked as low-capacity auxiliary mode, the central coordinator automatically triggers the global auxiliary mode. After automatically triggering the global auxiliary mode, cross-zone support instructions are generated, including active power support instructions and reactive power support instructions.
9. The urban energy intelligent management method based on deep learning according to claim 1, characterized in that: The specific process for adjusting the operational balance of the virtual power grid partition is as follows: The central coordinator will issue cross-zone support instructions to the local controllers of the high-capacity grid virtual zone and the local controllers of the high-capacity grid virtual zone that is electrically adjacent to the low-capacity grid virtual zone. Each local controller that receives the cross-zone support command drives the distributed power supply and reactive power compensation equipment to perform the corresponding power output adjustment. For the active power support command, the preset active power support amount generated by the high-capacity grid virtual zone is injected into the low-capacity grid virtual zone through the specified electrical connection path. Simultaneously, the reactive power support command causes the preset reactive power support amount provided by the electrically adjacent high-capacity grid virtual partition to be injected into the low-capacity grid virtual partition.
10. A deep learning-based intelligent urban energy management system, characterized in that, include: The operation evaluation module is used to divide the urban energy smart management grid into virtual partitions based on the physical and electrical structure parameters of the grid. Data acquisition units are deployed in each virtual partition to monitor and acquire the operation status parameters of each virtual partition and analyze the operation characteristic evaluation value of each virtual partition. The load-bearing capacity analysis module is used to synchronously collect the prediction data of each power grid virtual partition, process it to obtain the prediction data feature value of each power grid virtual partition, and comprehensively process the operation characteristic evaluation value and the prediction data feature value of each power grid virtual partition to obtain the dynamic load-bearing capacity index value of each power grid virtual partition. The mode determination module is used to dynamically determine the operating mode of each power grid virtual partition based on the dynamic carrying capacity index value of each power grid virtual partition. The operating modes of each power grid virtual partition include a high carrying capacity autonomous mode and a low carrying capacity auxiliary mode. The operation balancing module is used to grant high autonomy to virtual power grid partitions marked as high-capacity autonomous mode, automatically trigger global auxiliary mode for virtual power grid partitions marked as low-capacity auxiliary mode, and generate cross-partition support instructions to adjust the operation balance of virtual power grid partitions.