Automatic power regulation and control method for micro-grid
By building an integrated construction architecture and multi-source data fusion, combined with intelligent prediction and coordination controllers, the problems of new energy consumption and resource management in microgrids are solved, and efficient and reliable power supply and clean energy utilization are achieved.
Patent Information
- Application Number
- CN202511089761.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-05
- Publication Date
- 2025-10-14
- Estimated Expiration
- 2045-08-05
AI Technical Summary
Existing microgrids have problems such as insufficient capacity to absorb new energy, high difficulty in managing distributed resources, poor grid regulation flexibility, poor resource coordination and poor prediction accuracy, resulting in low energy utilization efficiency and insufficient power supply reliability.
Build an integrated construction architecture, achieve optimized coordinated control of sources, grids, loads and storage through multi-source data fusion and intelligent prediction, use coordinated controllers for real-time resource management and scheduling, combine energy storage systems and distributed power sources, optimize resource allocation and power supply strategies, and ensure frequency and voltage stability.
It has improved the capacity to absorb new energy, optimized distributed resource management, enhanced the flexibility of grid regulation and energy utilization efficiency, ensured power supply reliability and the utilization of clean energy, and achieved efficient and reliable power supply.
Smart Images

Figure CN120784858A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power regulation and resource management of microgrids, and in particular to a microgrid automatic power regulation method based on multi-source data fusion and intelligent prediction. Background Art
[0002] Under the global trend of actively responding to climate change and vigorously promoting energy transformation, many regions regard islands as important testing grounds for achieving large-scale application of renewable energy and reducing carbon emissions, and are committed to building them into "zero-carbon islands", "low-carbon islands" and even "negative carbon islands".
[0003] With the continued economic development and social progress, Island A and Island B have been designated as demonstration areas. Island B's load demand is experiencing rapid growth, and its load characteristics are undergoing significant changes. Currently, Island B's power grid consists of only two 35kV lines installed on the same tower. This structure severely inadequately controls transmission capacity, resulting in a weak overall grid structure and a purely AC network. While plans include improvements such as line voltage boosting and the construction of interconnecting channels, these projects require relatively long construction cycles. During this long transition period, grid reliability remains at significant risk.
[0004] Overall, current technology has obvious shortcomings in many aspects.
[0005] First, the capacity to absorb renewable energy is limited. In grid-connected mode, high levels of renewable energy generation can easily lead to power backflow and voltage exceeding upper limits. In off-grid mode, the resulting power is often difficult to absorb or insufficient to support the load. This leads to frequent wind and solar power curtailment, severely reducing energy efficiency and wasting energy resources.
[0006] Second, distributed resource management is quite difficult. Island B is home to a wide variety of distributed resources, each with its own unique communication methods and parameter configurations. Currently, there is no effective way to integrate and manage these independent distributed resources.
[0007] Third, grid control lacks flexibility. Faced with the complex and diverse operating scenarios of Island B, including different operating modes such as grid connection, off-grid operation, and black start, as well as different operating conditions such as normal, severe overload, fault, and maintenance, existing technologies have significant deficiencies in the adaptability and flexibility of control strategies.
[0008] Fourth, energy storage's synergy with other resources is poor. Energy storage has significant shortcomings in optimizing its coordination with new energy sources, traditional power sources, and loads. On a temporal scale, it's impossible to fully leverage the power generation and consumption characteristics and response speeds of different controllable resources to achieve precise charge and discharge control at different time stages—day-ahead, intraday, and in real time—making it difficult to maximize benefits. In terms of space, the zoning and layering coordination with other resources is incomplete, preventing energy storage from fully leveraging its key role in balancing local power and improving overall power supply reliability.
[0009] Fifth, forecast accuracy needs to be further improved. While artificial intelligence (AI) technology and related algorithms have been introduced for forecasting renewable energy power generation and load demand, current forecasting methods still struggle to achieve high levels of accuracy, as renewable energy generation is influenced by complex factors like weather, and load demand is constrained by numerous uncertainties like user behavior. Prediction errors can lead to inaccurate regulation strategies, impacting the economic and reliability of grid operations.
[0010] Therefore, there is a need for a microgrid automatic power control method that can improve the flexibility of microgrid control, enhance the absorption of new energy, optimize distributed resource management, and improve energy utilization efficiency. Summary of the Invention
[0011] In order to solve the defects of existing microgrid power control methods, such as poor new energy absorption capacity, high difficulty in distributed resource management, poor grid control flexibility, poor resource coordination effect, and poor power and load prediction accuracy, the present invention provides a microgrid automatic power control method that can improve microgrid control flexibility, enhance new energy absorption, optimize distributed resource management, and improve energy utilization efficiency.
[0012] The present invention provides a microgrid automatic power control method, comprising the following steps: S1. Build an integrated construction architecture to achieve switching and coordinated control of power grids A and B in grid-connected and off-grid modes; S2. Collect, process, and analyze multi-source data, integrate multi-source data through the data center, and obtain historical and real-time data; S3. Based on the historical data and real-time data, intelligently predict the power generation power of the new energy power generation equipment and the load demand of the energy storage system on a time scale; S4. In grid-connected mode, the integrated system is coordinated and dispatched according to the optimal target of the entire network. The B microgrid management system formulates the source-grid-load-storage dispatch plan based on the power generation forecast and energy storage load forecast data; S5. In off-grid mode, power supply management is carried out step by step according to power supply and load conditions and user priority to ensure frequency and voltage stability; S6. In grid-connected and off-grid modes, power allocation, load regulation, resource optimization configuration and scheduling are carried out according to load demand and power generation conditions, and distributed resources are controlled in real time through a coordination controller.
[0013] Further: in S6, the specific adjustment method for dynamically adjusting resources in time and space dimensions is: regulating resources on a time scale based on the power generation and consumption characteristics and response speed of different controllable resources, and making dynamic adjustments according to the time scale and resource status during the regulation process; in a spatial dimension, following the principles of local, nearby, and same voltage level balance and absorption to achieve partitioned and layered coordination of resources; wherein, in the grid-connected mode, optimized coordination of source, grid, load and storage is achieved based on the entire network, and in the off-grid mode, the B microgrid management system acts as a microgrid management system to independently control the resources on the island.
[0014] Furthermore, in S6, the real-time control of distributed resources by the coordination controller specifically includes the following steps: S61. During the load control process, optimal decisions are made based on load characteristics and grid status, guiding load adjustments at different time periods to assist in power balance of the microgrid. S62. Allocate power based on the power generation status of new energy power generation equipment and the load demand of the energy storage system, tap into the grid's support capacity for flexible and adjustable loads, and dynamically adjust resources in time and space based on the power generation and consumption characteristics of adjustable resources to achieve regional coordination and optimal allocation of resources. The dynamic resource adjustment includes energy storage and hydrogen production during periods of high renewable energy production, and charging of energy storage using large grid power during periods of low load. S63, the coordination controller performs coordinated control of distributed resources in off-grid mode through GOOSE communication, and automatically adjusts the frequency and voltage of the microgrid to ensure quasi-synchronous operation of the power grid; S64. The energy management system uses different methods to interact with different types of distributed resources and converts them into a unified method to interact with the upper-level dispatcher. According to the dynamic topology of the large power grid, independent distributed resource individuals are aggregated into virtual resource aggregates to realize the monitoring and management of distributed resources, and the resource aggregation and scheduling strategies are optimized in the distributed resource management process.
[0015] Further: in S63, the distributed resource coordinated control includes grid-connected power coordinated control, off-grid voltage and frequency control, off-grid power coordinated control, grid-connected and off-grid switching control, energy storage economic operation control, and black start control processing.
[0016] Furthermore, in S63, the switching and control process during the on-grid to off-grid and off-grid to on-grid transition specifically includes the following steps: S631, Initialization Phase: Set the initial control parameters of the adaptive control algorithm in the microgrid coordination controller and establish a reference model to describe the ideal values of the microgrid in different operating modes: in the grid-connected mode, the voltage and frequency in the reference model are consistent with those of the main grid; in the off-grid mode, the reference model is based on the local load demand and maintains a stable operating voltage and frequency range and power distribution method; The initial control parameters include the initial charge and discharge power coefficient of the energy storage system, the initial output power ratio of the distributed power supply, and the initial gain of frequency and voltage regulation; S632, monitoring and data acquisition phase: The microgrid coordination controller obtains the operating data of the microgrid and the large power grid in real time; and simultaneously obtains the status information of the distributed resources within the microgrid; S633, error calculation stage: the real-time monitored microgrid data is compared with the ideal value in the reference model to calculate the corresponding error; for power coordination control, the deviation between the actual power and the reference power is calculated, and the error is used to reflect the degree of deviation between the current operating state of the microgrid and the ideal state; S634, adjustment phase: automatically adjusting control parameters based on the calculated error and a pre-set algorithm. The algorithm dynamically changes the control parameters based on the error. In off-grid mode, the voltage error is positive, i.e., the measured voltage is higher than the reference voltage; the frequency error is also positive, i.e., the measured frequency is higher than the reference frequency. S635, Function Implementation Phase: When the microgrid coordination controller performs control functions, the control parameters adjusted by the algorithm are used to implement grid-connected power coordination control, off-grid voltage and frequency control, and off-grid power coordination control functions; in off-grid voltage and frequency control, the charging and discharging power of the energy storage and the output power of the distributed power supply are adjusted to stabilize the voltage and frequency of the microgrid within the range specified by the reference model; during the on-grid and off-grid switching control process, the algorithm dynamically adjusts the control parameters according to the data monitored during the switching process, thereby realizing switching and control during the on-grid to off-grid and off-grid to on-grid processes.
[0017] Furthermore, in S64, the optimization of resource aggregation and scheduling strategy specifically includes the following steps: S641. In the initial stage, the virtual "explorer" representing the resource allocation experiment will randomly attempt to connect various resource paths and leave a corresponding degree of "imprint" on the path based on the resource allocation effect produced by each attempt. The degree of the "imprint" left on the path is calculated based on the resource allocation effect. S642, over time and through repeated exploration, continuously optimize path selection and "imprint" level updates, resulting in continuous improvement of resource aggregation and scheduling strategies; S643. Effectively aggregate the dispersed distributed resource individuals into a virtual resource aggregate based on the real-time dynamic topology of the power grid.
[0018] Furthermore: in S3, the intelligent prediction specifically includes the following steps: S31, preprocessing the historical data and real-time data to build an algorithm network model; S32, dividing the pre-processed data into a training data group and a test data group to train the algorithm network model; S33. Update the model weights with the help of backpropagation and iterate repeatedly until the model converges.
[0019] Furthermore: in S4, the integrated system for dispatching in the grid-connected mode specifically includes the following steps: S41. Construct a microgrid model and use the power generation forecast of new energy power generation equipment and the load forecast data of the energy storage system to predict the power trend in the future period. Taking into account the optimal operation target of the entire network and the power forecast within the microgrid, solve an optimization problem in each control cycle to determine the optimal set value sequence of the interconnection line exchange power in the future preset time period, so as to achieve the operation target while satisfying the preset constraints. S42. Plan the charging and discharging power of energy storage at different times in advance based on power forecast information. Based on the predicted peak and valley periods of renewable energy power generation, charge the energy storage when there is excess power generation and discharge the energy storage when there is insufficient power generation or load peaks. At the same time, continuously optimize the scheduling plan, recalculate and adjust the interconnection line exchange power target and energy storage scheduling strategy to ensure that the microgrid always operates in the direction of maximizing green energy consumption and stabilizing frequency and voltage.
[0020] Further: in S5, the power supply management for the user in the off-grid mode specifically includes the following steps: S51: Balance basic power, frequency, and voltage. Grid-connected energy storage and diesel-generated power equipment are key control targets. Based on the microgrid's real-time power supply and demand, output power is automatically adjusted to maintain the microgrid's frequency and voltage within a preset range. S52. Power supply management is performed based on user priorities. Users within the microgrid are classified according to their priorities, and users are divided into important users and multi-level priority users. When the power output power of the power supply meets the power needs of important users, priority is given to ensuring continuous power supply to important users. When the power output power of the power supply still has surplus power after meeting the power needs of important users, power supply to users of different priorities is gradually restored according to a pre-set priority sequence. The power supply management is to monitor and control the power supply equipment and load of each user through the microgrid coordination controller.
[0021] Furthermore, in S51, the specific steps of automatically adjusting the output power are as follows: if a downward trend in the frequency of the microgrid is detected, indicating insufficient power generation, the diesel-generating power supply equipment will rapidly increase its output and the grid-connected energy storage will also release power in a timely manner according to preset rules to support the frequency to return to normal levels; conversely, if the frequency increases, the power generation output will be reduced accordingly.
[0022] The beneficial effects of the present invention are: The present invention sets Island A and Island B as the construction targets of relevant demonstration areas, and constructs an energy supply system that can ensure both efficient power supply and reliable and clean power supply. The system is based on the integrated architecture of the power grids of Island A and Island B. It is a comprehensive power supply and demand control platform composed of a flexible switching mechanism between grid connection and off-grid, a data middle platform, a microgrid coordination controller and multi-source collaborative control, aiming to adapt to the needs of the era of energy transformation and low-carbon development.
[0023] The microgrid automatic power control method described in the present invention can accurately respond to the volatility of renewable energy power generation by effectively coordinating intelligent prediction and coordinated control strategies in the absorption of new energy, greatly improve the absorption capacity, and reduce the wind and solar power curtailment rate; it can improve energy utilization efficiency, promote the development of clean energy, and assist in the optimization and transformation of the energy structure; in terms of energy storage, the optimized charging and discharging strategies and coordinated control not only extend the life of energy storage, but also enhance its effectiveness in balancing power, providing backup power and black start capabilities, thereby ensuring the stable operation of the microgrid; the flexibility of grid control is significantly enhanced, and it can adapt to a variety of modes and working conditions, ensuring smooth switching and stable control between grid and off-grid, and effectively ensuring power supply reliability; distributed resource management realizes unified monitoring and optimized configuration, fully taps load potential, assists grid power balance and improves overall economic efficiency; multi-source data fusion provides an accurate basis for control and enhances the scientific nature of decision-making; the platform has good scalability, compatibility and efficiency, supports the platform function expansion and algorithm iterative upgrade of the integrated coordination system and microgrid management system, facilitates integration and update, adapts to the development needs of microgrids, and thus effectively promotes energy transformation and sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1 It is the framework diagram of the integrated dispatching system; Figure 2 yes Figure 1 Framework diagram of the platform supported by the integrated dispatching system in the B microgrid and the B microgrid management system; Figure 3 It is a flow chart; Figure 4 This is a simplified diagram of the microgrid controller; Figure 5 It is the microgrid controller connection diagram; Figure 6is a microgrid controller function diagram; Figure 7 is a microgrid operation schematic diagram. DETAILED DESCRIPTION
[0025] The following merely illustrates the preferred embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. The following examples are used to explain the present application, but should not be interpreted as limiting the present application, the protection scope of the present application should be subject to the protection scope of the claims. The embodiments of the present application are described in detail below, in order to facilitate the description of the present application and simplify the description, the technical terms used in the specification of the present application should be interpreted broadly, including but not limited to the conventional replacement schemes not mentioned in the present application, and including direct implementation and indirect implementation.
[0026] Embodiment 1 Combined Figure 1 Figure 7 The present embodiment is disclosed, a microgrid automatic power regulation method based on multi-source data fusion and intelligent prediction, comprising the following steps: Reference to the attached Figure 1 And Figure 2 , an integrated construction architecture is constructed, and the regulation foundation is laid: emphasis is placed on creating a flexible, efficient power grid architecture. By constructing the integrated dual-active construction architecture of A and B power grids, and the unified construction scheme of III area, the flexible switching and collaborative operation between grid-connected and off-grid modes are realized. In the grid-connected mode, the depth optimization and coordination of source, grid, load and storage are carried out from the perspective of the whole grid, so as to promote the symbiosis of various elements and improve the overall efficiency; in the off-grid state, the B microgrid management system as a microgrid management system can independently control and manage the island resources, ensure the autonomy and stability of power supply, and lay a solid foundation for subsequent accurate regulation.
[0027] Integrate diverse data to provide decision-making basis: Data is collected extensively from various specialized systems (such as EMS (Energy Management System), DMS (Distribution Management System), AMI (Advanced Metering Infrastructure), and PMS (Power Production Management System), integrating meteorological, power generation, and load information to ensure data richness and accuracy. The data collection and processing processes ensure data accuracy and timeliness. At the same time, the data center plays a key role, integrating multi-source data to achieve unified data management and sharing. It supports the interaction of data including but not limited to distributed resource information, user information, and maintenance plan information to assist in power control decisions. The data center has data preprocessing capabilities to provide a standard data format for subsequent algorithms. The integration and management of this massive and complex data not only covers detailed information on distributed resources and users, but also incorporates important data such as maintenance plans. After preprocessing, the data is converted into a standard format suitable for subsequent complex algorithms, providing comprehensive and accurate data support for power control decisions.
[0028] Leveraging intelligent forecasting to gain insight into power grid trends: Advanced artificial intelligence technology leverages to fully exploit the underlying patterns in historical and real-time data. The algorithm builds a network model based on carefully prepared data (incorporating factors such as meteorology). Through rigorous training (data grouping, loss function selection, and multiple iterations until convergence), the algorithm achieves precise, intelligent forecasts of renewable energy generation power and load demand at various timescales (day-ahead, intraday, and real-time). These forecasts inform the development of power regulation strategies, with algorithm development, training, and optimization to improve forecast accuracy. This provides critical, forward-looking information for timely and effective regulation strategy development.
[0029] The specific process is as follows: First, in the data preparation phase, historical data and real-time data are comprehensively collected. The data covers but is not limited to key information such as renewable energy power generation and load demand, and is integrated with meteorological data. Preprocessing work is carried out on these collected data, and they are adjusted by means including but not limited to normalization and standardization. Suppose the historical data sequence of renewable energy power generation is ; in, Represents a power adjustment factor, t is the current time, n is the length of historical data, is the original renewable energy power generation power, and the load demand historical data sequence is , where m is the length of historical data, d represents the load change rate, Represents the load demand, and at the same time collects meteorological data series and preprocesses the data, using the normalization formula:
[0030] in, is the normalized renewable energy power generation power, and its value is between [0,1], which is convenient for use in algorithms (such as machine learning, control optimization, etc.). is the original new energy power generation capacity; is the maximum power generation in historical data; is the minimum power generation in historical data.
[0031] The value of n represents the length of the historical time window used to predict renewable energy power. It is usually set according to the requirements of the prediction model and the sampling period. For example, if the sampling frequency is 15 minutes, n=96 means reviewing the data of the previous 24 hours.
[0032] The value of m: represents the length of the historical data sequence used to predict load demand. It is usually consistent with n, but can also be set separately according to the load change characteristics. For example, m=48 represents the past 12 hours.
[0033] The meaning of t is the current prediction time point, usually the current sampling time, which can be dynamically advanced by the scheduling period (such as 5 minutes, 15 minutes).
[0034] Similar processing is performed on load demand and meteorological data. The data is then organized into time series according to different time scales, including the time steps corresponding to the day-ahead, intraday, and real-time. Real-time prediction is set at intervals based on actual demand. Then we enter the model construction phase and start to build the network model of the algorithm. According to the characteristic dimension of the input data, the number of input layer neural units is determined, and the number of hidden layer neural units is set in combination with the complexity of the problem and the amount of data. Then, the number of output layer neurons is set according to the prediction target, and the time step and sequence length adapted to different time scales are set to lay the foundation for subsequent accurate predictions. Suppose the characteristic dimension of the input data is k, including but not limited to the characteristics of renewable energy power generation, load demand, and meteorological data, then the number of input layer neural units is set to k; based on the complexity of the problem and the amount of data, the number of hidden layer neural units is set to h; the prediction targets are renewable energy power generation and load demand, and the number of output layer neurons is set to 2, corresponding to the predicted values of power generation and load demand, respectively; set the adaptive time step and sequence length for different time scales, represents the day-ahead step length, represents the intraday step length, represents the real-time step size, represents the length of the day-ahead sequence, represents the length of the intraday sequence, Indicates the real-time sequence length; Then the preprocessed data is divided into training data group and test data group according to a certain ratio: most of the data is used as the training data group, and the rest is used as the test data group; if the proportion of training data is α, then the number of training data groups is , Q is the total number of data, the number of test data groups is On this basis, we select but not limited to the mean square error loss function, input the training data into the algorithm network model, and set the input data as , the weight of the network model is , the bias is , then the forward propagation formula is , f is the activation function; calculate the loss value , is the predicted output vector, is the true value vector, and the mean square error MSE is selected as the loss function:
[0035] p represents the number of samples, i represents the current index, from 1 to p, used to traverse all data points; The model weights are updated with the help of backpropagation and iterated repeatedly until the model converges.
[0036] Reference Attachment Figure 3 , formulating precise strategies to ensure stable operation: Optimized coordination in grid-connected mode: The integrated dispatching system determines the interconnection line exchange power based on the optimal grid-wide target. The B microgrid management system plans a reasonable SOC level for energy storage based on renewable energy and load forecasts and develops a detailed on-island source-grid-load-storage scheduling plan. This process builds a microgrid model and utilizes algorithms to comprehensively consider multiple factors (such as equipment characteristics and power forecasts) to achieve an optimal balance between power generation costs and green energy consumption while meeting constraints. Simultaneously, plans are continuously optimized based on new data to ensure stable and efficient operation of the microgrid upon grid connection, continuously developing towards maximizing green energy utilization and stabilizing frequency and voltage.
[0037] The specific implementation is as follows: construct a model of the microgrid, which covers but is not limited to the operating characteristics and mutual relationships of new energy power generation equipment, energy storage systems, loads and interconnection lines. Based on this model, the algorithm uses the forecast data of new energy and loads to predict the power trend in the future period; for the determination of the interconnection line exchange power target, the algorithm comprehensively considers the optimal operation target of the entire network and the power forecast within the microgrid, and solves an optimization problem in each control cycle to determine the optimal set value sequence of the interconnection line exchange power in the future period, so as to achieve the lowest power generation cost and the maximum green energy consumption under the constraints including but not limited to the interconnection line power limit and equipment operation limit. The optimal operation target of the entire network; in terms of determining the SOC level and formulating the island's source-grid-load-storage scheduling plan, the algorithm also plans the charging and discharging power of energy storage at different times in advance based on power forecast information, and arranges energy storage charging to improve when there is excess power generation according to the predicted peak and off-peak periods of renewable energy power generation. When there is insufficient power generation or load peak, the energy storage is discharged to meet the load demand and track the superior scheduling instructions; at the same time, this scheduling plan is continuously optimized. As time goes by and new forecast data is obtained, the interconnection line exchange power target and energy storage scheduling strategy are recalculated and adjusted to ensure that the microgrid always operates in the direction of maximizing green energy consumption and stabilizing frequency and voltage, and realizing efficient and reliable grid-connected power regulation.
[0038] Reference Attachment Figure 4 - Figure 7 Emergency support in off-grid mode: Algorithms are used to combine grid-connected energy storage, diesel generators, and other equipment based on actual power and load conditions to prioritize power supply to critical users. Power output is adjusted in real time to ensure basic power balance and frequency and voltage stability. Load management is prioritized, with precise monitoring and control provided by the microgrid coordination controller. This ensures reliable power supply and efficient utilization of green energy for all island loads. Whether facing sudden off-grid situations or long-term independent operation, the microgrid ensures stable and reliable power service for users.
[0039] Figure 5 In Chinese, EMS: Energy Management System, energy management system; DMS: Distribution Management System, distribution management system; AMI: Advanced Metering Infrastructure, advanced metering system; PMS: Power Management System, power production management system; SOC: State of Charge, energy storage charge state; MGCC (Microgrid Central Controller); MGLC (Microgrid Local Controller); PQ control (Power–Reactive Power Control); VF control (Voltage–Frequency Control); Specifically: the first step of control focuses on basic power balance and frequency and voltage stability, and will include but not be limited to grid-type energy storage and diesel-generated main power supply equipment as key control objects, and use their rapid power regulation to automatically adjust the output power according to the real-time power supply and demand of the microgrid to maintain the frequency and voltage of the microgrid within a stable range: if the frequency of the microgrid is detected to be declining, indicating insufficient power generation, the diesel generator will quickly increase its output and the grid-type energy storage will also release electricity in time according to the preset rules to support the frequency to return to normal levels; conversely, if the frequency increases, the power generation output will be reduced accordingly; the core of the second step of control is orderly power supply management based on load priority, which classifies various loads in the microgrid in detail and clearly divides important users and low-priority users. users; when the power output meets the electricity needs of important users, priority is given to ensuring continuous power supply to important users including but not limited to hospitals, critical infrastructure, and important industrial production, to ensure that these users with extremely high requirements for power reliability are not affected by power fluctuations; when there is still surplus power after meeting the needs of important users, the power supply to low-priority loads is gradually restored according to the pre-set priority sequence. The entire process is carried out through the microgrid coordination controller to accurately monitor and control each power supply equipment and load, so that the microgrid can give priority to guaranteeing key power needs in the off-grid state, and meet the electricity needs of more users to the greatest extent when conditions permit, maintain the stability of the microgrid frequency and voltage to achieve reliable power supply to the entire island load and effective utilization of green energy, providing a solid guarantee for the stable and efficient operation of the off-grid microgrid.
[0040] Synergistic multi-resource, improve power supply efficiency: Focus on the resource coordination optimization between new energy and various types of power sources and loads. Through intelligent algorithms, the power supply power is dynamically allocated according to the new energy output and load demand, effectively suppressing new energy fluctuations, enhancing power supply reliability, and achieving optimal coordination between power sources. At the same time, the potential of flexible controllable loads is deeply tapped, and demand response is used to guide rational electricity consumption, so that it can provide strong support for microgrid power balance at different times. In this process, the algorithm makes intelligent decisions according to the load characteristics and real-time state of the grid to ensure that the power supply efficiency of the entire microgrid automatic power regulation system (including power sources, energy storage, loads, distribution networks, and control systems within the microgrid) is significantly improved, and resource utilization is more reasonable and efficient.
[0041] The resource coordination optimization step specifically includes but is not limited to the coordination optimization between wind, light, biomass power generation type new energy and energy storage, diesel generators and other power sources. According to the new energy output and load demand, the same algorithm as the S4 control strategy formulation step is used to reasonably allocate power supply power, suppress new energy fluctuations, improve power supply reliability, and achieve optimal coordination between power sources. The grid support capacity of flexible controllable loads including but not limited to charging piles, V2G, shore power, agricultural production, air conditioning, etc. is tapped, and demand response and other means are used to guide rational electricity consumption of loads at different times to assist microgrid power balance. In the process of load regulation, the algorithm makes optimal decisions according to the load characteristics and grid state.
[0042] Temporal and spatial coordination regulation to achieve optimal resource allocation: Dynamic adjustment in time dimension: Fully consider the power generation and consumption characteristics and response speed of different resources, and flexibly regulate resources in day-ahead, intra-day, real-time, etc. time scale. For example, store energy or hydrogen when new energy is generated, and use the large grid to charge the energy storage when the load is low. The algorithm optimally adjusts in real time according to the time change and resource state to ensure reasonable allocation of resources on the time axis and maximize benefits.
[0043] The temporal and spatial coordination regulation step specifically includes but is not limited to the reasonable regulation of resources in day-ahead, intra-day, real-time time scale based on the power generation and consumption characteristics and response speed of different controllable resources to maximize benefits, including but not limited to storing energy, hydrogen production during new energy generation period, and using large grid power to charge energy storage during load low period. During the regulation process, the algorithm dynamically adjusts according to the time scale and resource state using the same algorithm as the S4 control strategy formulation step. In the spatial dimension, the principle of local, nearby, same voltage level balance and consumption is followed to achieve the partition, hierarchical coordination of resources, including the hierarchical aggregation of adjustable capacity, hierarchical decomposition of regulation target, and mutual aid between partitions in the same layer. The new energy generation period usually concentrates in the noon to afternoon period, and the specific period varies due to regional policy differences.
[0044] Hierarchical coordination in spatial dimension: follow the principle of on-site, near, same voltage level balance and accommodation to achieve resource in high voltage transmission layer, medium voltage distribution network layer and low voltage area layer partition, hierarchical coordination. Through the hierarchical aggregation of adjustable capacity, hierarchical decomposition of adjustment target, and mutual aid of hierarchical partition autonomy, the complexity of microgrid control system is reduced, the adaptability between different levels and the stability of overall control strategy are improved, so as to realize the optimal configuration of resources in space and ensure the efficient and stable operation of microgrid.
[0045] Strengthen microgrid coordination and ensure stable operation: relying on the master-slave configuration and high-speed communication capability of microgrid coordination controller, the high-speed real-time coordination of distributed resources in off-grid mode is realized. In the initialization stage, the key control parameters are set and the reference model is established to provide a basic framework for subsequent regulation. Real-time monitoring of power grid operation data and distributed resource state calculation error and dynamic adjustment of control parameters according to algorithm rules make the actual operation of microgrid approach the ideal state. In terms of function implementation, by adjusting energy storage, distributed power and other devices, various control functions (grid-connected / off-grid power coordination, voltage frequency control, energy storage economic operation, black start, etc.) are realized to ensure the stable operation of microgrid in various operating conditions. Whether it is normal operation or switching process, it can smoothly transition to provide a solid guarantee for the reliable operation of microgrid.
[0046] The microgrid coordination control step is specifically: based on the microgrid coordination controller, the master-slave configuration is adopted, and the GOOSE high-speed communication is adopted between the master controller and the slave controller to realize the high-speed real-time coordination control of regional distributed resources in off-grid mode. It has processing functions including but not limited to grid-connected power coordination control, off-grid voltage frequency control, off-grid power coordination control, grid-connected / off-grid switching control, energy storage economic operation control, black start control, etc. The microgrid coordination controller works cooperatively with the overall regulation algorithm in executing control functions and accepts the decision instructions of the algorithm and executes the corresponding operations. According to the voltage and frequency monitoring of microgrid and power grid, the microgrid coordination controller automatically adjusts the frequency and voltage on the microgrid side to meet the quasi-synchronization requirements, realizes smooth switching and stable control in the process of grid-connected to off-grid and off-grid to grid-connected, and adjusts the control parameters in the switching process according to the monitoring data to make the switching process stable and reliable. Specifically: S7-1 initialization stage: set the initial control parameters of the adaptive control algorithm in the microgrid coordination controller. These parameters include but are not limited to the initial charge and discharge power coefficient of the energy storage system, the initial output power ratio of the distributed power supply, and the initial gain of frequency and voltage adjustment. At the same time, a reference model is established, which describes the ideal voltage, frequency and power state of the microgrid in different operating modes: in grid-connected mode, the voltage and frequency in the reference model remain consistent with the power grid; in off-grid mode, the reference model is based on local load demand and maintains a stable voltage and frequency range and power distribution mode; S7-2 Monitoring and Data Acquisition Phase: The microgrid coordination controller acquires key operating data, including but not limited to voltage and frequency, of the microgrid and the larger grid in real time. It also acquires status information on distributed resources within the microgrid, including but not limited to the SOC of energy storage and the output power of distributed power sources. This data serves as input to the algorithm for subsequent calculations and decision-making. S7-3 Error calculation stage: The real-time monitored microgrid voltage, frequency, and power data are compared with the ideal values in the reference model to calculate the corresponding errors. For power coordination control, the deviation between the actual power and the reference power is calculated. The error reflects the degree of deviation between the current operating state of the microgrid and the ideal state, which is an important basis for the algorithm to adjust the control parameters. S7-4 Adjustment Phase: Automatically adjust control parameters based on the calculated error and pre-set algorithmic rules. The algorithmic rule is a rule that dynamically changes control parameters based on the error. In off-grid mode, if the voltage error is positive (i.e., the measured voltage is higher than the reference voltage) and the frequency error is positive (i.e., the measured frequency is higher than the reference frequency), the algorithm reduces the charging power of the energy storage and the output power of the distributed generation to lower the voltage and frequency. By continuously adjusting the control parameters based on the error, the actual operating state of the microgrid approaches the reference model. S7-5 function realization stage: when the microgrid coordination controller performs the control function, the control parameters adjusted by the algorithm are used to realize functions including but not limited to grid-connected power coordination control, off-grid voltage and frequency control, and off-grid power coordination control; in the off-grid voltage and frequency control, the charging and discharging power of the energy storage and the output power of the distributed power supply are adjusted to stabilize the voltage and frequency of the microgrid within the range specified by the reference model; in the on-grid and off-grid switching control process, the algorithm dynamically adjusts the control parameters according to the data monitored during the switching process, so as to achieve smooth switching and stable control in the process of grid-connected to off-grid and off-grid to grid-connected; for the economic operation control of energy storage, the algorithm dynamically adjusts the charging and discharging strategy of energy storage according to the SOC of energy storage, electricity price information and power demand of the microgrid to maximize economic benefits while meeting the operation requirements of the microgrid; in the black start control, the algorithm can gradually adjust the power output of the starting equipment according to the state changes during the recovery process of the microgrid, so as to ensure the smooth progress of the black start process.
[0047] Optimizing distributed resource management and contributing to grid stability: The microgrid's automatic power control system uses differentiated interaction methods for different distributed resources and uniformly interfaces with higher-level dispatchers. It aggregates resources into virtual aggregates based on the grid's dynamic topology, enabling efficient monitoring and management. It utilizes an innovative algorithm that simulates the collaboration of biological communities to optimize resource aggregation and scheduling strategies. This algorithm achieves continuous improvement in resource allocation through repeated attempts by virtual "explorers" on resource connection paths and "imprint-based" path selection optimization. It also simplifies resource device modeling, improves overall operational economics, ensures distributed resources stably serve the grid, and assists in power control. The algorithm flexibly optimizes management strategies based on resource characteristics and grid demand, providing strong support for the stable operation of the microgrid.
[0048] The distributed resource management steps are specifically as follows: the microgrid automatic power control system interacts with different types of distributed resources using different methods and converts them into a unified method to interact with the upper-level dispatcher; based on the dynamic topology of the power grid, independent distributed resource individuals are aggregated into virtual resource aggregates to achieve effective monitoring and management of distributed resources; and an algorithm is used to optimize resource aggregation and scheduling strategies during the distributed resource management process. The algorithm is as follows: the algorithm simulates the group collaboration mode of social organisms in nature when exploring survival resources. These organisms leave a special "imprint" wherever they pass in the process of searching for survival resources. Subsequent similar organisms will be influenced by this "imprint" when moving, and are more inclined to move along paths with dense "imprints"; mapping this model to the field of distributed resource management, the connection links between distributed resources are like the movement trajectories of these organisms, and key indicators including but not limited to the efficiency, cost, and stability of resource transmission are abstracted as the significance of this "imprint"; in the initial stage, the virtual "explorer" representing the resource allocation attempt will randomly try on each resource connection path and leave a corresponding degree of "imprint" on the path based on the resource allocation effect produced by each attempt; let the distributed resource set be , m is the number of distributed resources, and the set of connection links between resources is ), including but not limited to the vector composed of key indicators of resource transmission efficiency, cost, and stability , define the “imprint” significance function = The virtual "explorer" k makes random attempts on the resource connection path, and assumes that the selected path , l is the path length, and the degree of “imprint” left on the path is calculated based on the resource allocation effect = , h is a function that calculates the degree of imprinting based on the resource allocation effect; in, is the i-th resource in the distributed resource set R; Represents a resource and The connection path between them; i is the index, indicating that the calculation is performed on the i-th sample; j is the index, indicating that the calculation is performed on the j-th sample; is the jth resource in the distributed resource set R; It is a resource and The transmission efficiency between them involves data transmission rate, bandwidth, etc. It is a resource and The transmission costs between them, such as energy consumption, fees, etc.; It is a resource and Stability between them, such as reliability index or packet loss rate; Resource connection path The prominence of the “mark”; A vector containing multiple indicators, defined as , used to comprehensively evaluate the performance of connection paths; Represents a power adjustment factor, which determines the power allocation priority; represents the input resource of a scheduling node; k is the number of trials, that is, the number of times a resource path is explored; A function that calculates the degree of imprinting based on the effect of resource allocation.
[0049] As time goes by and exploration is repeated many times, subsequent "explorers" are guided by the "imprints" accumulated in the early stage, and the probability of choosing a path increases. It is related to the prominence of the “mark” on the path; The total significance of the "marks" on , then select the path The probability formula is , is the set of all possible paths; when the explorer selects a path Then, the degree of "imprint" on the path is updated according to the resource allocation effect. = Continuously repeat the above exploration process to continuously optimize path selection and "imprint" level updates, so that resource aggregation and scheduling strategies are continuously improved; Finally, according to the real-time dynamic topology of the power grid, the dispersed distributed resource individuals are effectively aggregated into virtual resource aggregates; let the virtual resource aggregate set be , q is the number of aggregates, through specific aggregation rules , is an aggregate function, For the optimized path; aggregate related resources together to achieve accurate monitoring and reasonable management of distributed resources, providing strong guarantees for the scientific utilization of distributed resources and the stable operation of the power grid; simplify the modeling of distributed resource equipment through intelligent methods, shield its internal detailed characteristics, optimize distributed resource management, improve the overall operation economy, enable distributed resources to provide continuous and stable services to the power grid and assist in microgrid power regulation, and the algorithm optimizes the management strategy according to the characteristics of distributed resources and power grid needs.
[0050] Example 2 This embodiment is described in conjunction with Example 1. This embodiment discloses a microgrid automatic power control method based on multi-source data fusion and intelligent prediction. The grid-connected control during the generation of new energy specifically includes the following steps: Step 1. Data fusion and utilization: The microgrid automatic power control system collects real-time data on Island B's photovoltaic power generation, wind power generation, and load (including real-time power for various loads, such as residential and commercial electricity consumption), as well as non-real-time data (such as historical load curves), through systems such as EMS, DMS, AMI, and PMS. It also integrates meteorological data (such as light intensity, wind speed, and temperature). The data center integrates this data, for example, linking the location and capacity information of distributed photovoltaic power stations with their real-time power generation data. This information is combined with user information (such as electricity usage habits of users in different regions, peak and valley periods), and maintenance plans (such as the recent maintenance schedule for a transmission line). This provides a comprehensive and accurate data foundation for subsequent control, ensuring data accuracy and timeliness. Preprocessing converts the data into a format suitable for subsequent algorithm analysis.
[0051] Step 2. Smart Prediction Based on collected historical data (such as renewable energy power generation and load demand data for the past month) and real-time data, artificial intelligence algorithms (such as the long short-term memory (LSTM) network, a deep learning algorithm) are used to perform day-ahead, intraday, and real-time forecasts of renewable energy power generation and load demand. During the data preparation phase, key information, including renewable energy power generation and load demand, is comprehensively collected and integrated with meteorological data. The data is normalized and standardized, and organized into time series data at the day-ahead (24-hour, 1-hour time step), intraday (4-hour, 15-minute time step), and real-time (a number of time intervals after the current moment). When constructing the network model, the number of neural units in the input layer is determined based on the feature dimensions of the input data (such as the number of features in meteorological data and historical power generation data). The number of neural units in the hidden layer is determined based on the complexity of the problem (such as the complexity of Island B's power grid and the renewable energy access situation) and the amount of data. The number of neurons in the output layer is set to correspond to the forecasting objective (such as predicting renewable energy power generation and load demand at each future time step). The training data set (e.g., 80% of the data is used for training) and the test data set (20% of the data is used for testing) are divided. The mean squared error (MSE) is selected as the loss function. The training data is fed into the algorithm network model. The predicted output is obtained through forward propagation, and the loss value is calculated. The model weights are then updated using backpropagation, and the iterations are repeated until the model converges. Finally, real-time renewable energy power generation, load demand, and meteorological data are fed into the trained model to predict power generation and load demand for future periods. For example, it is predicted that renewable energy power generation will gradually increase in the morning and peak at noon.
[0052] Step 3. Regulatory strategy formulation In grid-connected mode, the integrated dispatching system determines the interconnection line exchange power target based on the optimal operational objectives of the entire grid (e.g., achieving the lowest power generation cost and maximizing green energy consumption while ensuring safe and stable grid operation). It also considers forecasts for renewable energy generation, allowing for some renewable energy generation to be transmitted externally. The B microgrid management system determines the SOC level for energy storage based on renewable energy and load forecasts. During peak renewable energy generation periods (e.g., 10:00 AM to 2:00 PM), it schedules energy storage charging or hydrogen production based on predicted power generation and load demand, pre-planning the charging and discharging power of energy storage at different times. For example, if renewable energy generation is predicted to exceed local load demand at 11:00 AM, energy storage will be charged at a certain power level starting at 11:00 AM. Simultaneously, a plan for source-grid-load-storage scheduling within the island is developed, tracking higher-level dispatch requirements. Using optimization algorithms, power is allocated rationally, increasing local renewable energy consumption (e.g., allocating excess renewable energy power to adjustable industrial loads) while reducing the output of traditional power sources (e.g., diesel generators), maximizing green energy consumption and stabilizing frequency and voltage.
[0053] Step 4. Resource Collaborative Optimization During periods of high renewable energy generation, renewable energy sources such as wind and solar power, along with energy storage and diesel generators, operate in a coordinated and optimized manner. Based on renewable energy output (such as the real-time power of wind and photovoltaic power generation) and load demand (such as the actual power requirements of various load types), an optimization algorithm is used to rationally allocate power to smooth fluctuations in renewable energy output. For example, when photovoltaic power suddenly increases, the algorithm controls the energy storage charging power to increase accordingly and the diesel generator power to decrease appropriately to maintain power balance in the microgrid. Furthermore, the grid's support capacity for flexible and controllable loads (such as charging stations and air conditioners) is leveraged, and demand response measures are used to guide loads to use electricity more efficiently during periods of high renewable energy generation. For example, control signals are sent to charging stations to encourage them to charge more during periods of high renewable energy generation. For air conditioner loads, temperature setpoints or operating hours are adjusted to assist in microgrid power balance.
[0054] Step 5. Spatiotemporal coordinated regulation On a temporal scale, based on the generation and consumption characteristics and response speeds of different controllable resources, day-ahead scheduling plans energy storage charging and adjustable load power usage according to predicted renewable energy generation. Intraday scheduling is further optimized and adjusted based on real-time monitoring of renewable energy generation and load changes. Real-time control ensures accurate execution of control strategies based on actual conditions within each time step. For example, during periods of high renewable energy generation, if real-time monitoring reveals that photovoltaic power generation exceeds expectations, energy storage charging power and load power usage strategies can be promptly adjusted. Spatially, adhering to the principles of local, nearby, and voltage-level balance and consumption, local photovoltaic power generation is prioritized for nearby loads. Excess power is stored in local energy storage equipment or transmitted to neighboring areas via appropriate transmission lines for consumption. For example, power generated by a distributed photovoltaic power station in a certain area of an island is prioritized for residential and commercial loads in that area. Any surplus power is transmitted via regional distribution lines to energy storage stations in neighboring areas for storage or directly supplied to the loads in that area. This achieves resource partitioning, hierarchical coordination, hierarchical aggregation of adjustable capacity, hierarchical decomposition of regulation targets, and intra-level autonomy and mutual assistance among regions.
[0055] Step 6. Microgrid coordinated control The microgrid coordination controller adopts a master-slave configuration, with GOOSE high-speed communication between the master and slave controllers. It monitors key operating data such as voltage and frequency of the microgrid and the main grid in real time, as well as status information such as the energy storage system's state of charge (SOC) and the output power of distributed generation (DGs). During the initialization phase, the adaptive control algorithm's initial control parameters are set, such as the initial charge and discharge power coefficient of the energy storage system, the initial output power ratio of the DGs, and the initial gains for frequency and voltage regulation. A reference model is also established to describe the ideal voltage, frequency, and power state of the microgrid in grid-connected mode (consistent with the main grid). During the monitoring and data acquisition phase, real-time operating data is continuously acquired. During the error calculation phase, the real-time monitored microgrid voltage, frequency, and power data are compared with the ideal values in the reference model to calculate the error. For power coordination control, the deviation between the actual power and the reference power is calculated. During the adjustment phase, the control parameters are automatically adjusted based on the calculated error and pre-set algorithm rules (for example, when the voltage and frequency errors are positive, the energy storage charging power is reduced and the DG output power is reduced), so that the actual operating state of the microgrid approaches the reference model. During the function realization stage, the adjusted control parameters are used to realize grid-connected power coordination control, off-grid voltage and frequency control and other functions, ensuring that the power exchange between the microgrid and the large grid is reasonable when renewable energy is generated, and the voltage and frequency in the microgrid are stable within the specified range. For example, the voltage of the microgrid is stabilized within the rated voltage range of ±5%, and the frequency is stabilized at 50Hz±0.2Hz.
[0056] Step 7. Distributed resource management The microgrid's automatic power control system interacts with different types of distributed resources (such as distributed photovoltaic power plants and decentralized wind farms) using different methods. For example, it communicates with distributed photovoltaic power plants via dedicated communication lines to obtain information such as their power generation and operating status, and sends control commands (such as adjusting power output). This system then converts these communications into a unified method for interaction with higher-level dispatchers. Based on the dynamic grid topology (such as changes in the access points of distributed power sources or the online addition of new distributed resources), it uses optimization algorithms to aggregate independent distributed resources into virtual resource aggregates. For example, multiple distributed photovoltaic power plants with similar geographical locations and power generation characteristics can be aggregated into a single virtual power source for unified management. This enables effective monitoring and management of distributed resources, optimizes resource aggregation and scheduling strategies, improves overall operational economics, ensures that distributed resources provide continuous and stable services to the grid, and assists in microgrid power control. For example, the output power of the virtual resource aggregate can be adjusted based on the microgrid's power balancing needs.
[0057] Example 3 This embodiment is described in conjunction with Example 1. This embodiment discloses a microgrid automatic power control method based on multi-source data fusion and intelligent prediction. The off-grid control when the load is peak and the output of new energy is insufficient specifically includes the following steps: 1. Data fusion and utilization The microgrid's automatic power control system utilizes various data sources, focusing on collecting load demand data during peak load periods (such as 7:00 PM to 10:00 PM) (including power demands for various loads, such as peak residential electricity consumption and concentrated commercial electricity consumption periods) and data on low renewable energy power generation, integrating meteorological data (such as the impact of low wind speeds at night on wind power generation). The data center integrates data, for example, linking load data from different regions with information about distributed resources in that region. It also combines user information (such as the distribution and power priority of important users) and maintenance plans (such as the recent maintenance schedule for a certain energy storage device) to provide accurate data support for off-grid control, ensure data accuracy and timeliness, and perform preprocessing to adapt to subsequent algorithm processing.
[0058] 2. Intelligent Prediction Based on data from historical peak load periods (such as the past week's nighttime peak load data), real-time monitored load growth trends, and renewable energy power generation forecasts (such as nighttime wind and photovoltaic power generation based on weather forecasts and current renewable energy equipment status), algorithms are used to generate accurate forecasts. During the data preparation phase, key information, including load demand and renewable energy power generation, is collected and integrated with meteorological data. Normalization and standardization are performed, and the data is organized into time series data at a time scale suitable for peak load forecasting (such as predicting load demand changes for the current nighttime period with a 30-minute time step). When constructing the network model, the number of neural units in the input layer is determined based on the characteristic dimensions of the input data (such as the number of features in historical load data and meteorological data). The number of neural units in the hidden layer is set based on the complexity of the problem (such as the complex electricity consumption during the nighttime peak load period on Island B) and the amount of data. The number of neurons in the output layer is set to correspond to the forecasting objective (such as predicting load demand and renewable energy power generation at each time step during the peak load period). The training and test data sets are divided, and an appropriate loss function (such as mean absolute error) is selected. The training data is fed into the model for training. Forward propagation, loss calculations, and backpropagation to update model weights are repeated, and the model is iterated until convergence. Finally, real-time load demand, renewable energy generation power, and meteorological data are fed into the trained model to predict power shortages during peak load periods.
[0059] 3. Regulatory strategy formulation In off-grid mode, based on power and load conditions, an algorithm-based, coordinated control of grid-connected energy storage, diesel generators, and other equipment prioritizes power supply to critical users, then gradually restores power to lower-priority loads. For example, this involves first ensuring continuous power supply to critical users like hospitals and critical infrastructure, and then gradually restoring lower-priority loads like commercial loads based on remaining power capacity. Before peak loads arrive, a suitable discharge plan is developed for the energy storage based on forecast information, such as starting the energy storage at a high power level at 6:30 pm to cope with the upcoming peak load. Furthermore, based on insufficient renewable energy generation, the startup and output of the diesel generator are appropriately scheduled to ensure power supply reliability. For example, if it is predicted that renewable energy generation will not be able to meet the power needs of critical users, the diesel generator is started in advance and its output is adjusted.
[0060] 4. Resource Collaborative Optimization When renewable energy generation is insufficient, the diesel generator starts up and adjusts its output based on load demand, while energy storage discharges to fill the power shortfall. The algorithm optimizes power distribution based on load characteristics (such as the importance and adjustability of different load types) and grid conditions. For example, power is prioritized for critical, non-interruptible loads (such as hospital equipment), while demand is reduced for adjustable loads (such as some commercial air conditioners) by controlling their operating conditions. Furthermore, the algorithm explores the potential of flexible and adjustable loads, such as by sending control signals to air conditioners to adjust their temperature setpoints or operating hours, reducing power consumption, assisting in microgrid power balance, and improving power supply reliability.
[0061] 5. Spatiotemporal Coordinated Control On a temporal scale, day-ahead dispatching plans strategies for nighttime peak loads, such as energy storage discharge plans and diesel generator backup arrangements. Intraday dispatching is further optimized based on real-time load fluctuations and renewable energy generation. Real-time control ensures timely adjustments to resource allocation during peak load periods. For example, during peak load periods, if real-time monitoring reveals that load demand exceeds expectations, diesel generator output can be promptly increased or energy storage discharge strategies adjusted. Spatially, following the principles of locality and proximity, local energy storage and nearby available diesel generators are prioritized to meet load demand, achieving regional autonomy. If local resources are insufficient, support is obtained from other regions through hierarchical coordination (e.g., resource sharing between adjacent regions through a microgrid coordination controller) to ensure balanced and efficient power consumption at the same voltage level. For example, during peak load in a particular region, energy storage and diesel generators are prioritized. If this still fails to meet demand, support is obtained from adjacent microgrids.
[0062] 6. Microgrid coordinated control The microgrid coordination controller monitors key operating data such as the microgrid's voltage and frequency, as well as status information such as the energy storage's state of charge (SOC) and the diesel generator's output in real time. During the initialization phase, control parameters suitable for off-grid operation are set, such as the energy storage system's charge and discharge power coefficient and the diesel generator's initial output setting. The monitoring and data acquisition phase continuously acquires real-time data. During the error calculation phase, the real-time monitored microgrid voltage, frequency, and power data are compared with the ideal values in the reference model (the voltage, frequency range, and power distribution method that maintains stable operation based on local load demand) to calculate the error. During the adjustment phase, control parameters are automatically adjusted based on the error and algorithmic rules. For example, if the measured voltage falls below the reference voltage and the frequency shows a downward trend, the energy storage discharge power or the diesel generator output is increased. During the functional implementation phase, off-grid voltage and frequency control and power coordination control are implemented by controlling the diesel generator, energy storage, and other equipment. This ensures stable operation of the microgrid during peak loads and when renewable energy output is insufficient. The voltage is stabilized within ±7% of the rated voltage and the frequency is stabilized within 50Hz ± 0.5Hz, guaranteeing reliable power supply to critical users.
[0063] 7. Distributed Resource Management The microgrid's automatic power control system interacts with distributed resources, obtains their status information, and manages them uniformly. For example, it monitors the remaining power of distributed energy storage and the operating status of distributed power sources in real time. It then uses algorithms to optimize the scheduling of distributed resources based on the grid's dynamic topology (e.g., distributed resources in certain areas may become critical support during peak load periods) and load demand. Distributed resources are aggregated into virtual resource aggregates, such as decentralized small-scale energy storage devices and distributed power sources. This improves the efficiency of monitoring and management of distributed resources, ensuring they can effectively assist the microgrid in power control during peak load periods and provide stable services to the grid. For example, the output power of the virtual resource aggregate is adjusted according to load demand to prioritize the power supply needs of critical loads.
[0064] Example 2 This embodiment is described in conjunction with Example 1. This embodiment discloses a microgrid automatic power control method based on multi-source data fusion and intelligent prediction, which specifically includes the following steps to cope with unplanned grid disconnection caused by extreme weather: 1. Data fusion and utilization When extreme weather (such as strong typhoons, heavy rain, etc.) occurs, the micro-grid automatic power regulation system quickly collects grid operation data, including the current load demand (such as the load that may suddenly increase or decrease due to weather), new energy power generation (such as strong winds that may affect wind power generation, heavy rain that may affect photovoltaic power generation), energy storage status (such as the remaining capacity of energy storage, charging and discharging status, etc.), combined with meteorological data (such as real-time wind speed, rainfall, air pressure, etc.) and previously collected B island grid infrastructure information (such as the wind resistance of transmission lines, the stability of towers, etc.). The data center integrates these data, for example, the load data of the area affected by the weather is associated with the distributed resources and the status of the transmission line in that area, combined with user information (such as user distribution and important user identification) and maintenance plans (such as scheduled preventive maintenance of equipment before extreme weather), to provide key data support for regulation during unplanned off-grid, ensuring the accuracy and timeliness of the data, and quickly preprocessing to adapt to the algorithm processing requirements in emergency situations.
[0065] 2. Intelligent prediction Based on historical data during extreme weather (such as grid operation data under similar weather conditions in the past) and real-time monitoring, the algorithm is used to predict the load demand changes and new energy power generation during unplanned off-grid. In the data preparation stage, key information including load demand, new energy power generation, etc. is collected and integrated with meteorological data, and fast normalization and standardization processing is performed, and time series data is arranged according to the time scale suitable for extreme weather conditions (such as predicting the grid operation in the next few hours with a 15-minute time step). When building a network model, the number of input layer neurons is determined according to the feature dimension of the input data (such as the number of features of meteorological data and grid operation data), the number of hidden layer neurons is set according to the complexity of the problem (such as the complex impact of extreme weather on the grid) and the amount of data, and the number of output layer neurons is set to correspond to the prediction target (such as predicting the load demand and new energy power generation at each time step during unplanned off-grid). Divide the training data set and the test data set, select the appropriate loss function (such as root mean square error), input the training data into the model for fast training (may use a simplified training algorithm or fine-tune the previously trained model), and update the model weights through forward propagation, loss calculation, and back propagation to make the model converge as soon as possible. Finally, input the real-time load demand, new energy power generation, and meteorological data into the trained model to predict the power balance during unplanned off-grid, for example, to predict whether the remaining grid can meet the load demand after some transmission lines are damaged by strong winds, and how much support new energy power generation can provide, etc.
[0066] 3. Regulation strategy development After an unplanned grid outage occurs, the integrated dispatching system rapidly adjusts the operating mode of distributed resources. For example, it adjusts the output settings of distributed power sources based on predicted renewable energy generation power and load demand (e.g., reducing unstable renewable energy output to prevent impact on the grid). Based on the current operating mode, the B microgrid management system rapidly adjusts protection settings and pressure plate status to ensure safe and stable grid operation during unplanned grid outages. Furthermore, based on power supply and load conditions, it prioritizes power supply to critical users, such as hospitals and emergency command centers, and formulates emergency load shedding plans, such as shutting off non-critical commercial loads to ensure power supply to critical loads. Microgrid voltage and frequency are established using grid-connected energy storage and diesel generators. Energy storage charging and discharging strategies are developed based on forecast information. For example, when renewable energy generation is unstable, energy storage charging and discharging are rationally controlled to maintain power balance in the microgrid and gradually restore power to low-priority loads. Simultaneously, the microgrid coordination controller achieves optimized coordinated control of the source, grid, load, and storage.
[0067] 4. Resource Collaborative Optimization During unplanned grid outages, renewable energy sources such as wind and solar power, along with energy storage and diesel generators, work together to address extreme weather challenges. Renewable energy sources generate power based on weather fluctuations (for example, wind power generation may fluctuate during unstable wind speeds). Energy storage rapidly adjusts charging and discharging based on load demand and renewable energy generation. Diesel generators serve as backup power, kicking in promptly when renewable energy generation is insufficient and storage capacity is low. An algorithm optimizes power allocation based on real-time resource status (such as remaining energy storage capacity and fluctuations in renewable energy generation power) and load demand (such as the power requirements of key users and remaining loads). This ensures stable grid operation during unplanned grid outages caused by extreme weather, mitigates the impact of renewable energy fluctuations on the automatic power control system of microgrids during off-grid conditions, and improves power supply reliability. For example, when wind power generation suddenly increases, the algorithm controls energy storage charging power to increase, while diesel generator output is appropriately reduced. When wind power generation decreases sharply, energy storage discharges to fill the power shortfall, while diesel generator output rapidly increases. At the same time, the grid support capacity of flexible and adjustable loads (such as air conditioners, some industrial loads, etc.) is explored, and demand response measures are used to guide the loads to use electricity reasonably in extreme weather conditions. For example, control signals are sent to air conditioners to adjust their temperature setting values or operating modes, reduce power consumption, and assist in microgrid power balance.
[0068] 5. Spatiotemporal Coordinated Control On a temporal scale, due to the uncertainty of extreme weather, real-time control after unplanned grid outages becomes critical. Algorithms rapidly adjust control strategies based on real-time monitored resource status and load demand. For example, they reassess renewable energy generation and load demand every 15 minutes, dynamically adjusting the charging and discharging power of energy storage and the output of diesel generators. In terms of space, adhering to the principles of locality and proximity, distributed resources that are undamaged or less impacted (such as local distributed energy storage and functional distributed power sources) are prioritized to meet the needs of critical local loads, achieving regional autonomy. For example, if a transmission line in a certain area is damaged by extreme weather, the distributed power sources and energy storage in that area form an independent power supply unit, prioritizing power supply to critical loads such as hospitals and emergency shelters. If local resources cannot meet all demand, the microgrid coordination controller coordinates resource sharing among neighboring areas to ensure voltage balance and absorption, and draws power from adjacent microgrids.
[0069] 6. Microgrid coordinated control The microgrid coordination controller plays a central role during unplanned grid outages. Its master-slave configuration uses GOOSE high-speed communication between the master and slave controllers to monitor key microgrid operating data such as voltage and frequency in real time, as well as information such as the energy storage system's state of charge (SOC), the output power of distributed generation (DGs), and the operating status of the diesel generators. During the initialization phase, control parameters suitable for unplanned grid outages are quickly set, such as the emergency charge and discharge power coefficient of the energy storage system, the rapid start-up of the DGs, and the maximum power output setting. The monitoring and data acquisition phase continuously and rapidly acquires real-time data. During the error calculation phase, the real-time monitored microgrid voltage, frequency, and power data are compared with the ideal values in the reference model (the voltage, frequency range, and power distribution method based on local critical load demand and stable operation in extreme weather conditions) to calculate the error. During the adjustment phase, control parameters are automatically adjusted based on the error and pre-defined emergency algorithm rules (for example, when the voltage drops sharply and the frequency becomes unstable, immediately start the DGs at maximum power output and adjust the energy storage discharge strategy), so that the actual operating status of the microgrid quickly approaches the reference model. During the function realization stage, by controlling diesel generators, energy storage, distributed power sources and other equipment, off-grid voltage and frequency control, power coordination control and other functions are realized to ensure the stable operation of the microgrid when extreme weather causes unplanned grid disconnection, stabilize the voltage as close to the rated voltage as possible (such as ±10% of the rated voltage), and stabilize the frequency within an acceptable range (such as 50Hz±1Hz), thus ensuring reliable power supply to important users.
[0070] 7. Distributed Resource Management The microgrid automatic power regulation system in off-grid state interacts with distributed resources in an emergency mode, ensuring that their status information can still be obtained and effectively managed under extreme weather conditions. For example, by using backup communication channels (such as wireless communication), the system maintains contact with distributed power sources and energy storage devices, obtains their operating status (such as whether they are malfunctioning, how much power is left, etc.), and sends emergency control instructions (such as adjusting power output, switching operating modes, etc.). According to the dynamic topology of the power grid (such as situations where some lines trip due to extreme weather, or distributed resources operate in island mode), the system uses optimization algorithms to aggregate distributed resources into virtual resource aggregates, improving the efficiency of monitoring and managing distributed resources. For example, by aggregating islanded distributed power sources and energy storage in a certain area into a virtual power source due to transmission line failures, the system can uniformly regulate their output power and prioritize the power supply needs of important loads in the region, ensuring that distributed resources can continue to provide stable services to the grid under extreme weather conditions and assist the microgrid in achieving power balance and stable operation. At the same time, by using intelligent methods to simplify the modeling of distributed resource devices and mask their internal details, the system optimizes the management of distributed resources and improves the overall economic efficiency of operation, such as quickly assessing the availability and reliability of distributed resources under extreme weather conditions and reasonably arranging their participation in the power regulation of the microgrid.
Claims
1. A microgrid automatic power control method, characterized in that: The steps include: S1. Build an integrated construction architecture to achieve switching and coordinated control of power grids A and B in grid-connected and off-grid modes; S2. Collect, process, and analyze multi-source data, integrate multi-source data through the data center, and obtain historical and real-time data; S3. Based on the historical data and real-time data, intelligently predict the power generation power of the new energy power generation equipment and the load demand of the energy storage system on a time scale; S4. In grid-connected mode, the integrated system is coordinated and dispatched according to the optimal target of the entire network. The B microgrid management system formulates the source-grid-load-storage dispatch plan based on the power generation forecast and energy storage load forecast data; S5. In off-grid mode, power supply management is carried out step by step according to power supply and load conditions and user priority to ensure frequency and voltage stability; S6. In grid-connected and off-grid modes, power allocation, load regulation, resource optimization configuration and scheduling are carried out according to load demand and power generation conditions, and distributed resources are controlled in real time through a coordination controller.
2. A microgrid automatic power control method according to claim 1, characterized in that: In S6, the specific adjustment method for dynamically adjusting resources in the time and space dimensions is: regulating resources on the time scale based on the power generation and consumption characteristics and response speed of different controllable resources, and making dynamic adjustments according to the time scale and resource status during the regulation process; in the spatial dimension, following the principles of local, nearby, and same voltage level balance and absorption to achieve partitioned and layered coordination of resources; wherein, in the grid-connected mode, the optimized coordination of source, grid, load and storage is achieved based on the entire network, and in the off-grid mode, the B microgrid management system acts as a microgrid management system to independently control the resources on the island.
3. A microgrid automatic power control method according to claim 2, characterized in that: In S6, the real-time control of distributed resources by the coordination controller specifically includes the following steps: S61. During the load control process, optimal decisions are made based on load characteristics and grid status, guiding load adjustments at different time periods to assist in power balance of the microgrid. S62. Allocate power based on the power generation status of new energy power generation equipment and the load demand of the energy storage system, tap into the grid's support capacity for flexible and adjustable loads, and dynamically adjust resources in time and space based on the power generation and consumption characteristics of adjustable resources to achieve regional coordination and optimal allocation of resources. The dynamic resource adjustment includes energy storage and hydrogen production during periods of high renewable energy production, and charging of energy storage using large grid power during periods of low load. S63, the coordination controller performs coordinated control of distributed resources in off-grid mode through GOOSE communication, and automatically adjusts the frequency and voltage of the microgrid to ensure quasi-synchronous operation of the power grid; S64. The energy management system uses different methods to interact with different types of distributed resources and converts them into a unified method to interact with the upper-level dispatcher. According to the dynamic topology of the large power grid, independent distributed resource individuals are aggregated into virtual resource aggregates to realize the monitoring and management of distributed resources, and the resource aggregation and scheduling strategies are optimized in the distributed resource management process.
4. A microgrid automatic power control method according to claim 3, characterized in that: In S63, the distributed resource coordinated control includes grid-connected power coordinated control, off-grid voltage and frequency control, off-grid power coordinated control, grid-connected and off-grid switching control, energy storage economic operation control, and black start control processing.
5. A microgrid automatic power control method according to claim 3, characterized in that: In S63, the switching and control process during the on-grid to off-grid and off-grid to on-grid transition specifically includes the following steps: S631, Initialization Phase: Set the initial control parameters of the adaptive control algorithm in the microgrid coordination controller and establish a reference model to describe the ideal values of the microgrid in different operating modes: in the grid-connected mode, the voltage and frequency in the reference model are consistent with those of the main grid; in the off-grid mode, the reference model is based on the local load demand and maintains a stable operating voltage and frequency range and power distribution method; The initial control parameters include the initial charge and discharge power coefficient of the energy storage system, the initial output power ratio of the distributed power supply, and the initial gain of frequency and voltage regulation; S632, monitoring and data acquisition phase: The microgrid coordination controller obtains the operating data of the microgrid and the large power grid in real time; and simultaneously obtains the status information of the distributed resources within the microgrid; S633, error calculation stage: the real-time monitored microgrid data is compared with the ideal value in the reference model to calculate the corresponding error; for power coordination control, the deviation between the actual power and the reference power is calculated, and the error is used to reflect the degree of deviation between the current operating state of the microgrid and the ideal state; S634, adjustment phase: automatically adjusting control parameters based on the calculated error and a pre-set algorithm. The algorithm dynamically changes the control parameters based on the error. In off-grid mode, the voltage error is positive, i.e., the measured voltage is higher than the reference voltage; the frequency error is also positive, i.e., the measured frequency is higher than the reference frequency. S635, Function Implementation Phase: When the microgrid coordination controller performs control functions, the control parameters adjusted by the adaptive control algorithm are used to implement grid-connected power coordination control, off-grid voltage and frequency control, and off-grid power coordination control functions; in off-grid voltage and frequency control, the charging and discharging power of the energy storage and the output power of the distributed power supply are adjusted to stabilize the voltage and frequency of the microgrid within the range specified by the reference model; in the on-grid and off-grid switching control process, the algorithm dynamically adjusts the control parameters according to the data monitored during the switching process, so as to realize switching and control during the on-grid to off-grid and off-grid to on-grid processes.
6. A microgrid automatic power control method according to claim 3, characterized in that: In S64, the optimization of resource aggregation and scheduling strategy specifically includes the following steps: S641. In the initial stage, the virtual "explorer" representing the resource allocation experiment will randomly attempt to connect various resource paths and leave a corresponding degree of "imprint" on the path based on the resource allocation effect produced by each attempt. The degree of the "imprint" left on the path is calculated based on the resource allocation effect. S642: As time goes by and exploration is repeated, path selection and imprinting are continuously optimized, leading to continuous improvement of resource aggregation and scheduling strategies. S643. Effectively aggregate the dispersed distributed resource individuals into a virtual resource aggregate based on the real-time dynamic topology of the power grid.
7. A microgrid automatic power control method according to claim 1, characterized in that: In S3, the intelligent prediction specifically includes the following steps: S31, preprocessing the historical data and real-time data to build an algorithm network model; S32, dividing the pre-processed data into a training data group and a test data group to train the algorithm network model; S33. Update the weights of the algorithm network model with the help of back propagation, and iterate repeatedly until the algorithm network model converges.
8. A microgrid automatic power control method according to claim 1, characterized in that: In S4, the integrated system for dispatching in grid-connected mode specifically includes the following steps: S41. Construct a microgrid model and use the power generation forecast of new energy power generation equipment and the load forecast data of the energy storage system to predict the power trend in the future period. Taking into account the optimal operation target of the entire network and the power forecast within the microgrid, solve an optimization problem in each control cycle to determine the optimal set value sequence of the interconnection line exchange power in the future preset time period, so as to achieve the operation target while satisfying the preset constraints. S42. Plan the charging and discharging power of energy storage at different times in advance based on power forecast information. Based on the predicted peak and valley periods of renewable energy power generation, charge the energy storage when there is excess power generation and discharge the energy storage when there is insufficient power generation or load peaks. At the same time, continuously optimize the scheduling plan, recalculate and adjust the interconnection line exchange power target and energy storage scheduling strategy to ensure that the microgrid always operates in the direction of maximizing green energy consumption and stabilizing frequency and voltage.
9. A microgrid automatic power control method according to claim 1, characterized in that: In S5, the power supply management for the user in the off-grid mode specifically includes the following steps: S51: Balance basic power, frequency, and voltage. Grid-connected energy storage and diesel-generated power equipment are key control targets. Based on the real-time power supply and demand of the microgrid, output power is automatically adjusted to maintain the frequency and voltage of the microgrid within a preset range. S52. Power supply management is performed based on user priorities. Users in the microgrid are classified according to their priorities, and users are divided into important users and multi-level priority users. When the power output power of the power supply meets the power demand of important users, priority is given to ensuring continuous power supply to important users. When the power output power is sufficient to meet the needs of important users and there is still power left, power supply to users of different priorities will be gradually restored according to the pre-set priority sequence; The power supply management is to monitor and control the power supply equipment and load of each user through the microgrid coordination controller.
10. A microgrid automatic power control method according to claim 9, characterized in that: In S51, the specific steps of automatically adjusting the output power are as follows: if a downward trend in the frequency of the microgrid is detected, indicating insufficient power generation, the diesel-generating power supply equipment will rapidly increase its output and the grid-connected energy storage will also release power in a timely manner according to preset rules to support the frequency to return to normal levels; conversely, if the frequency increases, the power generation output will be reduced accordingly.
Citation Information
Patent Citations
Real-time management control method of energy of microgrid
CN106992544A
Micro-grid source network charge coordination control method based on energy optimal dispatching
CN109004641A
Source-grid-load-storage integrated micro-grid multi-time-scale energy management optimization method
CN114884108A
Microgrid coordination control system based on hydrogen energy storage
CN116633022A
Micro-grid scheduling method and system
CN118763748A
Cited By
Micro-grid adaptive off-grid coordination control system and method based on multiple grid-connected points
CN120978860A
Staged dynamic balance control method for realizing independent supply guarantee of micro-grid
CN121417203A
Phase-based dynamic balance control method for realizing independent power supply of microgrid
CN121417203B
Generator set maintenance plan generation method and system based on multi-scale data fusion
CN121436965A
