New energy microgrid multi-objective optimization control method, device, equipment and medium

Abstract

The invention relates to a new energy micro-grid multi-objective optimization control method and device, equipment and a medium. The method comprises the following steps: firstly, acquiring real-time output data of renewable energy sources of the micro-grid, operation parameters of energy storage equipment and load demand records, and performing time sequence analysis to obtain an energy output variation rule and a load demand curve; constructing a multi-objective function model containing economy-stability double objectives and a coupling correction term based on the law and the curve, generating an initial balance scheme through linear programming, and iteratively determining resource collaborative allocation parameters through a particle swarm optimization algorithm; the distribution parameters and the energy storage available capacity are input into the agent model, a real-time optimization decision instruction is generated, and an economic evaluation value is calculated in combination with running log data; and if the evaluation value does not reach the power grid stability threshold value, recalculating a balance scheme through a multi-objective function model, and generating a standardized equipment instruction set. According to the method, collaborative optimization of economy and stability of the micro-grid is realized, and stable operation of the micro-grid in different scenes is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of microgrid operation and control technology, and in particular relates to multi-objective optimization control methods, devices, equipment and media for new energy microgrids. Background Technology

[0002] As the global energy transition deepens, renewable energy microgrids, due to their ability to efficiently integrate distributed photovoltaic and wind power, have become a core carrier for the local consumption of distributed energy, and their application is increasingly widespread in scenarios such as power supply to industrial parks and energy replenishment in remote areas. However, renewable energy microgrids face significant technical bottlenecks in operation: on the one hand, renewable energy output is highly volatile due to natural conditions (sunlight, wind speed, etc.), leading to insufficient energy supply stability; on the other hand, user loads (industrial impact loads, residential loads, etc.) exhibit diversified and uncertain characteristics, exacerbating the risk of supply-demand mismatch, and the energy storage system, as a key device for smoothing fluctuations, further increases the complexity of operation and control due to its charging and discharging efficiency and state of charge constraints. Traditional renewable energy microgrid control methods often focus on optimizing a single objective, either independently considering economic efficiency or unilaterally emphasizing stability, lacking a coordinated balance between the two objectives, easily leading to contradictions such as "stability exceeding limits when cost is optimal" or "energy consumption surging when stability is guaranteed"; at the same time, some methods have not established a dynamic correlation mechanism between real-time data and control strategies, making it difficult to adapt to the complex and ever-changing operating conditions of microgrids and failing to meet the actual needs of high reliability and high efficiency operation of large-scale renewable energy microgrids. Summary of the Invention

[0003] Therefore, it is necessary to provide a multi-objective optimization control method, device, equipment, and medium for new energy microgrids that can achieve coordinated control of the economy and stability of new energy microgrids, addressing the aforementioned technical problems.

[0004] Firstly, this application provides a multi-objective optimization control method for new energy microgrids, including:

[0005] Key operational data in the microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage devices, and load demand records, are acquired. Time series analysis is used to process the key operational data to obtain the variation pattern of energy output and the load demand curve.

[0006] Based on the variation pattern and load demand curve, a multi-objective function model is constructed to obtain a multi-objective balancing scheme. The multi-objective balancing scheme is iteratively calculated using the particle swarm optimization algorithm to determine the resource collaborative allocation parameters.

[0007] Resource collaborative allocation parameters and available capacity data of energy storage devices are input into an agent model trained based on machine learning algorithms to generate real-time optimization decision instructions. The device interaction control signals in the real-time optimization decision instructions are extracted and combined with operation log data to calculate the economic evaluation value.

[0008] If the economic assessment value is lower than the power grid stability threshold, the balance scheme data is recalculated through a multi-objective function model. Based on the balance scheme data, the final optimized control sequence is determined, and a standardized equipment instruction set is generated.

[0009] In one embodiment, key operational data of the microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage devices, and load demand records, are acquired. Time series analysis is then used to process the key operational data to obtain the variation patterns of energy output and the load demand curve, including:

[0010] Acquire real-time power output data of renewable energy generation, operating parameters of energy storage devices, and load demand records in the microgrid; real-time power output data includes instantaneous power generation values ​​in the continuous time dimension; device operating parameters include state of charge, real-time charging and discharging power, and device terminal voltage in the continuous time dimension; load demand records include historical load data and real-time load monitoring data of the microgrid.

[0011] A time series decomposition algorithm is used to synchronously decompose real-time power output data and equipment operating parameters to obtain the periodic characteristics, random variation characteristics and change characteristics of power output.

[0012] A joint analysis model is constructed using time series analysis to investigate periodicity, random variation, and change characteristics. By fitting and validating the joint analysis model, the variation patterns of energy output are determined. These variation patterns include fluctuation amplitude, frequency, and duration.

[0013] Key fluctuation parameters, including the maximum fluctuation difference, fluctuation frequency threshold, and continuous fluctuation duration, are extracted from the variation patterns. Correlation analysis is then conducted using historical load data of the microgrid to identify potential influencing factors affecting load uncertainty patterns. These potential influencing factors include the correlation factor between power fluctuation amplitude and load peak-valley difference.

[0014] If the quantified value of a potential influencing factor exceeds a preset threshold range, the real-time load monitoring data of the microgrid is called to dynamically correct the real-time load monitoring data and generate a corrected load demand curve; the load demand curve includes time nodes, load values, and curve confidence.

[0015] In one embodiment, a multi-objective function model is constructed based on the variation pattern and load demand curve to obtain a multi-objective balancing scheme. The multi-objective balancing scheme is iteratively calculated using a particle swarm optimization algorithm to determine resource collaborative allocation parameters, including:

[0016] A multi-objective function model is constructed based on the variation pattern and load demand curve; the multi-objective function model data includes economic index parameters and stability index parameters.

[0017] Based on the multi-objective function model, constraints for resource collaborative allocation are set, and an initial scheme for resource collaborative allocation is constructed using linear programming to obtain a preliminary multi-objective balance scheme. The multi-objective balance scheme includes the renewable energy output allocation ratio, the charging and discharging time and power of energy storage equipment, and the load priority allocation coefficient.

[0018] The multi-objective equilibrium scheme is iteratively calculated using the particle swarm optimization algorithm. The objective functions are minimizing the economic index and optimizing the stability index. The fitness function is used to evaluate the merits of each generation of the scheme. The optimization parameters of the multi-objective equilibrium scheme are determined by iterating until the convergence condition is met.

[0019] If the quantized values ​​of the optimization parameters all meet the corresponding preset balance thresholds, the final resource allocation scheme will be output; the resource allocation scheme specifies the specific resource allocation for each time period and each device.

[0020] Load adjustment instructions are generated based on the resource allocation scheme; the load adjustment instructions include the instruction execution subject, execution period, adjustment target value, and adjustment accuracy requirements.

[0021] The system acquires execution feedback data of load regulation commands and judges the feedback data based on the preset target value of the commands. The feedback data includes command execution completion rate, actual resource distribution data after execution, and power grid operation status data.

[0022] If the feedback data deviates from the expected target, the deviation from the target value in the feedback data is extracted, and the weight coefficients and constraints in the multi-objective function model are updated based on the deviation to obtain the updated multi-objective function model parameters.

[0023] Based on the updated multi-objective function model parameters, the particle swarm optimization algorithm is used again for iterative calculation to obtain the updated resource collaborative allocation parameters.

[0024] In one embodiment, the multi-objective function model is represented by the following formula:

[0025]

[0026] in, This represents the economic objective function. This represents the stability objective function. This represents the overall operating cost, which is calculated as: Overall operating cost = Equipment maintenance cost per unit time + Electricity purchase cost. This represents the cost of energy loss, which is calculated as: Renewable energy curtailment loss + Energy storage charging and discharging loss. The voltage stability coefficient is calculated as the reciprocal of the deviation rate between the actual voltage and the rated voltage. The frequency stability coefficient is calculated as the reciprocal of the deviation rate between the actual frequency and the rated frequency. This represents the power fluctuation coefficient, which is calculated as: maximum difference in renewable energy output per unit time / rated power. This represents the load forecast deviation rate, where the load forecast deviation rate = / Predicted load, , , , This represents the weighting coefficient, which is set according to the microgrid's operating scenario. , This represents the coupling correction coefficient, obtained by fitting historical data. This indicates the real-time output of renewable energy. Indicates the state of charge of the energy storage. This indicates the real-time load power of the microgrid. This indicates the real-time charging and discharging power of the energy storage device. This indicates the real-time power exchange between the microgrid and the main grid. Represents a time variable.

[0027] In one embodiment, resource collaborative allocation parameters and available capacity data of energy storage devices are input into a proxy model trained based on a machine learning algorithm to generate real-time optimization decision instructions. Device interaction control signals from these instructions are extracted and combined with operational log data to calculate an economic evaluation value, including:

[0028] Obtain available capacity data for energy storage devices; available capacity data includes current state of charge, maximum charge / discharge power, and remaining available capacity.

[0029] A dynamic adjustment model is constructed by combining available capacity data with resource collaborative allocation parameters. The model is then used to generate preliminary optimization decision instructions, which include the target operating power, adjustment period, and response delay requirements for each device.

[0030] A machine learning algorithm is used to train the agent model. The initial optimization decision instructions are input into the trained agent model. By simulating the execution effect of instructions under different operating scenarios, the corrected real-time optimization decision instructions are output. The training data of the agent model are sample pairs from the historical operation of the microgrid.

[0031] The system analyzes real-time optimization decision commands, extracts equipment interactive control signals, and determines the target operating status data of each device. The control signals are classified by equipment type, including output regulation signals of renewable energy power generation equipment, charging and discharging control signals of energy storage equipment, and load switching signals of load controllers.

[0032] Collect microgrid operation log data; the operation log data includes the actual operating status data of each device, grid operation parameters, energy consumption and cost data.

[0033] Based on the target operating status data, the deviation rate and unit electricity cost of the operating log data are calculated to obtain the economic evaluation value; the economic evaluation value includes unit electricity cost, renewable energy utilization rate, and energy storage operation and maintenance cost.

[0034] In one embodiment, if the economic assessment value is lower than the grid stability threshold, the balancing scheme data is recalculated iteratively using a multi-objective function model. Based on the balancing scheme data, the final optimized control sequence is determined, and a standardized equipment instruction set is generated, including:

[0035] The economic assessment value is compared with the preset power grid stability threshold to obtain the comparison result; the power grid stability threshold includes the allowable range of voltage fluctuation, the frequency deviation threshold, and the upper limit of the power deficit rate.

[0036] If the comparison result shows that the economic evaluation value is lower than the stability threshold, the updated real-time operation data of the microgrid will be used as input, and the updated balance scheme data will be obtained by iterating again through the multi-objective function model. The balance scheme data includes the adjusted renewable energy output allocation ratio, energy storage charging and discharging parameters, and load priority coefficient.

[0037] The balance scheme data is combined with the operating constraints of each device and the power limit of the power grid interaction, and integrated through time sequence arrangement to form the final optimized control sequence; the final optimized control sequence includes time nodes, device types, control parameters, and execution priorities.

[0038] The final optimized control sequence is analyzed, and a standardized equipment instruction set is generated according to the equipment type. The equipment instruction set includes instruction identifier, execution subject, execution time period, control target value, adjustment accuracy requirement, and safe operation boundary.

[0039] Secondly, this application also provides a multi-objective optimization control device for new energy microgrids, the device comprising:

[0040] The data acquisition and analysis module is used to acquire key operational data in the microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage equipment, and load demand records. Time series analysis is used to process the key operational data to obtain the variation pattern of energy output and the load demand curve.

[0041] The multi-objective model optimization module is used to construct a multi-objective function model based on the variation law and load demand curve to obtain a multi-objective balance scheme. The multi-objective balance scheme is iteratively calculated through the particle swarm optimization algorithm to determine the resource collaborative allocation parameters.

[0042] The economic evaluation module is used to input resource collaborative allocation parameters and available capacity data of energy storage devices into an agent model trained based on machine learning algorithms, generate real-time optimization decision instructions, extract equipment interaction control signals from the real-time optimization decision instructions and combine them with operation log data to calculate the economic evaluation value.

[0043] The equipment instruction generation module is used to recalculate the balance scheme data through a multi-objective function model if the economic evaluation value is lower than the power grid stability threshold, determine the final optimized control sequence based on the balance scheme data, and generate a standardized equipment instruction set.

[0044] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0045] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned method.

[0046] The aforementioned multi-objective optimization control method, device, computer equipment, and storage medium for new energy microgrids first acquire real-time renewable energy output data, energy storage device operating parameters, and load demand records. Time series analysis is then used to obtain the energy output variation pattern and load demand curve. Based on this pattern and curve, a multi-objective function model with dual economic and stability objectives and coupled correction terms is constructed. An initial balancing scheme is generated through linear programming, and resource collaborative allocation parameters are determined iteratively using a particle swarm optimization algorithm. The allocation parameters and available energy storage capacity are input into the trained surrogate model to generate real-time optimization decision instructions. Economic evaluation values ​​are calculated using operational log data. If the evaluation value does not reach the grid stability threshold, the balancing scheme is recalculated through the multi-objective function model to determine the final optimized control sequence and generate a standardized equipment instruction set. This method achieves coordinated optimization of microgrid economy and stability, solving the disconnect problem of single-objective optimization in traditional methods. The surrogate model and iterative mechanism adapt to dynamic operating conditions such as output fluctuations and load changes, ensuring stable operation under different scenarios. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart of a multi-objective optimization control method for new energy microgrids provided in an embodiment of the present invention;

[0049] Figure 2 The structural block diagram of the multi-objective optimization control device for new energy microgrids provided in the embodiments of the present invention is shown. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0051] In one embodiment, such as Figure 1 As shown, this application provides a multi-objective optimization control method for new energy microgrids, which may include the following steps:

[0052] Step S101: Obtain key operating data in the microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage devices, and load demand records. Use time series analysis to process the key operating data to obtain the variation pattern of energy output and the load demand curve.

[0053] First, three types of key operational data are acquired during the microgrid operation: real-time renewable energy power generation output data (including instantaneous power generation values ​​over a continuous time dimension, such as real-time power generation monitoring data from photovoltaic panels and wind turbines), energy storage device operating parameters (including state of charge (SOC), real-time charging and discharging power, and device terminal voltage over a continuous time dimension, collected from the energy storage converter PCS and energy storage management system), and load demand records (including historical load data and real-time load monitoring data of the microgrid, obtained through user-side smart meters and load controllers). Then, time series analysis is used to synchronously process these three types of data: first, time series decomposition algorithms (such as STL and wavelet decomposition) are used to decompose the real-time renewable energy power generation data and energy storage device operating parameters to obtain the periodic characteristics, random variation characteristics, and variation characteristics of energy storage device operating parameters; energy storage devices include battery-based energy storage devices such as lithium-ion batteries, lead-acid batteries, and flow batteries, as well as physical energy storage devices such as pumped hydro storage, compressed air storage, flywheel energy storage, and supercapacitors. Then, a joint analysis model is constructed using time series analysis methods (such as ARIMA and GARCH) to fit and verify the above characteristics, and to determine the variation pattern of energy output (including quantitative indicators such as fluctuation amplitude, frequency, and duration). Finally, key fluctuation parameters are extracted from the variation pattern and correlation analysis is performed in combination with historical load data in the load demand record. If potential influencing factors exceed the preset threshold, the load demand record is dynamically corrected, and finally, the variation pattern of energy output and the corrected load demand curve are output.

[0054] Step S102: Based on the variation law and load demand curve, construct a multi-objective function model to obtain a multi-objective balancing scheme. Iteratively calculate the multi-objective balancing scheme using the particle swarm optimization algorithm to determine the resource collaborative allocation parameters.

[0055] Furthermore, using the obtained energy output variation patterns and load demand curves as core inputs, a multi-objective function model is constructed. This model includes economic indicator parameters (comprehensive operating cost, energy loss cost) and stability indicator parameters (voltage stability coefficient, frequency stability coefficient), and introduces a coupled correction term for power fluctuation coefficient and load forecast deviation rate to achieve synergistic consideration of dual objectives. Based on this multi-objective function model, constraints for resource collaborative allocation are set (including upper limits for renewable energy output, upper and lower limits for the state of charge of energy storage devices, and power limits for interaction between microgrids and the main grid). An initial scheme for resource collaborative allocation is constructed using linear programming, resulting in a preliminary multi-objective balance scheme (including renewable energy output allocation ratio, energy storage device charging and discharging periods and power, and load priority allocation coefficient). Subsequently, the particle swarm optimization algorithm is used to iteratively calculate the initial balancing scheme. The objective functions are minimizing the economic index and optimizing the stability index. The fitness function is used to evaluate the merits of each generation of schemes. After iterating until the preset convergence condition is met (such as the objective function value deviation of 10 consecutive generations of schemes being less than 0.1%), the final resource coordination allocation parameters are output. These parameters clarify the specific resource allocation rules for renewable energy, energy storage equipment, and load in each time period.

[0056] Step S103: Input the resource collaborative allocation parameters and the available capacity data of the energy storage device into the agent model trained based on the machine learning algorithm to generate real-time optimization decision instructions. Extract the device interaction control signals in the real-time optimization decision instructions and combine them with the operation log data to calculate the economic evaluation value.

[0057] First, available capacity data is acquired for battery-type devices such as lithium-ion batteries, lead-acid batteries, and flow batteries, as well as energy storage devices such as pumped hydro storage, compressed air energy storage systems, flywheel energy storage devices, and supercapacitors. Specifically, this includes the current state of charge (SOC), maximum charge / discharge power, and remaining available capacity. This data is collected from the real-time monitoring module of the energy storage devices. The determined resource allocation parameters and the aforementioned available energy storage capacity data are input into a proxy model. This proxy model is trained using machine learning algorithms (such as random forest and LSTM), with training data consisting of sample pairs of "preliminary optimization decision instructions - actual operating effects" from the microgrid's historical operation (including instruction content, post-execution grid stability data, and economic data). The proxy model simulates the instruction execution effects under different operating scenarios and outputs corrected real-time optimization decision instructions. These real-time optimization decision instructions are then parsed to extract device interaction control signals categorized by device type (including output adjustment signals from renewable energy generation equipment, charge / discharge control signals from energy storage devices, and load switching signals from load controllers). Simultaneously, microgrid operation log data is collected, which includes actual operating status data of each device, grid operating parameters (voltage, frequency), energy consumption and cost data (electricity purchase cost, operation and maintenance energy consumption). Based on the target operating status data corresponding to the device interactive control signals, deviation analysis (calculating the deviation rate between actual value and target value) and cost accounting (calculating unit electricity cost based on actual energy consumption) are performed on the operation log data. Finally, an economic evaluation value including unit electricity cost, renewable energy utilization rate, and energy storage operation and maintenance cost is obtained.

[0058] Step S104: If the economic evaluation value is lower than the power grid stability threshold, the balance scheme data is recalculated through a multi-objective function model. Based on the balance scheme data, the final optimized control sequence is determined, and a standardized equipment instruction set is generated.

[0059] Specifically, the calculated economic evaluation value is compared with a preset grid stability threshold, which includes the allowable voltage fluctuation range, frequency deviation threshold, and power deficit rate upper limit. This threshold is preset based on microgrid operating standards and actual operating conditions. If the quantified value of the economic evaluation value is lower than the above stability threshold (i.e., it meets the conditions of "economic performance meets the standard but stability does not meet the standard" or "both objectives do not meet the standard"), the multi-objective balancing scheme recalculation process is triggered. The updated real-time operating data of the microgrid (including the latest renewable energy output, load demand, and energy storage status data) is used as input, and the multi-objective function model in step S102 is used to re-iterate the calculation to obtain the updated balancing scheme data (including the adjusted renewable energy output allocation ratio, energy storage charging and discharging parameters, and load priority coefficient). Combining the operating constraints of each device (such as the upper limit of distributed power output and energy storage SOC limit) and the grid interaction power limit, the updated balancing scheme data is time-series arranged and integrated to form the final optimized control sequence containing time nodes, device types, control parameters, and execution priorities. Finally, the final optimized control sequence is analyzed, and a standardized set of equipment instructions is generated according to the equipment type (distributed power source, energy storage system). This set of instructions includes instruction identifier, execution subject, execution time period, control target value, regulation accuracy requirements, and safe operation boundary, which is used to distribute to the corresponding equipment for execution to ensure the stable and efficient operation of the microgrid.

[0060] The aforementioned multi-objective optimization control method for new energy microgrids first acquires real-time renewable energy output data, energy storage device operating parameters, and load demand records. Time series analysis is then used to obtain the energy output variation pattern and load demand curve. Based on this pattern and curve, a multi-objective function model is constructed, incorporating both economic and stability objectives and coupled correction terms. An initial balancing scheme is generated through linear programming, and resource collaborative allocation parameters are determined iteratively using a particle swarm optimization algorithm. The allocation parameters and available energy storage capacity are input into the trained surrogate model to generate real-time optimization decision instructions. Economic evaluation values ​​are calculated using operational log data. If the evaluation value does not reach the grid stability threshold, the balancing scheme is recalculated through the multi-objective function model to determine the final optimized control sequence and generate a standardized equipment instruction set. This method achieves coordinated optimization of microgrid economy and stability, solving the disconnect problem of single-objective optimization in traditional methods. The surrogate model and iterative mechanism adapt to dynamic operating conditions such as output fluctuations and load changes, ensuring stable operation under different scenarios.

[0061] In one embodiment, acquiring key operational data from the microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage devices, and load demand records, and processing the key operational data using time series analysis to obtain the variation pattern of energy output and the load demand curve, may include the following steps:

[0062] Step S201: Obtain real-time power output data of renewable energy generation, operating parameters of energy storage equipment, and load demand records in the microgrid; real-time power output data includes instantaneous power generation values ​​in the continuous time dimension; equipment operating parameters include state of charge, real-time charging and discharging power, and equipment terminal voltage in the continuous time dimension; load demand records include historical load data and real-time load monitoring data of the microgrid.

[0063] Step S202: The real-time power output data and equipment operating parameters are synchronously decomposed using a time series decomposition algorithm to obtain the periodic characteristics, random variation characteristics and change characteristics of the power output.

[0064] Step S203: A joint analysis model is constructed using time series analysis to analyze the periodic characteristics, random variation characteristics, and change characteristics. The variation pattern of energy output is determined by fitting and verifying the joint analysis model. The variation pattern includes fluctuation amplitude, frequency, and duration.

[0065] Step S204: Extract key fluctuation parameters, including maximum fluctuation difference, fluctuation frequency threshold, and continuous fluctuation duration, from the variation pattern, and perform correlation analysis with historical load data of the microgrid to obtain potential influencing factors affecting the load uncertainty mode; potential influencing factors include the correlation factor between power fluctuation amplitude and load peak-valley difference.

[0066] Step S205: If the quantified value of the potential influencing factor exceeds the preset threshold range, the real-time load monitoring data of the microgrid is called to dynamically correct the real-time load monitoring data and generate a corrected load demand curve; the load demand curve includes time nodes, load values, and curve confidence.

[0067] Specifically, the process first acquires three core data types required for microgrid operation: real-time renewable energy generation output data, energy storage device operating parameters, and load demand records. The real-time renewable energy generation output data includes instantaneous power generation values ​​over a continuous time dimension; the energy storage device operating parameters include state of charge, real-time charging and discharging power, and device terminal voltage over a continuous time dimension; and the load demand records include historical load data and real-time load monitoring data of the microgrid. Subsequently, a time series decomposition algorithm is used to synchronously decompose the real-time renewable energy generation output data and the energy storage device operating parameters. This process yields the periodic characteristics and random variability of power output, as well as the variation characteristics of the energy storage device operating parameters. Next, a joint analysis model is constructed using time series analysis methods to address these periodic, random, and variability characteristics. Various features are substituted into the model for fitting and verification. After verification, the variation patterns of energy output are determined, specifically including quantitative indicators such as power fluctuation amplitude, fluctuation frequency, and fluctuation duration. Next, from the established energy output variation patterns, key fluctuation parameters, including the maximum fluctuation difference, fluctuation frequency threshold, and continuous fluctuation duration, are extracted. These key fluctuation parameters are then correlated with historical load data of the microgrid in the load demand record. Through analysis, potential influencing factors affecting load uncertainty patterns are identified. These potential influencing factors specifically include the correlation factor between power fluctuation amplitude and load peak-valley difference. Finally, it is determined whether the quantified values ​​of the potential influencing factors exceed a preset threshold range. If they do, the real-time load monitoring data of the microgrid in the load demand record is retrieved to dynamically correct the original real-time load monitoring data. After correction, a corrected load demand curve is generated, which includes specific time nodes, corresponding load values ​​at those time nodes, and a confidence index for the curve.

[0068] This embodiment achieves precise processing of core microgrid operational data through a coherent logic of hierarchical data acquisition, synchronous decomposition, joint modeling, correlation analysis, and dynamic correction. On one hand, the data acquisition process covers three key categories: renewable energy, energy storage devices, and loads, clearly defining the time dimension attributes and specific content of each data point. This provides comprehensive and standardized foundational data for subsequent processing, avoiding analytical biases caused by missing or non-standard data. On the other hand, the synchronous application of time series decomposition algorithms and the construction of joint analysis models simultaneously capture multi-dimensional characteristics of energy output and energy storage operation, ensuring the accuracy and completeness of energy output variation patterns. Furthermore, the correlation analysis between key fluctuation parameters and historical loads further clarifies the sources of load uncertainty, while the dynamic correction mechanism adjusts the load curve according to actual operating conditions, ensuring the real-time performance and reliability of the curve. The overall process provides accurate and reliable input data support for subsequent multi-objective optimization control of new energy microgrids, effectively solving the optimization control bias problems caused by "independent analysis of single data" and "statically fixed load curves" in traditional data processing. This lays a data foundation for the coordinated optimization of microgrid economy and stability.

[0069] In one embodiment, a multi-objective function model is constructed based on the variation pattern and load demand curve to obtain a multi-objective balancing scheme. The multi-objective balancing scheme is then iteratively calculated using a particle swarm optimization algorithm to determine the resource collaborative allocation parameters. This may include the following steps:

[0070] Step S301: Construct a multi-objective function model based on the variation pattern and load demand curve; the multi-objective function model data includes economic index parameters and stability index parameters.

[0071] Step S302: Based on the multi-objective function model, set the constraints for resource collaborative allocation, and construct the initial scheme for resource collaborative allocation through linear programming to obtain a preliminary multi-objective balance scheme. The multi-objective balance scheme includes the renewable energy output allocation ratio, the charging and discharging time and power of energy storage equipment, and the load priority allocation coefficient.

[0072] Step S303: Based on the multi-objective equilibrium scheme, the particle swarm optimization algorithm is used for iterative calculation. The objective functions are minimizing the economic index and optimizing the stability index. The fitness function is used to evaluate the merits of each generation of the scheme. The iteration is continued until the convergence condition is met to determine the optimization parameters of the multi-objective equilibrium scheme.

[0073] Step S304: If the quantized values ​​of the optimization parameters all meet the corresponding preset balance thresholds, the final resource allocation scheme is output; the resource allocation scheme specifies the specific resource allocation amount for each time period and each device.

[0074] Step S305: Generate load adjustment instructions based on the resource allocation scheme; the load adjustment instructions include the instruction execution subject, execution period, adjustment target value, and adjustment accuracy requirements.

[0075] Step S306: Obtain the execution feedback data of the load adjustment command, and judge the feedback data based on the preset target value of the command; the feedback data includes the command execution completion rate, the actual resource distribution data after execution, and the power grid operation status data.

[0076] Step S307: If the feedback data deviates from the expected target, the deviation from the target value in the feedback data is extracted, and the weight coefficients and constraints in the multi-objective function model are updated according to the deviation to obtain the updated multi-objective function model parameters.

[0077] Step S308: Based on the updated multi-objective function model parameters, the particle swarm optimization algorithm is used again for iterative calculation to obtain the updated resource collaborative allocation parameters.

[0078] First, a multi-objective function model is constructed using the energy output variation pattern and load demand curve as input. This model includes economic indicators (such as overall operating cost and energy loss cost) and stability indicators (such as voltage stability coefficient and frequency stability coefficient). Based on this multi-objective function model, constraints for resource collaborative allocation are set (such as the upper limit of renewable energy output and the state of charge range of energy storage equipment). An initial scheme for resource collaborative allocation is constructed using linear programming, resulting in a preliminary multi-objective balance scheme. This scheme specifically includes the renewable energy output allocation ratio, the charging and discharging periods and power of energy storage equipment, and the load priority allocation coefficient. Subsequently, based on the preliminary multi-objective balance scheme, a particle swarm optimization algorithm is used for iterative calculations. The objective functions are minimizing economic indicators and optimizing stability indicators. The merits of each generation of schemes are evaluated using a fitness function. The iteration continues until a preset convergence condition is met (such as the deviation of the objective function value of 10 consecutive generations of schemes being less than 0.1%), at which point the optimization parameters of the multi-objective balance scheme are determined. If the quantified values ​​of the optimized parameters all meet the corresponding preset balance thresholds (e.g., unit energy cost ≤ 0.5 yuan / kWh, voltage fluctuation range ≤ ±2%), the final resource allocation scheme is output. This scheme specifies the specific resource allocation for each time period and each device (e.g., photovoltaic, energy storage, load). Based on the final resource allocation scheme, a load adjustment command is generated. This command includes the command execution entity (e.g., a regional load controller, energy storage converter), execution time period (e.g., 9:00-10:00), adjustment target value (e.g., reducing commercial load from 1000kW to 800kW), and adjustment accuracy requirements (e.g., adjustment error ≤ 5%). The execution feedback data of this load adjustment command is obtained. The feedback data includes the command execution completion rate, actual resource distribution data after execution (e.g., actual load value, actual charging and discharging power of energy storage), and grid operation status data (e.g., actual voltage, frequency). The feedback data is then evaluated based on the preset target value of the command. If the feedback data deviates from the expected target (e.g., execution completion rate is only 80%, actual voltage fluctuation reaches ±3%), then the deviation from the target value in the feedback data is extracted (e.g., load regulation deviation of 200kW, voltage deviation of 1%). Based on the deviation, the weight coefficients in the multi-objective function model (e.g., increasing the weight of the stability index) and constraints (e.g., relaxing the energy storage charging and discharging power limit by 5%) are updated to obtain the updated multi-objective function model parameters. Finally, based on the updated multi-objective function model parameters, the particle swarm optimization algorithm is used again for iterative calculation to obtain the updated resource collaborative allocation parameters.

[0079] This embodiment achieves dynamic adaptation of multi-objective optimization for new energy microgrids: On the one hand, the multi-objective function model simultaneously covers economic and stability indicators, avoiding the contradictions caused by traditional single-objective optimization, such as "optimal cost but insufficient stability" or "stability priority but soaring energy consumption." The application of particle swarm optimization algorithm, combined with clear convergence conditions, ensures the accuracy and reproducibility of optimization parameters. The final output resource allocation scheme and load regulation command elements are complete and can directly guide equipment execution. On the other hand, the feedback update mechanism dynamically adjusts model parameters and constraints by capturing command execution deviations in real time, enabling the optimization strategy to adapt to the dynamic operating conditions of microgrids with fluctuations in renewable energy output and changes in load demand, solving the problem that traditional static optimization cannot cope with changes in operating conditions. At the same time, all parameters in the process (such as balance threshold and deviation) are set or calculated based on the microgrid's collectable data. The algorithms (linear programming and particle swarm optimization) are mature and controllable, ensuring the practical feasibility of the technical solution and providing reliable support for the long-term stable and efficient operation of the microgrid.

[0080] In one embodiment, the multi-objective function model can be represented by the following formula:

[0081]

[0082] in, This represents the economic objective function. This represents the stability objective function. This represents the overall operating cost, which is calculated as: Overall operating cost = Equipment maintenance cost per unit time + Electricity purchase cost. This represents the cost of energy loss, which is calculated as: Renewable energy curtailment loss + Energy storage charging and discharging loss. The voltage stability coefficient is calculated as the reciprocal of the deviation rate between the actual voltage and the rated voltage. The frequency stability coefficient is calculated as the reciprocal of the deviation rate between the actual frequency and the rated frequency. This represents the power fluctuation coefficient, which is calculated as: maximum difference in renewable energy output per unit time / rated power. This represents the load forecast deviation rate, where the load forecast deviation rate = / Predicted load, , , , This represents the weighting coefficient, which is set according to the microgrid's operating scenario. , This represents the coupling correction coefficient, obtained by fitting historical data. This indicates the real-time output of renewable energy. Indicates the state of charge of the energy storage. This indicates the real-time load power of the microgrid. This indicates the real-time charging and discharging power of the energy storage device. This indicates the real-time power exchange between the microgrid and the main grid. Represents a time variable.

[0083] In this embodiment, the multi-objective function model possesses significant technical advantages by constructing a dual-objective optimization framework of "economy-stability": on the one hand, it uses... Minimize the economic optimization To maximize stability optimization, the model deeply correlates energy output power fluctuations with load forecast deviations through coupling correction terms. This overcomes the limitations of traditional models that independently consider dual objectives and separate the impact of fluctuations and loads, effectively avoiding the contradiction of "exceeding stability limits when cost is optimal" or "surge in energy consumption when stability is guaranteed," and reducing operational risks caused by the superposition of fluctuations and load deviations. On the other hand, the model explicitly incorporates the upper limit of renewable energy output, the energy storage state of charge boundary, and power balance. Constraints are imposed to ensure that the optimization results conform to the operating capacity of the equipment and the balance between power grid supply and demand, avoiding the problem of theoretical feasibility but practical infeasibility; in addition, , , , The weighting coefficients can be flexibly set according to different microgrid scenarios such as grid-connected / off-grid, and all parameters are derived from the operational data that can be collected from the microgrid. This ensures both the adaptability and accuracy of the model, as well as the practical feasibility of the technical solution, providing core mathematical model support for the stable, efficient, and economical operation of new energy microgrids under dynamic operating conditions.

[0084] In one embodiment, resource collaborative allocation parameters and available capacity data of energy storage devices are input into a proxy model trained based on a machine learning algorithm to generate real-time optimization decision instructions. The device interaction control signals extracted from these instructions are then combined with operational log data to calculate an economic evaluation value. This process may include the following steps:

[0085] Step S401: Obtain available capacity data of the energy storage device; available capacity data includes current state of charge, maximum charging and discharging power, and remaining available capacity.

[0086] Step S402: Combine available capacity data with resource collaborative allocation parameters to construct a dynamic adjustment model, and generate preliminary optimization decision instructions through model calculation; the preliminary optimization decision instructions include the target operating power, adjustment period, and response delay requirements of each device.

[0087] Step S403: The agent model is trained using a machine learning algorithm. The preliminary optimization decision instructions are input into the trained agent model. By simulating the execution effect of instructions under different operating scenarios, the corrected real-time optimization decision instructions are output. The training data of the agent model are sample pairs from the historical operation of the microgrid.

[0088] Step S404: Analyze the real-time optimization decision instructions, extract the equipment interactive control signals, and determine the target operating status data of each device; the control signals are classified by equipment type, including the output adjustment signals of renewable energy power generation equipment, the charging and discharging control signals of energy storage equipment, and the load switching signals of load controllers.

[0089] Step S405: Collect microgrid operation log data; the operation log data includes the actual operating status data of each device, grid operation parameters, energy consumption and cost data.

[0090] Step S406: Calculate the deviation rate and unit electricity cost of the operation log data based on the target operation status data to obtain the economic evaluation value; the economic evaluation value includes unit electricity cost, renewable energy utilization rate, and energy storage operation and maintenance cost.

[0091] First, the available capacity data of the energy storage equipment is acquired. This data specifically includes the current state of charge (SOC), maximum charging and discharging power, and remaining available capacity, which is directly collected from the real-time monitoring module of the energy storage equipment. This available capacity data is then combined with previously determined resource allocation parameters (including renewable energy output allocation ratios, load priority coefficients, etc.) to construct a dynamic adjustment model. This model incorporates power balance constraints and equipment operating boundary conditions. Through model calculation, preliminary optimization decision instructions are output. These instructions specify the target operating power of each device (e.g., 120kW for photovoltaic output and 60kW for energy storage discharge power), the adjustment period (e.g., 14:00-15:00), and the response delay requirements (e.g., instruction execution delay ≤100ms). A proxy model is trained using machine learning algorithms (such as Random Forest and Long Short-Term Memory (LSTM) networks). The training data consists of sample pairs of "preliminary optimization decision instructions - actual operating effects" from the historical operation of the microgrid (each sample pair includes the content of the historical preliminary instructions, grid stability data after instruction execution, and economic data). After the model is trained, the currently generated preliminary optimization decision instructions are input into the proxy model. The model corrects the preliminary instructions by simulating the execution effects of instructions under different operating scenarios, such as high load fluctuations and sudden drops in renewable energy output, and outputs real-time optimization decision instructions adapted to the current operating conditions. The real-time optimization decision instructions are analyzed to extract equipment interaction control signals categorized by equipment type: for renewable energy generation equipment, output adjustment signals (such as the active power setpoint of photovoltaic inverters) are extracted; for energy storage equipment, charge and discharge control signals (such as start / stop instructions and power instructions of energy storage converters PCS) are extracted; for load control equipment, load switching signals (such as switching instructions for adjustable industrial loads) are extracted. Based on these control signals, the target operating status data for each device is determined (such as maintaining the target state of charge of energy storage at 40%-80% and stabilizing the target power of commercial loads at 500kW). Microgrid operation log data is collected, encompassing actual operating status data for each device (e.g., actual photovoltaic output, actual energy storage charging and discharging power, and actual load power consumption), grid operating parameters (e.g., grid connection point voltage and frequency), and energy consumption and cost data (e.g., electricity purchase costs and energy consumption for energy storage device operation and maintenance). Based on the target operating status data for each device, the operation log data is processed: firstly, the deviation rate between actual operating status data and target values ​​is calculated (e.g., the deviation rate between actual load power and target power); secondly, the unit cost of electricity is calculated based on energy consumption and cost data (unit cost of electricity = total operating cost / total power generation). Finally, an economic assessment value is obtained, including unit cost of electricity, renewable energy utilization rate (renewable energy utilization rate = actual renewable energy consumption / total power generation × 100%), and energy storage operation and maintenance costs.

[0092] In this embodiment, the dynamic adjustment model combines the actual available capacity of the energy storage device with resource coordination and allocation parameters to generate initial instructions, avoiding execution failures caused by instructions being out of sync with the device's operational capabilities. The proxy model corrects instructions based on historical samples and multi-scenario simulations, improving the adaptability of real-time optimization decision instructions to dynamic microgrid operating conditions (such as renewable energy output fluctuations and load surges), and solving the problem of insufficient flexibility of traditional fixed instructions in responding to changes in operating conditions. Control signals are extracted according to device type to ensure clear operational objectives for each device. The economic evaluation uses target operating status data as a benchmark, and quantifies the evaluation results through deviation rate and cost accounting, reflecting both the device execution accuracy and the microgrid's operational efficiency. This provides a reliable basis for subsequent judgments on whether iterative optimization of the multi-objective balance scheme is necessary, effectively supporting the maximization of economic efficiency in new energy microgrids under the premise of safe operation.

[0093] In one embodiment, if the economic assessment value is lower than the grid stability threshold, the balancing scheme data is recalculated iteratively using a multi-objective function model. Based on the balancing scheme data, the final optimized control sequence is determined, and a standardized equipment instruction set is generated. This may include the following steps:

[0094] Step S501: Compare the economic evaluation value with the preset power grid stability threshold to obtain the comparison result; the power grid stability threshold includes the allowable range of voltage fluctuation, the frequency deviation threshold, and the upper limit of the power deficit rate.

[0095] Step S502: If the comparison result shows that the economic evaluation value is lower than the stability threshold, then the updated real-time operation data of the microgrid is used as input, and the updated balance scheme data is obtained by iterating again through the multi-objective function model. The balance scheme data includes the adjusted renewable energy output allocation ratio, energy storage charging and discharging parameters, and load priority coefficient.

[0096] Step S503: Combine the balancing scheme data with the operating constraints of each device and the power interaction limit of the power grid, and integrate them through time sequence arrangement to form the final optimized control sequence; the final optimized control sequence includes time nodes, device types, control parameters, and execution priorities.

[0097] Step S504: Analyze the final optimized control sequence and generate a standardized equipment instruction set according to the equipment type; the equipment instruction set includes instruction identifier, execution subject, execution time period, control target value, adjustment accuracy requirement, and safe operation boundary.

[0098] Specifically, the economic assessment values ​​(including unit electricity cost, renewable energy utilization rate, and energy storage operation and maintenance cost) calculated in the early stage are first quantitatively compared with the preset grid stability threshold to generate comparison results. Among them, the grid stability threshold is the core safety and performance standard for microgrid operation, which specifically includes the allowable range of voltage fluctuation (such as ±2% of rated voltage), frequency deviation threshold (such as ±0.2Hz of rated frequency), and upper limit of power deficit rate (such as 5%), all of which are set based on microgrid design specifications and actual operation requirements. If the comparison results show that the economic assessment value is lower than the stability threshold (i.e., there is a situation where "economic performance meets the standard but stability exceeds the standard" or "neither objective is met"), then the scheme is re-iterated: the updated real-time operation data of the microgrid (including the latest renewable energy output, load demand, and current status data of energy storage equipment) is used as input, and the previously constructed multi-objective function model is substituted into iterative calculation to output the updated balance scheme data; the data specifically includes the adjusted renewable energy output allocation ratio (e.g., the proportion of photovoltaic output allocated to industrial load is increased from 70% to 80%), energy storage charging and discharging parameters (e.g., the charging and discharging power is adjusted from 500kW to 600kW, and the charging and discharging period is adjusted from 12:00-14:00 to 11:00-13:00), and load priority coefficient (e.g., the medical load priority coefficient is increased from 0.9 to 0.95). The updated balance scheme data is combined with the operating constraints of each device (such as the upper limit of renewable energy output, the upper and lower limits of energy storage state of charge, and the load regulation capacity limit) and the power interaction limit of the grid (such as the upper limit of power purchase and sale with the main grid ±1000kW). The control requirements of each device are arranged and integrated in time sequence according to the time dimension to form the final optimized control sequence. This sequence must clearly include specific time nodes (such as one control node every 15 minutes), device type (such as photovoltaic inverter, energy storage converter, industrial load controller), control parameters (such as target output, charging and discharging power, load regulation), and execution priority (such as emergency load control has higher priority than ordinary load). Finally, the final optimized control sequence is analyzed, and standardized equipment instruction sets are generated according to equipment type. Each instruction set must include an instruction identifier (e.g., instruction number 2025001), the executing entity (e.g., #1 photovoltaic inverter, #2 energy storage converter), the execution time period (e.g., 09:00-09:15), the control target value (e.g., target output of #1 photovoltaic inverter 800kW, target discharge power of #2 energy storage converter 300kW), the adjustment accuracy requirement (e.g., control error ≤3%), and the safe operation boundary (e.g., overcurrent protection threshold of #2 energy storage converter 1200A), to ensure that the instructions can be directly distributed to the corresponding equipment for execution.

[0099] In this embodiment, the quantitative comparison between the economic evaluation value and the stability threshold clarifies whether the optimization scheme meets the standards, avoiding the neglect of grid safety and stability due to a singular focus on economics, or the surge in operating costs caused by excessive stability assurance, thus ensuring the synergistic optimization of dual objectives. The scheme recalculation process uses real-time operating data as input, enabling the updated balancing scheme to adapt to dynamic operating conditions such as renewable energy output fluctuations and load abrupt changes, solving the problem that traditional static schemes cannot cope with changes in operating conditions. Finally, the timing integration of the optimized control sequence organizes the dispersed equipment control requirements according to the time dimension, ensuring that the operating instructions of each device are coordinated and consistent in time, avoiding operational conflicts between devices. The standardized equipment instruction set clarifies the execution subject, accuracy requirements, and safety boundaries, ensuring the accuracy of equipment execution while reducing safety risks during instruction execution. At the same time, all thresholds and constraints in the process are set based on the actual operating standards of the microgrid, and the instruction generation logic is reproducible, ensuring that the technical solution has strong practical operability and providing reliable control support for the long-term stable and economical operation of the new energy microgrid.

[0100] In one embodiment, such as Figure 2 As shown, this application also provides a multi-objective optimization control device for new energy microgrids, which may include:

[0101] The data acquisition and analysis module 601 is used to acquire key operating data in the microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage equipment, and load demand records. The key operating data is processed by time series analysis to obtain the variation pattern of energy output and the load demand curve.

[0102] The multi-objective model optimization module 602 is used to construct a multi-objective function model based on the variation law and load demand curve to obtain a multi-objective balance scheme. The multi-objective balance scheme is iteratively calculated through the particle swarm optimization algorithm to determine the resource collaborative allocation parameters.

[0103] The economic evaluation module 603 is used to input resource collaborative allocation parameters and available capacity data of energy storage equipment into a proxy model trained based on machine learning algorithms, generate real-time optimization decision instructions, extract equipment interaction control signals from the real-time optimization decision instructions and combine them with operation log data to calculate the economic evaluation value.

[0104] The equipment instruction generation module 604 is used to recalculate the balance scheme data through a multi-objective function model if the economic evaluation value is lower than the power grid stability threshold, determine the final optimized control sequence based on the balance scheme data, and generate a standardized equipment instruction set.

[0105] The aforementioned multi-objective optimization control device for new energy microgrids employs a data acquisition and analysis module responsible for obtaining three types of core data from the microgrid monitoring system: real-time renewable energy power generation output data (including instantaneous power generation values ​​over continuous time), energy storage equipment operating parameters (including state of charge, real-time charging and discharging power, and equipment terminal voltage), and load demand records (including historical load data and real-time load monitoring data). This data is processed using time series analysis methods, ultimately outputting the variation patterns of energy output and a corrected load demand curve. This output directly serves as the core input to the multi-objective model optimization module. Based on the variation patterns and load demand curve output by the data acquisition and analysis module, the multi-objective model optimization module constructs a multi-objective function model encompassing economic and stability indicators. It first generates a preliminary multi-objective balance scheme using linear programming, then iterates the preliminary scheme using a particle swarm optimization algorithm (aiming for optimal dual-objective balance, evaluating performance using a fitness function, and iterating until convergence). Finally, it outputs resource collaborative allocation parameters, which serve as one of the key inputs to the economic evaluation module. The economic evaluation module inputs the resource collaborative allocation parameters output by the multi-objective model optimization module and the available capacity data of energy storage devices obtained from the energy storage monitoring module into a surrogate model trained based on machine learning algorithms (such as random forest and LSTM) to generate real-time optimization decision instructions. Subsequently, the instructions are parsed to extract interactive control signals classified by equipment type, and combined with the collected microgrid operation log data, through deviation rate calculation and cost accounting, an economic evaluation value including unit electricity cost, renewable energy utilization rate, and energy storage operation and maintenance cost is output. This evaluation value provides the basis for judgment for the equipment instruction generation module. The equipment instruction generation module compares the economic evaluation value output by the economic evaluation module with the preset grid stability threshold. If the evaluation value is lower than the threshold, the updated microgrid real-time data is used as input to call the multi-objective function model to re-iterate and calculate the updated balance scheme data (including the adjusted output allocation ratio, energy storage parameters, and load priority coefficient). Then, combined with equipment operation constraints and grid interaction power limits, the balance scheme data time sequence is integrated into the final optimized control sequence containing time nodes, equipment types, control parameters, and execution priorities. Finally, the sequence is parsed to generate a standardized equipment instruction set containing instruction identifier, execution subject, execution period, control target value, regulation accuracy, and safety boundary according to equipment type.

[0106] In one embodiment, the data acquisition and analysis module can also be used for:

[0107] Acquire real-time power output data of renewable energy generation, operating parameters of energy storage devices, and load demand records in the microgrid; real-time power output data includes instantaneous power generation values ​​in the continuous time dimension; device operating parameters include state of charge, real-time charging and discharging power, and device terminal voltage in the continuous time dimension; load demand records include historical load data and real-time load monitoring data of the microgrid.

[0108] A time series decomposition algorithm is used to synchronously decompose real-time power output data and equipment operating parameters to obtain the periodic characteristics, random variation characteristics and change characteristics of power output.

[0109] A joint analysis model is constructed using time series analysis to investigate periodicity, random variation, and change characteristics. By fitting and validating the joint analysis model, the variation patterns of energy output are determined. These variation patterns include fluctuation amplitude, frequency, and duration.

[0110] Key fluctuation parameters, including the maximum fluctuation difference, fluctuation frequency threshold, and continuous fluctuation duration, are extracted from the variation patterns. Correlation analysis is then conducted using historical load data of the microgrid to identify potential influencing factors affecting load uncertainty patterns. These potential influencing factors include the correlation factor between power fluctuation amplitude and load peak-valley difference.

[0111] If the quantified value of a potential influencing factor exceeds a preset threshold range, the real-time load monitoring data of the microgrid is called to dynamically correct the real-time load monitoring data and generate a corrected load demand curve; the load demand curve includes time nodes, load values, and curve confidence.

[0112] In one embodiment, the multi-objective model optimization module can also be used for:

[0113] A multi-objective function model is constructed based on the variation pattern and load demand curve; the multi-objective function model data includes economic index parameters and stability index parameters.

[0114] Based on the multi-objective function model, constraints for resource collaborative allocation are set, and an initial scheme for resource collaborative allocation is constructed using linear programming to obtain a preliminary multi-objective balance scheme. The multi-objective balance scheme includes the renewable energy output allocation ratio, the charging and discharging time and power of energy storage equipment, and the load priority allocation coefficient.

[0115] The multi-objective equilibrium scheme is iteratively calculated using the particle swarm optimization algorithm. The objective functions are minimizing the economic index and optimizing the stability index. The fitness function is used to evaluate the merits of each generation of the scheme. The optimization parameters of the multi-objective equilibrium scheme are determined by iterating until the convergence condition is met.

[0116] If the quantized values ​​of the optimization parameters all meet the corresponding preset balance thresholds, the final resource allocation scheme will be output; the resource allocation scheme specifies the specific resource allocation for each time period and each device.

[0117] Load adjustment instructions are generated based on the resource allocation scheme; the load adjustment instructions include the instruction execution subject, execution period, adjustment target value, and adjustment accuracy requirements.

[0118] The system acquires execution feedback data of load regulation commands and judges the feedback data based on the preset target value of the commands. The feedback data includes command execution completion rate, actual resource distribution data after execution, and power grid operation status data.

[0119] If the feedback data deviates from the expected target, the deviation from the target value in the feedback data is extracted, and the weight coefficients and constraints in the multi-objective function model are updated based on the deviation to obtain the updated multi-objective function model parameters.

[0120] Based on the updated multi-objective function model parameters, the particle swarm optimization algorithm is used again for iterative calculation to obtain the updated resource collaborative allocation parameters.

[0121] In one embodiment, the economic evaluation module can also be used for:

[0122] Obtain available capacity data for energy storage devices; available capacity data includes current state of charge, maximum charge / discharge power, and remaining available capacity.

[0123] A dynamic adjustment model is constructed by combining available capacity data with resource collaborative allocation parameters. The model is then used to generate preliminary optimization decision instructions, which include the target operating power, adjustment period, and response delay requirements for each device.

[0124] A machine learning algorithm is used to train the agent model. The initial optimization decision instructions are input into the trained agent model. By simulating the execution effect of instructions under different operating scenarios, the corrected real-time optimization decision instructions are output. The training data of the agent model are sample pairs from the historical operation of the microgrid.

[0125] The system analyzes real-time optimization decision commands, extracts equipment interactive control signals, and determines the target operating status data of each device. The control signals are classified by equipment type, including output regulation signals of renewable energy power generation equipment, charging and discharging control signals of energy storage equipment, and load switching signals of load controllers.

[0126] Collect microgrid operation log data; the operation log data includes the actual operating status data of each device, grid operation parameters, energy consumption and cost data.

[0127] Based on the target operating status data, the deviation rate and unit electricity cost of the operating log data are calculated to obtain the economic evaluation value; the economic evaluation value includes unit electricity cost, renewable energy utilization rate, and energy storage operation and maintenance cost.

[0128] In one embodiment, the device instruction generation module can also be used to:

[0129] The economic assessment value is compared with the preset power grid stability threshold to obtain the comparison result; the power grid stability threshold includes the allowable range of voltage fluctuation, the frequency deviation threshold, and the upper limit of the power deficit rate.

[0130] If the comparison result shows that the economic evaluation value is lower than the stability threshold, the updated real-time operation data of the microgrid will be used as input, and the updated balance scheme data will be obtained by iterating again through the multi-objective function model. The balance scheme data includes the adjusted renewable energy output allocation ratio, energy storage charging and discharging parameters, and load priority coefficient.

[0131] The balance scheme data is combined with the operating constraints of each device and the power limit of the power grid interaction, and integrated through time sequence arrangement to form the final optimized control sequence; the final optimized control sequence includes time nodes, device types, control parameters, and execution priorities.

[0132] The final optimized control sequence is analyzed, and a standardized equipment instruction set is generated according to the equipment type. The equipment instruction set includes instruction identifier, execution subject, execution time period, control target value, adjustment accuracy requirement, and safe operation boundary.

[0133] In this embodiment, four modules enable precise implementation of the entire process of new energy microgrid control: Firstly, the data acquisition and analysis module ensures the integrity and processing accuracy of basic data, providing reliable input for subsequent optimization models and avoiding optimization inaccuracies caused by data deviations; the multi-objective model optimization module balances economic and stability requirements through dual-objective functions and particle swarm optimization, overcoming the limitations of traditional single-objective optimization; the economic evaluation module, combining proxy models and operation logs, improves the real-time adaptability of decision commands and ensures the measurability of operational benefits through quantitative evaluation; the threshold comparison and scheme recalculation mechanism of the equipment command generation module enables the control strategy to dynamically respond to changes in microgrid operating conditions, avoiding the rigidity problem of static commands. On the other hand, the input and output data of each module all come from the existing monitoring system of the microgrid. The algorithms used (time series analysis, particle swarm optimization, machine learning) are mature and reproducible. The functional boundaries of the modules are clear and closely coordinated, which not only ensures the practical operability of the technical solution, but also effectively supports the new energy microgrid to achieve stable and economical operation under different operating scenarios (grid-connected / off-grid, high fluctuation / stable operating conditions), providing modular and scalable control architecture support for the large-scale application of microgrids.

[0134] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0135] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the multi-objective optimization control method for new energy microgrids as described above.

[0136] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0137] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0138] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A multi-objective optimization control method for new energy microgrids, characterized in that, The method includes: Key operational data in a microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage devices, and load demand records, are acquired. Time series analysis is then used to process the key operational data to obtain the variation pattern of energy output and the load demand curve. Based on the aforementioned variation patterns and load demand curves, a multi-objective function model is constructed to obtain a multi-objective balancing scheme. The multi-objective balancing scheme is then iteratively calculated using a particle swarm optimization algorithm to determine resource collaborative allocation parameters. The resource collaborative allocation parameters and the available capacity data of the energy storage device are input into the agent model trained based on the machine learning algorithm to generate real-time optimization decision instructions. The device interaction control signals in the real-time optimization decision instructions are extracted and combined with the operation log data to calculate the economic evaluation value. If the economic evaluation value is lower than the power grid stability threshold, the balance scheme data is recalculated iteratively using the multi-objective function model. Based on the balance scheme data, the final optimized control sequence is determined, and a standardized equipment instruction set is generated.

2. The method according to claim 1, characterized in that, The process involves acquiring key operational data from the microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage devices, and load demand records. Time series analysis is then used to process this key operational data to obtain the variation patterns of energy output and the load demand curve, including: The system acquires real-time power output data of renewable energy generation, operating parameters of energy storage devices, and load demand records in the microgrid. The real-time power output data includes instantaneous power generation values ​​over a continuous time dimension. The operating parameters of the devices include state of charge, real-time charging and discharging power, and device terminal voltage over a continuous time dimension. The load demand records include historical load data and real-time load monitoring data of the microgrid. The real-time power output data and the equipment operating parameters are synchronously decomposed using a time series decomposition algorithm to obtain the periodic characteristics, random variation characteristics and change characteristics of the power output. A joint analysis model is constructed using time series analysis to analyze the periodicity, random variation, and change characteristics. The variation pattern of energy output is determined by fitting and validating the joint analysis model. The variation pattern includes fluctuation amplitude, frequency, and duration. Key fluctuation parameters, including the maximum fluctuation difference, fluctuation frequency threshold, and continuous fluctuation duration, are extracted from the aforementioned variation patterns. Correlation analysis is then performed using the historical load data of the microgrid to identify potential influencing factors affecting load uncertainty patterns. These potential influencing factors include the correlation factor between power fluctuation amplitude and load peak-valley difference. If the quantified value of the potential influencing factor exceeds the preset threshold range, the real-time load monitoring data of the microgrid is invoked to dynamically correct the real-time load monitoring data and generate a corrected load demand curve; the load demand curve includes time nodes, load values, and curve confidence.

3. The method according to claim 1, characterized in that, The process involves constructing a multi-objective function model based on the variation pattern and load demand curve to obtain a multi-objective balancing scheme. The multi-objective balancing scheme is then iteratively calculated using a particle swarm optimization algorithm to determine resource collaborative allocation parameters, including: A multi-objective function model is constructed based on the aforementioned variation patterns and load demand curves; the multi-objective function model data includes economic index parameters and stability index parameters; Based on the multi-objective function model, constraints for resource collaborative allocation are set, and an initial scheme for resource collaborative allocation is constructed using linear programming to obtain a preliminary multi-objective balance scheme. The multi-objective balance scheme includes the renewable energy output allocation ratio, the charging and discharging time and power of energy storage equipment, and the load priority allocation coefficient. The multi-objective balancing scheme is iteratively calculated using a particle swarm optimization algorithm, with the objective functions being minimization of the economic index and optimization of the stability index. The fitness function is used to evaluate the merits of each generation of the scheme, and the optimization parameters of the multi-objective balancing scheme are determined by iterating until the convergence condition is met. If the quantized values ​​of the optimization parameters all meet the corresponding preset balance thresholds, the final resource allocation scheme is output; the resource allocation scheme specifies the specific resource allocation amount for each time period and each device. A load adjustment instruction is generated based on the resource allocation scheme; the load adjustment instruction includes the instruction execution subject, execution period, adjustment target value, and adjustment accuracy requirements. The execution feedback data of the load adjustment command is obtained, and the feedback data is judged based on the target value preset by the command; the feedback data includes the command execution completion rate, the actual resource distribution data after execution, and the power grid operation status data; If the feedback data deviates from the expected target, the deviation from the target value in the feedback data is extracted, and the weight coefficients and constraints in the multi-objective function model are updated according to the deviation to obtain the updated multi-objective function model parameters. Based on the updated multi-objective function model parameters, the particle swarm optimization algorithm is used again for iterative calculation to obtain the updated resource collaborative allocation parameters.

4. The method according to claim 3, characterized in that, The multi-objective function model is expressed by the following formula: in, This represents the economic objective function. This represents the stability objective function. This represents the overall operating cost, which is calculated as: Overall operating cost = Equipment maintenance cost per unit time + Electricity purchase cost. This represents the cost of energy loss, which is calculated as: Energy loss cost = Renewable energy curtailment loss + Energy storage charging and discharging loss. The voltage stability coefficient is calculated as the reciprocal of the deviation rate between the actual voltage and the rated voltage. The frequency stability coefficient is calculated as the reciprocal of the deviation rate between the actual frequency and the rated frequency. This represents the power fluctuation coefficient, which is calculated as: maximum difference in renewable energy output per unit time / rated power. This represents the load forecast deviation rate, where the load forecast deviation rate = / Predicted load, , , , This represents the weighting coefficient, which is set according to the microgrid's operating scenario. , This represents the coupling correction coefficient, obtained by fitting historical data. This indicates the real-time output of renewable energy. Indicates the state of charge of the energy storage. This indicates the real-time load power of the microgrid. This indicates the real-time charging and discharging power of the energy storage device. This indicates the real-time power exchange between the microgrid and the main grid. Represents a time variable.

5. The method according to claim 1, characterized in that, The process involves inputting the resource collaborative allocation parameters and available capacity data of the energy storage device into a proxy model trained based on a machine learning algorithm to generate real-time optimization decision instructions. The process then extracts the device interaction control signals from these instructions and combines them with operational log data to calculate an economic evaluation value, including: Obtain available capacity data of the energy storage device; the available capacity data includes the current state of charge, maximum charge / discharge power, and remaining available capacity. The available capacity data is combined with the resource collaborative allocation parameters to construct a dynamic adjustment model, and preliminary optimization decision instructions are generated through model calculations. The preliminary optimization decision instructions include the target operating power, adjustment period, and response delay requirements for each device. A machine learning algorithm is used to train a proxy model. The initial optimization decision command is input into the trained proxy model. By simulating the execution effect of the command under different operating scenarios, the corrected real-time optimization decision command is output. The training data of the proxy model are sample pairs from the historical operation of the microgrid. The real-time optimization decision instructions are analyzed, the equipment interaction control signals are extracted, and the target operating status data of each device is determined. The control signals are classified according to equipment type, including output adjustment signals of renewable energy power generation equipment, charging and discharging control signals of energy storage equipment, and load switching signals of load controllers. Collect microgrid operation log data; the operation log data includes the actual operating status data of each device, grid operating parameters, energy consumption and cost data; Based on the target operating status data, the deviation rate and unit electricity cost of the operating log data are calculated to obtain an economic evaluation value; the economic evaluation value includes unit electricity cost, renewable energy utilization rate, and energy storage operation and maintenance cost.

6. The method according to claim 1, characterized in that, If the economic evaluation value is lower than the power grid stability threshold, the balance scheme data is recalculated iteratively using the multi-objective function model. Based on the balance scheme data, the final optimized control sequence is determined, and a standardized equipment instruction set is generated, including: The economic evaluation value is compared with the preset power grid stability threshold to obtain the comparison result; the power grid stability threshold includes the allowable range of voltage fluctuation, the frequency deviation threshold, and the upper limit of the power deficit rate; If the comparison result shows that the economic evaluation value is lower than the stability threshold, then the updated microgrid real-time operation data is used as input, and the multi-objective function model is iterated again to obtain the updated balance scheme data; the balance scheme data includes the adjusted renewable energy output allocation ratio, energy storage charging and discharging parameters, and load priority coefficient; The data from the balancing scheme are combined with the operating constraints of each device and the power limit of the power grid interaction, and integrated through time-series arrangement to form the final optimized control sequence; the final optimized control sequence includes time nodes, device types, control parameters, and execution priorities; The final optimized control sequence is analyzed, and a standardized equipment instruction set is generated according to the equipment type. The equipment instruction set includes instruction identifier, execution subject, execution time period, control target value, adjustment accuracy requirement, and safe operation boundary.

7. A multi-objective optimization control device for new energy microgrids, characterized in that, The device includes: The data acquisition and analysis module is used to acquire key operating data in the microgrid, including real-time output data of renewable energy generation, operating parameters of energy storage equipment, and load demand records. The key operating data is processed by time series analysis to obtain the variation pattern of energy output and the load demand curve. The multi-objective model optimization module is used to construct a multi-objective function model based on the variation law and load demand curve to obtain a multi-objective balance scheme, and to iteratively calculate the multi-objective balance scheme through the particle swarm optimization algorithm to determine the resource collaborative allocation parameters. The economic evaluation module is used to input the resource collaborative allocation parameters and the available capacity data of the energy storage device into the agent model trained based on machine learning algorithm, generate real-time optimization decision instructions, extract the device interaction control signals in the real-time optimization decision instructions and combine them with the operation log data to calculate the economic evaluation value. The equipment instruction generation module is used to recalculate the balance scheme data through the multi-objective function model if the economic evaluation value is lower than the power grid stability threshold, determine the final optimized control sequence based on the balance scheme data, and generate a standardized equipment instruction set.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images