Series-parallel micro-grid coordination control system for stabilizing load fluctuation

Abstract

The invention discloses a series-parallel micro-grid coordination control system for load fluctuation stabilization, and relates to the technical field of grid control, and the system comprises the steps: constructing a load prediction channel, and obtaining a load prediction result; formulating a power distribution table according to the load prediction result in combination with a power constraint interval; executing the first power distribution scheme and the second power distribution scheme to monitor the hybrid microgrid in real time, and drawing a power fluctuation curve graph; carrying out feedback response according to the power fluctuation curve graph, and carrying out dynamic adjustment on the hybrid microgrid according to a response instruction to generate an adjustment data set; and constructing an adjustment strategy according to the adjustment data set to intelligently stabilize the load fluctuation of the hybrid microgrid. According to the invention, the technical problems of unstable operation, low energy utilization efficiency and poor adaptability of the hybrid microgrid in the prior art are solved, and the technical effects of improving the stability of the hybrid microgrid, optimizing the energy utilization efficiency and enhancing the adaptability of the system are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid control, and in particular to a hybrid microgrid coordinated control system for load fluctuation smoothing. Background Art

[0002] With the continuous growth of energy demand and the increasing emphasis on renewable energy, hybrid microgrids are increasingly used in power systems. However, due to the intermittent and uncertain nature of renewable energy and the complexity and variability of user electricity demand, the load fluctuation of hybrid microgrids has become a key issue that needs to be addressed. Traditional power systems often have problems such as untimely response, low efficiency and waste of resources when dealing with load fluctuations.

[0003] The existing technology has technical problems such as unstable operation of hybrid microgrids, low energy utilization efficiency and poor adaptability. Summary of the invention

[0004] The present application provides a hybrid microgrid coordinated control system for load fluctuation smoothing, which is used to solve the technical problems of unstable operation, low energy utilization efficiency and poor adaptability of hybrid microgrids in the prior art.

[0005] In view of the above problems, the present application provides a hybrid microgrid coordinated control system for load fluctuation smoothing.

[0006] The present application provides a hybrid microgrid coordinated control system for load fluctuation smoothing, the system comprising:

[0007] A load forecasting result acquisition module, the load forecasting result acquisition module is used to construct a load forecasting channel, synchronize the hybrid microgrid to the load forecasting channel, and obtain a load forecasting result; a power allocation table formulation module, the power allocation table formulation module is used to formulate a power allocation table according to the load forecasting result combined with the power constraint interval, and the power allocation table includes a first power allocation scheme and a second power allocation scheme; a power fluctuation curve drawing module, the power fluctuation curve drawing module is used to execute the first power allocation scheme and the second power allocation scheme to perform real-time monitoring of the hybrid microgrid, and draw a power fluctuation curve according to the monitoring results; an adjustment data set generation module, the adjustment data set generation module is used to perform feedback response according to the power fluctuation curve, dynamically adjust the hybrid microgrid according to the response instruction, and generate an adjustment data set; an intelligent smoothing module, the intelligent smoothing module is used to construct an adjustment strategy according to the adjustment data set to intelligently smooth the load fluctuation of the hybrid microgrid.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] The load forecast result acquisition module is used to construct a load forecast channel, synchronize the hybrid microgrid to the load forecast channel, and obtain the load forecast result; the power allocation table formulation module is used to formulate a power allocation table according to the load forecast result combined with the power constraint interval; the power fluctuation curve drawing module is used to execute the first power allocation plan and the second power allocation plan to monitor the hybrid microgrid in real time, and draw a power fluctuation curve according to the monitoring results; the adjustment data set generation module is used to respond to feedback according to the power fluctuation curve, dynamically adjust the hybrid microgrid according to the response instruction, and generate an adjustment data set; the intelligent smoothing module is used to construct an adjustment strategy according to the adjustment data set to intelligently smooth the load fluctuation of the hybrid microgrid. The technical effect of improving the stability of the hybrid microgrid, optimizing energy utilization efficiency, and enhancing the adaptability of the system is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A structural schematic diagram of a hybrid microgrid coordinated control system for load fluctuation smoothing is provided for an embodiment of the present application;

[0012] Figure 2 A structural execution schematic diagram of a power allocation table formulation module of a hybrid microgrid coordinated control system for load fluctuation smoothing is provided for an embodiment of the present application.

[0013] Explanation of the reference numerals: load forecast result acquisition module 10 , power allocation table formulation module 20 , power fluctuation curve drawing module 30 , adjustment data set generation module 40 , intelligent smoothing module 50 . DETAILED DESCRIPTION

[0014] The present application provides a hybrid microgrid coordination control system for load fluctuation smoothing, which is used to solve the technical problems of unstable operation, low energy utilization efficiency and poor adaptability of hybrid microgrids in the prior art. The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0015] Embodiment 1, as Figure 1As shown, the present application provides a hybrid microgrid coordinated control system for load fluctuation smoothing, the system comprising:

[0016] The load forecasting result acquisition module 10 is used to construct a load forecasting channel, synchronize the hybrid microgrid to the load forecasting channel, and obtain the load forecasting result.

[0017] Specifically, the main function of the load forecast result acquisition module is to build a load forecast channel for the hybrid microgrid in order to obtain accurate load forecast results. First, it is necessary to collect various relevant data of the hybrid microgrid, including but not limited to historical load data (power consumption in different time periods), output power of distributed power sources, usage of various power-consuming equipment, season and weather information, etc. Specific algorithms and models are used to analyze and process these data. For example, time series analysis methods can be used to infer future load conditions by studying the changing trends of historical load data, and predict the changes in residents' power consumption and solar power generation in different time periods on a certain day in the future, so as to obtain the load forecast results of the entire microgrid. Such prediction results can help grid managers to reasonably arrange power resources, ensure the stable operation of microgrids, and enhance system adaptability.

[0018] The power allocation table formulation module 20 is used to formulate a power allocation table according to the load forecast result combined with the power constraint interval, and the power allocation table includes a first power allocation plan and a second power allocation plan.

[0019] Specifically, first obtain the load forecast results to understand the expected load of the hybrid microgrid. Then, combine the set power constraint interval to determine how to allocate power among the various components or devices. The first power allocation scheme and the second power allocation scheme are formulated according to different conditions, strategies or optimization goals. For example, the first scheme may focus more on achieving a certain performance indicator (such as maximizing efficiency, minimizing costs, etc.) under the premise of meeting the basic load demand; while the second scheme may be designed to deal with some special situations or backup situations. When formulating a power allocation scheme, many factors need to be considered. For example, the power capacity, operating status, efficiency characteristics of different devices, and the stability and reliability requirements of the entire hybrid microgrid. By comprehensively analyzing these factors, the power share that each part or device should bear is determined within the power constraint interval to achieve the purpose of reasonable and efficient power allocation. For example, the power allocation table based on the greedy algorithm is used as an example to obtain the load forecast results and clarify the expected total load of the hybrid microgrid. Then, the power upper limit, lower limit, efficiency and other parameters of each power generation equipment and energy storage equipment are obtained. For the first power allocation scheme, power is allocated in order from high to low equipment efficiency. First, power is allocated to the most efficient device until it reaches the power limit, and then the next most efficient device is allocated in turn until the total load demand is met. For the second power allocation scheme, the power requirements of key devices or devices with specific requirements are met first. For example, first ensure that the power supply equipment of important loads is met, and then allocate it according to the priority of other devices and the remaining load. During the allocation process, if the power exceeds the constraint range, the allocation strategy is readjusted until a feasible power allocation plan is obtained within the constraint range. Make more efficient use of the power generation and energy storage equipment in the hybrid microgrid, improve the operating efficiency of the microgrid, and enhance the stability of the system.

[0020] The power fluctuation curve drawing module 30 is used to execute the first power allocation scheme and the second power allocation scheme to monitor the hybrid microgrid in real time, and draw a power fluctuation curve according to the monitoring results.

[0021] Specifically, firstly, the first power allocation scheme and the second power allocation scheme are obtained and executed. During the execution process, the sensors and monitoring devices installed at various key positions of the hybrid microgrid collect power-related data in real time, including the output power of each power generation equipment, the charging and discharging power of the energy storage equipment, the power consumption of the load, etc. These real-time collected data will be continuously transmitted to the module for processing and analysis, and the data will be sampled and recorded at a certain time interval (such as every second, every minute or every hour). Then, based on these sampled data, the total power of the hybrid microgrid at each time point is calculated. By using time as the horizontal axis and total power as the vertical axis, a power fluctuation curve is drawn using a drawing tool or software. In the curve, the change of power over time can be clearly seen. For example, there may be a sudden increase or decrease in power, which is caused by a sudden increase or decrease in load, or a failure or startup of a power generation equipment. By observing the power fluctuation curve, the operating stability and dynamic response characteristics of the hybrid microgrid under different power allocation schemes can be intuitively understood. The law and trend of power fluctuation can be discovered, so as to evaluate the effect of the power allocation scheme and provide a basis for further optimization and adjustment.

[0022] The adjustment data set generation module 40 is used to perform feedback response according to the power fluctuation curve diagram, dynamically adjust the hybrid microgrid according to the response instruction, and generate an adjustment data set.

[0023] Specifically, after receiving the power fluctuation curve, it will analyze and interpret it in detail, and it will identify the parts where the power fluctuation exceeds the normal range, the trend is unstable or does not meet expectations. According to these abnormal situations, the corresponding feedback response is generated. For example, if the power fluctuation is too large and lasts for a long time, the response instruction may be to increase the output power of a certain energy storage device, or to reduce the access of a certain unstable power generation device. In the process of executing the response instruction to dynamically adjust the hybrid microgrid, the module will record the specific actions of the adjustment and the changes in related parameters in real time. These records include the time of adjustment, the object of adjustment, the amplitude of adjustment, and the power data before and after adjustment. All these detailed adjustment information will be sorted and summarized to form an adjustment data set. This data set not only contains specific adjustment operations, but also reflects the response and changes of the microgrid under different circumstances. By generating an adjustment data set, the system can accumulate rich operating experience and data samples, and use these data for more in-depth analysis in the future, such as finding out the links that need to be adjusted frequently and evaluating the effects of different adjustment strategies, so as to further optimize the control strategy and power allocation scheme of the entire microgrid and improve the stability and adaptability of the system.

[0024] The intelligent smoothing module 50 is used to construct an adjustment strategy according to the adjustment data set to intelligently smooth the load fluctuation of the hybrid microgrid.

[0025] Specifically, first, the adjustment data set generated by the adjustment data set generation module 40 is obtained, and these data sets contain detailed information and results of the dynamic adjustment of the hybrid microgrid in the past. In-depth analysis and mining of these data will identify effective adjustment measures and strategy patterns under different load fluctuations. For example, it may be found that when the load suddenly increases significantly, increasing the output power of a specific energy storage device can effectively smooth the fluctuation; or when a certain power generation device has unstable output, adjusting the power distribution of other devices can maintain system stability. Based on these analysis results, a set of targeted adjustment strategies is constructed. These strategies are based on data-driven and experience learning, and can automatically select and execute the most appropriate adjustment action according to the load fluctuations monitored in real time. In actual operation, when the load fluctuates in the hybrid microgrid, the intelligent smoothing module 50 will respond quickly, select and execute the corresponding strategy from the constructed adjustment strategy according to the current load fluctuation characteristics, so as to quickly and accurately smooth the load fluctuation. By continuously accumulating and updating the adjustment data set and optimizing the adjustment strategy, the intelligent smoothing module 50 can make the operation of the hybrid microgrid more stable and reliable, improve the power quality, and reduce the risk of system operation.

[0026] In a possible implementation, the load forecasting result acquisition module 10 further includes:

[0027] The load forecasting channel includes a first load forecasting branch and a second load forecasting branch; the hybrid microgrid is analyzed to obtain an AC subnet and a DC subnet; a microgrid synchronous phasor measurement unit is used to collect data from the AC subnet and the DC subnet to obtain multiple AC data and multiple DC data; based on the multiple AC data, the AC subnet is synchronized to the first load forecasting branch to generate a first load forecasting result; based on the multiple DC data, the DC subnet is synchronized to the second load forecasting branch to generate a second load forecasting result; the first load forecasting result and the second load forecasting result are integrated to generate the load forecasting result.

[0028] Specifically, this description points out that the load forecasting channel consists of two branches, namely the first load forecasting branch and the second load forecasting branch. The two branches are set up to process data of different types or sources respectively, so as to make load forecasting more targeted. For example, the first load forecasting branch focuses on processing data related to the AC part, while the second load forecasting branch mainly predicts data related to the DC part. This branch structure helps to improve the accuracy and efficiency of load forecasting, making the entire load forecasting process more refined and professional.

[0029] A hybrid microgrid is usually composed of a variety of distributed energy sources, energy storage devices, and loads, which may exist in the form of AC or DC and are interconnected. Through the parsing process, the part that operates and supplies power in the form of AC can be clearly distinguished, namely the AC subnet. This part includes AC generators, AC loads, and AC-related converters and other equipment. At the same time, the part that operates in the form of DC can also be determined, namely the DC subnet. The DC subnet includes DC power sources (such as photovoltaic panels), DC energy storage devices (such as battery packs), and DC loads.

[0030] Using the microgrid synchronous phasor measurement unit, various data of the AC subnet and DC subnet are collected in real time through RTU, sensors, etc., such as load data, weather data, socio-economic data, etc., to obtain multiple AC data and multiple DC data. For the AC subnet, the multiple AC data collected include but are not limited to the real-time load value of the AC line, such as the specific values ​​of the power consumption, current, voltage, etc. of each node; weather data related to the operation of the AC subnet is also collected, such as temperature, humidity, light intensity, etc. In addition, socio-economic data such as the industrial structure of the region, enterprise production plans, and residents' consumption levels will also be collected, because these factors will indirectly affect the changes in power load. For the DC subnet, similar multiple DC data will also be collected, such as load data such as power, current, voltage, etc. of the DC line, as well as corresponding weather and socio-economic data. These large amounts of multi-dimensional data collected constitute multiple AC data and multiple DC data.

[0031] Preprocess and extract features from the collected multiple AC data. For example, calculate the characteristic values ​​such as active power and reactive power from the AC voltage and current data, and organize the data at different time points into a time series. Based on the collected multiple AC data, connect the AC subnet to the first load forecasting branch, and use the time series analysis method to find the periodic change trend of the AC subnet load. Combine these AC data to perform multivariate regression analysis and obtain a data variable regression parameter set. Finally, according to the load cycle trend found previously, use this parameter set to perform load forecasting, thereby obtaining the first load forecasting result.

[0032] The acquired DC data are processed and analyzed in detail, including removing outliers, smoothing data fluctuations, and extracting key feature information, such as the mean and standard deviation of DC voltage, the trend of DC current, and the peak and valley values ​​of power. Then, the processed DC data is input into the second load forecasting branch. This branch uses a prediction model designed specifically for the characteristics of the DC subnet, such as a long short-term memory network (LSTM) model based on deep learning or an integrated learning model based on a decision tree. The model will conduct in-depth mining and learning of the input DC data to find the inherent laws and periodic patterns of the DC subnet load, while considering external factors that may affect the load, such as the addition or retirement of DC equipment, and changes in DC load types. Combined with the historical DC subnet load data and the current DC data characteristics, the load situation of the DC subnet in the future is predicted to generate the second load forecasting result.

[0033] The formats and units of the first load forecast result and the second load forecast result are unified to ensure effective comparison and integration. The two forecast results are weighted averaged according to the weight or importance of the AC subnet and the DC subnet in the hybrid microgrid. For example, if the AC subnet accounts for a large proportion of the entire microgrid, the first load forecast result may be given a higher weight in the integration. The first load forecast result and the second load forecast result processed as above are merged to generate a comprehensive and accurate load forecast result. Assume that the first load forecast result shows that the average load of the AC subnet in the next week is 500 kilowatts, and the second load forecast result shows that the average load of the DC subnet is 200 kilowatts, and the weight of the AC subnet is determined to be 70% and the DC subnet is 30%. Then, by weighted average calculation: (500×70%)+(200×30%)=410 kilowatts, this 410 kilowatts is the integrated load forecast result.

[0034] In a possible implementation, the load forecasting result acquisition module 10 further includes:

[0035] The AC subnet is synchronized to the first load prediction branch, and the load cycle trend of the AC subnet is captured by using a time series analysis method; a multivariate regression analysis is performed on the multiple AC data to generate a data variable regression parameter set; load prediction is performed on the data variable regression parameter set according to the load cycle trend to generate the first load prediction result.

[0036] Specifically, first, collect the long-term load data of the AC subnet, which should cover different time periods and various operating conditions. Then, preprocess the data, including removing outliers and filling missing values, to ensure the quality and reliability of the data. Next, use the time series analysis method to analyze the past load data of the AC subnet in chronological order. It can identify the periodic change patterns of the load in different time periods (such as days, weeks, months, seasons, years, etc.), such as whether there are daily peak and trough periods, higher loads on certain days of the week, or seasonal load increase and decrease trends.

[0037] Combine the collected AC data to perform multivariate regression analysis, which includes multiple variables such as voltage, current, power factor, equipment operation status, etc. Through multivariate regression analysis, the mathematical relationship between these variables and the load is established, thereby generating a data variable regression parameter set. This parameter set reflects the degree and method of the impact of each variable on the load.

[0038] Carefully analyze the load cycle trends captured previously. For example, if it is found that the load of the AC subnet has a clear and predictable growth trend in the summer every year, or shows different regular changes on weekdays and weekends, use this load cycle trend as an important reference framework. Use the data variable regression parameter set, which contains the precise impact values ​​of multiple variables such as voltage, current, and power factor on the load. In the actual prediction process, if the current time is a certain day in summer, we will first set an initial prediction benchmark based on the overall load increase trend in summer. Then, obtain the real-time data of this day, such as the voltage measurement value of the day is X1, the current value is X2, the power factor is X3, and so on. According to the corresponding parameters in the data variable regression parameter set, assume that the voltage parameter is b1, the current parameter is b2, and the power factor parameter is b3. By calculating b1X1+b2X2+b3X3, the direct impact values ​​of these variables on the load are obtained. But this is not the final prediction result. It is also necessary to adjust this preliminary calculation result according to whether the day is a weekday or a weekend, and the specific time period (peak time or valley time). After comprehensive consideration of load cycle trends, current real-time data, and complex calculations and fine-tuning of data variable regression parameters, we ultimately arrive at a fully revised and optimized load forecast value, which is the first load forecast result we generated.

[0039] In one possible implementation, Figure 2 As shown, the power allocation table formulation module 20 also includes:

[0040] A power allocation model is constructed based on the load forecast result in combination with the power constraint interval; the operation cycle of the hybrid microgrid is retrieved to determine multiple time scales; power is allocated to the hybrid microgrid according to the multiple time scales through the power allocation model to obtain multiple power allocation data; the multiple power allocation data are mapped and associated to obtain data association factors; a power allocation table is constructed based on the data association factors in combination with the multiple power allocation data; the power allocation table is traversed and matched according to the first load forecast result to obtain a first power allocation scheme for the AC subnet; the power allocation table is traversed and matched according to the second load forecast result to obtain a second power allocation scheme for the DC subnet.

[0041] Specifically, the load forecast results are analyzed in depth. The load forecast results include the expected load of the AC subnet and the DC subnet in different time periods, the trend of load changes, and other information. At the same time, the specific scope and restrictions of the power constraint interval are clarified. The power constraint interval is determined by many factors, such as the maximum output power of the power generation equipment, the capacity limit of the transmission line, the charging and discharging capacity of the energy storage equipment, and related safety and operation standards. Then, the power allocation model is constructed by comprehensively considering the load forecast results and the power constraint interval. In the model, the load forecast results are used as demand input to ensure that future power demand is met. For example, if the load forecast results show that the load will rise sharply in a certain period of time, and the power constraint interval shows that the maximum output power of the current power generation equipment is limited, then the model needs to consider how to reasonably allocate existing resources, including increasing the discharge of energy storage equipment, optimizing the operating efficiency of power generation equipment, or reducing the power supply of non-critical loads when necessary. The model is constructed by linear programming, nonlinear programming, dynamic programming and other methods. The goal is to minimize the cost, maximize the efficiency, or maximize the reliability of power allocation under the premise of meeting the power constraint interval. In summary, the constructed power distribution model is a complex system that comprehensively considers load demand and various constraints, and realizes the reasonable, efficient and safe distribution of power in the hybrid microgrid.

[0042] Hybrid microgrids have certain operating rules and cycles. In order to analyze and manage them more effectively, it is necessary to retrieve relevant data of their operating cycles. The operating cycle covers a range from milliseconds to hours, days, months and even years. For example, the rapid response and switching of some power equipment may occur in milliseconds, while seasonal changes in electricity demand may be in years. When determining multiple time scales, they are divided according to specific needs and analysis purposes. Determine the hourly scale for real-time monitoring and power adjustment in a short period of time; the daily scale can be used for daily operation scheduling and planning; the weekly scale is suitable for evaluating the balance of power supply and demand and resource allocation within a week; the monthly scale can be used to analyze monthly energy consumption and cost accounting; the annual scale is helpful for long-term strategic planning and equipment renewal plans. By carefully retrieving the operating cycle and determining multiple time scales, it is possible to fully understand the operating characteristics of the hybrid microgrid from different time dimensions, providing a more accurate and comprehensive basis for subsequent power allocation and optimization.

[0043] The constructed power allocation model is ready to run. Then, multiple time scales set before are determined, such as hourly, daily, weekly, etc. When the power allocation model starts to run, calculations and planning are performed for each time scale. For the hourly time scale, the model will calculate how to allocate power between the AC subgrid and the DC subgrid within this hour based on the current load forecast, power constraint interval, and other related factors, including how much power to obtain from each power source, how much power to deliver to each load, and the charging and discharging power of the energy storage device, so as to obtain the power allocation data for this hour. Similarly, for the daily time scale, the model will comprehensively consider the load changes at different times of the day, formulate a full-day power allocation plan, and form the corresponding power allocation data. For the weekly time scale, the power allocation strategy and data for the week will be obtained based on the different load patterns of working days and rest days in a week, as well as possible seasonal changes and other factors. By calculating and planning at different time scales, multiple power allocation data covering various time scales are finally obtained.

[0044] Collect multiple power allocation data, which contain specific values ​​and related information of power allocation for hybrid microgrids under different time scales and different operating conditions. Conduct in-depth analysis and comparison of these data, considering multiple dimensions such as time sequence, object of power allocation (AC subnet or DC subnet), grid state at allocation (such as load level, energy storage state, etc.). Through the K-Means clustering method, divide the data into K different clusters, so that the data points within the cluster are more similar, while the data points between clusters are less similar. Cluster similar power allocation data patterns into one category, and find related features from them. Get a quantitative indicator that can reflect the degree and pattern of association between data, which is the data association factor. The data association factor can be a value, a set of coefficients, or a function expression. Its role is to be able to concisely summarize the complex relationship between multiple power allocation data, and provide key basis and clues for the subsequent construction of power allocation tables and more in-depth analysis.

[0045] The inherent connections and rules between the power allocation data reflected by the data association factors are clarified, and multiple power allocation data are classified and sorted according to different conditions and scenarios. These conditions include different time periods, grid load levels, energy supply conditions, etc. Then, according to the relationship revealed by the data association factors, the corresponding allocation strategy and specific values ​​are determined for each type of sorted power allocation data. When constructing the power allocation table, a clear table structure is created, and the rows and columns of the table correspond to different classification conditions and specific parameters of power allocation. For example, the rows can represent different time periods (such as different time periods every hour or every day), and the columns can represent various power allocation parameters of the AC subnet and the DC subnet (such as input power, output power, charging and discharging power of energy storage equipment, etc.). For each table cell, fill in the corresponding power value or allocation strategy description according to the data association factor and the corresponding power allocation data. The power allocation table constructed in this way can clearly and intuitively show how power should be reasonably allocated in the hybrid microgrid under various different situations. Through such a power allocation table, clear guidance can be provided for the operation and management of the microgrid to ensure efficient and stable supply of electricity.

[0046] Carefully analyze the first load forecast results to clarify the expected load situation of the AC subnet in a specific time period in the future, including specific information such as the size of the load, the trend of change, and possible peak and valley periods. Then, start traversing the power allocation table. Starting from the first row of the power allocation table, check the conditions and parameters in the table row by row. Compare the conditions in the power allocation table with the first load forecast results. These conditions include different time ranges, load level ranges, power grid operating status, etc. During the traversal process, continue to look for the row or group of rows that best matches the first load forecast result. When a combination of conditions that fully meets or is closest to the first load forecast result is found, the corresponding AC subnet power allocation data is obtained, which constitutes the first power allocation plan for the AC subnet.

[0047] Study the second load forecast results, grasp the load forecast of the DC subnet in a specific time period in the future, such as the specific value of the load, the direction of change, possible load peaks and valleys, and other key information, and compare them row by row starting from the table header according to the arrangement order of the power allocation table. Compare the conditions set for each row in the power allocation table, such as the time interval, the amplitude range of the load, the specific operating conditions of the DC subnet, etc., with the second load forecast results one by one. During the traversal process, continue to look for the row or group of rows that best matches the second load forecast result. When a combination of conditions that fully meets or is closest to the second load forecast result is found, the corresponding DC subnet power allocation data is obtained, which constitutes the second power allocation plan for the DC subnet.

[0048] In a possible implementation, the power fluctuation curve drawing module 30 further includes:

[0049] Execute the first power allocation scheme to perform demand analysis on the AC subnet and generate a stable power demand and a peak power demand; perform real-time monitoring of the AC subnet based on the stable power demand and the peak power demand to generate first power consumption data; execute the second power allocation scheme to perform demand analysis on the DC subnet and generate a power generation demand and a load power demand; perform real-time monitoring of the DC subnet based on the power generation demand and the load power demand to generate second power consumption data; add the first power consumption data and the second power consumption data to the monitoring results of the hybrid microgrid.

[0050] Specifically, when the first power allocation plan formulated for the AC subnet is implemented, the operation and load characteristics of the AC subnet will be carefully studied and evaluated. First, observe the power consumption of various devices and loads in the AC subnet under normal and continuous working conditions. Considering the continuous operation of some basic equipment and the relatively stable power consumption behavior, a relatively stable power level is determined by monitoring and statistical analysis of power consumption over a period of time. This is the stable power demand. It is also necessary to pay attention to special situations or high-load scenarios, such as the startup moment of large equipment and multiple devices running at full load at the same time. At this time, the power demand of the AC subnet will rise sharply. By simulating and analyzing these possible extreme situations, the maximum power demand that the AC subnet can achieve is predicted, which is the peak power demand. Through such demand analysis, the stable power demand and peak power demand of the AC subnet in different working scenarios are clarified, which provides an important reference for subsequent monitoring and management.

[0051] After clarifying the stable power demand and peak power demand of the AC subnet, special monitoring equipment and systems are used to monitor the power of the AC subnet in real time. The monitoring equipment will continuously obtain the power values ​​of each part of the AC subnet, accurately measure parameters such as current and voltage, and calculate the real-time power. During the monitoring process, the real-time power data is compared and analyzed with the previously determined stable power demand and peak power demand. If the real-time power fluctuates within the range of the stable power demand, these data are recorded as the power consumption under normal operating conditions. When the real-time power approaches or exceeds the peak power demand, the relevant data at this time will be specially marked and recorded in detail, including the specific power value, the time of occurrence, and the operation of the equipment in the AC subnet at that time. Through a period of continuous monitoring, all the recorded power data are sorted and summarized to generate the first power consumption data.

[0052] After executing the second power allocation plan specially formulated for the DC subnet, it is necessary to conduct a demand analysis on the DC subnet. Analyze the power generation equipment of the DC subnet, consider factors such as the type, capacity, operating efficiency, and current working status of the power generation equipment, and through evaluation and calculation of these aspects, determine the power output of the power generation equipment to meet the power demand of various loads in the DC subnet, which results in the power generation demand. Analyze various loads in the DC subnet, including the number of loads such as different types of electronic equipment and DC motors, the working mode, and the probability of simultaneous operation. By studying and calculating the power consumption characteristics of these loads, it is clear that the total power required in different working scenarios is the load power demand.

[0053] After clarifying the power generation and load power requirements of the DC subnet, we began to use special monitoring equipment and technology to track and record the power situation of the DC subnet in real time. The monitoring equipment will obtain various power-related data in the DC subnet at very short time intervals. For example, the actual output power of the power generation equipment and the power consumed by each load are measured in real time. These real-time collected data are compared and analyzed with the previously determined power generation and load power requirements. If the actual output power of the power generation equipment is close to or reaches the power generation demand, the specific value, time and related operating conditions at this time are recorded in detail. For load power consumption, if the actual load power fluctuates within the range of the load power demand, the data of these fluctuations are also recorded, including the power value, the time of occurrence and the load situation that was running at that time. During the monitoring process, these data are continuously collected and sorted. After a period of time, all the recorded real-time data on the power of the DC subnet are summarized and processed to generate the second power consumption data.

[0054] After obtaining the first power consumption data of the AC subnet and the second power consumption data of the DC subnet respectively, they need to be integrated into the overall monitoring results of the hybrid microgrid. First, the monitoring system of the hybrid microgrid reserves a specific data interface or storage space for receiving and storing power consumption data from each subnet. Then, the first power consumption data and the second power consumption data are accurately transmitted to the monitoring system of the hybrid microgrid in accordance with the prescribed data format and transmission method. After receiving these data, the monitoring system will classify and mark them so that it can clearly distinguish which data comes from the AC subnet and which comes from the DC subnet. Then, these data are associated and integrated with other relevant data already in the monitoring system (such as the overall power input and output of the power grid, the status of the energy storage device, etc.). Through this addition operation, the monitoring results of the hybrid microgrid are more comprehensive and complete. In the subsequent evaluation, analysis and optimization of the operating status of the hybrid microgrid, the power consumption of the AC subnet and the DC subnet can be comprehensively considered to make more accurate and effective decisions.

[0055] In a possible implementation, the adjustment data set generating module 40 further includes:

[0056] Based on the peak power demand and the load power demand, a feature analysis is performed on the power fluctuation curve to obtain multiple power fluctuation features; based on the multiple power fluctuation features, fluctuation identification is performed to determine power fluctuation period information, power fluctuation frequency information, and power fluctuation extreme value information; the historical power fluctuation record log is retrieved to set a fluctuation threshold interval, and a feedback mechanism is constructed; the power fluctuation period information, the power fluctuation frequency information, and the power fluctuation extreme value information are synchronized to the feedback mechanism to determine whether the fluctuation value of the hybrid microgrid is within the fluctuation threshold interval; if the fluctuation value of the hybrid microgrid is not within the fluctuation threshold interval, a feedback response is triggered to identify a power fluctuation mode, and the power fluctuation mode is added to the response instruction.

[0057] Specifically, obtain a power fluctuation curve, which intuitively shows how power changes over time, and uses the previously determined peak power demand and load power demand as an important reference. Observe the rise and fall of the power value on the curve, compare it with the peak power demand and load power demand, and determine the magnitude of the fluctuation. Determine the repetition period of power fluctuations, that is, the time interval between two adjacent similar fluctuation patterns. Find the time period in the curve where the power is relatively stable and the fluctuation is small, and calculate its duration. Compare the power value in the curve with the peak power demand and load power demand, and calculate the size of the deviation and the frequency of occurrence. Through observation and calculation of these aspects, it is possible to obtain multiple features that describe the characteristics of power fluctuations, providing a basis for subsequent fluctuation identification and processing.

[0058] After obtaining multiple power fluctuation features, more in-depth analysis and identification are started. The power fluctuation cycle information is determined by observing the time intervals at which similar fluctuation patterns recur in the fluctuation curve. For example, if it is found that the power goes through a complete process from trough to peak and then to trough at regular intervals, this time interval is the power fluctuation cycle. The power fluctuation frequency information is determined by calculating the number of power fluctuations per unit time. For example, if the power completes 5 complete fluctuations within an hour, the power fluctuation frequency is 5 times / hour. The power fluctuation extreme value information, including the maximum and minimum values, is the maximum and minimum values ​​of the power extracted directly from the fluctuation curve. During the identification process, special attention will be paid to those maximum and minimum values ​​that obviously deviate from the normal fluctuation range.

[0059] Obtain the past historical power fluctuation record logs from the system. These logs contain detailed data on power fluctuations over a period of time. Then, conduct in-depth analysis and statistics on these historical data. For example, calculate the average, standard deviation, maximum and minimum values ​​of historical power fluctuations. Based on the results of these statistical analyses, set a reasonable fluctuation threshold range, which is usually determined based on actual needs and system tolerance. For example, set the threshold range to the average plus or minus two standard deviations, or set a specific range based on experience and safety requirements. The feedback mechanism constructed is a system that can monitor the current power fluctuation in real time and compare and judge it with the set threshold range. When the current power fluctuation value is within the threshold range, the feedback mechanism will not take special actions, or only perform routine monitoring and recording. However, if the current power fluctuation value exceeds the threshold range, the feedback mechanism will be triggered and initiate corresponding response measures, such as issuing an alarm, adjusting power allocation, and starting the backup power supply.

[0060] The previously determined key data such as power fluctuation cycle information, frequency information and extreme value information are accurately transmitted to the feedback mechanism that has been constructed. After receiving this information, the feedback mechanism will immediately compare and analyze it with the fluctuation threshold interval set in advance. For the power fluctuation cycle, the feedback mechanism will determine whether it is within the reasonable cycle range specified by the threshold interval. The power fluctuation frequency will also be judged similarly to check whether it meets the allowable frequency range set by the threshold interval. For the power fluctuation extreme value, the feedback mechanism will compare whether the obtained extreme value information exceeds the maximum and minimum value range set by the threshold interval. Through such a comprehensive comparison and judgment, it is determined whether the current power fluctuation value of the hybrid microgrid is within the set fluctuation threshold interval as a whole.

[0061] When it is found that the power fluctuation value of the hybrid microgrid is not within the pre-set fluctuation threshold range, a specific feedback response mechanism will be activated. In this feedback response process, the current power fluctuation mode is further analyzed and identified. The power fluctuation mode includes sudden large jumps or drops, continuous slow rises or falls, periodic violent fluctuations, irregular random fluctuations, etc. The current specific power fluctuation mode is determined by analyzing the shape, change trend, duration and other characteristics of the power fluctuation curve. Once the power fluctuation mode is identified, it will be added to the corresponding response instruction. This response instruction is used to trigger a series of countermeasures, such as adjusting the output power of the power generation equipment, changing the working state of the energy storage equipment, reasonably controlling the load or starting the backup power supply, etc. In this way, it is possible to deal with abnormal power fluctuations in a more targeted manner and ensure the stable operation of the hybrid microgrid.

[0062] In a possible implementation, the adjustment data set generating module 40 further includes:

[0063] Based on the power fluctuation period information and the power fluctuation frequency information, the power fluctuation extreme value information is identified to obtain multiple peak points and multiple valley points; based on the multiple peak points and the multiple valley points, the response instruction is executed to perform dynamic energy storage peak regulation to obtain dynamic peak regulation results; based on the multiple valley points, dynamic energy storage leveling is performed in combination with the dynamic peak regulation results to obtain dynamic leveling results; the dynamic peak regulation results and the dynamic leveling results are normalized to generate the adjustment data set.

[0064] Specifically, the power fluctuation cycle information is the time required for the power fluctuation to complete a complete cycle from the trough to the peak and then back to the trough, while the power fluctuation frequency information indicates the number of such cycles per unit time. The power fluctuation extreme value information is a detailed analysis of the curve of power change over time, and the highest and lowest points in each fluctuation cycle are determined according to the laws of cycle and frequency. By comprehensively marking the entire power fluctuation curve, multiple peak points and multiple valley points can be obtained.

[0065] After multiple peak points and valley points are determined, the system will take action according to the preset response instructions to perform dynamic energy storage peak regulation. When the power load is at a peak, the energy storage system releases the previously stored electric energy to supplement the power supply capacity of the power system; when the load is at a low point, the energy storage system absorbs excess electric energy for use during peak hours. This method can effectively smooth out the fluctuations in power load. By operating in this way at multiple peak points and valley points, dynamic allocation and balance of electric energy can be achieved. After completing a series of peak regulation operations, a result on dynamic peak regulation will be obtained. This result includes detailed data and analysis on the charge and discharge capacity of the energy storage equipment, the actual power changes of the power grid at each moment, and the impact of the peak regulation operation on the stability and efficiency of the power grid.

[0066] After obtaining the data of multiple valley points, it is combined with the previous dynamic peak-shaving results to carry out dynamic energy storage leveling. Multiple valley points reflect the specific time and degree of low power in the power system, while the dynamic peak-shaving results contain relevant information about the previous energy storage at the peak point and partial discharge at the valley point. At the valley point, due to the relative excess power supply or low demand, the energy storage system will perform energy storage operations. The dynamic peak-shaving results can tell us the energy storage and discharge at the peak point before, as well as the impact on the power system. Combined with this result, energy storage can be arranged more accurately at the valley point. During operation, the energy storage system can effectively smooth out the fluctuations of renewable energy generation. For example, solar power generation reaches a peak at noon on a sunny day and then drops rapidly. The energy storage system can store part of the electricity at the peak and release it at the valley, making the power supply more stable. At the same time, the energy storage system can also cooperate with traditional thermal power generation and hydropower generation. For example, the output of thermal power generation is relatively stable, but it may be excessive in certain periods of time, and the energy storage system can store energy at this time; hydropower generation is large during the flood season, and it can also be adjusted through the energy storage system to better control the load of the power system. After such a series of operations based on valley points and combined with dynamic peak-shaving results, the final result is the dynamic leveling result. This result can reflect the stability and efficiency improvement of the power system after energy storage leveling, as well as the operating status and effect of the energy storage system itself.

[0067] Normalization is a method of unifying data of different magnitudes and dimensions into a specific range or standard form. The data of dynamic peak-shaving results and dynamic leveling results are obtained separately. These data contain values ​​of various types and ranges such as power value, power change of energy storage equipment, time, etc. These data are mapped to a unified interval through the minimum-maximum normalization method. The purpose of this is to eliminate the impact of differences in magnitude and dimension between data, so that different types of data can be compared and comprehensively analyzed under the same standard. In the normalization process, the data will be converted according to the maximum, minimum, average and other characteristic values. After normalization, the processed dynamic peak-shaving results and dynamic leveling results are integrated to form a complete, unified and easy-to-analyze and use data set, that is, the adjustment data set.

[0068] In a possible implementation, the intelligent smoothing module 50 further includes:

[0069] Preprocessing is performed based on the adjustment data set to generate adjustment processing data; the adjustment processing data is divided into a training set, a validation set, and a test set; data enhancement is performed on the adjustment processing data according to the training set and the validation set to generate training adjustment enhancement data; the training adjustment enhancement data is tested based on the test set, and adjustment result prediction is performed based on the test result to determine the prediction result; the load fluctuation of the prediction result is evaluated to generate a prediction evaluation coefficient; and the adjustment strategy is formulated according to the prediction evaluation coefficient.

[0070] Specifically, after obtaining the adjusted data set, the preprocessing operation begins. The first step is to remove outliers and handle missing values ​​in the data, and then extract the key features of voltage fluctuations. For voltage fluctuations, focus on the amplitude (i.e., the difference between the maximum and minimum values ​​of the voltage fluctuation), frequency (the number of voltage fluctuations per unit time) and duration (the duration of each voltage fluctuation), and accurately extract these characteristic values ​​from the original data. After the above series of preprocessing operations, the adjusted processed data after cleaning, supplementation and feature extraction are obtained. These data are more accurate, complete and representative, and can provide a more reliable basis for subsequent analysis, modeling and decision-making.

[0071] After having the pre-processed and adjusted data, the next step is to reasonably divide it into training set, validation set and test set. The training set usually occupies most of the data, such as about 70%. It is used for model learning and parameter optimization. The model gradually grasps the rules and patterns in the data by repeatedly learning and adjusting the data in the training set. The proportion of the validation set is generally small, such as 20%. During the training process of the model, the validation set is periodically used to evaluate the performance of the model on unseen data. By comparing the performance indicators of the model on the validation set at different training stages, such as accuracy, loss value, etc., it can be determined whether the model is overfitting or underfitting, so as to adjust the model's hyperparameters, such as learning rate, number of layers, number of nodes, etc., to optimize the performance of the model. The proportion of the test set is usually about 10%. It is used to finally evaluate the generalization ability and actual performance of the model after the model training and hyperparameter adjustment are completed. The data of the test set has not been touched by the model at all during the model training process, and can more realistically reflect the performance of the model on new data. When dividing the data, random sampling is usually used to ensure that each subset can represent the characteristics and distribution of the original data. At the same time, in order to ensure the stability and reliability of the results, different random seeds can be used for division in multiple experiments, and the results can be comprehensively evaluated. Select a suitable model architecture, such as a neural network, decision tree, etc., and perform initialization settings. Use the data of the training set to input the model, calculate the error between the predicted value and the true value, and use the optimization algorithm to continuously adjust the model parameters to optimize the model's performance and achieve model construction.

[0072] Data augmentation is a technical means to increase data diversity by randomly transforming and expanding the original data. First, analyze the data characteristics and distribution in the training set and the validation set, and then, based on these characteristics, use a series of data augmentation methods to adjust the processed data. Data augmentation methods include: random rotation, for image data, it can be randomly rotated by a certain angle; random scaling: randomly enlarging or reducing the data; flipping, such as horizontal flipping and vertical flipping; adding noise, adding random noise to the data; color transformation, for image data, adjusting the value of the color channel. Through these data augmentation methods, the original data is expanded and diversified. Training adjustment enhanced data is obtained. These enhanced data enable the model to learn more changes, improve the generalization ability and robustness of the model, and reduce the risk of overfitting.

[0073] After completing data augmentation and obtaining training-adjusted augmented data, use a specially reserved test set to evaluate and test it. Input the data in the test set into the model trained on the training-adjusted augmented data. The model will calculate and reason based on the input data and output the corresponding results. Then, based on these test results, perform predictions of the adjusted results. This may involve comparing and analyzing the output of the model with the actual expected results, considering various indicators and factors, such as the size of the error, the degree of trend conformity, etc. By comprehensively evaluating and analyzing the test results, the prediction result is finally determined. This prediction result reflects the performance and ability of the model when facing new and unseen data.

[0074] After obtaining the load data in the forecast results, in order to comprehensively and accurately evaluate the load fluctuation, a variety of quantitative indicators and analysis methods will be used. Calculate the absolute fluctuation amplitude of the load. This can be obtained by subtracting the minimum load value from the maximum load value. Secondly, calculate the relative fluctuation amplitude, which can be obtained by dividing the absolute fluctuation amplitude by the average load value. Then, analyze the frequency of load fluctuations and observe the number of times the load has increased or decreased significantly in a specific time period. The periodicity of load fluctuations will also be considered. Check whether there is a certain pattern, such as similar fluctuation patterns appearing at fixed time intervals. In addition, calculate the rate of load change, that is, the size of the load change per unit time. Combine the indicators of the above multiple aspects, set weights for these indicators, integrate and calculate them, and finally get the prediction evaluation coefficient.

[0075] If the prediction evaluation coefficient indicates that the load fluctuation is small and within an acceptable range, the adjustment strategy is to maintain the current operating state and control parameters unchanged, or to make some slight optimizations, such as fine-tuning the charging and discharging thresholds of the energy storage system. Assuming that the prediction evaluation coefficient shows that the load fluctuation is close to zero, it means that the load is very stable. At this time, there may be no need for major adjustments, and only the system status needs to be monitored. When the prediction evaluation coefficient is large, it means that the load fluctuation exceeds the expected or acceptable level, and a more active adjustment strategy is needed. This includes increasing the capacity of energy storage equipment to enhance the buffering capacity for load fluctuations; optimizing the power distribution algorithm to allocate power resources more accurately; adjusting the output power of power generation equipment to make it more adaptable to load changes; or introducing more advanced control technology to improve the response speed and stability of the system. Flexible formulation of corresponding adjustment strategies to achieve stable and optimized operation of the hybrid microgrid load.

[0076] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0077] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0078] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A hybrid microgrid coordinated control system for load fluctuation smoothing, characterized in that: The system comprises: A load forecasting result acquisition module, which is used to construct a load forecasting channel, synchronize the hybrid microgrid to the load forecasting channel, and obtain a load forecasting result; A power allocation table formulation module, wherein the power allocation table formulation module is used to formulate a power allocation table according to the load forecast result combined with the power constraint interval, wherein the power allocation table includes a first power allocation scheme and a second power allocation scheme; A power fluctuation curve drawing module, wherein the power fluctuation curve drawing module is used to execute the first power allocation scheme and the second power allocation scheme to perform real-time monitoring on the hybrid microgrid, and draw a power fluctuation curve according to the monitoring results; An adjustment data set generation module, the adjustment data set generation module is used to perform feedback response according to the power fluctuation curve diagram, dynamically adjust the hybrid microgrid according to the response instruction, and generate an adjustment data set; An intelligent smoothing module is used to construct an adjustment strategy according to the adjustment data set to intelligently smooth the load fluctuation of the hybrid microgrid.

2. The hybrid microgrid coordinated control system for load fluctuation smoothing according to claim 1, characterized in that: The load forecasting result acquisition module also includes: The load prediction channel includes a first load prediction branch and a second load prediction branch; Analyze the hybrid microgrid to obtain the AC subnet and DC subnet; Using a microgrid synchronous phasor measurement unit to collect data from the AC subnet and the DC subnet to obtain a plurality of AC data and a plurality of DC data; Synchronizing the AC subnet to the first load prediction branch based on the multiple AC data to generate a first load prediction result; Synchronizing the DC subnet to the second load forecasting branch based on the multiple DC data to generate a second load forecasting result; The first load forecasting result and the second load forecasting result are integrated to generate the load forecasting result.

3. The hybrid microgrid coordinated control system for load fluctuation smoothing according to claim 2, characterized in that: The load forecasting result acquisition module also includes: Synchronizing the AC subnetwork to the first load forecasting branch, and capturing the load cycle trend of the AC subnetwork using a time series analysis method; Performing multivariate regression analysis on the plurality of communication data to generate a data variable regression parameter set; Perform load forecasting on the data variable regression parameter set according to the load cycle trend to generate the first load forecasting result.

4. The hybrid microgrid coordinated control system for load fluctuation smoothing according to claim 2, characterized in that: The power allocation table formulation module also includes: Constructing a power allocation model based on the load forecast result and the power constraint interval; The operation cycle of the hybrid microgrid is retrieved to determine multiple time scales; Performing power distribution on the hybrid microgrid according to the multiple time scales by using the power distribution model to obtain multiple power distribution data; Mapping and associating the plurality of power allocation data to obtain a data association factor; Constructing a power allocation table based on the data association factor and the plurality of power allocation data; According to the first load forecast result, the power allocation table is traversed for matching to obtain a first power allocation scheme for the AC subnet; The power allocation table is traversed and matched according to the second load forecast result to obtain a second power allocation scheme for the DC subnet.

5. The hybrid microgrid coordinated control system for load fluctuation smoothing according to claim 2, characterized in that: The power fluctuation curve drawing module also includes: Execute the first power allocation scheme to perform demand analysis on the AC subnet to generate a stable power demand and a peak power demand; Based on the stable power demand and the peak power demand, the AC subnet is monitored in real time to generate first power consumption data; Executing the second power allocation scheme to perform demand analysis on the DC subnet to generate power demand and load power demand; Based on the power generation demand and the load power demand, the DC subnet is monitored in real time to generate second power consumption data; The first power consumption data and the second power consumption data are added to the monitoring results of the hybrid microgrid.

6. The hybrid microgrid coordinated control system for load fluctuation smoothing according to claim 5, characterized in that: The adjustment data set generation module also includes: Performing feature analysis on the power fluctuation curve diagram based on the peak power demand and the load power demand to obtain a plurality of power fluctuation features; Perform fluctuation identification based on the multiple power fluctuation characteristics to determine power fluctuation period information, power fluctuation frequency information, and power fluctuation extreme value information; Retrieve historical power fluctuation logs to set fluctuation thresholds and build a feedback mechanism; Synchronize the power fluctuation period information, the power fluctuation frequency information, and the power fluctuation extreme value information to the feedback mechanism to determine whether the fluctuation value of the hybrid microgrid is within the fluctuation threshold range; If the fluctuation value of the hybrid microgrid is not within the fluctuation threshold range, a feedback response is triggered to identify a power fluctuation mode, and the power fluctuation mode is added to the response instruction.

7. The hybrid microgrid coordinated control system for load fluctuation smoothing according to claim 6, characterized in that: The adjustment data set generation module also includes: Based on the power fluctuation period information and the power fluctuation frequency information, the power fluctuation extreme value information is marked to obtain a plurality of peak points and a plurality of valley points; Executing the response instruction based on the multiple peak points and the multiple valley points to perform dynamic energy storage peak regulation and obtain a dynamic peak regulation result; Based on the multiple valley points, dynamic energy storage leveling is performed in combination with the dynamic peak load regulation result to obtain a dynamic leveling result; The dynamic peak adjustment result and the dynamic leveling result are normalized to generate the adjustment data set.

8. The hybrid microgrid coordinated control system for load fluctuation smoothing according to claim 1, characterized in that: The intelligent smoothing module also includes: Performing preprocessing based on the adjustment data set to generate adjustment processing data; Dividing the adjusted processed data into a training set, a validation set, and a test set; Performing data enhancement on the adjusted processed data according to the training set and the validation set to generate training adjusted enhanced data; Testing the training adjustment and enhancement data based on the test set, and performing adjustment result prediction according to the test result to determine the prediction result; evaluating load fluctuations of the prediction results to generate a prediction evaluation coefficient; The adjustment strategy is formulated according to the prediction evaluation coefficient.

Images