New energy micro-grid intelligent autonomous operation method and system

Abstract

This invention relates to the field of autonomous operation technology for power grids, specifically a method and system for intelligent autonomous operation of a new energy microgrid. By deploying sensors and edge computing nodes to monitor power generation, consumption, grid status, and external factors in real time, a real-time operating topology of the new energy microgrid is constructed. Based on this real-time topology, the optimal operating scheme is calculated from a preset operating strategy. Virtual simulation is used to verify the feasibility of the optimal operating scheme for steady-state power control. The verified optimal operating scheme is then distributed to all grid facilities to ensure consistent execution. When a grid facility fails, neighboring facilities proactively compensate for the interruption, achieving autonomous and collaborative control. Furthermore, by continuously learning from the actual operating data, the control strategy can be continuously optimized, improving the microgrid's operating efficiency, stability, and environmental adaptability.

Description

Technical Field

[0001] This invention relates to the field of autonomous operation technology of power grids, and in particular to a method and system for intelligent autonomous operation of new energy microgrids. Background Technology

[0002] With the rapid development of renewable energy, wind and solar power are increasingly being used in microgrid systems. As an independent power system integrating power generation, consumption, energy storage, and energy management, microgrids play a crucial role in improving energy efficiency and enhancing grid resilience. However, the autonomous operation of existing renewable energy microgrids largely relies on centralized monitoring. Data acquisition and transmission suffer from latency and are prone to single-point bottlenecks, making it difficult to reflect the dynamic correlations between power generation, load, grid connection status, and external variables (weather, electricity prices) in a timely manner. Secondly, traditional autonomous operation methods are mostly based on static rules or empirical models, lacking deep learning and online optimization for large-scale, multivariate systems. This makes it difficult to formulate optimal autonomous operation schemes, easily leading to slow strategy adjustments when dealing with renewable energy fluctuations and sudden changes in electricity consumption. Furthermore, if core control equipment or communication links fail, the entire microgrid may be paralyzed; existing solutions struggle to achieve dynamic complementarity and local self-healing between devices, affecting the safe and stable operation of the system. Furthermore, current autonomous operation methods are mostly closed-loop "set-up" processes, lacking in-depth feedback on execution results and online learning capabilities. This prevents continuous optimization of scheduling strategies based on operational data, hindering long-term improvement in the power grid system's operational efficiency. Therefore, it is necessary to develop an autonomous operation method that can enhance the rationality, robustness, and continuous evolution of autonomous renewable energy microgrids to address these issues. Summary of the Invention

[0003] This invention overcomes the shortcomings of existing technologies and provides a method and system for intelligent autonomous operation of new energy microgrids.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] The first aspect of this invention provides a method for intelligent autonomous operation of a new energy microgrid, comprising the following steps:

[0006] S102: Real-time data of new energy microgrid in power generation, power consumption, grid status and external factors are monitored in real time through distributed sensors and edge computing. The real-time data is used to perform topology calculation for message transmission to obtain the real-time operating topology of the new energy microgrid.

[0007] S104: Construct a decision body memory bank using the potential decision solutions and corresponding feasible values ​​that meet the autonomous operation requirements in the preset operation strategy. Analyze and replace the values ​​in the decision body memory bank according to the real-time chaotic sequence of the power flow vector describing the real-time operation topology field to generate the optimal operation scheme.

[0008] S106: The optimal operating scheme is tested through virtual simulation. The feasibility of the optimal operating scheme is verified on the power operation sliding surface of the system power and under the applied external disturbance resistance based on the simulated operation test parameter set. The optimal operating scheme is then issued to all power facilities.

[0009] S108: Divide each power grid facility into multiple sub-jurisdiction areas, analyze the power operation steady-state lock-in problem facilities in each sub-jurisdiction area, and robustly design steady-state structured gain compensators for adjacent power grid facilities to actively compensate for the stable power control of problem facilities.

[0010] S110: Obtain real-time operation data of each power grid facility executing preset operation strategies, continuously learn from the real-time operation data to optimize strategies, and improve the adaptability of new energy microgrids.

[0011] More specifically, step S102 includes the following steps:

[0012] Real-time monitoring of the new energy microgrid through distributed sensors acquires real-time distributed sensing data of each grid facility in the new energy microgrid at a preset time and space sequence, as well as real-time distributed edge data output by each grid edge node.

[0013] The edge message communication protocol of the new energy microgrid is obtained. Based on the edge message communication protocol, the neighboring collaborative nodes of each grid edge node are identified and connected. An edge communication blueprint is constructed, and the communication trajectory layout between each grid edge node is extracted based on the edge communication blueprint.

[0014] According to the communication trajectory layout, the attraction centroids corresponding to each power grid edge node are randomly initialized and allocated. Real-time distributed edge data is injected into the attraction centroids near the corresponding output power grid edge node and the distribution is iteratively updated to form a sub-power grid communication edge distribution with each power grid edge node as the edge radiation core.

[0015] Obtain a schematic diagram of the grid facility layout of the new energy microgrid. Based on the schematic diagram, each grid facility is designated as a variable point. According to the conditional transfer function that generates real-time distributed sensor data based on the correlation control between each grid facility, a working condition transition probability is preset. The working condition transition condition is defined as a factor constraint node. Each variable point and its corresponding factor constraint node are linked together to generate a Belief message channel diagram.

[0016] Based on the edge distribution of subgrid communication, perturbation variable messages of different predetermined scales are constructed and transmitted to the message channel connecting the link variable points and adjacent factor constraint nodes in the Belief message channel diagram. This enables the nodes to be updated iteratively and the current scale of each pair of nodes to be obtained synchronously.

[0017] During the node iteration update process, if the size deviation between the current disturbance message size and the predetermined disturbance message size is less than a preset threshold, the message transmission operation of the node pair is stopped, and the edge probability of the power grid facility corresponding to the link variable point in each node pair is finally output.

[0018] By introducing Kirchhoff's laws and using edge probability and subgrid communication edge distribution as the basis, the power generation, power consumption, grid status and external factors of the power grid are plotted in the topology of Kirchhoff's laws, thus obtaining the real-time operating topology field of the new energy microgrid.

[0019] More specifically, S104 includes the following steps:

[0020] Obtain the standardized operation process, autonomous operation requirements, and preset operation strategies of the new energy microgrid, and extract multiple related operation decision indicators based on the standardized operation process;

[0021] Based on the preset operation strategy, extract several sets of potential decision solutions for new energy microgrids under the premise of meeting the autonomous operation requirements, and the feasible values ​​of each potential decision solution for each operation decision index discrete programming. Obtain the original affinity of each potential decision solution, and map and bind the potential decision solution, the original affinity and the corresponding feasible value to construct a decision body memory bank.

[0022] The real-time chaotic sequence of the power flow vector during the operation of the new energy microgrid is extracted through the real-time running topology field, and a real-time Logistic chaotic mapping model is constructed based on the real-time chaotic sequence.

[0023] The desired Logistic chaotic mapping model is obtained through autonomous operation requirements. If the mapping deviation between the actual Logistic chaotic mapping model and the desired Logistic chaotic mapping model is greater than the preset mapping deviation, the memory consideration rate and the mutation adjustment rate are set based on the mapping deviation.

[0024] Based on the real-time chaotic sequence, fine-tuning mapping values ​​are obtained by using the memory consideration rate analysis in the decision-making body memory bank. New decision solutions are obtained, and the new decision solutions are iteratively replaced by the mutation adjustment rate to obtain the optimal operation scheme of the new energy microgrid.

[0025] More specifically, the step of using the memory consideration rate analysis based on the real-time chaotic sequence to perform fine-tuning mapping values ​​in the decision-making body's memory bank to obtain new decision solutions, and then performing iterative replacement on the new decision solutions through the mutation adjustment rate to obtain the optimal operation scheme of the new energy microgrid, specifically includes the following steps:

[0026] By mapping the real-time chaotic sequence to values ​​in the decision-making body's memory bank, if the memory consideration rate is greater than the preset memory consideration rate, the feasible value is marked and selected as the main candidate feasible value in the decision-making body's memory bank, and the main candidate feasible value is finely adjusted according to the mutation adjustment rate to generate the first candidate community situation.

[0027] If the memory consideration rate is less than the preset memory consideration rate, the feasible value is marked as a secondary candidate feasible value, no fine-tuning is performed, and a second candidate community situation is generated.

[0028] By combining the situations of the first and second candidate communities, a new decision solution is output that enables the new energy microgrid to meet the requirements of autonomous operation, and the mutation affinity of the new decision solution is calculated.

[0029] If the mutation affinity is greater than the original affinity, then the new decision solution replaces the potential decision solution where the feasible value is located; otherwise, discard the new decision solution and keep the decision unchanged to obtain the optimal operating decision regarding the operating decision index.

[0030] By following the standardized operation process and merging all optimal operation decisions regarding operation decision indicators, the optimal operation scheme for the new energy microgrid is obtained.

[0031] More specifically, S106 includes the following steps:

[0032] Obtain the power design specifications and standardized operation process of the new energy microgrid, and use Simulink power simulation software to input and analyze the power design data recorded in the power design specifications to construct a virtual power simulation model of the new energy microgrid;

[0033] Based on the autonomous operation requirements, an ideal operation reference value is preset. The optimal operation scheme is then simulated and tested using a virtual power simulation model with the ideal operation reference value as the arrival constraint condition. A system power structured state chain of the new energy microgrid is created, and the set of simulation operation test parameters is recorded and obtained.

[0034] During the simulation process, a power operation sliding surface with error tracking concept is designed for the new energy microgrid based on the autonomous operation requirements. On the power operation sliding surface, the control error trajectory of the system's power structured state chain is continuously monitored and tracked based on the simulated operation test parameter set. In this way, the simulated Lyapunov function that makes the system's power structured state chain approach the power operation sliding surface is evaluated and determined.

[0035] Obtain one or more infeasible events of the new energy microgrid and the expected Lyapunov function of each infeasible event, construct event responders for the infeasible events based on the expected Lyapunov function, and form an event responder array;

[0036] The event responder array is deployed on the power operation sliding surface. During simulation, it is determined whether there is any expected Lyapunov function in the simulated Lyapunov function. If there is, the event responder array is triggered to eliminate the optimal operating scheme and re-select the next optimal operating scheme in descending order.

[0037] If it does not exist, then design an ideal sliding mode control law based on the equivalent control and approximation solution of the ideal operating reference value. Apply external disturbance resistance events to the ideal sliding mode control law and analyze the resistance performance within a quantitative range to verify the feasibility of the optimal operating scheme. If the feasibility is high, then distribute the optimal operating scheme to all power facilities.

[0038] More specifically, if the aforementioned does not exist, then an ideal sliding mode control law is designed based on the equivalent control and approximation solution of the ideal operating reference value. An external disturbance resistance event is applied to the ideal sliding mode control law, and the resistance performance within a quantitative range is analyzed to verify the feasibility of the optimal operating scheme. If the feasibility is high, the optimal operating scheme is distributed to all power facilities. This specifically includes the following steps:

[0039] If it does not exist, then the equivalent control principle and the approach law are introduced. Based on the ideal operating reference value, the simulation operation test parameter set is solved in the equivalent control principle and the approach law to make the system state control maintain the autonomous operation requirements that approximate the power operation sliding surface, and the ideal sliding mode control law is designed.

[0040] External disturbance resistance events of the optimal operation scheme are extracted from the work log of the new energy microgrid. Interference operation terms are created based on the external disturbance resistance events and applied to the ideal sliding mode control law.

[0041] Calculate the coverage ratio between the disturbance boundary of the ideal sliding mode control law and the disturbance limit of the interference operation term, and determine the resistance index of the ideal sliding mode control law to the interference operation term based on the coverage ratio;

[0042] If the resistance index exceeds the preset resistance index range, the optimal operating plan is eliminated; if the resistance index does not exceed the preset resistance index range, the current optimal operating plan is distributed to all power facilities.

[0043] More specifically, S108 includes the following steps:

[0044] Based on the standardized operation process charter, the work weight coefficients of each power grid facility for the standardized normal operation of the new energy microgrid are obtained. The process control scope of each power grid facility within the new energy microgrid is divided according to the work weight coefficients, resulting in N sub-jurisdiction areas.

[0045] By optimizing the actual operation scheme of each power grid facility and monitoring the power operation steady state of the new energy microgrid in real time, the actual Lyapunov function of the system power control in each sub-jurisdiction area is obtained;

[0046] Extract the topological boundary regions where there is direct power continuity interaction between the power grid facilities of adjacent sub-jurisdictional areas in the real-time running topological field;

[0047] If the actual Lyapunov function corresponding to the power grid facility within the sub-jurisdiction area is greater than the actual Lyapunov function corresponding to the neighboring power grid facility within the topological boundary area, then the power facility within the sub-jurisdiction area is marked as a problem facility, and multiple fluctuation frequency inflection points of the actual power control interruption in the time sequence of the problem facility are obtained.

[0048] A stable gain control architecture corresponding to the initial closed loop of the system's power structured state chain is constructed using the PID underlying control logic. A frequency scanning algorithm is introduced to find a scaling weight function that can transform the uncertainty structure of the closed loop system at each fluctuation frequency inflection point, and the frequency domain singular value of the closed loop system is calculated once when the scaling weight function is found.

[0049] Based on the scaling weight function, a weighted transformation is performed on the topological boundary domain of the problematic facilities in the closed-loop system of the new energy microgrid, so that the gain feedback of the stable gain control architecture is reconfigured and adjusted.

[0050] If the frequency domain singular value is less than the preset frequency domain singular value, the gain feedback adjustment operation of the stable gain control architecture is stopped, and a steady-state structured gain compensator is finally generated. The steady-state structured gain compensator is sent and applied to the power grid facilities adjacent to the problematic facility, thereby actively compensating for the stable power control of the problematic facility.

[0051] More specifically, S110 includes the following steps:

[0052] Acquire real-time operation data of each power grid facility executing a preset operation strategy at a uniform time step, and construct a strategy trajectory network by combining the real-time operation data with the strategy trajectory of the interaction sampling actions, status and rewards of the preset operation strategy.

[0053] The dominance function of each strategy trajectory at continuous equilibrium time steps is evaluated through the power state evaluation system. If the dominance function is greater than the preset dominance function, the strategy trajectory is marked as a positive strategy trajectory; if it is less than the preset dominance function, it is marked as a negative strategy trajectory.

[0054] By using a pre-defined running strategy to weight the probability ratio of learning the positive strategy trajectory compared to learning the negative strategy trajectory in the strategy trajectory network, the strategy learning ratio is obtained. The strategy learning ratio is then used to plan and formulate the gradient direction of a proxy strategy that continuously learns from real-time running data.

[0055] Based on the autonomous operation requirements, the allowable adjustment range of the preset operation strategy is obtained. Based on the allowable adjustment range, a KL divergence constraint trust region is constructed. The conjugate gradient method is introduced to approximate the gradient direction of the proxy strategy, and the natural gradient direction is obtained.

[0056] A continuous linear search is performed on the real-time operating data along the natural gradient direction at uniform time steps, outputting several strategy optimization parameters and corresponding KL divergences. If the KL divergence is always within the KL divergence constraint trust region, the strategy optimization parameters are used to modify and optimize the preset operating strategy, thereby improving the adaptability of the new energy microgrid.

[0057] The second aspect of the present invention provides a smart autonomous operation system for a new energy microgrid, the smart autonomous operation system for a new energy microgrid includes a memory and a processor, the memory stores a program for a smart autonomous operation method for a new energy microgrid, and when the program for a smart autonomous operation method for a new energy microgrid is executed by the processor, the steps of the smart autonomous operation method for a new energy microgrid as described in any one of the present invention are implemented.

[0058] This invention addresses the technical deficiencies in the prior art, and its beneficial technical effects are as follows:

[0059] Real-time monitoring of power generation, consumption, grid status, and external factors is achieved through sensors and edge computing. A global power topology is constructed for further visualization, effectively improving the timeliness and accuracy of information acquisition and providing data support for subsequent scheduling. Virtual simulation is used to verify the feasibility of the operational plan, avoiding system failures caused by unreasonable scheduling. Devices can mutually recognize commands and respond collaboratively, possessing self-repair and fault-tolerance capabilities. When one device malfunctions, adjacent devices can proactively take over its function, ensuring continuous and stable system operation. The power system has the ability to continuously learn from operational data, adjusting and optimizing control strategies based on historical operating results, gradually improving environmental adaptability and intelligent decision-making capabilities. In summary, the intelligent autonomous mechanism of this invention reduces the need for manual intervention, significantly improves the system's autonomous operation capability, reduces operation and maintenance complexity and costs, and makes the power operation of the microgrid more efficient and stable. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0061] Figure 1A flowchart of the first method for intelligent autonomous operation of a new energy microgrid is shown;

[0062] Figure 2 A flowchart of the second method for a smart autonomous operation method of a new energy microgrid is shown;

[0063] Figure 3 A system framework diagram of a smart autonomous operation system for a new energy microgrid is shown. Detailed Implementation

[0064] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0065] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0066] The first aspect of this invention provides a method for intelligent autonomous operation of a new energy microgrid, such as... Figure 1 As shown, it includes the following steps:

[0067] S102: Real-time data of new energy microgrid in power generation, power consumption, grid status and external factors are monitored in real time through distributed sensors and edge computing. The real-time data is used to perform topology calculation for message transmission to obtain the real-time operating topology of the new energy microgrid.

[0068] S104: Construct a decision body memory bank using the potential decision solutions and corresponding feasible values ​​that meet the autonomous operation requirements in the preset operation strategy. Analyze and replace the values ​​in the decision body memory bank according to the real-time chaotic sequence of the power flow vector describing the real-time operation topology field to generate the optimal operation scheme.

[0069] S106: The optimal operating scheme is tested through virtual simulation. The feasibility of the optimal operating scheme is verified on the power operation sliding surface of the system power and under the applied external disturbance resistance based on the simulated operation test parameter set. The optimal operating scheme is then issued to all power facilities.

[0070] S108: Divide each power grid facility into multiple sub-jurisdiction areas, analyze the power operation steady-state lock-in problem facilities in each sub-jurisdiction area, and robustly design steady-state structured gain compensators for adjacent power grid facilities to actively compensate for the stable power control of problem facilities.

[0071] S110: Obtain real-time operation data of each power grid facility executing preset operation strategies, continuously learn from the real-time operation data to optimize strategies, and improve the adaptability of new energy microgrids.

[0072] More specifically, step S102 includes the following steps:

[0073] Real-time monitoring of the new energy microgrid through distributed sensors acquires real-time distributed sensing data of each grid facility in the new energy microgrid at a preset time and space sequence, as well as real-time distributed edge data output by each grid edge node.

[0074] The edge message communication protocol of the new energy microgrid is obtained. Based on the edge message communication protocol, the neighboring collaborative nodes of each grid edge node are identified and connected. An edge communication blueprint is constructed, and the communication trajectory layout between each grid edge node is extracted based on the edge communication blueprint.

[0075] According to the communication trajectory layout, the attraction centroids corresponding to each power grid edge node are randomly initialized and allocated. Real-time distributed edge data is injected into the attraction centroids near the corresponding output power grid edge node and the distribution is iteratively updated to form a sub-power grid communication edge distribution with each power grid edge node as the edge radiation core.

[0076] Obtain a schematic diagram of the grid facility layout of the new energy microgrid. Based on the schematic diagram, each grid facility is designated as a variable point. According to the conditional transfer function that generates real-time distributed sensor data based on the correlation control between each grid facility, a working condition transition probability is preset. The working condition transition condition is defined as a factor constraint node. Each variable point and its corresponding factor constraint node are linked together to generate a Belief message channel diagram.

[0077] Based on the edge distribution of subgrid communication, perturbation variable messages of different predetermined scales are constructed and transmitted to the message channel connecting the link variable points and adjacent factor constraint nodes in the Belief message channel diagram. This enables the nodes to be updated iteratively and the current scale of each pair of nodes to be obtained synchronously.

[0078] During the node iteration update process, if the size deviation between the current disturbance message size and the predetermined disturbance message size is less than a preset threshold, the message transmission operation of the node pair is stopped, and the edge probability of the power grid facility corresponding to the link variable point in each node pair is finally output.

[0079] By introducing Kirchhoff's laws and using edge probability and subgrid communication edge distribution as the basis, the power generation, power consumption, grid status and external factors of the power grid are plotted in the topology of Kirchhoff's laws, thus obtaining the real-time operating topology field of the new energy microgrid.

[0080] It should be noted that while distributed sensors and edge devices deployed in new energy microgrids enable real-time monitoring of their power operation, existing methods struggle to provide comprehensive and visual feedback on the real-time power data, including the microgrid's power generation, power consumption, grid status, and external interference factors. This lack of reliable observation data hinders the autonomous operation of the microgrid, potentially leading to control chaos, numerical deviations, and power behavior errors. To address this, this method first acquires real-time distributed edge data from grid edge nodes (edge ​​devices). Given the potential collaborative observation relationships between different edge devices in the microgrid, this method identifies and connects potentially related adjacent grid edge node pairs based on the microgrid's edge message communication protocol, constructing an edge communication blueprint. This enables rapid discovery of collaborative edge nodes and establishment of information channels, ensuring the effectiveness and high consistency of subsequent edge computing. Subsequently, based on the communication trajectory layout of the edge communication blueprint, attractive centroids are allocated, and the acquired real-time distributed edge data is injected into nearby attractive centroids for several rounds of iterative updates. This emphasizes the potential collaborative distribution of local data, forming a preliminary aggregated form of real-time edge data. The subgrid communication edge distribution reflects the collaborative position and pattern of the communication link in which the edge data provided by each grid edge node is located, reflects the temporal structure and communication scheduling of real-time edge data, and enables each grid edge node to accurately share the power communication information associated with its respective aggregation center, forming a power knowledge collaboration body for edge computing, and providing a global shared foundation for real-time feedback of the power status of new energy microgrids.

[0081] It should be noted that external factors include weather, electricity prices, geographical features, external damage, and electromagnetic interference. Different grid facilities in a new energy microgrid typically exhibit interconnected control behaviors according to the requirements of the entire production process. For example, the electricity generated by distributed power sources needs to be converted into a form of electricity suitable for the microgrid load through energy conversion devices (such as inverters and transformers). Therefore, there is a close link between these two grid facility nodes. Real-time distributed sensor data serves as the probabilistic constraint condition for power conversion communication between facility nodes. This method uses the pre-set operating condition change probabilities of the conditional transfer function of the real-time distributed sensor data to jointly establish the link variable points with power control connections, generating a Belief message channel diagram. The Belief message channel diagram, also known as the belief-based message channel diagram, represents the joint probability distribution as a local factor function, enabling the mutual transmission of probability information (beliefs) between power facility nodes. Furthermore, based on the edge distribution of subgrid communication, different predetermined scales of disturbance variable messages are constructed to quantify the global dependency initial value of the dynamically changing grid operating state, ensuring the fidelity of collaborative reasoning in real-time status feedback. The mutual transmission of perturbation variable messages between link variable points and adjacent factor constraint nodes ensures that the global collaborative information of edge data calculation gradually converges to the local variables of the power grid facilities. This allows for a refined solution of the local edge probability distribution of each power facility, enabling the topology nodes of the new energy microgrid to express the power generation, power consumption, grid status, and external interference factors of the facilities in a real-time and concrete manner based on power data. Compared with traditional power grid operation monitoring methods, this method can visualize the real-time operation status of the topology grid at both the global macroscopic and local microscopic levels. It replaces the cumbersome steps and large amount of calculation errors caused by manual intervention, improves the analysis and response efficiency of autonomous operation, and enables the new energy microgrid to make more reasonable and accurate autonomous control and regulation decisions, reducing the chaos in autonomous operation.

[0082] More specifically, S104 includes the following steps:

[0083] Obtain the standardized operation process, autonomous operation requirements, and preset operation strategies of the new energy microgrid, and extract multiple related operation decision indicators based on the standardized operation process;

[0084] Based on the preset operation strategy, extract several sets of potential decision solutions for new energy microgrids under the premise of meeting the autonomous operation requirements, and the feasible values ​​of each potential decision solution for each operation decision index discrete programming. Obtain the original affinity of each potential decision solution, and map and bind the potential decision solution, the original affinity and the corresponding feasible value to construct a decision body memory bank.

[0085] The real-time chaotic sequence of the power flow vector during the operation of the new energy microgrid is extracted through the real-time running topology field, and a real-time Logistic chaotic mapping model is constructed based on the real-time chaotic sequence.

[0086] The desired Logistic chaotic mapping model is obtained through autonomous operation requirements. If the mapping deviation between the actual Logistic chaotic mapping model and the desired Logistic chaotic mapping model is greater than the preset mapping deviation, the memory consideration rate and the mutation adjustment rate are set based on the mapping deviation.

[0087] Based on the real-time chaotic sequence, fine-tuning mapping values ​​are obtained by using the memory consideration rate analysis in the decision-making body memory bank. New decision solutions are obtained, and the new decision solutions are iteratively replaced by the mutation adjustment rate to obtain the optimal operation scheme of the new energy microgrid.

[0088] It should be noted that the preset operation strategies of new energy microgrids typically formulate highly adaptable operating parameter values ​​for different real-time operating conditions, providing accurate control inputs for the autonomous operation of the grid. However, existing autonomous operation methods struggle to accurately identify and determine the optimal parameter values ​​that enable the grid to operate as required, leading to a high likelihood of deviation from the expected power output demand in the autonomous operation control of new energy microgrids, and failing to guarantee optimal grid operation control performance. To address this, this method constructs a decision body memory bank by extracting potential decision solutions from the preset operation strategy and discretizing the feasible values ​​of each operational decision index for each potential decision solution. These operational decision indices include voltage, current, frequency, harmonics, power factor, and short-circuit capacity. Native affinity is used to measure the rationality and adaptability of the potential decision solutions to the current operating conditions of the new energy microgrid. Since the real-time operating topology records the power flow situation of the new energy microgrid in real time, including the direction and magnitude of the power flow, these flows illustrate the general trends of problems that the current grid operation needs to address and the tendency of system chaos disturbances, providing autonomous guidance with a constant vector. Therefore, this method constructs a real-time Logistic chaotic mapping model to represent the ecological diversity of the current power grid system based on the real-time chaotic sequence with uniform and aperiodic power flow vector distribution. This clarifies the location of the decision solution required for the initialization of current power grid operation, enabling the preset operation strategy to escape local optima in the selection of dynamic autonomous operation schemes for the power grid. This avoids the clustering or bias caused by ordinary random optimal solution exploration and effectively improves the mapping coverage of the memory bank. If the mapping deviation between the real-time Logistic chaotic mapping model and the expected Logistic chaotic mapping model is greater than the preset mapping deviation, it indicates that the current decision solution location of the power grid operation state has shifted locally, causing it to fail to meet the expected goal of autonomous operation. This means that the operation decision parameter values ​​provided for solving a certain link of the current power grid need to be redefined. The mapping deviation magnitude measures the degree of consideration for the necessity of reselecting the operation parameter from the existing solution and the range of variation in adjusting it from the existing parameter values. This enables the control of each parameter of the operation scheme to have a high degree of global optimality, improving the accuracy and reliability of the self-operation of the new energy microgrid. This method enables AI-powered decision-making for optimal operation of new energy microgrids, achieving intelligent dynamic matching of energy supply and demand, maximizing the stability and matching degree of power operation, optimizing energy utilization and reducing energy waste, and improving the overall energy efficiency of the microgrid system.

[0089] More specifically, the step of using the memory consideration rate analysis based on the real-time chaotic sequence to perform fine-tuning mapping values ​​in the decision-making body's memory bank to obtain new decision solutions, and then performing iterative replacement on the new decision solutions through the mutation adjustment rate to obtain the optimal operation scheme of the new energy microgrid, specifically includes the following steps:

[0090] By mapping the real-time chaotic sequence to values ​​in the decision-making body's memory bank, if the memory consideration rate is greater than the preset memory consideration rate, the feasible value is marked and selected as the main candidate feasible value in the decision-making body's memory bank, and the main candidate feasible value is finely adjusted according to the mutation adjustment rate to generate the first candidate community situation.

[0091] If the memory consideration rate is less than the preset memory consideration rate, the feasible value is marked as a secondary candidate feasible value, no fine-tuning is performed, and a second candidate community situation is generated.

[0092] By combining the situations of the first and second candidate communities, a new decision solution is output that enables the new energy microgrid to meet the requirements of autonomous operation, and the mutation affinity of the new decision solution is calculated.

[0093] If the mutation affinity is greater than the original affinity, then the new decision solution replaces the potential decision solution where the feasible value is located; otherwise, discard the new decision solution and keep the decision unchanged to obtain the optimal operating decision regarding the operating decision index.

[0094] By following the standardized operation process and merging all optimal operation decisions regarding operation decision indicators, the optimal operation scheme for the new energy microgrid is obtained.

[0095] It should be noted that mapping values ​​based on the real-world chaotic sequence can greatly enhance the ability to explore the solution space of the decision-making body's memory bank, reducing the probability of getting trapped in local optima. If the memory consideration rate is greater than the preset memory consideration rate, it indicates that the current decision solution of a certain link in the power grid has many non-optimal local operating parameter values, which cannot enable the autonomous operation of the microgrid to meet the expected goals. Therefore, a feasible value that tends towards the optimum is reselected from the decision-making body's memory bank as the candidate value of the main theme, i.e., the main candidate feasible value, and fine-tuned according to the degree of variation adjustment. Conversely, it indicates that the current operating parameter value is infinitely close to the optimum, and no fine-tuning is required; it can be directly retained as the continuation control solution. The generation of the first candidate community situation and the second candidate community situation relies on the experience and innovation exploration of the preset operating strategy, realizing the random creation of the optimal solution for the impromptu operation of the microgrid, and greatly improving the robustness and adaptability of the optimal operating scheme. Among them, the memory consideration rate controls the inheritance of the decision solution with the preset operating strategy, tending to retain the good genes planned by the strategy, ensuring that the autonomous operation goals are met while following the operating strategy. The mutation adjustment rate introduces fine-grained local perturbations, which helps to escape the local minima of the preset operating strategy, thereby ensuring a certain degree of global exploration in random generation. If the mutation affinity is greater than the original affinity, it means that the new decision solution has extremely high adaptability and control stability for the current power grid operation, and can be preferentially accepted by the current power grid, enabling the current power grid to achieve the target requirements. Therefore, the new decision solution replaces the potential decision solution where the feasible value is located, realizing the survival of the fittest solution evolution mechanism.

[0096] More specifically, S106, as Figure 2 As shown, the specific steps include:

[0097] S202: Obtain the power design specifications and standardized operation process of the new energy microgrid, and use Simulink power simulation software to input and analyze the power design data recorded in the power design specifications to construct a virtual power simulation model of the new energy microgrid;

[0098] S204: Based on the autonomous operation requirements, preset ideal operation reference values, and use a virtual power simulation model to perform virtual simulation tests on the optimal operation scheme with the ideal operation reference values ​​as the arrival constraint conditions. Create a system power structured state chain of the new energy microgrid, and record and obtain the set of simulation operation test parameters.

[0099] S206: During the simulation process, a power operation sliding surface with error tracking concept is designed for the new energy microgrid according to the autonomous operation requirements. On the power operation sliding surface, the control error trajectory of the system's power structured state chain is continuously monitored and tracked based on the simulated operation test parameter set. In this way, the simulated Lyapunov function that makes the system's power structured state chain approach the power operation sliding surface is evaluated and determined.

[0100] S208: Obtain one or more infeasible events of the new energy microgrid and the expected Lyapunov function of each infeasible event, construct an event responder for the infeasible event based on the expected Lyapunov function, and form an event responder array;

[0101] S210: Deploy the event responder array onto the power operation sliding surface. During simulation, determine whether there is any desired Lyapunov function in the simulated Lyapunov function. If so, trigger the event responder array to eliminate the optimal operating scheme and re-select the next optimal operating scheme in descending order.

[0102] S212: If it does not exist, design an ideal sliding mode control law based on the equivalent control and approximation solution of the ideal operating reference value. Apply external disturbance resistance events to the ideal sliding mode control law and analyze the resistance performance within the quantitative range. Verify the feasibility of the optimal operating scheme. If the feasibility is high, distribute the optimal operating scheme to all power facilities.

[0103] It should be noted that traditional autonomous operation methods typically require manual implementation of the calculated optimal operating scheme to verify its feasibility. This significantly reduces the efficiency and accuracy of feasibility verification, easily leading to uncertain control fluctuations and errors in the microgrid's power system. It also increases the damage rate during on-site microgrid testing and raises maintenance costs. To address this, this method utilizes Simulink power simulation software to construct a virtual power simulation model of the new energy microgrid to virtually simulate the optimal operating scheme and obtain key data for feasibility verification. This includes creating a structured state chain for the new energy microgrid's system power and a set of simulated operation test parameters. Compared to using on-site operation schemes for new energy microgrid facilities, this method significantly improves verification efficiency, providing reliable data support for subsequent verification and analysis. Furthermore, it replaces the tedious steps of traditional manual operations, saving time and effort, reducing equipment operating frequency and aging, and decreasing the probability of damage and maintenance costs. The structured state chain for the system power is a mathematical system model of the current microgrid power control, clearly reflecting the dynamic behavior of the system power, including state variables, control inputs, and disturbances, revealing the characteristics of the controlled object. A power system operating sliding surface is a hypersurface representing the desired behavior of a power system. When the system's state trajectory reaches and moves along this sliding surface, it exhibits strong robustness to external disturbances and changes in internal parameters, thus demonstrating good dynamic performance. Therefore, this method utilizes this power system operating sliding surface to continuously monitor and track the structured state chain of the power system to generate the control error trajectory when simulating the test parameter set. This dynamically characterizes the system's power control error, enabling further evaluation of the steady-state performance of the current power system's control response to the optimal operating scheme, i.e., by simulating the Lyapunov function description.

[0104] It should be noted that infeasible events include voltage fluctuations, voltage flicker, frequency drift, excessive power outages, power outage timeouts, and power imbalances. These infeasible events represent a series of factors that lead to unstable control in new energy microgrids. If one or more infeasible events occur, the Lyapunov function of the microgrid will fluctuate. Therefore, this method introduces an array of event responders constructed from infeasible events into the power operation sliding surface during the simulation process to verify whether the power control under the current optimal operating scheme of the power grid system still tends to the power operation sliding surface. If any desired Lyapunov function exists in the simulated Lyapunov function, it indicates that the equivalent dynamics of the power system on the sliding surface fluctuate, indicating that the optimal operating scheme can no longer maintain stable control operation of the power system on the sliding surface. Therefore, the optimal operating scheme is not adopted, and a new scheme is selected from the secondary alternatives in descending order to replace it. If the optimal operating scheme does not exist, it allows the power system to move stably along the sliding mode surface without divergence. Therefore, further verification of the external disturbance control of the optimal operating scheme is conducted to ensure the theoretical correctness of the scheme's control strategy. This method can eliminate infeasible or risky schemes for autonomous operation of new energy microgrids in advance, enhancing the stability and security of power system control and operation.

[0105] More specifically, if the aforementioned does not exist, then an ideal sliding mode control law is designed based on the equivalent control and approximation solution of the ideal operating reference value. An external disturbance resistance event is applied to the ideal sliding mode control law, and the resistance performance within a quantitative range is analyzed to verify the feasibility of the optimal operating scheme. If the feasibility is high, the optimal operating scheme is distributed to all power facilities. This specifically includes the following steps:

[0106] If it does not exist, then the equivalent control principle and the approach law are introduced. Based on the ideal operating reference value, the simulation operation test parameter set is solved in the equivalent control principle and the approach law to make the system state control maintain the autonomous operation requirements that approximate the power operation sliding surface, and the ideal sliding mode control law is designed.

[0107] External disturbance resistance events of the optimal operation scheme are extracted from the work log of the new energy microgrid. Interference operation terms are created based on the external disturbance resistance events and applied to the ideal sliding mode control law.

[0108] Calculate the coverage ratio between the disturbance boundary of the ideal sliding mode control law and the disturbance limit of the interference operation term, and determine the resistance index of the ideal sliding mode control law to the interference operation term based on the coverage ratio;

[0109] If the resistance index exceeds the preset resistance index range, the optimal operating plan is eliminated; if the resistance index does not exceed the preset resistance index range, the current optimal operating plan is distributed to all power facilities.

[0110] It should be noted that the simulated operation test parameter set is solved by using the equivalent control principle and the reaching law, and an ideal sliding mode control law is designed with the ideal operating reference value as the objective function. This ensures that the system state control always converges on the sliding surface to meet the execution constraints during power simulation. This ideal sliding mode control law ensures high robustness to disturbances and parameter uncertainties introduced by the optimal operating scheme during power system simulation. Therefore, this method uses this ideal sliding mode control law to further verify the control steady state of the optimal operating scheme. The coverage quantifies the robustness of the ideal sliding mode control law against uncertainties and disturbances in power system simulation, responding to whether the power system of the new energy microgrid can stably operate in the actual complex environment brought about by the optimal operating scheme, and thus determining the resistance index. If the resistance index exceeds the preset resistance index range, it indicates that the control law cannot resist external disturbances introduced within the subcritical range, proving that the optimal operating scheme is not feasible for the current autonomous operation of the new energy microgrid, and therefore the optimal operating scheme is eliminated and reselected in descending order. Conversely, if the resistance index is within the preset range, it indicates that the optimal operating scheme is feasible.

[0111] More specifically, S108 includes the following steps:

[0112] Based on the standardized operation process charter, the work weight coefficients of each power grid facility for the standardized normal operation of the new energy microgrid are obtained. The process control scope of each power grid facility within the new energy microgrid is divided according to the work weight coefficients, resulting in N sub-jurisdiction areas.

[0113] By optimizing the actual operation scheme of each power grid facility and monitoring the power operation steady state of the new energy microgrid in real time, the actual Lyapunov function of the system power control in each sub-jurisdiction area is obtained;

[0114] Extract the topological boundary regions where there is direct power continuity interaction between the power grid facilities of adjacent sub-jurisdictional areas in the real-time running topological field;

[0115] If the actual Lyapunov function corresponding to the power grid facility within the sub-jurisdiction area is greater than the actual Lyapunov function corresponding to the neighboring power grid facility within the topological boundary area, then the power facility within the sub-jurisdiction area is marked as a problem facility, and multiple fluctuation frequency inflection points of the actual power control interruption in the time sequence of the problem facility are obtained.

[0116] A stable gain control architecture corresponding to the initial closed loop of the system's power structured state chain is constructed using the PID underlying control logic. A frequency scanning algorithm is introduced to find a scaling weight function that can transform the uncertainty structure of the closed loop system at each fluctuation frequency inflection point, and the frequency domain singular value of the closed loop system is calculated once when the scaling weight function is found.

[0117] Based on the scaling weight function, a weighted transformation is performed on the topological boundary domain of the problematic facilities in the closed-loop system of the new energy microgrid, so that the gain feedback of the stable gain control architecture is reconfigured and adjusted.

[0118] If the frequency domain singular value is less than the preset frequency domain singular value, the gain feedback adjustment operation of the stable gain control architecture is stopped, and a steady-state structured gain compensator is finally generated. The steady-state structured gain compensator is sent and applied to the power grid facilities adjacent to the problematic facility, thereby actively compensating for the stable power control of the problematic facility.

[0119] It should be noted that the new energy microgrid sends the confirmed optimal operating plan to all grid facilities, and the devices execute the plan after mutually confirming the instructions. However, existing new energy microgrids struggle to accurately locate and troubleshoot real-time operational faults in grid facilities during the execution of the optimal operating plan, leading to the inability to detect power control anomalies in a timely manner when some grid facilities malfunction. To address this, this method divides the functional management area of ​​each grid facility by standardizing the normal operation work weight coefficients of the grid facilities in the new energy microgrid. This blocks the overall macroscopic new energy microgrid structure into different sub-management areas, making the troubleshooting of power control anomalies in grid facilities more precise and improving the accuracy of fault location. Specifically, the topology boundary region is a sensitive switching area where the Lyapunov function value is prone to sudden jumps during power transmission between adjacent grid facilities. Determining the Lyapunov function of adjacent grid facilities in this topology boundary region ensures the continuity of control stability analysis, thereby capturing energy discontinuities that occur during power switching and preventing misjudgments due to energy jumps. If the actual Lyapunov function corresponding to a power grid facility within a sub-jurisdiction area is greater than the actual Lyapunov function corresponding to a neighboring power grid facility within its topological boundary region, it indicates that there is an unstable power control situation between the power grid facility within the sub-jurisdiction area and the neighboring power grid facility. This means that there is a discontinuity or disconnection between the two power grid facilities in the power control switching region, indicating that there is a fault in the current power grid facility. Therefore, it is marked as a problematic facility, thereby achieving the effect of anomaly investigation and fault location of power grid facilities.

[0120] It should be noted that the inflection point of the fluctuation frequency is the control mutation point where the frequency fluctuates continuously due to the interruption of the actual power operation of the problematic facility in the time sequence. This method constructs a closed-loop structure of the new energy microgrid through the PID underlying control logic, that is, the stable gain control architecture of the initial closed loop corresponding to the system's power structured state chain, for closed-loop performance analysis. Subsequently, under the action of the current controller, the robustness of the power system to structured uncertainties under different fluctuation frequency points is analyzed, forming a weight matrix for scaling the uncertainty structure of the power system, i.e., the scaling weight function, and frequency domain singular values ​​that measure the robust stability and scaling performance of the uncertainty structure (control interruption) brought about by the reversal of the problematic facility in the closed-loop system. Next, a weighted transformation is performed on the topological boundary region corresponding to the problematic facility in the closed-loop system of the new energy microgrid according to the scaling weight function. This minimizes the frequency domain singularity of the gain of the stable gain control architecture. If the frequency domain singularity is less than the preset frequency domain singularity, it indicates that the frequency domain singularity has been minimized. At this time, the gain feedback adjustment of the stable gain control architecture can compensate for the power interruption control deviation introduced by the problematic facility. Therefore, the steady-state structured gain compensator is sent and applied to the grid facilities adjacent to the problematic facility. The gain compensation of the stable gain control architecture can more accurately correct the robust drift of the uncertain structure, thereby ensuring the robust stability of the power system against the control interruption introduced by the problematic facility and achieving the active compensation effect of stable power control for the problematic facility. This method can enhance the coordination and fault tolerance of the new energy microgrid, achieve seamless switching of fault control, and improve the robustness and reliability of the system.

[0121] More specifically, S110 includes the following steps:

[0122] Acquire real-time operation data of each power grid facility executing a preset operation strategy at a uniform time step, and construct a strategy trajectory network by combining the real-time operation data with the strategy trajectory of the interaction sampling actions, status and rewards of the preset operation strategy.

[0123] The dominance function of each strategy trajectory at continuous equilibrium time steps is evaluated through the power state evaluation system. If the dominance function is greater than the preset dominance function, the strategy trajectory is marked as a positive strategy trajectory; if it is less than the preset dominance function, it is marked as a negative strategy trajectory.

[0124] By using a pre-defined running strategy to weight the probability ratio of learning the positive strategy trajectory compared to learning the negative strategy trajectory in the strategy trajectory network, the strategy learning ratio is obtained. The strategy learning ratio is then used to plan and formulate the gradient direction of a proxy strategy that continuously learns from real-time running data.

[0125] Based on the autonomous operation requirements, the allowable adjustment range of the preset operation strategy is obtained. Based on the allowable adjustment range, a KL divergence constraint trust region is constructed. The conjugate gradient method is introduced to approximate the gradient direction of the proxy strategy, and the natural gradient direction is obtained.

[0126] A continuous linear search is performed on the real-time operating data along the natural gradient direction at uniform time steps, outputting several strategy optimization parameters and corresponding KL divergences. If the KL divergence is always within the KL divergence constraint trust region, the strategy optimization parameters are used to modify and optimize the preset operating strategy, thereby improving the adaptability of the new energy microgrid.

[0127] It should be noted that existing autonomous operation methods for renewable energy microgrids typically lack adaptive learning capabilities, resulting in operational strategies that cannot be updated in real time, thus reducing the accuracy of grid operation. To address this, this method constructs a strategy trajectory network that interactively samples real-time operational data and preset operational strategies. This network determines the agent behavior of the renewable energy microgrid based on the learned actual operational data according to the preset operational strategies. The dominance function measures the relative merit of a particular action compared to a strategy behavior. If the dominance function is greater than the preset dominance function, it indicates that the action for a certain real-time operational parameter is superior to a certain strategy behavior, i.e., a positive strategy trajectory; conversely, it is inferior, i.e., a negative strategy trajectory. It provides the direction and magnitude of the strategy gradient, helping to identify the target strategy behavior that should be strengthened. This method generates a natural gradient direction by approximating the gradient direction of the proxy policy. Then, it performs a continuous linear search along this natural gradient direction on the real-time running data at uniform time steps. This achieves the effect of continuously learning and optimizing the operating strategy from each output of real-time running data. Simultaneously, a KL divergence-constrained trust region is added. This KL divergence-constrained trust region limits the magnitude of policy updates, ensuring that the new optimized strategy does not deviate significantly from the preset operating strategy and preventing instability in the learning optimization process. This method can collect and analyze feedback data during each operation, thereby continuously learning and optimizing the operating strategy. This allows the operating strategy to adapt to environmental changes, achieving long-term dynamic intelligent optimization and enhancing the adaptability and sustainable operation capability of the new energy microgrid.

[0128] The second aspect of this invention provides a smart autonomous operation system for new energy microgrids, such as... Figure 3 As shown, the intelligent autonomous operation system of the new energy microgrid includes a memory 31 and a processor 32. The memory 31 stores a program for an intelligent autonomous operation method of the new energy microgrid. When the program for the intelligent autonomous operation method of the new energy microgrid is executed by the processor 32, the steps of the intelligent autonomous operation method of the new energy microgrid described in any one of the above descriptions are implemented.

[0129] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for intelligent autonomous operation of a new energy microgrid, characterized in that, Includes the following steps: S102: Real-time data of new energy microgrid in power generation, power consumption, grid status and external factors are monitored in real time through distributed sensors and edge computing. The real-time data is used to perform topology calculation for message transmission to obtain the real-time operating topology of the new energy microgrid. S104: Construct a decision body memory bank using the potential decision solutions and corresponding feasible values ​​that meet the autonomous operation requirements in the preset operation strategy. Analyze and replace the values ​​in the decision body memory bank according to the real-time chaotic sequence of the power flow vector describing the real-time operation topology field to generate the optimal operation scheme. S106: The optimal operating scheme is tested through virtual simulation. The feasibility of the optimal operating scheme is verified on the power operation sliding surface of the system power and under the applied external disturbance resistance based on the simulated operation test parameter set. The optimal operating scheme is then issued to all power facilities. S108: Divide each power grid facility into multiple sub-jurisdiction areas, analyze the power operation steady-state lock-in problem facilities in each sub-jurisdiction area, and robustly design steady-state structured gain compensators for adjacent power grid facilities to actively compensate for the stable power control of problem facilities. S110: Obtain real-time operation data of each power grid facility executing preset operation strategies, continuously learn from the real-time operation data to optimize strategies, and improve the adaptability of new energy microgrids; Specifically, S108 includes the following steps: Based on the standardized operation process charter, the work weight coefficients of each power grid facility for the standardized normal operation of the new energy microgrid are obtained. The process control scope of each power grid facility within the new energy microgrid is divided according to the work weight coefficients, resulting in N sub-jurisdiction areas. By optimizing the actual operation scheme of each power grid facility and monitoring the power operation steady state of the new energy microgrid in real time, the actual Lyapunov function of the power control system in each sub-jurisdiction area is obtained; Extract the topological boundary regions where there is direct power continuity interaction between the power grid facilities of adjacent sub-jurisdictional areas in the real-time running topological field; If the actual Lyapunov function corresponding to the power grid facility within the sub-jurisdiction area is greater than the actual Lyapunov function corresponding to the neighboring power grid facility within the topological boundary area, then the power facility within the sub-jurisdiction area is marked as a problem facility, and multiple fluctuation frequency inflection points of the actual power control interruption in the time sequence of the problem facility are obtained. A stable gain control architecture corresponding to the initial closed loop of the system's power structured state chain is constructed using the PID underlying control logic. A frequency scanning algorithm is introduced to find a scaling weight function that can transform the uncertainty structure of the closed loop system at each fluctuation frequency inflection point, and the frequency domain singular value of the closed loop system is calculated once when the scaling weight function is found. Based on the scaling weight function, a weighted transformation is performed on the topological boundary domain of the problematic facilities in the closed-loop system of the new energy microgrid, so that the gain feedback of the stable gain control architecture is reconfigured and adjusted. If the frequency domain singular value is less than the preset frequency domain singular value, the gain feedback adjustment operation of the stable gain control architecture is stopped, and a steady-state structured gain compensator is finally generated. The steady-state structured gain compensator is sent and applied to the power grid facilities adjacent to the problematic facility, thereby actively compensating for the stable power control of the problematic facility.

2. The intelligent autonomous operation method for new energy microgrids according to claim 1, characterized in that, S102 specifically includes the following steps: Real-time monitoring of the new energy microgrid through distributed sensors acquires real-time distributed sensing data of each grid facility in the new energy microgrid at a preset time and space sequence, as well as real-time distributed edge data output by each grid edge node. The edge message communication protocol of the new energy microgrid is obtained. Based on the edge message communication protocol, the neighboring collaborative nodes of each grid edge node are identified and connected. An edge communication blueprint is constructed, and the communication trajectory layout between each grid edge node is extracted based on the edge communication blueprint. According to the communication trajectory layout, the attraction centroids corresponding to each power grid edge node are randomly initialized and allocated. Real-time distributed edge data is injected into the attraction centroids near the corresponding output power grid edge node and the distribution is iteratively updated to form a sub-power grid communication edge distribution with each power grid edge node as the edge radiation core. Obtain a schematic diagram of the grid facility layout of the new energy microgrid. Based on the schematic diagram, each grid facility is designated as a variable point. According to the conditional transfer function that generates real-time distributed sensor data based on the correlation control between each grid facility, a working condition transition probability is preset. The working condition transition condition is defined as a factor constraint node. Each variable point and its corresponding factor constraint node are linked together to generate a Belief message channel diagram. Based on the edge distribution of subgrid communication, perturbation variable messages of different predetermined scales are constructed and transmitted to the message channel connecting the link variable points and adjacent factor constraint nodes in the Belief message channel diagram. This enables the nodes to be updated iteratively and the current scale of each pair of nodes to be obtained synchronously. During the node iteration update process, if the size deviation between the current disturbance message size and the predetermined disturbance message size is less than a preset threshold, the message transmission operation of the node pair is stopped, and the edge probability of the power grid facility corresponding to the link variable point in each node pair is finally output. By introducing Kirchhoff's laws and using edge probability and subgrid communication edge distribution as the basis, the power generation, power consumption, grid status and external factors of the power grid are plotted in the topology of Kirchhoff's laws, thus obtaining the real-time operating topology field of the new energy microgrid.

3. The intelligent autonomous operation method for new energy microgrids according to claim 1, characterized in that, S104 specifically includes the following steps: Obtain the standardized operation process, autonomous operation requirements, and preset operation strategies of the new energy microgrid, and extract multiple related operation decision indicators based on the standardized operation process; Based on the preset operation strategy, extract several sets of potential decision solutions for new energy microgrids under the premise of meeting the autonomous operation requirements, and the feasible values ​​of each potential decision solution for each operation decision index discrete programming. Obtain the original affinity of each potential decision solution, and map and bind the potential decision solution, the original affinity and the corresponding feasible value to construct a decision body memory bank. The real-time chaotic sequence of the power flow vector during the operation of the new energy microgrid is extracted through the real-time running topology field, and a real-time Logistic chaotic mapping model is constructed based on the real-time chaotic sequence. The desired Logistic chaotic mapping model is obtained through autonomous operation requirements. If the mapping deviation between the actual Logistic chaotic mapping model and the desired Logistic chaotic mapping model is greater than the preset mapping deviation, the memory consideration rate and the mutation adjustment rate are set based on the mapping deviation. Based on the real-time chaotic sequence, fine-tuning mapping values ​​are obtained by using the memory consideration rate analysis in the decision-making body memory bank. New decision solutions are obtained, and the new decision solutions are iteratively replaced by the mutation adjustment rate to obtain the optimal operation scheme of the new energy microgrid.

4. The intelligent autonomous operation method for new energy microgrids according to claim 3, characterized in that, The process of using the memory consideration rate analysis based on the real-time chaotic sequence to perform fine-tuning mapping values ​​in the decision-making body's memory bank, obtaining new decision solutions, and then iteratively replacing these new decision solutions using the mutation adjustment rate to obtain the optimal operation scheme for the new energy microgrid, specifically includes the following steps: By mapping the real-time chaotic sequence to values ​​in the decision-making body's memory bank, if the memory consideration rate is greater than the preset memory consideration rate, the feasible value is marked and selected as the main candidate feasible value in the decision-making body's memory bank, and the main candidate feasible value is finely adjusted according to the mutation adjustment rate to generate the first candidate community situation. If the memory consideration rate is less than the preset memory consideration rate, the feasible value is marked as a secondary candidate feasible value, no fine-tuning is performed, and a second candidate community situation is generated. By combining the situations of the first and second candidate communities, a new decision solution is output that enables the new energy microgrid to meet the requirements of autonomous operation, and the mutation affinity of the new decision solution is calculated. If the mutation affinity is greater than the original affinity, then the new decision solution replaces the potential decision solution where the feasible value is located; otherwise, discard the new decision solution and keep the decision unchanged to obtain the optimal operating decision regarding the operating decision index. By following the standardized operation process and merging all optimal operation decisions regarding operation decision indicators, the optimal operation scheme for the new energy microgrid is obtained.

5. The intelligent autonomous operation method for new energy microgrids according to claim 1, characterized in that, S106 specifically includes the following steps: Obtain the power design specifications and standardized operation process of the new energy microgrid, and use Simulink power simulation software to input and analyze the power design data recorded in the power design specifications to construct a virtual power simulation model of the new energy microgrid; Based on the autonomous operation requirements, an ideal operation reference value is preset. The optimal operation scheme is then simulated and tested using a virtual power simulation model with the ideal operation reference value as the arrival constraint condition. A system power structured state chain of the new energy microgrid is created, and the set of simulation operation test parameters is recorded and obtained. During the simulation process, a power operation sliding surface with error tracking concept is designed for the new energy microgrid based on the autonomous operation requirements. On the power operation sliding surface, the control error trajectory of the system's power structured state chain is continuously monitored and tracked based on the simulated operation test parameter set. In this way, the simulated Lyapunov function that makes the system's power structured state chain approach the power operation sliding surface is evaluated and determined. Obtain one or more infeasible events of the new energy microgrid and the expected Lyapunov function of each infeasible event, construct event responders for the infeasible events based on the expected Lyapunov function, and form an event responder array; The event responder array is deployed on the power operation sliding surface. During simulation, it is determined whether there is any expected Lyapunov function in the simulated Lyapunov function. If there is, the event responder array is triggered to eliminate the optimal operating scheme and re-select the next optimal operating scheme in descending order. If it does not exist, then design an ideal sliding mode control law based on the equivalent control and approximation solution of the ideal operating reference value. Apply external disturbance resistance events to the ideal sliding mode control law and analyze the resistance performance within a quantitative range to verify the feasibility of the optimal operating scheme. If the feasibility is high, then distribute the optimal operating scheme to all power facilities.

6. The intelligent autonomous operation method for new energy microgrids according to claim 5, characterized in that, If the aforementioned does not exist, then an ideal sliding mode control law is designed based on the equivalent control and approximation solution of the ideal operating reference value. An external disturbance resistance event is applied to the ideal sliding mode control law, and the resistance performance within a quantitative range is analyzed to verify the feasibility of the optimal operating scheme. If the feasibility is high, the optimal operating scheme is distributed to all power facilities. Specifically, this includes the following steps: If it does not exist, then the equivalent control principle and the approach law are introduced. Based on the ideal operating reference value, the simulation operation test parameter set is solved in the equivalent control principle and the approach law to make the system state control maintain the autonomous operation requirements that approximate the power operation sliding surface, and the ideal sliding mode control law is designed. External disturbance resistance events of the optimal operation scheme are extracted from the work log of the new energy microgrid. Interference operation terms are created based on the external disturbance resistance events and applied to the ideal sliding mode control law. Calculate the coverage ratio between the disturbance boundary of the ideal sliding mode control law and the disturbance limit of the interference operation term, and determine the resistance index of the ideal sliding mode control law to the interference operation term based on the coverage ratio; If the resistance index exceeds the preset resistance index range, the optimal operating plan is eliminated; if the resistance index does not exceed the preset resistance index range, the current optimal operating plan is distributed to all power facilities.

7. The intelligent autonomous operation method for new energy microgrids according to claim 1, characterized in that, S110 specifically includes the following steps: Acquire real-time operation data of each power grid facility executing a preset operation strategy at a uniform time step, and construct a strategy trajectory network by combining the real-time operation data with the strategy trajectory of the interaction sampling actions, status and rewards of the preset operation strategy. The dominance function of each strategy trajectory at continuous equilibrium time steps is evaluated through the power state evaluation system. If the dominance function is greater than the preset dominance function, the strategy trajectory is marked as a positive strategy trajectory; if it is less than the preset dominance function, it is marked as a negative strategy trajectory. By using a pre-defined running strategy to weight the probability ratio of learning the positive strategy trajectory compared to learning the negative strategy trajectory in the strategy trajectory network, the strategy learning ratio is obtained. The strategy learning ratio is then used to plan and formulate the gradient direction of a proxy strategy that continuously learns from real-time running data. Based on the autonomous operation requirements, the allowable adjustment range of the preset operation strategy is obtained. Based on the allowable adjustment range, a KL divergence constraint trust region is constructed. The conjugate gradient method is introduced to approximate the gradient direction of the proxy strategy to obtain the natural gradient direction. A continuous linear search is performed on the real-time operating data along the natural gradient direction at uniform time steps, outputting several strategy optimization parameters and corresponding KL divergences. If the KL divergence is always within the KL divergence constraint trust region, the strategy optimization parameters are used to modify and optimize the preset operating strategy, thereby improving the adaptability of the new energy microgrid.

8. A smart autonomous operation system for a new energy microgrid, characterized in that, The intelligent autonomous operation system for new energy microgrids includes a memory and a processor. The memory stores a program for an intelligent autonomous operation method for new energy microgrids. When the program for the intelligent autonomous operation method for new energy microgrids is executed by the processor, the steps of the intelligent autonomous operation method for new energy microgrids as described in any one of claims 1-7 are implemented.

Images