Source network load storage collaborative optimization method for energy atom modeling and valence calibration

Abstract

The invention relates to the technical field of energy management and collaborative optimization, particularly discloses a source network load storage collaborative optimization method for energy atom modeling and valence calibration aiming at the high-dynamic energy guarantee requirement of a battlefield, and aims to solve the problems that in the prior art, the attributes of source network load storage units are fixed, and collaborative response is delayed. According to the method, source, network, load and storage units are abstracted into energy atoms with dynamic chemical properties, and the adjustment capability and interaction willingness of the energy atoms under the battlefield condition are corrected in real time through a valence dynamic calibration model; analyzing battlefield energy history and real-time data based on a long-short-term memory neural network, predicting a high-field-intensity region and situation gradient evolution trend, and providing guidance for atom deployment; according to the method, self-organization and self-adaptive cooperation of a battlefield energy system in a dynamic complex environment is realized, and the accuracy, the response speed and the overall robustness of energy allocation are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of learning optimization technology, specifically to a source-grid-load-storage collaborative optimization method based on energy atom modeling and valence calibration. Background Technology

[0002] The energy security system of modern battlefields is developing towards a distributed, multi-source heterogeneous, and highly dynamic direction. Its "source-grid-load-storage" components are becoming increasingly complex, covering everything from traditional solar panels and wind turbines to armored vehicles with multi-mode switching capabilities, as well as various cutting-edge individual soldier equipment and medical devices. Such systems not only need to meet the stable power supply requirements of routine tasks such as daily patrols, mobile support, and emergency medical care, but also need to achieve rapid, accurate, and reliable energy coordination in extreme scenarios such as sudden mission changes, equipment damage, or emergency medical care. The inherent dynamism, urgency, and spatial heterogeneity of the battlefield environment pose unprecedented challenges to the real-time response capability, environmental adaptability, and system resilience of energy management.

[0003] Existing collaborative optimization methods are generally based on static or simplified models, pre-setting core attributes such as the adjustment capability and energy interaction willingness of each unit as fixed parameters. This approach cannot dynamically correct based on the real-time power output and health status of the equipment, nor can it respond to multi-dimensional dynamic factors such as combat scenario switching, changes in battlefield demand gradients, and power deviations. This rigid representation method leads to a serious disconnect between the system model and the actual physical state, resulting in delayed decision-making. Furthermore, when dealing with multi-functional composite units such as dual-mode armored vehicles, custom-designed algorithms are often required, leading to high adaptation costs and integration complexity. In addition, traditional methods lack in-depth mining and forward-looking guidance of the spatiotemporal correlation of the battlefield situation, making it difficult to achieve second-level self-healing and resource reorganization when demand changes or failures occur. Therefore, how to construct an optimization framework that can calibrate unit attributes in real time, accurately perceive and predict the battlefield situation, and support autonomous collaboration of intelligent agents has become a key challenge in improving the agility and survivability of battlefield energy systems. To this end, this invention proposes a source-grid-load-storage collaborative optimization method based on energy atom modeling and dynamic valence calibration. Summary of the Invention

[0004] The purpose of this invention is to provide a source-grid-charge-storage collaborative optimization method for energy atom modeling and valence calibration.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a source-grid-charge-storage collaborative optimization method based on energy atom modeling and valence calibration, comprising atom modeling, field strength fusion, hierarchical decision-making, pre-preparation mechanism, and simulation optimization. The specific operation steps of the source-grid-charge-storage collaborative optimization method based on energy atom modeling and valence calibration are as follows:

[0006] Step 1: Define the source, network, load, and storage intelligent agent units as energy atoms with different chemical properties, and establish a dynamic calibration model for valence.

[0007] Step 2: Introduce a spatiotemporal correlation factor into the field strength calculation model, adaptively adjust the weights according to the task type, analyze historical situation data based on long short-term memory neural network, and predict the evolution trend of high field strength areas in advance.

[0008] Step 3: Construct a hierarchical decision-making architecture that includes local edge nodes and a central system. Local nodes are responsible for quickly matching temporary bonds, while the central system is responsible for optimizing stable bonds.

[0009] Step 4: Establish a chemical bond pre-breaking and pre-bonding prediction mechanism. When the rate of change of the state field exceeds a set threshold, the connection preparation of adjacent atoms is triggered in advance.

[0010] Step 5: Build a digital twin simulation platform to simulate diverse combat scenarios, quantitatively evaluate system performance indicators, and establish a closed-loop iterative mechanism to continuously optimize strategies.

[0011] As a further aspect of the present invention: In step one, the source-grid-load-storage intelligent body unit is defined as three types of basic atoms and one type of hybrid atom according to the energy interaction characteristics. The basic atoms include energy-supplying atoms, energy-consuming atoms, and buffer atoms. The electronegativity of energy-supplying atoms ranges from 0.1 to 0.3, and the number of valence electrons is ≥80. Specifically, they include solar panels, wind turbines, and armored vehicles in power supply mode. The electronegativity of energy-consuming atoms ranges from 0.7 to 0.9, and the number of valence electrons is ≤30. Specifically, they include individual soldier communication equipment and battlefield medical equipment. The electronegativity of buffer atoms ranges from 0.4 to 0.6, and the number of valence electrons is 40 to 60. Specifically, they include lithium battery packs, supercapacitors, and armored vehicles in energy storage mode. The criteria for determining hybrid atoms are that they simultaneously possess two or more functions of supply, consumption, and storage, and the proportion of a single function is <70%. Their electronegativity is calculated by weighting the proportion of each function, and the number of valence electrons is weighted and summed according to the functional adjustment capability.

[0012] As a further aspect of the present invention: in step one, a dynamic calibration model for valence is constructed: ,in, represent Real-time valence electron count after time calibration The nominal number of valence electrons representing an intelligent agent unit. Represents the power deviation correction factor. Represents the equipment health status coefficient. This represents the cumulative deviation correction amount within the calibration period. represent Real-time electronegativity after time calibration The baseline electronegativity representing the type of atom, Represents the weighting coefficient of the combat scenario. The gradient response coefficient of the representative morphogen, The atom type modification term defines the flexibility, valence electron number, and electronegativity of the intelligent agent unit. The valence electron number represents the regulatory ability of the intelligent agent unit, and electronegativity represents the strength of the intelligent agent unit's willingness to attract and provide flexibility. Based on the differences in valence electrons and electronegativity, intelligent agents form different types of virtual chemical bonds to self-organize into stable energy molecules. Energy molecules are dynamic; when the environment changes, old chemical bonds break and new energy molecules are formed with other intelligent agents.

[0013] As a further aspect of the present invention: In step two, a morphogenetic gradient field is constructed using the battlefield situation as the driving force, transforming battlefield requirements into situation intensity values. ,in, Representative area At any moment The strength of the situation, Represents the weighting coefficient. Representative moment area The Quantitative values ​​of situational elements The field strength calculation model is formed by fusing the battlefield situation field and the morphogenetic gradient field to represent the time decay coefficient. ,in, Representing coordinates At any moment The fusion field strength, Representing coordinates The strength of the situation, Representing coordinates The morphogenetic gradient magnitude, Represents spatiotemporal correlation factors. and Representing the fusion weight, the morphogenetic gradient field is updated synchronously with the situation field, so that the update of the morphogenetic gradient field meets the response to changes in the battlefield situation.

[0014] As a further aspect of the present invention: in step two, with As the core environmental state input of the Long Short-Term Memory (LSTM) model, it guides the agent units to learn strategies for migrating along the gradient direction, matching bonding objects, and selecting bond types. Based on the original historical situation data, time series data of morphogenetic concentration and gradient values ​​from the past 72 hours are added. The dataset is divided into 5-minute steps. In addition to outputting the evolution of high field strength regions in the next 30 minutes, the LTM model also outputs the migration path of the morphogenetic gradient center, guiding energy atoms to pre-deploy along the gradient path. When the LTM model predicts that the field strength change rate is >30% and the morphogenetic gradient G mutation is >0.3, it triggers the reconstruction of energy molecules. The agent units match bonding objects along the morphogenetic gradient direction, and each agent unit learns bonding through a multi-agent reinforcement learning model to form stable energy molecules.

[0015] As a further aspect of the present invention: In step three, a hierarchical decision-making architecture comprising local edge nodes and a central system is constructed. The local edge nodes are deployed at the company-level combat unit to process temporary bonding, real-time data acquisition, and preliminary fault diagnosis, and synchronously perceive local morphogenetic gradient values ​​and feed them back to the intelligent agent unit. The central system is deployed at the brigade-level command center to optimize stable bonding and global resource scheduling, and dynamically update morphogenetic gradient field parameters. The intelligent agent unit combines the local observation data fed back by the edge nodes with the global gradient field information issued by the central system to autonomously select the bonding type, which includes covalent bonds, ionic bonds, and coordinate bonds. Covalent bonds are suitable for conventional power supply and local load balancing, ionic bonds are suitable for emergency power supply and temporary fault replacement, and coordinate bonds are suitable for long-term stable power supply and core equipment protection.

[0016] As a further aspect of the present invention: In step four, the multi-agent reinforcement learning model monitors the rate of change of the situation field and the mutation rate of the morphogenetic gradient in real time. When any rate of change is greater than 50% / min, the central system sends a pre-preparation instruction. The pre-preparation instruction includes the edge nodes pre-storing the bonding parameter templates of five adjacent agent units, the buffer atoms adjusting their remaining power to 50%-70% in advance and pre-migrating along the morphogenetic gradient direction, and the power supply atoms starting the standby unit in advance. After triggering the pre-preparation, valuable adjacent atoms are selected based on the matching degree between the gradient field direction and the atom attributes. The edge nodes then send the list of selected candidate atoms to the corresponding agents.

[0017] As a further aspect of the present invention: In step four, the multi-agent reinforcement learning model predicts that the change in situation will cause the existing bonding to fail and needs to prepare for the breakage in advance. The candidate atom enters low-power standby mode, maintains communication with the edge node, and updates its own attributes in real time. If the situation deteriorates further, the central system issues an immediate bonding instruction, and the edge node directly calls the pre-stored bonding parameter template to enable the candidate atom to complete the bonding. If the situation eases, the central system issues a preparation to release instruction, and the agent returns to the normal state.

[0018] As a further aspect of the present invention: In step five, a morphogenetic gradient field simulation module is added to the digital twin simulation platform. The morphogenetic gradient field simulation module works in conjunction with the original module to customize the morphogenetic concentration threshold, gradient update frequency, and weight parameters in the gradient field, simulate the gradient distribution under different battlefield situations, and achieve synchronous simulation of the situation field and gradient field through timestamp alignment and data feedback, so that the simulation process is consistent with the changes in the real battlefield situation, and displays the migration trajectory of atoms along the gradient direction, the bonding position and the matching degree of the gradient center in real time. Closed-loop iterative optimization continuously optimizes the system strategy and gradient field parameters based on the simulation results.

[0019] Compared with the prior art, the beneficial effects of the present invention by adopting the above technical solution are as follows:

[0020] 1. This invention subdivides the source-grid-load-storage intelligent agent unit into three categories of basic atoms and hybrid atoms: energy supply, energy consumption, and buffer. Through a clear functional proportion determination standard and weighted calculation method, it is perfectly adapted to multi-functional composite units such as dual-mode armored vehicles. There is no need to develop a separate collaborative algorithm for such devices, which greatly reduces the adaptation cost of new devices. At the same time, the constructed dynamic calibration model of chemical valence can combine multiple dimensions such as power deviation, equipment health status, combat scenario and gradient response to correct the number of valence electrons (adjustment ability) and electronegativity (energy interaction willingness) of atoms in real time, so that the bonding matching between atoms is upgraded from traditional empirical judgment to quantitative calculation.

[0021] 2. This invention constructs a fusion model of the battlefield situation field and the morphogenetic gradient field, introduces spatiotemporal correlation factors and adaptive weights, and transforms abstract battlefield requirements into calculable field strength values. This provides precise spatial guidance for atomic migration, enabling atoms to move directionally along the gradient direction. Combined with the analysis of historical situation data and 72-hour morphogenetic related data by a long short-term memory neural network, it can predict the evolution of high field strength regions and the migration path of gradient centers 30 minutes in advance, guiding the pre-deployment of atoms. The chemical bond pre-breaking and pre-bonding prediction mechanism, by monitoring the comprehensive change rate of the situation field, triggers pre-preparation actions when a set threshold is reached, reducing the bonding response time under sudden demands. In the face of equipment damage and other failures, it can achieve rapid bonding and power restoration, fully meeting the second-level support requirements of emergency scenarios such as battlefield medical care and command and communication. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the source-grid-load-storage collaborative optimization method based on energy atom modeling and valence calibration. Detailed Implementation

[0023] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0024] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0025] This invention discloses a source-grid-charge-storage collaborative optimization method based on energy atom modeling and valence calibration, comprising atom modeling, field strength fusion, hierarchical decision-making, pre-preparation mechanism, and simulation optimization. The specific operation steps of the source-grid-charge-storage collaborative optimization method based on energy atom modeling and valence calibration are as follows:

[0026] Step 1: Define the source, network, load, and storage intelligent agent units as energy atoms with different chemical properties, and establish a dynamic calibration model for valence.

[0027] Step 2: Introduce a spatiotemporal correlation factor into the field strength calculation model, adaptively adjust the weights according to the task type, and analyze historical situation data based on a long short-term memory (LSTM) neural network to predict the evolution trend of high field strength areas in advance.

[0028] Step 3: Construct a hierarchical decision-making architecture that includes local edge nodes and a central system. Local nodes are responsible for quickly matching temporary bonds, while the central system is responsible for optimizing stable bonds.

[0029] Step 4: Establish a chemical bond pre-breaking and pre-bonding prediction mechanism. When the rate of change of the state field exceeds a set threshold, the connection preparation of adjacent atoms is triggered in advance.

[0030] Step 5: Build a digital twin simulation platform to simulate diverse combat scenarios, quantitatively evaluate system performance indicators, and establish a closed-loop iterative mechanism to continuously optimize strategies.

[0031] Example 1: Energy Atom Modeling and Collaborative Response in Emergency Medical Scenarios

[0032] A forward emergency medical post at the company level (deployed in plain terrain with no geographical barriers) needs to provide continuous power to 3 field ventilators and 1 electrocardiogram monitor. The following source-grid-load-storage intelligent agent units are required on site:

[0033] Power supply unit A: 2 x 100 kW solar panels (rated valence electron number 90, reference electronegativity 0.2, rated power generation efficiency 90%, current real-time power output 75 kW, actual efficiency 83%).

[0034] Energy-consuming atom B: 3 battlefield ventilators (each with a rated power of 20 kW, a rated number of valence electrons of 25, and a reference electronegativity of 0.8) + 1 electrocardiogram monitor (rated power of 5 kW, a rated number of valence electrons of 10, and a reference electronegativity of 0.85).

[0035] Buffer atom C: 1 50 kWh lithium battery pack (rated valence electron count 50, reference electronegativity 0.5, remaining capacity upper limit 80%, lower limit 20%, current remaining capacity 55%).

[0036] Hybrid Atom D: 1 dual-mode armored vehicle (can switch between power supply / energy storage modes; currently, the actual output of the power supply function is 32 kW, accounting for 40%; the current remaining power of the energy storage function is 65%, accounting for 60%; in the power supply mode, the rated number of valence electrons is 85 and the reference electronegativity is 0.2; in the energy storage mode, the rated number of valence electrons is 55 and the reference electronegativity is 0.5).

[0037] Property calibration of mixed atoms D

[0038] Baseline attribute calculation: Weighted by functional proportion, the baseline electronegativity of the hybrid atom D is "40% × 0.2 for energy supply function + 60% × 0.5 for energy storage function", and the rated valence electron number is "40% × 85 for energy supply function + 60% × 55 for energy storage function".

[0039] Real-time attribute calibration:

[0040] The combat scenario is emergency medical care, with a scenario weight coefficient of 1.4. The gradient value of the morphogenetic elements around the current medical point is 0.8. The gradient response coefficient is calculated as "1 + 0.4 × gradient value". The atom type correction term is calculated based on the proportion of energy supply function and energy storage function. The real-time electronegativity of the mixed atom D is obtained by substituting it into the valence dynamic calibration model.

[0041] The power deviation correction factor is "total real-time output of mixed atoms ÷ total rated output", the equipment health factor is "power supply efficiency × 40% + energy storage health × 60%", and the cumulative deviation correction is the average deviation of the first 5 calibrations × 0.1. Substituting these values ​​into the formula yields the real-time valence electron number of mixed atom D.

[0042] Situation intensity calculation:

[0043] The task priority is emergency medical care, with a quantitative value of 1.0 and a weight of 0.4; the total demand of the medical point is 85 kilowatts, and the current total energy supply is 129.5 kilowatts (75 kilowatts of solar power + 32 kilowatts of armored power supply + 22.5 kilowatts of energy storage), with a supply-demand gap quantitative value of 0.2 (no power shortage, calculated as a low gap), and a weight of 0.3; one backup solar panel has failed, with an equipment damage rate quantitative value of 0.3 (number of failures ÷ total number of devices), and a weight of 0.2; the geographical barrier coefficient of the plain terrain has a quantitative value of 0.1, and a weight of 0.1;

[0044] The situation was updated 3 minutes ago. The effective period of the emergency scenario situation is 5 minutes. The situation intensity of the medical point area is finally obtained based on the time decay coefficient.

[0045] Gradient field and field strength calculation:

[0046] The medical site has a situation intensity of 0.48, belonging to a secondary morphogen source, with an initial concentration of 0.7. The emergency scenario diffusion coefficient is taken as 0.5 km. The distance of buffer atom C from the medical site is 0.8 km, and the geographical barrier coefficient is 0.1. Calculate the morphogen concentration at the location of buffer atom C.

[0047] In emergency scenarios, the fusion weights of the situation field and gradient field are α=0.6 and β=0.4, and the spatiotemporal correlation factor is 1.03 (stable temporal correlation and small spatial gradient difference). The fused field strength at the location of the buffer atom C is calculated by substituting the weights into the field strength calculation model.

[0048] Key type selection and preparatory actions

[0049] Bond type matching: The difference in valence electrons between buffer atom C (52 valence electrons, electronegativity 0.48) and battlefield respirator B (25 valence electrons, electronegativity 0.8) is (52-25)÷25=108%, and the difference in electronegativity is 0.32 (0.3-0.5 range). Considering the needs of the emergency scenario, it is determined to be an ionic bond. The difference in valence electrons between energy-providing atom A (82 valence electrons, electronegativity 0.21) and buffer atom C is (82-52)÷52=57.7%, and the difference in electronegativity is 0.27 (0.3-0.5 range). It is determined to be a coordinate bond.

[0050] Pre-preparation trigger: Real-time monitoring of the overall change rate of the situation field. If the need for an additional ventilator at a medical facility increases the temporal change rate, spatial change rate, and overall change rate, and this continues for 20 seconds, the central system will issue a pre-preparation command:

[0051] The bonding parameter templates of the pre-stored buffer atom C at the cascade edge node and the surrounding 5 atoms (2 powered, 2 power-consuming, and 1 hybrid) include the interface protocol, power adjustment threshold (20-50 kW), and communication encryption key;

[0052] Buffer atom C adjusts its remaining charge and slowly pre-migrates 600 meters along the gradient direction (pointing towards the medical point), with the movement speed controlled at 2 km / h (to avoid collisions).

[0053] Power supply atom A starts the backup inverter and enters hot standby mode, with the output power stabilizing at 80 kilowatts (80% of the rated value), while completing fault self-check (no abnormalities).

[0054] Example 2: Self-healing and Digital Twin Verification After Equipment Damage

[0055] Fault occurrence and response

[0056] Fault detection: A solar panel in power supply atom A suddenly short-circuited, with the current abnormally dropping to 70% of the rated value and the voltage abnormally dropping to 82% of the rated value. The cascade edge node detected the fault within 0.9 seconds, immediately cut off the bonding link between the atom and other units, and sent fault location information (coordinates, fault type) to the brigade-level central system.

[0057] Global Update and Scheduling: The central system synchronously updates the situation field and gradient field of the fault area within 1.8 seconds. The situation intensity of the fault area increases from 0.48 to 0.72 (emergency level upgrade), the morphogenetic gradient center migrates to a 1-kilometer radius around the fault point, and the diffusion coefficient is adjusted to 0.4 kilometers; resources are scheduled according to redundancy backup priority.

[0058] Priority 1: Buffer atom C (0.8 km from the fault area, gradient direction matching degree 93%, remaining power 65%)

[0059] Priority 2: Hybrid Atom D (1.3 km from the fault area, gradient direction matching degree 89%, real-time output 54.5 kW);

[0060] Rapid bond formation recovery: At 3.8 seconds, hybrid atom D rapidly forms ionic bonds with the two affected ventilators based on the bonding parameter template pre-stored at the edge node (matching parameters with battlefield ventilator B), thus completing power restoration;

[0061] Digital twin simulation optimization

[0062] Simulation parameter configuration: Import actual data of the fault scenario into the digital twin platform, customize the morphogenetic concentration threshold to 0.1, the gradient update frequency to 5 Hz, and the initial value of the diffusion coefficient to 0.4 km; Simulate the situation field and gradient field simultaneously, align the timestamps, and display the atomic migration trajectory and bonding process in real time;

[0063] Simulation result evaluation:

[0064] Atomic migration trajectories: The migration direction of buffer atom C matches the gradient direction 93% and that of mixed atom D 89%, with no invalid migrations;

[0065] Bonding matching degree: The bonding position of the mixed atom D and the ventilator B is 0.4 km away from the gradient center, and the matching degree is "1 - (0.4 km ÷ gradient center coverage radius 0.6 km)";

[0066] Performance indicators: self-healing response time 4.3 seconds (meets the standard), bonding success rate 92% (close to the qualified threshold of 95%), resource utilization rate 86% (meets the standard).

[0067] Closed-loop iterative optimization: Based on simulation results, the gradient field diffusion coefficient and the "bonding matching degree weight" in the multi-agent reinforcement learning reward function are adjusted through particle swarm optimization algorithm. After resimulation, the matching degree between the bonding position and the gradient center increases, the bonding power increases, and all indicators meet the standards. The optimized parameters are used as the default configuration for this type of fault scenario.

[0068] Example 3: Stable Bonding Optimization in Conventional Patrol Scenarios

[0069] A brigade-level routine patrol area (covering 60 square kilometers, including 5 patrol squad outposts, with terrain mainly hilly) is equipped with 12 power supply units (50 kW wind turbines), 25 energy consumption units (individual soldier radios, 5 kW each), and 6 buffer units (20 kWh supercapacitors). Company-level edge nodes are deployed at the 5 outposts, and the brigade-level central system is deployed at the rear command center, responsible for stable bonding optimization.

[0070] Global modeling: The central system collects real-time data from each node—patrol station situation intensity 0.3-0.4 (normal scenario), morphogenetic gradient value 0.2-0.3 (low gradient stable region), diffusion coefficient 2.0 km; calculates the real-time properties of each atom (energy-supplying atom real-time valence electron count 75-82, electronegativity 0.18-0.22, buffer atom real-time valence electron count 45-52, electronegativity 0.45-0.53, energy-consuming atom real-time valence electron count 8-12, electronegativity 0.75-0.85);

[0071] Stable bonding configuration: Stable bonding is planned according to a cluster pattern of "1 energy-supplying atom + 1 buffer atom + 5 energy-consuming atoms" (lifetime 8 hours):

[0072] Energy supply atoms and buffer atoms: Based on the difference in valence electrons and electronegativity, they are determined to be coordinate bonds to ensure a stable energy supply to the buffer unit;

[0073] Buffer atoms and energy-consuming atoms: Based on the difference in valence electrons and electronegativity, they are determined to be ionic bonds, which can meet the distributed power supply needs of multiple radio stations;

[0074] Resource scheduling and balancing: In response to the situation of excess energy supply at station 3 and insufficient energy supply at station 5, the central system schedules the buffer atoms at station 3 to migrate to station 5 along the gradient direction to supplement the stable bonded clusters, and finally controls the energy supply and demand difference in the whole region within ±5%.

[0075] Long-term operation and simulation iteration

[0076] Operational results: The stable bonding cluster ran continuously for 8 hours, during which it experienced two wind speed fluctuations (power output ±15%). The buffer atoms dynamically adjusted and smoothed out the fluctuations using their remaining power.

[0077] Simulation optimization: In the scenario of "the movement of the patrol team causing the energy demand to migrate" simulated on the digital twin platform, it was found that the original gradient update frequency (2 Hz) could not respond to the change in demand in time. After iterative adjustment to 3 Hz and re-simulation, the atomic migration response latency was shortened from 12 seconds to 8 seconds, the resource utilization was improved, and the optimization strategy was applied to the actual system.

[0078] As attached Figure 1 As shown, with energy atom modeling as the core, the source, network, load and storage are abstracted into energy atoms and a dynamic calibration model of chemical valence is constructed. The battlefield situation field and the gradient field of morphogenetic elements are integrated to form a field strength model. A hierarchical decision-making architecture of company-level edge-brigade-level central is established, and a pre-fracture-pre-bonding mechanism is set up. Through digital twin simulation closed-loop optimization, the speed, accuracy and reliability of battlefield energy coordinated response are improved.

[0079] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Any variations and modifications can be made by those skilled in the art without departing from the spirit and scope of the invention. Therefore, any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention, without departing from the scope of the invention, fall within the protection scope defined by the claims of the present invention.

Claims

1. A source-grid-charge-storage collaborative optimization method based on energy atom modeling and valence calibration, comprising atom modeling, field strength fusion, hierarchical decision-making, pre-preparation mechanism, and simulation optimization, characterized in that, The specific operation steps of the source-grid-load-storage collaborative optimization method based on energy atom modeling and valence calibration are as follows: Step 1: Define the source, network, load, and storage intelligent agent units as energy atoms with different chemical properties, and establish a dynamic calibration model for valence. Step 2: Introduce a spatiotemporal correlation factor into the field strength calculation model, adaptively adjust the weights according to the task type, analyze historical situation data based on long short-term memory neural network, and predict the evolution trend of high field strength areas in advance. Step 3: Construct a hierarchical decision-making architecture that includes local edge nodes and a central system. Local nodes are responsible for quickly matching temporary bonds, while the central system is responsible for optimizing stable bonds. Step 4: Establish a chemical bond pre-breaking and pre-bonding prediction mechanism. When the rate of change of the state field exceeds a set threshold, the connection preparation of adjacent atoms is triggered in advance. Step 5: Build a digital twin simulation platform to simulate diverse combat scenarios, quantitatively evaluate system performance indicators, and establish a closed-loop iterative mechanism to continuously optimize strategies.

2. The source-grid-charge-storage collaborative optimization method for energy atom modeling and valence calibration according to claim 1, characterized in that: In step one, the source-grid-load-storage intelligent body unit is defined as three types of basic atoms and one type of hybrid atoms according to the energy interaction characteristics. The basic atoms include energy-supplying atoms, energy-consuming atoms and buffer atoms. The electronegativity of energy-supplying atoms ranges from 0.1 to 0.3 and the number of valence electrons is ≥80. Specifically, they include solar panels, wind turbines, and armored vehicles with power supply modes. The electronegativity of energy-consuming atoms ranges from 0.7 to 0.9, and the number of valence electrons is ≤30. Specifically, these include individual soldier communication equipment and battlefield medical equipment. The electronegativity of buffer atoms ranges from 0.4 to 0.6, and the number of valence electrons is 40 to 60. Specifically, they include lithium battery packs, supercapacitors, and armored vehicles with energy storage modes. The criteria for determining hybrid atoms are that they have two or more functions of supply, consumption, and storage, and the proportion of a single function is less than 70%. Their electronegativity is calculated by weighting the proportion of each function, and the number of valence electrons is weighted and summed according to the functional adjustment capability.

3. The source-grid-charge-storage co-optimization method for energy atom modeling and valence calibration according to claim 2, characterized in that: In step one, a dynamic valence calibration model is constructed: ,in, represent Real-time valence electron count after time calibration The nominal number of valence electrons representing an intelligent agent unit. Represents the power deviation correction factor. Represents the equipment health status coefficient. This represents the cumulative deviation correction amount within the calibration period. represent Real-time electronegativity after time calibration The baseline electronegativity representing the type of atom, Represents the weighting coefficient of the combat scenario. The gradient response coefficient of the representative morphogen, The atom type modification term defines the flexibility, valence electron number, and electronegativity of the intelligent agent unit. The valence electron number represents the regulatory ability of the intelligent agent unit, and electronegativity represents the strength of the intelligent agent unit's willingness to attract and provide flexibility. Based on the differences in valence electrons and electronegativity, intelligent agents form different types of virtual chemical bonds to self-organize into stable energy molecules. Energy molecules are dynamic; when the environment changes, old chemical bonds break and new energy molecules are formed with other intelligent agents.

4. The source-network-charge-storage co-optimization method for energy atom modeling and valence calibration according to claim 3, characterized in that: In step two, a morphogenetic gradient field is constructed using the battlefield situation as the driving force, transforming battlefield requirements into situation intensity values: ,in, Representative area At any moment The strength of the situation, Represents the weighting coefficient. Representative moment area The Quantitative values ​​of situational elements The field strength calculation model is formed by fusing the battlefield situation field and the morphogenetic gradient field to represent the time decay coefficient. ,in, Representing coordinates At any moment The fusion field strength, Representing coordinates The strength of the situation, Representing coordinates The morphogenetic gradient magnitude, Represents spatiotemporal correlation factors. and Representing the fusion weight, the morphogenetic gradient field is updated synchronously with the situation field, so that the update of the morphogenetic gradient field meets the response to changes in the battlefield situation.

5. The source-network-charge-storage collaborative optimization method for energy atom modeling and valence calibration according to claim 4, characterized in that: In step two, with As the core environmental state input of the Long Short-Term Memory (LSTM) model, it guides the agent units to learn strategies for migrating along the gradient direction, matching bonding objects, and selecting bond types. Based on the original historical situation data, time series data of morphogenetic concentration and gradient values ​​from the past 72 hours are added. The dataset is divided into 5-minute steps. In addition to outputting the evolution of high field strength regions in the next 30 minutes, the LTM model also outputs the migration path of the morphogenetic gradient center, guiding energy atoms to pre-deploy along the gradient path. When the LTM model predicts that the field strength change rate is >30% and the morphogenetic gradient G mutation is >0.3, it triggers the reconstruction of energy molecules. The agent units match bonding objects along the morphogenetic gradient direction, and each agent unit learns bonding through a multi-agent reinforcement learning model to form stable energy molecules.

6. The source-network-charge-storage collaborative optimization method for energy atom modeling and valence calibration according to claim 5, characterized in that: In step three, a hierarchical decision-making architecture including local edge nodes and a central system is built. The local edge nodes are deployed in the company-level combat units to handle temporary bonding, real-time data acquisition, and preliminary fault diagnosis, and synchronously perceive the local morphogenetic gradient value and feed it back to the intelligent agent unit. The central system is deployed in the brigade-level command center, optimizing stable bonding and global resource scheduling, and dynamically updating the morphogenetic gradient field parameters. The intelligent agent unit combines the local observation data fed back by the edge nodes with the global gradient field information issued by the central system to autonomously select the bond type, which includes covalent bonds, ionic bonds and coordinate bonds. Covalent bonds are suitable for conventional power supply and local load balancing, ionic bonds are suitable for emergency power supply and temporary fault replacement, and coordinate bonds are suitable for long-term stable power supply and core equipment protection.

7. The source-network-charge-storage co-optimization method for energy atom modeling and valence calibration according to claim 6, characterized in that: In step four, the multi-agent reinforcement learning model monitors the rate of change of the situation field and the mutation rate of the morphogenetic gradient in real time. When any rate of change is greater than 50% / min, the central system sends a pre-preparation instruction. The pre-preparation instruction includes the edge nodes pre-storing the bonding parameter templates of five adjacent agent units, the buffer atoms adjusting their remaining power to 50%-70% in advance and pre-migrating along the morphogenetic gradient direction, and the power supply atoms starting the standby unit in advance. After the pre-preparation is triggered, valuable adjacent atoms are selected based on the matching degree between the gradient field direction and the atom attributes. The edge nodes then send the list of selected candidate atoms to the corresponding agents.

8. The source-network-charge-storage co-optimization method for energy atom modeling and valence calibration according to claim 7, characterized in that: In step four, the multi-agent reinforcement learning model predicts that the change in situation will cause the existing bond to fail and needs to prepare for the breakage in advance. The candidate atom enters low-power standby, maintains communication with the edge node, and updates its own attributes in real time. If the situation deteriorates further, the central system issues an immediate bonding instruction, and the edge node directly calls the pre-stored bonding parameter template to enable the candidate atom to complete the bonding. If the situation eases, the central system issues a preparation to release the bond, and the agent returns to the normal state.

9. The source-network-charge-storage collaborative optimization method for energy atom modeling and valence calibration according to claim 8, characterized in that: In step five, a morphogenetic gradient field simulation module is added to the digital twin simulation platform. This module works in conjunction with the existing module to customize the morphogenetic concentration threshold, gradient update frequency, and weight parameters in the gradient field. It simulates the gradient distribution under different battlefield situations and achieves synchronous simulation of the situation field and gradient field through timestamp alignment and data feedback. This ensures that the simulation process is consistent with the changes in the real battlefield situation and displays the migration trajectory of atoms along the gradient direction, the bonding position, and the matching degree of the gradient center in real time. Closed-loop iterative optimization continuously optimizes the system strategy and gradient field parameters based on the simulation results.

Images