A source-load-storage system real-time control method and device based on a micro-increase rate of equal-loss value

Abstract

The application discloses a source-load-storage system real-time control method and device based on a slight increase rate of loss value, relates to the field of power system dispatching, and comprises the following steps: performing multi-level anomaly detection and cleaning on multi-source heterogeneous real-time data, and outputting device adjustable state and verification data; calculating regional control bias ACE based on the verification data; calculating real-time adjustable margin of each station, constructing a dynamic loss value slight increase rate model of unified quantification source-load-storage regulation loss, performing power distribution to obtain a preliminary instruction; performing multi-constraint safety boundary verification and fast suboptimal correction on the preliminary instruction; judging functions and safety interlocks based on device state and system safety indexes; and issuing control instructions by adopting a time-event double triggering mechanism combined with a multi-period scheduling strategy. The application realizes unified quantification and dynamic optimal distribution of heterogeneous resource generalized regulation loss, breaks through the bottleneck of separate operation of multiple AGC systems, and improves the collaborative control precision of the source-load-storage system under complex working conditions.

Description

Technical Field

[0001] This application relates to the field of power system dispatching, and in particular to a real-time control method and device for a source-load-storage system based on the incremental rate of equal loss value. Background Technology

[0002] Vigorously developing new energy sources is an inevitable choice for current development. To date, China's installed capacity of wind power, photovoltaics, and other new energy sources ranks first in the world, becoming the world's largest power source. However, the inherent randomness, volatility, and intermittency of new energy power generation pose severe challenges to the real-time control and safe operation of power systems. The regulation resources of new power systems have expanded from dozens of conventional units to over ten thousand heterogeneous entities, including wind power, photovoltaics, energy storage, solar thermal, and controllable loads. Their regulation characteristics and losses differ significantly, and coupled with power control deviation assessment constraints, it becomes difficult to achieve low-loss, high-precision integrated real-time collaborative control under complex operating conditions.

[0003] Traditional power systems rely primarily on conventional power sources with high inertia and adjustability, such as coal and hydropower, using proportional-integral control to smooth out small load disturbances and maintain stable system frequency and tie-line power. However, with the increasing installed capacity of new energy sources, the regulation resources of new power systems have expanded from dozens of conventional units to over ten thousand heterogeneous entities, including wind power, photovoltaics, energy storage, solar thermal, and controllable loads. The regulation characteristics and losses of these entities differ significantly. Power fluctuations in new energy sources can exceed 10% of installed capacity within minutes, and there are thousands of configurations for megawatt-level distributed resource clusters. Response times range from milliseconds to minutes, and the constraint of a 2% power control deviation makes it difficult to achieve low-loss, high-precision integrated real-time collaborative control under complex operating conditions. This requires real-time control systems to possess comprehensive multi-resource collaborative control strategies, extremely fast response speeds, strong anti-disturbance capabilities, and high adaptability.

[0004] Existing automatic generation control technologies mainly face the following technical bottlenecks: 1. Separate control architectures, difficulties in coordinating heterogeneous resources, and control risks. To circumvent the coordination challenges of heterogeneous resources such as new energy, energy storage, and controllable loads in terms of regulation costs and response characteristics, existing AGC systems often set up independent control systems according to the type of regulation resource. This results in the grid dispatching system needing to execute multiple decomposed strategies simultaneously during real-time control, facing coordination failures and control risks under complex operating conditions and real-time constraints. For example, the Inner Mongolia West Provincial Dispatch Center operates three independent AGC systems: conventional thermal power AGC (Beijing Kedong), new energy AGC (Beijing Zhongke Furui), and energy storage AGC (Beijing Kedong). Under this architecture, thermal power, as the main power source, primarily undertakes frequency regulation and tie-line deviation adjustment functions; the new energy AGC adjusts the output of wind and photovoltaic power plants at the minute level based on the spot market clearing results to participate in real-time balancing; while energy storage power plants are currently mainly used as a temporary dispatching means to support peak evening periods when new energy output is low. This discrete model leads to the need for grid dispatch to execute multiple decomposed strategies simultaneously during real-time control, each independently executing the spot market clearing results, lacking an integrated and coordinated control strategy for wind, solar, thermal, and energy storage. 2. The optimization model is statically rigid and difficult to adapt to the dynamic and time-varying characteristics of the source-load-storage system. Traditional economic dispatch and AGC control models are mostly built based on the "equal consumption incremental rate" criterion, whose core assumption is that the cost characteristics and regulation capabilities of power generation resources are relatively fixed. These models are mainly designed for thermal power units whose coal consumption curves have undergone long-term operational calibration and whose parameters change slowly. Once the model parameters are tuned, they often remain unchanged for a considerable timescale. However, in power systems with a high proportion of new energy sources, operating conditions exhibit significant dynamic and time-varying characteristics: the output of wind and solar power fluctuates dramatically on a minute or even second basis due to weather conditions; the regulation capability of the energy storage system is coupled with its state of charge (SOC) in real time, and when the SOC is too high or too low, the available regulation capacity and power support capability will change nonlinearly; the response margin of the controllable load changes instantaneously with factors such as user energy consumption behavior and ambient temperature. Traditional models have static parameters and cannot dynamically respond to complex operating conditions such as drastic fluctuations in renewable energy output, real-time changes in the state of charge (SOC) of energy storage, and instantaneous changes in the controllable load adjustment margin. This leads to a disconnect between optimization results and reality in scenarios with a high proportion of renewable energy, resulting in decreased control accuracy and adaptability. 3. Insufficient ability to cope with abnormal operating conditions and massive amounts of data limits the reliability and agility of the control system.Existing AGC systems lack a systematic real-time data anomaly detection and cleaning mechanism at the data quality and status awareness level. This makes it impossible to effectively identify and eliminate erroneous data, communication delays, and abnormal equipment states, resulting in a weak foundation for control decisions and directly affecting control accuracy and safety. At the level of massive resource management, facing heterogeneous controlled objects of "over ten thousand" scale, traditional centralized optimization methods suffer from the disaster of computational dimensionality, making it difficult to complete the optimal solution within the millisecond-level control cycle. The lack of effective dynamic clustering and dimensionality reduction methods leads to slow instruction generation and an inability to match the real-time response characteristics of various resources.

[0005] Therefore, there is an urgent need for a real-time control method that can uniformly quantify the adjustment of heterogeneous resource losses, dynamically adapt to complex working conditions, and possess adaptive fault tolerance and massive resource dimensionality reduction management capabilities. Summary of the Invention

[0006] The purpose of this application is to provide a real-time control method and device for a source-load-storage system based on the infinitesimal increase rate of equal loss value, so as to overcome the defects of existing real-time control technologies for source-load-storage systems, such as discrete control architecture, static and rigid optimization models, difficulty in managing massive objects, and poor adaptability to abnormal operating conditions.

[0007] To achieve the above objectives, this application provides the following solution: This application provides a real-time control method for a source-load-storage system based on the infinitesimal rate of increase in value with equal loss, including: S1: Performs multi-level anomaly detection and cleaning on real-time operating data of multi-source heterogeneous systems from new energy sources, energy storage, thermal power, and controllable loads, and outputs real-time data after adjusting the status markers of the equipment and verification. S2: Calculate the regional control deviation ACE based on the verified real-time data and by introducing a dynamic frequency deviation coefficient; S3: Perform low-pass filtering and proportional-integral control on the area control deviation ACE to generate a total regulation power demand command, and obtain the total power to be allocated after intelligent dead zone judgment; S4: Calculate the real-time adjustability margin of each station based on the verified real-time data; S5: Construct the physical output boundaries of equipment at each station; S6: Determine the feasible domain of commands based on the real-time adjustability margin of each station and the physical output boundary of the equipment at each station; S7: Construct a dynamic loss value incremental rate model; S8: Using the total power to be allocated as the allocation target, within the instruction feasible domain, iteratively allocate power according to the dynamic loss value incremental rate model and the principle of equal loss value incremental rate to obtain a preliminary instruction set; S9: Based on the adjustable status flag of the device and the verified real-time data, perform multi-constraint safety boundary verification and real-time correction on the preliminary instruction set. If the verification fails, calculate the instruction correction amount and return to step S3 to regenerate the total adjustment power demand instruction. If the verification passes, output the safety instruction set. S10: Based on the adjustable status flag of the device and the system security indicators, perform functional interlocking and security interlocking judgment on the set of security instructions, eliminate unadjustable resource instructions, and generate valid issuance instructions; S11: Based on the effective issued instructions and the real-time operating status of the system, control instructions are issued to each station or virtual control unit using a time-event dual-trigger mechanism combined with a multi-cycle scheduling strategy.

[0008] Secondly, this application provides a real-time control device for a source-load-storage system based on the infinitesimal rate of increase in value with equal loss, comprising: The anomaly detection and cleaning module is used to perform multi-level anomaly detection and cleaning on real-time operating data of multi-source heterogeneous systems from new energy sources, energy storage, thermal power, and controllable loads, and outputs adjustable status markers and verified real-time data. The area control deviation calculation module is used to calculate the area control deviation ACE based on the verified real-time data and by introducing a dynamic frequency deviation coefficient. The module for determining the total power to be allocated is used to perform low-pass filtering and proportional-integral control on the area control deviation ACE, generate a total regulation power demand command, and obtain the total power to be allocated after intelligent dead zone judgment. The real-time adjustable margin calculation module is used to calculate the real-time adjustable margin of each station based on the verified real-time data. The equipment physical output boundary construction module is used to construct the equipment physical output boundaries of each site; The instruction feasible region determination module is used to determine the instruction feasible region based on the real-time adjustability margin of each station and the physical output boundary of the equipment of each station; The Dynamic Loss Value Increment Rate Model Construction Module is used to construct a dynamic loss value incremental rate model. The preliminary instruction set determination module is used to take the total power to be allocated as the allocation target, and within the instruction feasible domain, iteratively allocate power according to the dynamic loss value incremental rate model and the principle of equal loss value incremental rate to obtain the preliminary instruction set. The correction module is used to perform multi-constraint safety boundary verification and real-time correction on the preliminary instruction set based on the adjustable status flag of the device and the verified real-time data. If the verification fails, the instruction correction amount is calculated and returned to the total power to be allocated determination module to regenerate the total adjustment power demand instruction. If the verification passes, the safety instruction set is output. The judgment module is used to perform functional interlocking and security interlocking judgments on the set of security instructions based on the adjustable status flags of the device and system security indicators, eliminate instructions for unadjustable resources, and generate valid instructions to be issued. The control command issuing module, based on the valid issued commands and the real-time operating status of the system, uses a time-event dual-trigger mechanism combined with a multi-cycle scheduling strategy to issue control commands to each station or virtual control unit.

[0009] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a real-time control method and device for a source-load-storage system based on the infinitesimal rate of increase in value with equal loss, which has the following significant advantages: 1. Breakthrough in achieving unified economic synergy for heterogeneous resources. By establishing a dynamic loss value incremental rate model, this paper for the first time unifies and quantifies multi-dimensional losses such as the cost of new energy curtailment, energy storage cycle life loss, and controllable load service quality loss. It introduces a real-time operation status correction coefficient to dynamically characterize the marginal cost of unit active power adjustment, providing a comparable economic decision-making benchmark for source-load-storage resources with distinct characteristics, and minimizing the overall system adjustment loss.

[0010] 2. The technical challenge of high-precision collaborative control of multi-resource clusters under dynamic operating conditions has been overcome. The proposed equal-loss value incremental rate allocation method, combined with multi-level anomaly detection and adaptive fault-tolerant strategy, enables the control model to respond in real time to drastic fluctuations in new energy sources and changes in energy storage SOC; the time-event dual-trigger command issuance mechanism takes into account both the agility of the system's emergency response and the economy of different equipment physical characteristics, significantly improving control accuracy and anti-disturbance capability.

[0011] 3. It provides an engineering-implementable, independently controllable equipment solution. By using dynamic clustering and dimensionality reduction technology, the problem of coordinating resources of "over 10,000 units" is reduced to virtual units of hundreds of variables. Combined with the online hot-start solution and fast suboptimal correction module for the Lagrange equation, it achieves real-time optimal solution for multi-megawatt clusters within millisecond cycles, fundamentally solving the risk of collaborative failure and computational dimensionality disaster caused by the separate operation of multiple systems. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating a real-time control method for a source-load-storage system based on the infinitesimal rate of increase in value with equal loss, provided as an embodiment of this application; Figure 2 A framework diagram for calculating regional control deviations provided in an embodiment of this application; Figure 3 A diagram illustrating the principle of allocating incremental rates of equal loss value according to an embodiment of this application; Figure 4 This is a flowchart illustrating the optimal deviation allocation based on the incremental rate of equal loss value, provided as an embodiment of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0015] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0016] In one exemplary embodiment, such as Figure 1 As shown, a real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method includes the following steps: S1: Performs multi-level anomaly detection and cleaning on real-time operating data of multi-source heterogeneous systems such as new energy, energy storage, thermal power and controllable loads, and outputs real-time data after adjustable status marking and verification of equipment.

[0017] To ensure the reliability, consistency, and timeliness of the data upon which control decisions are based, this application establishes a multi-layered, adaptive, and integrated real-time data anomaly detection mechanism that integrates system operating status and data-driven processes. This mechanism performs online cleaning, verification, and status assessment of multi-source heterogeneous real-time operating data collected from the power plant, including data from new energy sources, energy storage, thermal power, and controllable loads. The output adjustable equipment status flags and anomaly information will directly serve as important criteria for subsequent steps, ensuring that only healthy and controllable equipment can receive and execute commands. Through this mechanism, end-to-end data quality assurance is achieved from data input to command correction and secure command issuance.

[0018] The real-time data set detected in this step is represented by a multi-dimensional vector. This indicates that it covers the key operational states of various controllable resources at different time points: ; Where N is the total number of stations; for any one of the stations The verification data is as follows: ; in, These represent the active and reactive power outputs of station k at time t, respectively. Let k be the voltage at the power station at time t. The AGC system operating status of station k is a binary 0 / 1 variable. The system is running normally, when System exit, This indicates that the data is abnormal.

[0019] 1. Single-point data verification Dead data verification: When the system experiences communication interruptions or other issues, the change in data points within T consecutive sampling periods must be less than a minimum threshold. If the data is deemed dead, it is marked as invalid data.

[0020] ; in, This represents the problem data; t Indicates the current time; T Indicates the inspection cycle; k Indicates station k; Representing various types of data: ; Threshold exceedance verification: Checks whether the real-time data of the site exceeds physical limits or the rated limits of the equipment. Among these, This refers to the lower limit of various monitoring data, such as the minimum active power of the power station. Minimum voltage at the station Minimum reactive power of the station ; This represents the upper limit of the monitoring data for the site, such as the maximum active power. The highest voltage at the station The highest reactive power of the station .

[0021] ; in, This indicates data that exceeds the limit; Rate of change / gradient anomaly check: For power data, the rate of change should conform to the physical inertia of the equipment. If the rate of change... Exceeding the maximum permissible grade ratio If so, it is determined to be a data anomaly.

[0022] ; ; in, This indicates gradient anomaly data.

[0023] New energy power output verification: measured wind and solar power With ultra-short-term predictive power , and theoretical maximum power output Compare the data; if the measured data deviates significantly from the prediction band or exceeds the theoretical limit, the data is considered abnormal.

[0024] ; ; .

[0025] in, This indicates a deviation from the equipment's maximum output data; This indicates a power output below the device's predicted minimum power capacity. This indicates that the power output exceeds the equipment's predicted maximum output power. This indicates the deviation margin.

[0026] 2. Power balance residual verification: total injected power of the power station group It should be basically balanced with the outflow power and the load power. Residual If the threshold is exceeded If so, it is determined to be a data anomaly.

[0027] ; ; in, Indicates the station’s efforts; This indicates the load on the connecting lines in the area; Indicates line loss power; This represents the unbalanced quantity, when it is greater than... This means the data is abnormal.

[0028] 3. Delay judgment An abnormal signal must persist for a sufficiently long time before the system recognizes it as a genuine and valid anomaly, rather than a momentary interference or jitter. Therefore, after the real-time data verification, a delay is used to further determine whether it is a valid anomaly. Only when both the real-time verification data and the delayed verification data are abnormal is the data considered abnormal. ; in, This indicates the finally confirmed abnormal data; This indicates the abnormal signal at station k at time t; k Indicates station k; Indicates the duration of the signal.

[0029] 4. Comprehensive judgment of system operating status Based on the above data verification results, combined with the hard node status signals sent by the equipment (such as AGC activation / deactivation, communication interruption, manual mode, etc.), the adjustable status of each station is comprehensively determined. When an AGC abnormality occurs at a station... Communication abnormality In manual mode Data anomalies At that time, the station was determined to be unadjustable.

[0030] in, Indicates an uncontrollable state; This indicates that AGC has exited. Indicates a communication failure; This indicates that the system is in manual mode. This indicates that the data is abnormal.

[0031] Multi-level real-time data anomaly detection enables front-end data quality assurance and adjustable state determination. The cleaned and verified data lays a solid foundation for the calculation of area control deviation (ACE) and the generation of closed-loop control signals.

[0032] S2: Calculate the regional control deviation (ACE) based on the verified real-time data and by introducing a dynamic frequency deviation coefficient.

[0033] For detailed procedures, please refer to [link / document / documentation]. Figure 2 In area control, Area Control Error (ACE) measures the imbalance of a control area (e.g., a provincial power grid) at a given moment. It visually represents the "imbalance" caused by the control area to the interconnected power grid. When ACE > 0, it indicates insufficient power generation or excessive load in the area. In this case, the AGC system needs to increase power generation or reduce load in the area to bring ACE to zero. When ACE < 0, it indicates excessive power generation or insufficient load in the area. In this case, the AGC system needs to reduce power generation or increase load to bring ACE to zero. The specific formula for calculating ACE is as follows: ; ; in, The measured frequency at time t at the base station; The system reference frequency; The dynamic frequency deviation coefficient represents the power generation that the region needs to adjust for every 1 Hz change in system frequency. To collect real-time measured power of the tie line; Planned power for tie lines; Represents the reference frequency deviation coefficient Traditional fixed-frequency deviation coefficients do not account for the significant changes in real-time inertia. When the proportion of renewable energy is high, the system inertia is low, and the same frequency deviation means a more severe power deficit. This application introduces a dynamic frequency deviation coefficient based on the real-time penetration rate of renewable energy. Its expression is shown in the above formula; where This is the reference frequency deviation coefficient, which is a fixed value tuned under the system's theoretical model; This represents the total active power output of all synchronous generators in the current system. This refers to the total generating capacity or total active load of the system at this moment. When thermal power plants are operating at full capacity... When the value is 1, the system operates in the traditional inertial mode; when photovoltaic power generation is high, the system inertia is low and the frequency changes rapidly. This avoids drastic fluctuations in the ACE value, allowing the ACE to approach zero in a more stable and optimal way, thus achieving active real-time control.

[0034] The calculated ACE is then low-pass filtered to eliminate high-frequency noise in the signal, resulting in a more stable control signal. A proportional-integral operation is then performed on the filtered ACE to generate a continuously adjustable total power demand command. In this stage, the patent introduces an adaptive proportional gain coefficient. This ensures that the dynamic performance of the control loop remains stable and consistent under different inertial conditions; finally, it is integrated with the feedforward control signal to obtain the total system regulation power. .

[0035] ; ; ; ; in, This represents the initial regulation power calculated on the source side, without considering the regulation potential of controllable loads. Indicates regional control deviation; The transfer function representing series correction; Represents the transfer function of the PI controller; Indicates the time-varying proportionality coefficient; Represents the time-varying gain coefficient; Represents a variable in the complex frequency domain; This indicates the controllable load adjustment power.

[0036] S3: Perform low-pass filtering and proportional-integral control on the area control deviation ACE to generate a total regulation power demand command, and obtain the total power to be allocated after intelligent dead zone judgment.

[0037] To ensure safe, economical, and stable operation in high-proportion source-load-storage systems, The instruction utilizes a smart dead zone. The introduction of this dead zone not only prevents equipment wear due to frequent responses to minute noises but also embeds the concept of economical operation into the final stage of control. The specific expression is as follows: ; ; ; in, The dynamic threshold for the intelligent dead zone; To fix the technical dead zone; This is the economic weighting coefficient; Marginal cost, i.e., the rate of incremental increase in the equivalent loss value of the called-up unit; The baseline marginal cost; if If the current available resources are too large and the cost is too high, then... Increased size makes the system more tolerant of deviations and reduces the need to use expensive equipment; if Smaller size, lower cost of currently available resources. If the value is reduced, the system will actively adjust to improve control accuracy.

[0038] After intelligent dead zones are implemented, resources are allocated to the system for adjustment; the adjustable resources considered in this application include: Power source side: Wind power Photovoltaics Light and heat Hydropower thermal power Energy storage wait; Load side: Controllable load clusters such as industrial and commercial loads, building air conditioning, electric vehicles, and data centers aggregated through virtual power plants.

[0039] Virtual power plants aggregate massive amounts of distributed, adjustable loads through a digital platform, combining numerous scattered and adjustable loads into a unified whole to participate in grid dispatch and electricity market transactions. This forms a "virtual regulation unit" with a unified dispatch interface, and its regulation potential is significant. It can serve as an important flexible adjustment resource for the system in real-time control. For example, when the frequency of the control area is low and it is necessary to increase the output of generating units while reducing the load, the virtual power plant can reduce the power of building air conditioners and the charging rate of electric vehicle charging piles according to the current adjustable range of resources to complete the process of controllable load adjustment.

[0040] ; ; in, It provides power to various controllable resources on the load side aggregated by the virtual power plant. The minimum effort required for all types of resources.

[0041] S4: Calculate the real-time adjustability margin of each station based on the verified real-time data.

[0042] 1. Adjustable resource margin and constraint analysis Based on the reliable real-time data of each power station provided in step 1, including the maximum power generation capacity... Actual output State of charge And so on, as well as the abnormal information returned by the anomaly detection, such as data anomalies. Communication abnormality Exit AGC mode Manual mode Quantitative analysis of the real-time adjustability of various resources is used to clarify the boundaries of power allocation.

[0043] First, the output boundary is calculated by combining real-time data from the station operation to clarify the real-time adjustable margin.

[0044] New energy power plants (wind power, photovoltaic): Their up and down adjustment margins depend on the difference between their real-time maximum / minimum generating power and the current actual output.

[0045] ; ; in, For new energy power stations The upward adjustment margin is calculated based on the theoretical maximum generating power of the power station. Compared with the current actual output difference, For station The utility coefficient. For station The downward adjustment margin, For the actual power output of station k, At the minimum output level, when no penalty for curtailment of wind and solar power is considered, the theoretical minimum output of a renewable energy power station is 0, meaning that it can completely curtail electricity.

[0046] Thermal power plants and energy storage facilities: Their regulation capacity is limited by both the upper and lower limits of equipment output and the ramp-up rate. Therefore, their actual adjustable margin is taken as the smaller value between "power margin" and "ramp-up capacity" to ensure that the commands are feasible.

[0047] ; ; in, (t) represents the margin for adjusting k at thermal power or energy storage plants, and its value is the maximum unit output. With current output The difference must not exceed the maximum climbing power. Lowering the margin similar.

[0048] Virtual power plant load aggregation: Its upward adjustment margin (increased electricity consumption) and downward adjustment margin (reduced electricity consumption) are determined by the adjustment baseline of each controllable load within the aggregation range and its allowable adjustment boundary.

[0049] ; ; in, To aggregate controllable load margin for virtual power plants, To maximize the output of each piece of equipment, This is where the virtual power plant contributes its power; To reduce the margin, M represents the minimum output of each device, indicating the total number of adjustable resources or distributed units aggregated in the virtual power plant (VPP). The first virtual power plant aggregation Distributed resources.

[0050] S5: Construct the physical output boundaries of equipment at each station In addition to the aforementioned upward and downward adjustment margin constraints, the power output of the station should also meet the technical boundary constraints: the power output command of the station must be strictly within the upper and lower limits of the physical output of the equipment.

[0051] ; in, Indicates the minimum power output of the generator unit; Indicates the actual power of the unit; This indicates the maximum power output of the generator unit.

[0052] S6: Determine the feasible domain of commands based on the real-time adjustable margin of each station and the physical output boundary of the equipment at each station.

[0053] The feasible command domain is a range consisting of the real-time adjustability margin and the upper and lower limits of the equipment's physical output.

[0054] S7: Construct a dynamic loss value incremental rate model.

[0055] By taking into account factors such as coal consumption of thermal power, curtailment of new energy / hydropower, energy storage life loss, and service quality of controllable load, a unified quantitative model of the incremental rate of loss value for source-load-storage regulation loss is created. This model dynamically characterizes the unit active power regulation cost under different operating conditions of source, load, and storage, expanding the quantification of regulation loss from the single-dimensional incremental rate of coal consumption of thermal power to the incremental rate of loss value of multiple resources (source, load, and storage).

[0056] Thermal power unit loss model: Using the classic quadratic coal consumption cost function, for each thermal power unit j, the following loss cost exists, where... Let j be the cost coefficient of the unit. It provides power to thermal power units.

[0057] ; in, This indicates the cost of loss.

[0058] The loss cost of each power station can be aggregated from the loss costs of several thermal power units, as expressed by the following formula: ; in, For thermal power plant The cost of loss, For station Cost coefficient, The power output of the power station; the power station cost coefficient is obtained by weighted average aggregation of the unit cost coefficients, as shown in the following expression: ; .

[0059] Energy storage loss model: Energy storage system losses are a complex process, mainly stemming from battery aging caused by charge-discharge cycles. The magnitude of these losses depends not only on the charge-discharge power but also on the battery's state of charge (SOC) and state of health (SOH). This application constructs an energy storage loss value function that comprehensively considers multiple factors. Includes cyclic aging losses SOC offset penalty and charge / discharge efficiency loss .

[0060] ; ; ; ; in, This is the cost coefficient. For the quadratic and linear coefficients of the power term in the aggregated SOC offset penalty cost, For electricity sales price, For charging and discharging efficiency, here we use or To distinguish the charging and discharging directions, This indicates the health status of battery j in generator set. Indicates the state of charge of unit j. This represents the reference value for the state of charge of the energy storage unit (usually 0.5, near which battery loss is minimal). This represents the penalty coefficient in the aggregated SOC offset penalty cost related to the degree of SOC deviation.

[0061] Similarly, the energy storage site loss cost function is derived from the above cost model: ; ; ; ; in, This represents the total loss cost of the k-th energy storage station. This represents the aggregate cycle aging loss cost of the k-th energy storage station. This represents the aggregate SOC offset penalty cost for the k-th energy storage station. This represents the aggregate charge / discharge efficiency loss cost of the k-th energy storage station. This represents the aggregate equivalent health status of the k-th energy storage station. This represents the coefficient of the quadratic term in the cyclic aging cost function after the k-th energy storage station is aggregated. This represents the coefficient of the first-order term in the cyclic aging cost function after the k-th energy storage station is aggregated. Indicates the output of energy storage station k. Represents the coefficient of the constant term. This represents the charging and discharging efficiency of the station after k aggregation. This represents the equivalent state of charge after the aggregation of station k. This represents the reference state of charge, which is the baseline value that the energy storage battery is expected to maintain at its SOC. and This represents the quadratic and linear coefficients of the power term in the aggregated SOC offset penalty cost; This represents the penalty coefficient in the aggregated SOC offset penalty cost related to the degree of SOC deviation; Indicates the electricity price. The cost of deviation penalty is represented in exponential form; when the battery is in a bad state (large SOC deviation) and still needs to output high power, the total loss cost will soar exponentially, thus forcing the scheduling algorithm to avoid such dangerous operating conditions.

[0062] New energy generating units: Their cost is mainly "opportunity cost", that is, the loss of wind and solar power due to curtailment, which is related to the difference between real-time available power and actual output.

[0063] Photovoltaic units: Photovoltaic units Loss cost The formula is as follows: ; in, For the curtailment cost parameters of unit j, For photovoltaic units Operating cost parameters This represents the theoretical maximum power output of the j-th photovoltaic unit. This represents the actual active power output of the j-th photovoltaic unit.

[0064] Photovoltaic power station (obtained by combining photovoltaic units): Photovoltaic power station Loss Costs The formula is as follows: ; in, For station Curtailment cost parameters For photovoltaic power stations Operating cost parameters are obtained by weighted average of unit parameters; This represents the actual active power output of the k-th photovoltaic power station; This represents the theoretical maximum power output of the k-th photovoltaic power station.

[0065] Solar thermal power units: solar thermal power units Loss Costs The expression is as follows: ; in, For the unit Curtailment cost parameters; This represents the maximum usable solar thermal energy collected by the j-th solar thermal power unit; This represents the actual thermal energy utilized by the j-th solar thermal power unit; This represents the thermoelectric conversion loss coefficient of a solar thermal power unit; This represents the actual power generation of unit j.

[0066] ; Solar thermal power plant (obtained by polymer solar thermal power units): Solar thermal power plant Loss Costs The formula is as follows: ; in, For station The cost parameters for curtailed solar power are obtained by weighted average of the unit parameters. This represents the maximum usable solar thermal energy collected by solar thermal power station k; This represents the actual thermal energy utilized by station k; This represents the actual power generation of station k.

[0067] Wind turbine: Wind turbine Loss Costs The expression is as follows: ; in, For the unit Wind curtailment cost parameters; For wind turbines Operating cost parameters; This indicates the maximum power output of wind turbine j; This indicates the actual active power output of unit j.

[0068] Wind farm (obtained by aggregating wind turbine units): Wind turbine units Loss Costs The expression is as follows: ; in, For the curtailment cost parameter of wind farms, For wind farms Operating cost parameters are obtained by weighted average of unit parameters; This indicates the maximum generating capacity of wind farm station k; This represents the actual power output of wind farm station k.

[0069] Virtual power plant aggregation of controllable loads: Various controllable loads aggregated by virtual power plants are important regulation resources; this application establishes a unique loss value function for each type of controllable load based on its characteristics.

[0070] Electric vehicles: When electric vehicles participate in demand response, their "losses" are mainly reflected in the cost of battery cycle aging. This cost is closely related to charge and discharge power, current state of charge (SOC), and state of battery health (SOH).

[0071] ; ; in, Battery aging cost factor Actual charging power This is the lowest charging power. Rated charge / discharge power; This is the state-of-charge penalty factor, reflecting the degree of influence of regulation on battery life under different SOCs; Let $\frac{i}{i}$ represent the adjustment loss cost of the $i$-th electric vehicle. This represents the adjustable power increment of the i-th electric vehicle; Let represent the coefficient of the linear term of the adjustment cost of the i-th electric vehicle.

[0072] Air conditioning: When air conditioning load participates in demand response, "losses" are mainly reflected in the reduction of equipment lifespan and the reduction of user comfort.

[0073] ; ; in, This is the power regulation lifespan loss coefficient; This refers to the operating power of the air conditioner. Minimum power (MW); Rated power; This is the thermal comfort loss coefficient; This refers to the actual indoor temperature. Set the temperature for the user; This represents the adjustment loss cost of i air conditioners; This represents the adjustable power increment of the i-th air conditioner.

[0074] Industrial and commercial interruptible loads: When industrial and commercial interruptible loads (such as factory production lines, large commercial air conditioning, lighting systems, motor equipment, etc.) participate in demand response, their core "losses" are reflected in the compensation costs for production losses and service quality decline.

[0075] ; in, This is the cost coefficient. For actual power reduction; This is the rated interruptible capacity; To reduce time; This represents the adjustment loss cost of the i-th interruptible industrial and commercial load.

[0076] By aggregation, a loss cost model for the controllable resources of a virtual power plant is formed: ; in, for , , sum; From the above The weighted average aggregation yields the result. For a fixed linear term, its value is all sum.

[0077] S8: Using the total power to be allocated as the allocation target, within the feasible domain of the instructions, iteratively allocate power according to the dynamic loss value incremental rate model and the principle of equal loss value incremental rate to obtain a preliminary instruction set.

[0078] Taking the first derivative of the cost function with respect to power yields the "depletion rate of value" for each resource. It accurately represents the marginal economic cost incurred by each unit increase or decrease in power adjustment at the current operating point of the resource.

[0079] The principle of allocating incremental values ​​based on equal loss rates: The ideal optimization objective is that all resources actually involved in the adjustment have the same incremental rate after allocation, at which point the total adjustment loss of the system is minimized. For example... Figure 3 As shown: When At this point, the marginal costs of multiple generators are equal, and the total cost is minimized when all units operate at equal marginal costs.

[0080] ; ; ; ; ; ; ; in, For station Injected active power, For load power, This is due to network loss in the system. For the slight increase in network loss; branch road Admittance; branch road Current, The branch resistance is denoted by ; the nodes at both ends of the branch are denoted by . node, They are nodes The voltage; Energy storage stations Total loss cost function, thermal power plant Total loss cost function, virtual power plant Total loss cost function, new energy power station The total loss cost function; This represents the incremental rate of the equivalent loss value of the k-th energy storage station after accounting for network losses; This represents the incremental rate of the equivalent loss value of the k-th thermal power plant after accounting for network losses; This represents the incremental rate of the equal loss value of the k-th virtual power plant (aggregating controllable loads) after accounting for network losses; This represents the incremental rate of the equivalent loss value of the k-th renewable energy power station after accounting for network losses; This represents the voltage phase angle difference between node i and node j; This indicates the power output of station k.

[0081] After initial allocation, it is verified whether the instructions for each resource exceed its "instruction feasible region". If a resource instruction exceeds the limit (e.g., it has been allocated to its maximum output), its instruction is fixed at that limit, and the resource is temporarily removed from the current adjustable resource list. Subsequently, the remaining unmet adjustment demands are recalculated, and a new round of equal incremental rate allocation is repeated in the remaining set of adjustable resources. The above iterative process continues until all adjustment demands are met, or all available resources are exhausted, ultimately outputting an optimal power instruction set that satisfies all constraints and minimizes (or nearly minimizes) the total economic cost of system adjustment. The process is as follows: Figure 4 As shown.

[0082] Steps S4-S8 follow the total regulation power demand calculated in step S3. The core task is to allocate total demand economically and accurately to all controllable resources, including new energy, thermal power, energy storage, and virtual power plants, while meeting the physical and security constraints of various resources. To achieve this goal, this method innovatively introduces the "equal loss value incremental rate" criterion into the unified quantitative analysis of diversified resource costs, thus constructing a unified optimization framework.

[0083] S9: Based on the adjustable status flag of the device and the verified real-time data, perform multi-constraint safety boundary verification and real-time correction on the preliminary instruction set. If the verification fails, calculate the instruction correction amount and return to step S3 to regenerate the total adjustment power demand instruction. If the verification passes, output the safety instruction set.

[0084] Initial instructions from the "optimal deviation allocation stage" Multiple safety checks must be performed before issuing the command to ensure that it does not exceed the physical limits and adjustment capabilities of the equipment. If the check fails, the command must be corrected and the process must return to the deviation calculation stage.

[0085] New energy instruction verification: For photovoltaic power plants, the control commands solved earlier... Should meet ,like or Then the following corrections will be made: ; in, This is the maximum power generation capacity of the photovoltaic power station. To preliminarily calculate the power output commanded by the photovoltaic power station, To amend the directives for photovoltaic power plants.

[0086] Similarly, the correction formulas for wind power, solar thermal power, and hydropower are as follows: ; ; in, This is the maximum power generation capacity of the wind farm. To preliminarily calculate the commanded power output of the wind farm, To amend the instructions for wind farms.

[0087] ; ; in, This is the maximum power generation capacity of a solar thermal power plant. To preliminarily calculate the power output commanded by the solar thermal power plant, To amend the directives for solar thermal power plants.

[0088] ; ; in, The maximum power generation capacity of the hydroelectric power station This represents the minimum generating capacity of the hydroelectric power station. To preliminarily calculate the power output commanded by the hydropower station, To amend the instructions for hydroelectric power plants.

[0089] For thermal power plants, control commands should meet the requirements for thermal power command boundaries and ramp rate verification: ; ; ; ; ; in, This is the minimum power generation capacity of the thermal power plant. For the maximum power generation of the thermal power plant, k Initial instructions were issued to thermal power plant K. The maximum ramping power of station k. To meet the correction commands for maximum and minimum power, To satisfy the correction instructions for the climbing constraint, This is the final correction instruction.

[0090] Energy storage status verification: ; ; ; The derivation yields: ; make: , ; but: ; ; ; ; in, This represents the maximum charging power of the energy storage unit. This is the unit's maximum discharge power. To initially issue the energy storage unit k command, For energy storage unit k, the maximum ramp power is... This represents the maximum state of charge of the energy storage unit. This represents the minimum state of charge for the energy storage unit. The rated capacity of the energy storage unit, For the unit k Real-time state of charge, unit power It should be kept between the minimum and maximum capacity; For the energy storage unit's instruction execution time, The minimum power constraint for equivalent energy, derived from the energy constraint, has a value of [value missing]. , This represents the maximum equivalent power derived from the energy constraints. This is a correction command under power constraints; This is a correction command under the climbing rate constraint; This is a correction command under power constraints; This is the final correction instruction.

[0091] Controllable load command correction: ; ; in, Maximize the output of controllable resources aggregated in the virtual power plant. Minimize the output of controllable resources aggregated in the virtual power plant; To issue initial power commands, This is the revised instruction.

[0092] Cross-sectional power flow verification: ; ; ; The transfer factor SF is used to represent the value of the node when the node is in its current state. k When the output changes by 1 MW, the branch The change in current on the surface is , Indicates when the line When the power flow exceeds the limit, the power output of station k needs to be adjusted; Let be the active power transmitted by line ij. branch road electrical conductivity, Let i be the electrical conductance to ground. For branch circuit susceptance, They are nodes The voltage; Let be the phase angle difference between node i and node j; This represents the maximum active power transmitted by the branch.

[0093] S10: Based on the adjustable status flag of the device and the system security indicators, perform functional interlocking and security interlocking judgments on the set of security instructions, eliminate unadjustable resource instructions, and generate valid issuance instructions.

[0094] When the proportion of new energy sources is high and the number of controlled objects is large, if a single equipment anomaly (such as communication interruption or AGC exit) or a system-level anomaly (such as sudden power changes in the tie line or manual intervention) is not identified and the command is blocked in time, it may lead to the issuance of erroneous commands, causing equipment overload, system oscillation, or even cascading failures.

[0095] Functional interlock: When station communication is abnormal Exit AGC mode Manual mode Data anomalies In situations such as these, a locking action is triggered to prevent erroneous operation.

[0096] ; ; ; ; ; Safety interlock: When a major safety issue occurs in the system, such as frequency fluctuations leading to system crash, The trend exceeds the limits of the equipment. Wait, at which point a global or local latch is triggered.

[0097] ; ; ; ; The system continuously monitors the interlocking conditions. When all interlocking conditions are not met, the system automatically releases the interlocking state and resumes command issuance. Simultaneously, the system records interlocking events and reports them to the monitoring interface for operators to trace.

[0098] S11: Based on the effective issued instructions and the real-time operating status of the system, control instructions are issued to each station or virtual control unit using a time-event dual-trigger mechanism combined with a multi-cycle scheduling strategy.

[0099] The command issuance adopts a fusion architecture combining "periodic baseline scheduling" and "event-driven emergency response." Once a command passes verification, correction, and interlock checks, it enters the issuance phase. At this point, according to... Assess the urgency of the event: If If the load is too high, the system faces risks such as frequency collapse and generator overload, which is considered an emergency and requires immediate instructions; if If the area is under normal control, instructions will be issued according to the original normal cycle; if If the instruction is located in the dead zone, no instruction will be issued; this achieves the goal of issuing instructions with high reliability, low latency, and strong security.

[0100] ; When issuing commands periodically, the physical characteristics, response rates, and control requirements of different devices are all different. For example, energy storage has the fastest response and is suitable for high-frequency regulation, thermal power has a slow ramp-up and is suitable for low-frequency regulation, virtual power plants have a certain delay in load aggregation response, and wind and solar power are greatly affected by weather. Therefore, issuing commands periodically requires designing multi-cycle scheduling strategies based on the characteristics of each type of power source.

[0101] ; ; The control cycles of different types of equipment are not the same, among which, For device type d, the distribution cycle is... This is the response cycle of a thermal power unit. Thermal power units respond slowly and have a long cycle, typically 3-10 seconds. The response cycle of a photovoltaic power station is relatively fast; The photothermal response cycle is relatively slow, typically requiring more than 20 seconds. The response cycle of the hydropower unit. For the response cycle of the energy storage unit, The controllable load response cycle is aggregated for virtual power plants.

[0102] By combining event triggering with multi-cycle scheduling, the system can ensure rapid response in emergency situations while taking into account the physical characteristics and control economy of various devices, thus achieving a highly reliable, low-latency, and highly adaptable command issuance mechanism.

[0103] In another embodiment of this application, a real-time control device for a source-load-storage system based on the constant loss value incremental rate is provided, comprising: The anomaly detection and cleaning module is used to perform multi-level anomaly detection and cleaning on real-time operating data of multi-source heterogeneous systems from new energy sources, energy storage, thermal power, and controllable loads, and outputs adjustable status markers and verified real-time data. The area control deviation calculation module is used to calculate the area control deviation ACE based on the verified real-time data and by introducing a dynamic frequency deviation coefficient. The module for determining the total power to be allocated is used to perform low-pass filtering and proportional-integral control on the area control deviation ACE, generate a total regulation power demand command, and obtain the total power to be allocated after intelligent dead zone judgment. The real-time adjustable margin calculation module is used to calculate the real-time adjustable margin of each station based on the verified real-time data. The equipment physical output boundary construction module is used to construct the equipment physical output boundaries of each site; The instruction feasible region determination module is used to determine the instruction feasible region based on the real-time adjustability margin of each station and the physical output boundary of the equipment of each station; The Dynamic Loss Value Increment Rate Model Construction Module is used to construct a dynamic loss value incremental rate model. The preliminary instruction set determination module is used to take the total power to be allocated as the allocation target, and within the instruction feasible domain, iteratively allocate power according to the dynamic loss value incremental rate model and the principle of equal loss value incremental rate to obtain the preliminary instruction set. The correction module is used to perform multi-constraint safety boundary verification and real-time correction on the preliminary instruction set based on the adjustable status flag of the device and the verified real-time data. If the verification fails, the instruction correction amount is calculated and returned to the total power to be allocated determination module to regenerate the total adjustment power demand instruction. If the verification passes, the safety instruction set is output. The judgment module is used to perform functional interlocking and security interlocking judgments on the set of security instructions based on the adjustable status flags of the device and system security indicators, eliminate instructions for unadjustable resources, and generate valid instructions to be issued. The control command issuing module, based on the valid issued commands and the real-time operating status of the system, uses a time-event dual-trigger mechanism combined with a multi-cycle scheduling strategy to issue control commands to each station or virtual control unit.

[0104] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0105] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A real-time control method for a source-load-storage system based on the infinitesimal rate of increase in value with equal loss, characterized in that, The real-time control method for the source-load-storage system based on the constant loss value increment rate includes: S1: Performs multi-level anomaly detection and cleaning on real-time operating data of multi-source heterogeneous systems from new energy sources, energy storage, thermal power, and controllable loads, and outputs real-time data after adjusting the status markers of the equipment and verification. S2: Calculate the regional control deviation ACE based on the verified real-time data and by introducing a dynamic frequency deviation coefficient; S3: Perform low-pass filtering and proportional-integral control on the area control deviation ACE to generate a total regulation power demand command, and obtain the total power to be allocated after intelligent dead zone judgment; S4: Calculate the real-time adjustability margin of each station based on the verified real-time data; S5: Construct the physical output boundaries of equipment at each station; S6: Determine the feasible command domain based on the real-time adjustability margin of each station and the physical output boundary of the equipment at each station; S7: Construct a dynamic loss value incremental rate model; S8: Using the total power to be allocated as the allocation target, within the instruction feasible domain, iteratively allocate power according to the dynamic loss value incremental rate model and the principle of equal loss value incremental rate to obtain a preliminary instruction set; S9: Based on the adjustable status flag of the device and the verified real-time data, perform multi-constraint safety boundary verification and real-time correction on the preliminary instruction set. If the verification fails, calculate the instruction correction amount and return to step S3 to regenerate the total adjustment power demand instruction. If the verification passes, output the safety instruction set. S10: Based on the adjustable status flag of the device and the system security indicators, perform functional interlocking and security interlocking judgment on the set of security instructions, eliminate unadjustable resource instructions, and generate valid issuance instructions; S11: Based on the effective issued instructions and the real-time operating status of the system, control instructions are issued to each station or virtual control unit using a time-event dual-trigger mechanism combined with a multi-cycle scheduling strategy.

2. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, The multi-level anomaly detection in step S1 includes: single-point data verification, multi-source data correlation verification, delay judgment, and comprehensive judgment of system operating status; The single-point data verification is used for dead data verification, threshold over-limit verification, and rate of change gradient anomaly verification. The multi-source data correlation verification is used for power balance residual verification and new energy output verification; The delay judgment and system operation status comprehensive judgment are used to mark the corresponding station as unadjustable when both real-time verification data and delayed verification data are abnormal, and the hard node status signal is combined to determine that AGC is abnormal, communication is abnormal, manual mode is abnormal or data is abnormal.

3. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, Based on the verified real-time data and introducing a dynamic frequency deviation coefficient, the area control deviation (ACE) is calculated using the following formula: ； in, Indicates base Real-time measured frequency; Indicates the system reference frequency; Indicates the dynamic frequency deviation coefficient; This indicates the real-time acquisition of measured power of the tie line; Indicates the planned power of the tie line; ； in, Indicates the reference frequency deviation coefficient; This represents the total active power output of all synchronous generators in the current system. This indicates the total power generation or total active load in the system at the current moment.

4. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, The total power to be allocated is obtained through intelligent dead-zone determination as follows: ； in, Indicates the dynamic threshold of the intelligent dead zone; Indicates a fixed technical dead zone; This represents the economic weighting coefficient; Indicates marginal cost; Indicates the baseline marginal cost; ； ； in, This indicates that resources are allocated to the system's controllable resources after passing through the intelligent dead zone; Indicates photovoltaic resources; Indicates wind power resources; Indicates solar thermal resources; Indicates hydropower resources; Indicates energy storage resources; It refers to thermal power resources.

5. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, The real-time adjustability margin of each station is calculated based on the verified real-time data using the following formula: ； ； ； ； ； ； in, For new energy power stations The upward adjustment margin is calculated based on the theoretical maximum generating power of the power station. Compared with the current actual output difference, For station The utility coefficient, For station The downward adjustment margin, For station Actual effort Minimum output; For thermal power or energy storage facilities The upward adjustment margin is the value of the maximum unit output. With current output The difference must not exceed the maximum climbing power. ; For thermal power or energy storage facilities Reduce margin; To aggregate controllable load margin for virtual power plants, To maximize the output of each piece of equipment, This is where the virtual power plant contributes its power; To reduce the margin, M represents the minimum output of each device, indicating the total number of adjustable resources or distributed units aggregated in the virtual power plant (VPP). The first virtual power plant aggregation Distributed resources.

6. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, The expression for the physical output boundary of the device is as follows: ； in, Indicates the minimum output of the equipment; Indicates the actual output of the equipment; This indicates the maximum output of the equipment.

7. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, The construction of the dynamic loss value incremental rate model includes: The secondary coal consumption cost function of thermal power units, the loss value function of energy storage system including cyclic aging and SOC offset penalty, the curtailment opportunity cost function of new energy power station, and the loss value function of virtual power plant aggregated load are established respectively. Based on the secondary coal consumption cost function of thermal power units, the loss value function of energy storage systems including cycle aging and SOC offset penalties, the curtailment opportunity cost function of new energy power plants, and the loss value function of virtual power plant aggregated load, a weighted average clustering method is used to aggregate the unit parameters of each power plant into a power plant-level cost function: ; ; ; ;in, This represents the total loss cost of the k-th energy storage station. This represents the aggregate cycle aging loss cost of the k-th energy storage station. This represents the aggregate SOC offset penalty cost for the k-th energy storage station. This represents the aggregate charge / discharge efficiency loss cost of the k-th energy storage station. This represents the aggregate equivalent health status of the k-th energy storage station. This represents the coefficient of the quadratic term in the cyclic aging cost function after the k-th energy storage station is aggregated. This represents the coefficient of the first-order term in the cyclic aging cost function after the k-th energy storage station is aggregated. Indicates the output of energy storage station k. Represents the coefficient of the constant term. This represents the charging and discharging efficiency of the station after k aggregation. This represents the equivalent state of charge after the aggregation of station k. This represents the reference state of charge, which is the baseline value that the energy storage battery is expected to maintain at its SOC. and This represents the quadratic and linear coefficients of the power term in the aggregated SOC offset penalty cost; This represents the penalty coefficient in the aggregated SOC offset penalty cost related to the degree of SOC deviation; Indicates the electricity price. The cost of deviation from the penalty is expressed in exponential form; The first derivative of the cost function at each station level with respect to power is obtained to obtain the incremental rate of loss value of the corresponding resource. The incremental rate of loss value is used to characterize the marginal economic cost of adjusting the unit active power of each resource at the current operating point.

8. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, Using the total power to be allocated as the allocation target, within the feasible region of the instructions, iterative power allocation is performed according to the dynamic loss value incremental rate model and the principle of equal loss value incremental rate to obtain the preliminary instruction set. The specific process is as follows: The initial allocation is carried out by ensuring that the rate of increase in the value of all resources actually involved in the adjustment is equal. Check whether each resource instruction exceeds the instruction feasible range. If a resource instruction exceeds the limit, fix the instruction at the limit boundary and remove it from the current adjustable list. Recalculate the remaining unmet adjustment needs and repeat the equal incremental rate allocation in the remaining adjustable resource set; Iterate until the total adjustment requirements are allocated or all available resources are exhausted, and output a preliminary set of instructions; The initial allocation, which ensures that the incremental rate of loss value of all resources actually involved in the adjustment is equal, is specifically implemented using the following formula: ； ； ； ； ； ； ； in, For station Injected active power, For load power, This is due to network loss in the system. For the slight increase in network loss; branch road Admittance; branch road Current, The branch resistance is denoted by ; the nodes at both ends of the branch are denoted by . node, They are nodes The voltage; Energy storage stations Total loss cost function, thermal power plant Total loss cost function, virtual power plant Total loss cost function, new energy power station The total loss cost function; This represents the incremental rate of the equivalent loss value of the k-th energy storage station after accounting for network losses; This represents the incremental rate of the equivalent loss value of the k-th thermal power plant after accounting for network losses; This represents the incremental rate of the equal loss value of the kth virtual power plant after accounting for network losses; This represents the incremental rate of the equivalent loss value of the k-th renewable energy power station after accounting for network losses; This represents the voltage phase angle difference between node i and node j; N represents the power output of station k; N represents the total number of nodes; and L represents the total number of branches.

9. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, The functional interlock is triggered when communication is abnormal, AGC mode is exited, manual mode is activated, or data is abnormal; the safety interlock is triggered when the system frequency fluctuates and faces the risk of collapse or when the cross-sectional power flow exceeds the equipment limit; the interlock is automatically released and the command issuance is resumed when the interlock conditions are not met.

10. The real-time control method for a source-load-storage system based on the infinitesimal increase rate of equal loss value according to claim 1, characterized in that, The time-event dual-trigger mechanism is as follows: when the absolute value of the area control deviation exceeds the safety threshold, it is determined to be an emergency event and an instruction is immediately triggered; when the area control deviation is within the intelligent dead zone, it is locked and no instruction is issued; otherwise, it is issued according to the multi-cycle reference cycle set by the response characteristics of each device.

11. A real-time control device for a source-load-storage system based on the infinitesimal rate of increase in value with equal loss, characterized in that, include: The anomaly detection and cleaning module is used to perform multi-level anomaly detection and cleaning on real-time operating data of multi-source heterogeneous systems from new energy sources, energy storage, thermal power, and controllable loads, and outputs adjustable status markers and verified real-time data. The area control deviation calculation module is used to calculate the area control deviation ACE based on the verified real-time data and by introducing a dynamic frequency deviation coefficient. The module for determining the total power to be allocated is used to perform low-pass filtering and proportional-integral control on the area control deviation ACE, generate a total regulation power demand command, and obtain the total power to be allocated after intelligent dead zone judgment. The real-time adjustable margin calculation module is used to calculate the real-time adjustable margin of each station based on the verified real-time data. The equipment physical output boundary construction module is used to construct the equipment physical output boundaries of each site; The instruction feasible region determination module is used to determine the instruction feasible region based on the real-time adjustability margin of each station and the physical output boundary of the equipment of each station; The Dynamic Loss Value Increment Rate Model Construction Module is used to construct a dynamic loss value incremental rate model. The preliminary instruction set determination module is used to take the total power to be allocated as the allocation target, and within the instruction feasible domain, iteratively allocate power according to the dynamic loss value incremental rate model and the principle of equal loss value incremental rate to obtain the preliminary instruction set. The correction module is used to perform multi-constraint safety boundary verification and real-time correction on the preliminary instruction set based on the adjustable status flag of the device and the verified real-time data. If the verification fails, the instruction correction amount is calculated and returned to the total power to be allocated determination module to regenerate the total adjustment power demand instruction. If the verification passes, the safety instruction set is output. The judgment module is used to perform functional interlocking and security interlocking judgments on the set of security instructions based on the adjustable status flags of the device and system security indicators, eliminate instructions for unadjustable resources, and generate valid instructions to be issued. The control command issuing module, based on the valid issued commands and the real-time operating status of the system, uses a time-event dual-trigger mechanism combined with a multi-cycle scheduling strategy to issue control commands to each station or virtual control unit.

Images