Smart energy storage system multi-target hierarchical scheduling method and system oriented to source network load storage cooperation

Abstract

The invention discloses an intelligent energy storage system multi-target hierarchical scheduling method and system oriented to source network load storage cooperation, and belongs to the technical field of energy storage system optimization control. The method comprises three levels of day-ahead layer multi-objective game optimization, intra-day layer rolling correction optimization and real-time layer adaptive droop control. The day-ahead layer establishes three objective functions of economy, environmental protection and smoothness, and solves and outputs a day-ahead charging and discharging power plan by using a Nash negotiation algorithm. And the intra-day layer obtains ultra-short-term prediction data of the source load, performs rolling correction on the day-ahead plan by adopting a model prediction control method, and outputs a corrected real-time power instruction. The real-time layer collects power grid frequency deviation and a battery health state value, calculates an adaptive droop coefficient according to the health state value, and superposes and outputs primary frequency modulation response power and a real-time power instruction. According to the invention, source network load storage collaborative optimization is realized through multi-time scale hierarchical scheduling, and the service life of an energy storage system is prolonged through adaptive droop control based on health state perception.

Description

Technical Field

[0001] This invention belongs to the field of energy storage system optimization and control technology, specifically involving a multi-objective hierarchical scheduling method and system for intelligent energy storage systems oriented towards source-grid-load-storage coordination. Background Technology

[0002] With the advancement of global energy structure transformation and green and low-carbon goals, the penetration rate of new energy sources, represented by photovoltaic power generation, in the power system is continuously increasing. However, new energy power generation is characterized by intermittency, volatility, and randomness, and its large-scale grid connection poses a severe challenge to the safe and stable operation of the power system. Energy storage systems, as a key link connecting power sources, the grid, and loads, play a crucial role in mitigating fluctuations in new energy output, improving the absorption of new energy, and participating in grid frequency regulation and peak shaving. How to achieve optimized scheduling of energy storage systems, balancing economic benefits and new energy absorption while ensuring grid safety and stability, has become a research hotspot in the energy and power sector.

[0003] Existing energy storage system scheduling methods suffer from the following technical problems. In terms of time scale, existing methods typically employ a single time scale for scheduling decisions, making it difficult to simultaneously meet the requirements of global optimization and real-time response. Simple day-ahead optimization cannot address the plan deviation caused by intraday source-load forecasting errors, while simple real-time control lacks a global perspective, making it difficult to achieve overall optimization. Regarding multi-objective optimization, existing methods typically use weighted summation to merge multiple objectives into a single objective for optimization. This requires manually pre-setting the weight coefficients for each objective; improper weight settings can lead to the excessive sacrifice of some objectives, making it difficult to achieve a reasonable balance between economy, environmental protection, and grid security.

[0004] Regarding forecast error handling, the accuracy of day-ahead forecasts is limited due to the influence of various factors such as weather and user behavior on photovoltaic power generation and load demand. A deviation between the day-ahead optimization plan and the actual operating status is inevitable. Existing methods lack effective error correction mechanisms during the intraday phase, leading to potential deviations in the actual operating status of the energy storage system from expectations, and even issues affecting system safety such as exceeding the state-of-charge limit. In terms of battery life protection, existing energy storage systems typically use a fixed droop coefficient when participating in primary grid frequency regulation. All battery banks respond to grid frequency changes with the same power-frequency characteristics, without considering the differences in the health status of each battery bank. This approach causes battery banks with poor health to bear the same power surges as healthy battery banks, accelerating the aging of weaker battery banks, shortening the overall lifespan of the energy storage system, and increasing the total lifespan cost of the energy storage system. Summary of the Invention

[0005] To address the problems existing in the background technology, this invention provides a multi-objective hierarchical scheduling method for smart energy storage systems oriented towards source-grid-load-storage coordination, comprising the following steps: S1. Day-ahead multi-objective game optimization: Collect photovoltaic power generation forecast data, load power forecast data and time-of-use electricity price data for the next 24 hours, establish economic objective function, environmental objective function and smoothing objective function, use Nash negotiation algorithm to solve the three objective functions together, and output the day-ahead charging and discharging power plan of the energy storage system. S2, Intra-day Rolling Correction Optimization: At the current operating moment, obtain the ultra-short-term source-load forecast data in the forecast time domain, construct a rolling optimization model with the goal of tracking the day-ahead charging and discharging power plan, set energy storage ramp rate constraints and state of charge constraints, solve the rolling optimization model and output the corrected real-time power command; S3. Real-time adaptive droop control: The energy storage converter receives the corrected real-time power command, collects the grid frequency deviation and the health status value of the energy storage battery pack, calculates the adaptive droop coefficient based on the health status value, and superimposes the primary frequency regulation response power with the corrected real-time power command to output the final power of the energy storage converter.

[0006] Further, step S1 includes the following sub-steps: S11. Data Acquisition: Collect the photovoltaic power generation forecast values ​​for each time period of the next 24 hours through the photovoltaic inverter data interface, collect the load power forecast values ​​for each time period of the next 24 hours through the smart meter, and obtain the time-of-use electricity price for each time period of the next 24 hours through the power grid dispatch system. S12. Objective Function Construction: Three objective functions are established based on the collected data: The economic objective function calculates the system operating cost using the following formula: ; In the formula, The economic objective function value (in yuan); To Summation operators from 1 to 24; For time period Time-of-use electricity price (RMB / kWh); For time period The exchange power (kW) with the power grid, with positive values ​​for purchasing electricity and negative values ​​for selling electricity; The duration is 1 (h). This is the depreciation cost factor for energy storage batteries (yuan / kWh). This is the absolute value operator; For time period The charging and discharging power (kW) of the energy storage system; The environmental protection objective function is used to calculate the renewable energy absorption rate. The calculation formula is as follows: ; In the formula, The objective function value is the environmental friendliness. To Summation operators from 1 to 24; For time period Actual photovoltaic power generation absorbed (kW); For time period Forecasted photovoltaic power generation (kW); The smoothness objective function is used to calculate the variance of tie-line power fluctuations. The calculation formula is as follows: ; In the formula, The objective function value for smoothness is (kW²). To Summation operators from 1 to 24; For time period Power exchanged with the power grid (kW); This represents the average daily switching power of the tie line (kW). S13. Nash Negotiation Solution: Optimize the economic objective function, environmental objective function, and smoothness objective function separately to obtain the ideal point value for each objective. , , Construct the Nash product objective function as follows: The sequential quadratic programming algorithm is used to solve for the maximum value of the Nash product objective function, and the day-ahead charging and discharging power plan of the energy storage system for each time period is output. .

[0007] Further, step S2 includes the following sub-steps: S21. Acquisition of ultra-short-term forecast data: at the current runtime Get from the current moment At that time Source load ultra-short-term forecast data, The number of steps to predict the time domain; S22. Rolling Optimization Model Construction: With the objective of minimizing the deviation from the day-ahead charging and discharging power plan, the objective function of the rolling optimization model is constructed as follows: ; In the formula, The minimum value operator; To optimize the objective function value (kW²) for rolling. To from to The summation operator; For a moment Energy storage charging and discharging power (kW) to be optimized; For a moment The planned daytime charge and discharge power (kW); This is the current running time; To predict the number of time-domain steps; S23. Constraint settings: Set the ramp rate constraint and the state of charge constraint; The expression for the gradeability constraint is: ; In the formula, This is the absolute value operator; For a moment Energy storage charging and discharging power (kW); For a moment Energy storage charging and discharging power (kW); The maximum ramp rate (kW / h) of the energy storage system. To control the time step length (h); The expression for the state-of-charge constraint is: ; In the formula, This is the upper limit value under charged state; For a moment The state of charge value; This represents the upper limit of the state of charge. S24. Solution and Output: The rolling optimization model is solved using a quadratic programming algorithm under the constraints of ramp rate and state of charge, yielding the optimal charging and discharging power sequence in the predicted time domain. The first component of this sequence is taken as the corrected real-time power command for the current moment. The data is sent to the energy storage converter for execution; at the next control moment, the initial state is updated with the new measured value, and steps S21 to S24 are repeated.

[0008] Furthermore, step S3 includes the following sub-steps: S31. Status Acquisition: The energy storage converter receives the corrected real-time power command. The frequency detection module collects the real-time frequency of the power grid and calculates the frequency deviation. The battery management module obtains the current health status value of the energy storage battery pack. and the average health status value of the energy storage battery pack ; S32. Calculation of Adaptive Sag Coefficient: The adaptive sag coefficient is calculated based on the health status value. The calculation formula is as follows: ; In the formula, For a moment The adaptive droop factor (kW / Hz); The baseline droop factor (kW / Hz) is used. As a regulator of health status; For a moment Health status values ​​of the energy storage battery pack; This represents the average health status value of the energy storage battery pack. For a moment Frequency deviation penalty factor; The formula for calculating the frequency deviation penalty factor is: ; In the formula, For a moment Frequency deviation penalty factor; This is the penalty intensity coefficient; This is the absolute value operator; For a moment The power grid frequency deviation (Hz); The permissible frequency deviation limit (Hz); S33. Power Superposition Output: The primary frequency regulation response power is superimposed with the corrected real-time power command to calculate the final power of the energy storage converter. The calculation formula is as follows: ; In the formula, For a moment Final power (kW) of the energy storage converter; For a moment Revised real-time power command (kW); For a moment The adaptive droop factor (kW / Hz); For a moment Grid frequency deviation (Hz); control the energy storage converter according to the final power To output power.

[0009] Furthermore, in step S31, the health status value Obtain it through the following steps: The battery management module collects the terminal voltage, charging / discharging current, and temperature of the energy storage battery pack in real time; it calculates the state of charge (SOC) of the battery pack by integrating the charging / discharging current over time and correcting it with the open-circuit voltage; and it calculates the health status value based on the ratio of the current maximum release capacity to the rated capacity of the energy storage battery pack. The calculation formula is as follows: ; In the formula, For a moment Health status value; For a moment Maximum release capacity of the energy storage battery pack (Ah); This refers to the rated capacity (Ah) of the energy storage battery pack.

[0010] This invention also provides a multi-objective hierarchical scheduling system for smart energy storage systems oriented towards source-grid-load-storage coordination, comprising: The source-load panoramic perception module communicates with photovoltaic inverters, smart meters and power grid dispatching systems to collect photovoltaic power generation prediction data, load power prediction data and time-of-use electricity price data. The cloud-based game scheduling platform communicates with the source-load panoramic perception module to receive photovoltaic power generation forecast data, load power forecast data, and time-of-use electricity price data. It constructs economic objective functions, environmental objective functions, and smoothing objective functions, and runs the Nash negotiation algorithm to output the day-ahead charging and discharging power plan. The edge-side collaborative controller communicates with the cloud-based game scheduling platform and the source-load panoramic perception module to receive day-ahead charging and discharging power plans and source-load ultra-short-term forecast data, and runs a rolling optimization model to output corrected real-time power commands. The intelligent converter execution unit is signal-connected to the edge-side co-controller and electrically connected to the power grid and energy storage battery pack. It is used to receive the corrected real-time power command, collect the power grid frequency deviation and the health status value of the energy storage battery pack, calculate the adaptive droop coefficient and perform power superposition, and output the final power of the energy storage converter.

[0011] In a preferred embodiment, the edge-side collaborative controller includes: The data receiving unit is connected to the cloud-based game scheduling platform and the source-load panoramic perception module to receive day-ahead charging and discharging power plans and source-load ultra-short-term forecast data. The MPC computing unit is connected to the data receiving unit and is used to build a rolling optimization model and set the ramp rate constraint and state of charge constraint. It solves the rolling optimization model to generate the corrected real-time power command. The instruction issuing unit is connected to the MPC calculation unit and the intelligent converter execution unit via signals, and is used to issue the corrected real-time power instruction to the intelligent converter execution unit.

[0012] In the preferred embodiment, the intelligent converter execution unit includes: The battery management module is connected to the energy storage battery pack and is used to collect the terminal voltage, charging and discharging current and temperature of the energy storage battery pack, calculate and output the health status value and average health status value of the energy storage battery pack. The frequency detection module is electrically connected to the power grid and is used to detect the power grid frequency and calculate the frequency deviation. The control calculation module is connected to the battery management module, frequency detection module and edge-side collaborative controller. It is used to receive the corrected real-time power command, health status value and frequency deviation, calculate the adaptive droop coefficient, and perform power superposition calculation to calculate the final power of the energy storage converter.

[0013] The power conversion module is signal-connected to the control calculation module and electrically connected to the energy storage battery pack and the power grid. It is used to control the energy storage converter to perform power conversion based on the final power output of the control calculation module.

[0014] In the preferred embodiment, a two-way communication connection is formed between the cloud-based game scheduling platform and the edge-side collaborative controller. The cloud-based game scheduling platform sends the day-ahead charging and discharging power plan to the edge-side collaborative controller, and the edge-side collaborative controller uploads the energy storage system operation status data to the cloud-based game scheduling platform. A bidirectional signal connection is formed between the edge-side collaborative controller and the intelligent converter execution unit. The edge-side collaborative controller sends a corrected real-time power command to the intelligent converter execution unit, and the intelligent converter execution unit feeds back the operating status of the energy storage converter to the edge-side collaborative controller.

[0015] In the preferred embodiment, the control calculation module is integrated into the DSP chip inside the energy storage converter. The control calculation module is signal-connected to the IGBT drive circuit of the power conversion module. The control calculation module generates a PWM control signal based on the final power and outputs it to the IGBT drive circuit of the power conversion module.

[0016] The beneficial effects achieved by this invention are as follows: This invention employs a three-layer hierarchical scheduling architecture combining day-ahead multi-objective game optimization, intra-day rolling correction optimization, and real-time adaptive droop control. It decomposes the energy storage system scheduling problem into sub-problems at different time scales, handling them separately. The layers coordinate with each other through power command transmission. The day-ahead layer performs global optimization at an hourly time granularity, the intra-day layer performs rolling correction at a minute-level time granularity, and the real-time layer provides rapid response at a millisecond-level time granularity. These three layers form a progressive relationship in terms of time scale, with the upper layer providing a reference benchmark for the lower layer, and the lower layer finely adjusting the commands from the upper layer. This multi-time-scale coordinated hierarchical architecture enables the energy storage system to achieve both global optimization and rapid response to source-load uncertainty and grid frequency fluctuations. It effectively resolves the contradiction between source-load uncertainty, grid stability, and energy storage lifetime in scenarios with high renewable energy penetration, achieving deep integration and collaborative optimization of multiple entities including source, grid, load, and storage.

[0017] This invention employs the Nash negotiation algorithm at the day-ahead layer to jointly solve the economic, environmental, and smoothness objective functions. It abstracts the interests of four stakeholders—source, grid, load, and storage—into three optimization objectives, and seeks a Pareto optimal solution that balances the interests of all parties by constructing a Nash product objective function. Compared to traditional weighted summation methods, the Nash negotiation algorithm does not require pre-setting fixed weights; instead, it automatically adjusts priorities based on the degree to which each objective deviates from its ideal point, better reflecting the urgency of different stakeholders at different times. The Nash product form ensures fair treatment of each objective during the optimization process, avoiding the excessive sacrifice of any one objective. This achieves a balance between economy, environmental protection, and smoothness in the optimization result, reducing system operating costs and battery depreciation costs, improving the absorption of new energy sources, and minimizing the impact of tie-line power fluctuations on the upstream grid.

[0018] This invention employs a rolling optimization method based on model predictive control (MMC) to perform real-time corrections to the day-ahead charge and discharge power plan at the intraday level. It continuously tracks the actual operating status and dynamically adjusts power commands by combining rolling prediction and feedback correction. Due to the strong randomness and volatility of photovoltaic power generation and load demand, a deviation inevitably exists between the day-ahead predicted values ​​and the actual values. The MMC method solves the optimal control problem within a finite time domain at each control moment based on the current state and ultra-short-term prediction information, and updates the initial state with new measurements for rolling optimization. This allows the system to promptly detect deviations between the actual and expected operating states and correct them through re-optimization. This rolling optimization mechanism gives the system strong adaptability and robustness, effectively eliminating the impact of source-load prediction errors on the day-ahead plan execution, avoiding problems such as the energy storage system deviating from the expected state or exceeding the state-of-charge limit, and ensuring that the energy storage system always operates within a safe and reasonable range.

[0019] This invention employs an adaptive droop control strategy based on battery health status awareness at the real-time layer. The droop coefficient is adjusted in real-time according to the health status of each battery pack, allowing battery packs with higher health status to undertake more frequency regulation tasks, while those with lower health status reduce their output. Traditional energy storage systems typically use a fixed droop coefficient when participating in primary grid frequency regulation, without considering the differences in the health status of each battery pack. This results in battery packs with poor health status bearing the same power surge as those with better health status, accelerating the aging of weaker battery packs. This invention introduces a health status adjustment factor to dynamically adjust the droop coefficient, achieving differentiated output allocation, effectively protecting weaker battery packs, and extending the lifespan of the entire energy storage system. Simultaneously, this invention introduces a frequency deviation penalty factor, automatically increasing the droop coefficient when the grid frequency deteriorates severely, ensuring that the energy storage system provides sufficient frequency regulation power to support grid frequency recovery. This achieves a dynamic balance between battery life protection and grid safety support, prioritizing the protection of weaker battery packs under normal operating conditions and prioritizing grid safety under emergency conditions. Attached Figure Description

[0020] Figure 1 This is a diagram of the multi-objective hierarchical scheduling system architecture of the intelligent energy storage system oriented towards source-grid-load-storage coordination according to the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figure 1 This invention provides a multi-objective hierarchical scheduling system for intelligent energy storage systems oriented towards source-grid-load-storage coordination. It comprises four main components: a source-load panoramic perception module, a cloud-based game-theoretic scheduling platform, an edge-side collaborative controller, and an intelligent converter execution unit. These four components are spatially distributed in different locations, forming an organic whole through communication networks and signal cables, collaboratively completing hierarchical scheduling tasks.

[0023] The source-load panoramic sensing module is the data acquisition front-end of the system, communicating with photovoltaic inverters, smart meters, and the power grid dispatching system. It collects photovoltaic power generation forecast data, load power forecast data, and time-of-use electricity price data through standard communication protocols, and transmits the collected data to the cloud-based game scheduling platform and the edge-side collaborative controller. The source-load panoramic sensing module plays a crucial role in information aggregation throughout the system, providing data support for upper-level optimization decisions. Communication between the source-load panoramic sensing module and the photovoltaic inverter preferably uses the Modbus or IEC61850 protocol; communication with the smart meter preferably uses the DL / T645 or Modbus protocol; and communication with the power grid dispatching system preferably uses the IEC104 protocol.

[0024] The cloud-based game theory scheduling platform serves as the system's day-ahead optimization decision center, communicating with the source-load panoramic perception module. Deployed on a cloud server, the platform boasts robust computing and storage resources, making it suitable for running computationally intensive optimization algorithms. It receives photovoltaic power generation forecast data, load power forecast data, and time-of-use electricity price data transmitted from the source-load panoramic perception module. Based on the collected data, it constructs economic, environmental, and smoothing objective functions, runs the Nash negotiation algorithm to solve the multi-objective optimization problem, and outputs a day-ahead charge-discharge power plan. This plan is then distributed to the edge-side collaborative controller for execution via the communication network. The cloud-based game theory scheduling platform also includes auxiliary functions such as historical data storage, operational status monitoring, and report generation.

[0025] The edge-side collaborative controller serves as the system's intraday optimization decision-making and command relay center, communicating with the cloud-based game scheduling platform and the source-load panoramic perception module. Deployed at the energy storage power station site, the edge-side collaborative controller is close to the energy storage equipment, resulting in low communication latency, making it suitable for running control algorithms with high real-time requirements. The edge-side collaborative controller receives the day-ahead charge / discharge power plan from the cloud-based game scheduling platform and the ultra-short-term source-load forecast data transmitted from the source-load panoramic perception module. It runs a rolling optimization model based on model predictive control to make real-time corrections to the day-ahead plan and outputs the corrected real-time power command. The corrected real-time power command is then transmitted to the intelligent converter execution unit via signal cables for execution. The edge-side collaborative controller also uploads the energy storage system's operating status data to the cloud-based game scheduling platform, forming a closed-loop information flow.

[0026] The edge-side collaborative controller includes a data receiving unit, an MPC calculation unit, and an instruction issuing unit. The data receiving unit communicates with the cloud-based game scheduling platform and the source-load panoramic perception module, receiving day-ahead charge / discharge power plans and ultra-short-term source-load forecast data, and parsing and preprocessing the received data. The MPC calculation unit is signal-connected to the data receiving unit, constructing a rolling optimization model and setting ramp rate and state-of-charge constraints. It uses a quadratic programming algorithm to solve the rolling optimization model and generate corrected real-time power instructions. The MPC calculation unit is preferably implemented using an industrial-grade embedded computer or a programmable logic controller (PLC), possessing strong real-time computing and anti-interference capabilities. The instruction issuing unit is signal-connected to the MPC calculation unit and the intelligent converter execution unit, encapsulating the corrected real-time power instructions according to a specified communication protocol and issuing them to the intelligent converter execution unit. Communication between the instruction issuing unit and the intelligent converter execution unit preferably uses a CAN bus or industrial Ethernet.

[0027] The intelligent converter execution unit (ICU) is the system's real-time control and power conversion execution terminal. It is signal-connected to the edge-side co-controller and electrically connected to the power grid and energy storage battery pack. Integrated within the energy storage converter, the ICU receives corrected real-time power commands from the edge-side co-controller, collects grid frequency deviation and energy storage battery pack health status values, calculates the adaptive droop coefficient, performs power superposition, and controls the energy storage converter to output the final power. The ICU upgrades the energy storage converter from a traditional passive execution device to an intelligent device with autonomous decision-making capabilities, ensuring the execution of upper-level commands while possessing rapid local response capabilities.

[0028] The intelligent converter execution unit includes a battery management module, a frequency detection module, a control calculation module, and a power conversion module. The battery management module connects to the energy storage battery pack and collects the battery pack's terminal voltage, charging / discharging current, and temperature. Based on the collected data, it calculates and outputs the energy storage battery pack's health status value and the average health status value of the battery pack. By monitoring battery operating parameters and estimating battery status parameters, the battery management module provides battery information support for upper-level control strategies and also has battery protection functions, promptly cutting off the charging / discharging circuit when battery abnormalities occur. The frequency detection module is electrically connected to the power grid and detects the grid frequency and calculates the frequency deviation. The frequency detection module uses a high-precision frequency measurement circuit to acquire accurate grid frequency information within milliseconds, providing frequency feedback for primary frequency regulation control. The control calculation module connects to the battery management module, the frequency detection module, and the edge-side collaborative controller. It receives corrected real-time power commands, health status values, and frequency deviations, calculates the adaptive droop coefficient based on the adaptive droop control strategy, performs power superposition to calculate the final power of the energy storage converter, and generates control signals for the power conversion module. The control and calculation module is integrated into the DSP chip inside the energy storage converter, possessing high-speed computing and precise control capabilities. The control and calculation module is signal-connected to the IGBT drive circuit of the power conversion module, generating PWM control signals based on the final power and outputting them to the IGBT drive circuit of the power conversion module. The power conversion module is signal-connected to the control and calculation module, and electrically connected to the energy storage battery pack and the power grid. It is used to control the on / off switching of the IGBT switching devices according to the control signals output by the control and calculation module, realizing bidirectional power conversion between the energy storage system and the power grid. The power conversion module includes a DC-DC converter and a DC-AC converter. The DC-DC converter is used to regulate the DC bus voltage, and the DC-AC converter is used to realize the mutual conversion between DC and AC power.

[0029] A bidirectional communication connection is established between the cloud-based scheduling platform and the edge-side co-controller. The cloud-based scheduling platform sends daily charging and discharging power plans to the edge-side co-controller, while the edge-side co-controller uploads energy storage system operating status data to the cloud-based scheduling platform. This bidirectional communication connection enables the cloud platform to monitor the energy storage system's operating status in real time and make necessary adjustments to the daily plans based on actual operating conditions. The communication between the cloud-based scheduling platform and the edge-side co-controller preferably uses 4G / 5G wireless communication or fiber optic wired communication to ensure communication reliability and bandwidth.

[0030] A bidirectional signal connection is established between the edge-side collaborative controller and the intelligent converter execution unit. The edge-side collaborative controller sends corrected real-time power commands to the intelligent converter execution unit, while the intelligent converter execution unit feeds back the operating status of the energy storage converter to the edge-side collaborative controller. The feedback operating status includes information such as actual output power, DC bus voltage, AC side voltage and current, equipment temperature, and fault alarms. This bidirectional signal connection forms a closed-loop control between the intraday layer and the real-time layer, ensuring accurate execution of power commands and timely response to abnormal situations. The signal connection between the edge-side collaborative controller and the intelligent converter execution unit preferably uses a CAN bus or industrial Ethernet to ensure the real-time performance and reliability of signal transmission.

[0031] This invention presents a multi-objective hierarchical scheduling method and system for intelligent energy storage systems oriented towards source-grid-load-storage coordination. Through three layers of coordinated operation—day-ahead multi-objective game optimization, intraday rolling correction optimization, and real-time adaptive droop control—it achieves optimized control of the energy storage system at different time scales. The day-ahead layer employs the Nash negotiation algorithm to solve the multi-objective optimization problem, balancing economic, environmental, and smoothness considerations. The intraday layer uses model predictive control to perform rolling corrections on the day-ahead plan, effectively eliminating the impact of source-load prediction errors. The real-time layer adopts an adaptive droop control strategy based on health status awareness, protecting weak battery banks while ensuring grid frequency support and extending the lifespan of the energy storage system. The system adopts a cloud-edge-device collaborative architecture, with the cloud responsible for global optimization, the edge responsible for real-time correction, and the terminal responsible for rapid response. The clear division of labor and coordinated cooperation among each layer achieve deep integration and collaborative optimization of source, grid, load, and storage.

[0032] This invention also provides a multi-objective hierarchical scheduling method and system for intelligent energy storage systems oriented towards source-grid-load-storage coordination. It adopts a three-layer architecture of day-ahead global optimization, intraday rolling correction, and real-time adaptive control. Through multi-timescale coordination, it resolves the contradiction between source-load uncertainty, grid stability, and energy storage lifespan under high new energy penetration.

[0033] This invention decomposes the energy storage scheduling problem into sub-problems at different time scales and processes them separately. At the day-ahead layer, a multi-objective game theory optimization method is employed, comprehensively considering three objectives: economy, environmental friendliness, and smoothness. The Nash negotiation algorithm is used to solve for the Pareto optimal solution, generating a charging and discharging power plan for the next 24 hours. At the intra-day layer, a model predictive control method is used to continuously revise the day-ahead plan in real time through rolling optimization, eliminating the impact of source-load prediction errors. At the real-time layer, an adaptive droop control strategy based on battery health status perception is adopted. The droop coefficient is dynamically adjusted according to the health status of each battery pack, allowing batteries with higher health status to undertake more frequency regulation tasks, thereby extending the overall battery life and ensuring grid frequency stability.

[0034] The multi-objective hierarchical scheduling method for a smart energy storage system oriented towards source-grid-load-storage coordination, as described in this invention, includes steps S1 (day-ahead multi-objective game optimization), S2 (intra-day rolling correction optimization), and S3 (real-time adaptive droop control). These three steps are progressive in time scale: the day-ahead layer performs global optimization with a 1-hour time granularity; the intra-day layer performs rolling correction with a minute-level time granularity; and the real-time layer performs rapid response with a millisecond-level time granularity. The layers coordinate with each other through power command transmission, with the upper layer providing a reference benchmark for the lower layer, and the lower layer finely adjusting the commands from the upper layer.

[0035] Step S1, the multi-objective game optimization at the day-ahead stage, aims to generate a charging and discharging power plan for the energy storage system for the next 24 hours based on forecast information. This step abstracts the interests of the four stakeholders—source, grid, load, and storage—into three optimization objectives. The Nash negotiation algorithm is then used to solve the multi-objective optimization problem, yielding a Pareto optimal solution that balances the interests of all parties. Compared to traditional weighted summation methods, the Nash negotiation algorithm does not require pre-setting fixed weights; instead, it automatically adjusts priorities based on the degree to which each objective deviates from its ideal point, better reflecting the urgency of different stakeholders at different times.

[0036] Step S1 includes sub-step S11 data acquisition, sub-step S12 objective function construction, and sub-step S13 Nash negotiation solution.

[0037] Sub-step S11, data acquisition, is the data preparation stage for day-ahead optimization. Predicted photovoltaic (PV) power generation for each time period of the next 24 hours is collected via the PV inverter data interface. PV power generation forecasting typically employs a combination of numerical weather prediction and historical power generation data; the accuracy of the forecast directly impacts the feasibility of the day-ahead plan. Predicted load power for each time period of the next 24 hours is collected via smart meters. Load forecasting usually considers historical load curves, meteorological factors, and holiday factors. Time-of-use (TOU) electricity prices for each time period of the next 24 hours are obtained through the grid dispatch system. TOU reflects the supply and demand situation of the grid at different times and is an important basis for peak-valley arbitrage in energy storage systems. The three types of data collected constitute the input information set for day-ahead optimization.

[0038] Sub-step S12, constructing the objective function, is the process of mathematizing the interests of multiple parties. This invention establishes three objective functions: an economic objective function, an environmental objective function, and a smoothness objective function, which respectively reflect the economic interests of users, the consumption needs of renewable energy generators, and the safety and stability requirements of the power grid.

[0039] The economic objective function is used to calculate the system operating cost, and its calculation formula is as follows: ; In the formula, This represents the economic objective function value, and its unit is yuan. To Summation operators from 1 to 24; For time period The time-of-use electricity price is expressed in yuan / kWh. For time period The power exchanged with the power grid, measured in kW. A positive value indicates that electricity is purchased from the power grid, while a negative value indicates that electricity is sold to the power grid. This represents the duration of the time period, with a value of 1, and its unit is hours (h). This is the depreciation cost coefficient for energy storage batteries, with the unit being yuan / kWh. This coefficient reflects the depreciation loss corresponding to each kWh of charge / discharge capacity of the energy storage battery. This is the absolute value operator; For time period The charging and discharging power of an energy storage system is measured in kW. The optimization objective of the economic objective function is to make... Minimization means reducing the cost of electricity purchase and battery depreciation as much as possible while meeting electricity demand.

[0040] The environmental protection objective function is used to calculate the renewable energy absorption rate, and its calculation formula is as follows: ; In the formula, The objective function value for environmental protection is a dimensionless ratio value, ranging from 0 to 1. To Summation operators from 1 to 24; For time period The actual photovoltaic power generation absorbed is measured in kW. For time period The predicted photovoltaic power generation is expressed in kW. The optimization objective of the environmental friendliness objective function is to make... Maximizing the utilization of photovoltaic power generation means absorbing as much solar power as possible and reducing curtailment. Energy storage systems charge and store excess energy during peak photovoltaic power generation periods and release energy during peak load periods, thereby improving the utilization rate of new energy sources.

[0041] The smoothness objective function is used to calculate the variance of tie-line power fluctuations, and its calculation formula is as follows: ; In the formula, The value of the smoothness objective function is expressed in kW². To Summation operators from 1 to 24; For time period The power exchanged with the power grid, measured in kW; This represents the average daily switching power of the tie line, measured in kW. The optimization objective of the smoothness objective function is to make... Minimizing power fluctuations in tie lines is crucial. Excessive power fluctuations can impact the upstream power grid, affecting its safe and stable operation. Regulating power flow through energy storage systems can effectively mitigate power fluctuations and improve grid friendliness.

[0042] Sub-step S13, Nash negotiation, is the process of transforming a multi-objective optimization problem into a single-objective optimization problem and solving it. The Nash negotiation solution originates from game theory, and its core idea is to find a fair solution acceptable to all parties involved. In this invention, the three objectives of economy, environmental protection, and smoothness are considered as three game participants, and the Nash negotiation mechanism is used to find an equilibrium solution that takes into account the interests of all parties.

[0043] First, the economic objective function, the environmental objective function, and the smoothness objective function are optimized separately to obtain the ideal point value for each objective. , , The ideal point refers to the optimal value that can be achieved when each objective is optimized individually, representing the theoretical best state of each objective. Since there are interdependent relationships among the three objectives, it is impossible to reach all ideal points simultaneously in actual operation; therefore, trade-offs need to be made among the objectives.

[0044] Then, the Nash product objective function is constructed. The Nash product form ensures that all objectives are treated fairly during the optimization process, preventing any one objective from being excessively sacrificed. The expression for the Nash product objective function is as follows: For economic and smoothness objectives, since their optimization direction is minimization, the objective value is expressed as the ideal value minus the current value. For environmental objectives, since their optimization direction is maximization, the objective value is expressed as the current value minus the ideal value. The larger the Nash product objective function value, the more balanced the deviation of each objective from its ideal point.

[0045] Finally, a sequential quadratic programming algorithm is used to find the maximum value of the Nash product objective function. Sequential quadratic programming is an effective method for solving nonlinear constrained optimization problems. This algorithm decomposes the original problem into a series of quadratic programming subproblems for iterative solving, offering advantages such as fast convergence speed and high solution accuracy. After solving, the day-ahead charging and discharging power plan of the energy storage system for each time period is output. This plan serves as a reference benchmark for intraday rolling optimization.

[0046] The purpose of step S2, intraday rolling correction optimization, is to perform real-time corrections to the day-ahead charge and discharge power plan during the intraday operation phase, eliminating the impact of source-load forecast errors on plan execution. Due to the strong randomness and volatility of photovoltaic power generation and load demand, deviations between day-ahead forecasts and actual values ​​are inevitable. Strict adherence to the day-ahead plan may lead to deviations in the energy storage system's operation from the expected state, or even problems such as exceeding the state-of-charge limit. The model predictive control method, through a combination of rolling prediction and feedback correction, continuously tracks the actual operating status and dynamically adjusts the power command to ensure that the energy storage system always operates within a safe and reasonable range.

[0047] Model predictive control (MMC) is an advanced model-based control strategy. Its core idea is to solve an optimal control problem within a finite time domain at each control time step based on the current state and future predictions. The first component of the optimal control sequence is then used as the current control variable, and this process is repeated at the next control time step. This rolling optimization mechanism gives the system strong adaptability and robustness, effectively addressing model uncertainties and external disturbances.

[0048] Step S2 includes sub-step S21 acquiring ultra-short-term forecast data, sub-step S22 constructing a rolling optimization model, sub-step S23 setting constraints, and sub-step S24 solving and outputting the solution.

[0049] Sub-step S21, ultra-short-term forecast data acquisition, is the data preparation stage for intraday rolling optimization. At the current runtime... Get from the current moment At that time Source load ultra-short-term forecast data, The number of steps in the prediction time domain. Ultra-short-term (USST) forecasting typically employs methods such as time series analysis and neural networks to predict power changes over a short period based on recently measured data. Compared to day-ahead forecasting, USST forecasting has a shorter time span, typically 15 minutes to 1 hour, but offers higher accuracy and better reflects the real-time trends of source load changes. Prediction time domain. The selection of [a specific method / mechanism] requires a trade-off between prediction accuracy and computational burden. Too small a size will result in insufficient field of view for optimal optimization. Setting the time domain too high will increase computational load and reduce prediction accuracy. Based on engineering experience, the preferred prediction time domain is 15 to 30 minutes.

[0050] Sub-step S22, "Rolling Optimization Model Construction," is the process of establishing an intraday optimization mathematical model. The rolling optimization model aims to minimize the deviation from the day-ahead charging and discharging power plan, ensuring that the intraday revised power command is as close as possible to the day-ahead plan, maintaining the continuity and consistency of scheduling. The objective function of the rolling optimization model is: ; In the formula, The minimum value operator; The objective function value is used for rolling optimization, and its unit is kW². To from to The summation operator; For a moment The energy storage charging and discharging power to be optimized is expressed in kW, and this variable is the optimization decision variable. For a moment The day-ahead charge and discharge power plan, in kW, is given by the day-ahead optimization results; This is the current running time; To predict the number of time-domain steps, the objective function is in quadratic form, which imposes a greater penalty on steps with larger deviations, thus improving the overall tracking performance.

[0051] Setting constraints in sub-step S23 is a crucial step to ensure that the optimization results meet the physical limitations of the energy storage system. In actual operation, energy storage systems are subject to various physical constraints, primarily including ramp rate constraints and state of charge constraints.

[0052] The ramp rate constraint is used to limit the rate of power change in an energy storage system, preventing sudden power surges from impacting batteries and power electronic devices. The expression for the ramp rate constraint is: ; In the formula, This is the absolute value operator; For a moment The energy storage charging and discharging power, its unit is kW; For a moment The energy storage charging and discharging power, its unit is kW; The maximum ramp rate of the energy storage system is expressed in kW / h. This parameter is determined by the hardware characteristics of the energy storage system. The time step length is controlled by a unit of hours (h). This constraint ensures that the power change between two adjacent control moments does not exceed the maximum allowable rate of change, thus achieving a smooth power transition.

[0053] State of charge (SCC) constraints are used to limit the SCC of energy storage batteries to operate within a safe range, preventing damage from overcharging or over-discharging. The expression for SCC constraints is: ; In the formula, This is the limit value under charged state, and it is a dimensionless ratio. For a moment The state of charge value, which is a dimensionless ratio, ranging from 0 to 1; This is the upper limit of the state of charge (SOC), a dimensionless ratio. The upper and lower limits of SOC are typically set based on the battery type and operating strategy. For lithium-ion batteries, the upper limit is preferably set to 0.9 to 0.95, and the lower limit is preferably set to 0.1 to 0.2, to ensure the battery operates within its optimal operating range and extend its lifespan. State of Charge With charge and discharge power The two are related by the law of conservation of energy. When solving optimization problems, the power variable needs to be converted into a charged state variable for constraint verification.

[0054] Sub-step S24, solving and outputting, is the process of performing optimization and outputting control commands. A quadratic programming algorithm is used to solve the rolling optimization model under the constraints of ramp rate and state of charge. Quadratic programming refers to optimization problems where the objective function is a quadratic function and the constraints are linear functions. This type of problem has mature solution algorithms and fast computation speed, making it suitable for online real-time solutions. After solving, the optimal charging and discharging power sequence in the predicted time domain is obtained. The first component of this sequence is taken as the corrected real-time power command for the current moment. The data is then sent to the energy storage converter for execution. At the next control moment, the initial state is updated with the new measured value, and steps S21 to S24 are repeated to form a closed-loop control with rolling optimization and feedback correction. This feedback correction mechanism enables the system to promptly detect deviations between the actual operating state and the expected state, and corrects them through re-optimization, ensuring that the energy storage system is always in the optimal operating state.

[0055] The purpose of step S3, real-time adaptive droop control, is to achieve rapid response of the energy storage system to grid frequency fluctuations on a millisecond timescale. Simultaneously, it dynamically adjusts the output distribution based on the health status of each battery pack, protecting battery packs with poor health and extending the overall battery life. Traditional energy storage systems typically use a fixed droop coefficient when participating in primary grid frequency regulation, meaning all battery packs respond to grid frequency changes with the same power-frequency characteristics. This approach does not consider the differences in the health status of each battery pack, causing battery packs with poor health to bear the same power surge as those with better health, accelerating the aging of weaker battery packs and ultimately shortening the lifespan of the entire energy storage system. The adaptive droop control strategy based on health status awareness proposed in this invention adjusts the droop coefficient in real time according to the health status of each battery pack, allowing battery packs with higher health to undertake more frequency regulation tasks, while battery packs with lower health reduce their output, achieving a dynamic balance between battery life protection and grid frequency support.

[0056] Droop control is a control strategy that simulates the power-frequency characteristics of a traditional synchronous generator. Its basic principle is to increase active power output when the grid frequency decreases and decrease active power output when the grid frequency increases, thereby achieving automatic adjustment of the grid frequency. The droop coefficient determines the proportional relationship between the frequency change and the power change; the larger the droop coefficient, the greater the power response to frequency changes.

[0057] Step S3 includes sub-step S31 state acquisition, sub-step S32 adaptive droop coefficient calculation, and sub-step S33 power superposition output.

[0058] Sub-step S31, status acquisition, is the data preparation stage for real-time layer control. The energy storage converter receives the corrected real-time power command. This instruction originates from the intraday rolling optimization results of the edge-side co-controller. The frequency detection module acquires the real-time grid frequency and calculates the frequency deviation. Frequency detection typically employs methods such as phase-locked loops or zero-crossing detection, enabling accurate acquisition of grid frequency information within milliseconds. Grid frequency deviation. Defined as the difference between the measured frequency and the rated frequency, when a sudden increase in load leads to insufficient power supply, the grid frequency will decrease, and the frequency deviation will be negative; when a sudden decrease in load leads to excess power supply, the grid frequency will increase, and the frequency deviation will be positive. The current health status value of the energy storage battery pack is obtained through the battery management module. and the average health status value of the energy storage battery pack The battery management module monitors parameters such as voltage, current, and temperature of each individual battery cell in real time and estimates the health status of each battery pack using a built-in algorithm.

[0059] Health status value The acquisition process is as follows: The battery management module collects the terminal voltage, charging / discharging current, and temperature of the energy storage battery pack in real time. The state of charge (SOC) of the energy storage battery pack is calculated by integrating the charging / discharging current over time and correcting it using the open-circuit voltage. The open-circuit voltage correction is to eliminate the accumulated error of the coulomb integration method. When the battery is in a resting state, the estimated SOC is calibrated based on the correspondence between the open-circuit voltage and the SOC. The health status value is calculated based on the ratio of the current maximum releaseable capacity to the rated capacity of the energy storage battery pack. The calculation formula is: ; In the formula, For a moment The health status value is a dimensionless ratio value that ranges from 0 to 1. The larger the value, the better the battery health status. For a moment The maximum release capacity of an energy storage battery pack is measured in Ah, and this value gradually decreases as the battery ages. This refers to the rated capacity of the energy storage battery pack, measured in Ah. This value represents the battery's nominal capacity at the time of manufacture. The state of health (SHS) is an important indicator of battery aging. New batteries have an SHS value close to 1, which gradually decreases with increasing charge-discharge cycles and calendar aging. When the SHS value drops below 0.8, the battery is generally considered to be in the late stages of aging and needs to be replaced.

[0060] Sub-step S32, the calculation of the adaptive droop coefficient, is one of the core innovations of this invention. The adaptive droop coefficient is calculated based on the battery health status value, allowing battery packs in different health states to undertake different frequency regulation tasks. The formula for calculating the adaptive droop coefficient is: ; In the formula, For a moment The adaptive droop coefficient, with units of kW / Hz, determines the strength of the energy storage system's response to changes in grid frequency. The baseline droop factor is measured in kW / Hz and is set based on the rated power of the energy storage system and the primary frequency regulation requirements of the power grid. This is a health status moderating factor, a dimensionless coefficient used to moderate the influence of differences in health status on the droop coefficient. The higher the value, the greater the impact of differences in health status on the droop coefficient. The preferred value range is 1 to 2; For a moment Health status values ​​of the energy storage battery pack; This is the average state of health value of the energy storage battery pack, which is obtained by taking the arithmetic mean of the state of health values ​​of all battery packs. For a moment Frequency deviation penalty factor.

[0061] As can be seen from the above formula, when the health status value of a certain battery pack... Above average When, its adaptive droop coefficient Greater than the benchmark value This battery pack will undertake more frequency regulation tasks; when the health status value of a certain battery pack... Below average When, its adaptive droop coefficient Less than the benchmark value This battery pack will reduce frequency regulation output. This differentiated output allocation strategy effectively protects the weaker battery packs and extends the lifespan of the entire energy storage system.

[0062] Frequency deviation penalty factor This is used to increase the droop factor when the grid frequency deteriorates significantly, ensuring that the energy storage system can provide sufficient frequency regulation power support. The formula for calculating the frequency deviation penalty factor is: ; In the formula, For a moment The frequency deviation penalty factor is a dimensionless coefficient ranging from 1 to... ; This is the penalty intensity coefficient, a dimensionless coefficient used to adjust the degree of influence of frequency deviation on the penalty factor. The preferred value range is 0.5 to 2; This is the absolute value operator; For a moment The power grid frequency deviation, in Hz; The allowable frequency deviation limit is expressed in Hz. This value is set according to the power grid operation regulations. In my country, the frequency deviation limit for the power grid is usually between 0.2Hz and 0.5Hz.

[0063] As can be seen from the above formula, when the frequency deviation is small, the penalty factor... When the coefficient is close to 1, the adaptive droop coefficient is mainly determined by the health status; when the frequency deviation is large, the penalty factor... As the droop factor increases, the energy storage system provides greater frequency regulation power to support grid frequency recovery. This design achieves a dynamic balance between battery life protection and grid safety support, prioritizing the protection of weaker battery banks under normal operating conditions and ensuring grid safety under emergency conditions.

[0064] Sub-step S33, power superposition output, is the process of superimposing the primary frequency regulation response power with the corrected real-time power command and outputting the result. The final power of the energy storage converter is calculated using the following formula: ; In the formula, For a moment The final power of the energy storage converter is measured in kW, with positive values ​​indicating discharge and negative values ​​indicating charging. For a moment The revised real-time power command, in kW, is given by intraday rolling optimization. For a moment The adaptive droop coefficient, with units of kW / Hz; For a moment The grid frequency deviation is expressed in Hz. The negative sign in the formula reflects the droop control characteristic; when the grid frequency decreases... When it is negative, When the value is positive, the energy storage system increases its discharge power to support the power grid; when the power grid frequency increases... When it is a positive value, When the value is negative, the energy storage system reduces its discharge power or increases its charging power to absorb excess electrical energy.

[0065] Control the energy storage converter according to the final power Power output is then achieved. After receiving the power command, the energy storage converter precisely controls the power by adjusting the switching devices of the power conversion module. The power control loop typically employs a dual closed-loop control structure with an inner current loop and an outer power loop, characterized by fast response speed and high steady-state accuracy. This completes all operations for one control cycle, and the system enters the next control cycle to continue executing steps S31 to S33, forming continuous real-time frequency modulation control.

[0066] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-objective hierarchical scheduling method for smart energy storage systems oriented towards source-grid-load-storage coordination, characterized in that, Includes the following steps: S1. Day-ahead multi-objective game optimization: Collect photovoltaic power generation forecast data, load power forecast data and time-of-use electricity price data for the next 24 hours, establish economic objective function, environmental objective function and smoothing objective function, use Nash negotiation algorithm to solve the three objective functions together, and output the day-ahead charging and discharging power plan of the energy storage system. S2, Intra-day Rolling Correction Optimization: At the current operating moment, obtain the ultra-short-term source-load forecast data in the forecast time domain, construct a rolling optimization model with the goal of tracking the day-ahead charging and discharging power plan, set energy storage ramp rate constraints and state of charge constraints, solve the rolling optimization model and output the corrected real-time power command; S3. Real-time adaptive droop control: The energy storage converter receives the corrected real-time power command, collects the grid frequency deviation and the health status value of the energy storage battery pack, calculates the adaptive droop coefficient based on the health status value, and superimposes the primary frequency regulation response power with the corrected real-time power command to output the final power of the energy storage converter.

2. The multi-objective hierarchical scheduling method for smart energy storage systems oriented towards source-grid-load-storage coordination as described in claim 1, characterized in that, Step S1 includes the following sub-steps: S11. Data Acquisition: Collect the photovoltaic power generation forecast values ​​for each time period of the next 24 hours through the photovoltaic inverter data interface, collect the load power forecast values ​​for each time period of the next 24 hours through the smart meter, and obtain the time-of-use electricity price for each time period of the next 24 hours through the power grid dispatch system. S12. Objective Function Construction: Three objective functions are established based on the collected data: The economic objective function calculates the system operating cost using the following formula: ； In the formula, The economic objective function value (in yuan); To Summation operators from 1 to 24; For time period Time-of-use electricity price (RMB / kWh); For time period The exchange power (kW) with the power grid, with positive values ​​for purchasing electricity and negative values ​​for selling electricity; The duration is 1 (h). This is the depreciation cost factor for energy storage batteries (yuan / kWh). This is the absolute value operator; For time period The charging and discharging power (kW) of the energy storage system; The environmental protection objective function is used to calculate the renewable energy absorption rate. The calculation formula is as follows: ； In the formula, The objective function value is the environmental friendliness. To Summation operators from 1 to 24; For time period Actual photovoltaic power generation absorbed (kW); For time period Forecasted photovoltaic power generation (kW); The smoothness objective function is used to calculate the variance of tie-line power fluctuations. The calculation formula is as follows: ； In the formula, The objective function value for smoothness is (kW²). To Summation operators from 1 to 24; For time period Power exchanged with the power grid (kW); This represents the average daily switching power of the tie line (kW). S13. Nash Negotiation Solution: Optimize the economic objective function, environmental objective function, and smoothness objective function separately to obtain the ideal point value for each objective. , , ; Construct the Nash product objective function as follows: The sequential quadratic programming algorithm is used to solve for the maximum value of the Nash product objective function, and the day-ahead charging and discharging power plan of the energy storage system for each time period is output. .

3. The multi-objective hierarchical scheduling method for smart energy storage systems oriented towards source-grid-load-storage coordination as described in claim 1, characterized in that, Step S2 includes the following sub-steps: S21. Acquisition of ultra-short-term forecast data: at the current runtime Get from the current moment At that time Source load ultra-short-term forecast data, The number of steps to predict the time domain; S22. Rolling Optimization Model Construction: With the objective of minimizing the deviation from the day-ahead charging and discharging power plan, the objective function of the rolling optimization model is constructed as follows: ； In the formula, The minimum value operator; To optimize the objective function value (kW²) for rolling. To from to The summation operator; For a moment Energy storage charging and discharging power (kW) to be optimized; For a moment The planned daytime charge and discharge power (kW); This is the current running time; To predict the number of time-domain steps; S23. Constraint settings: Set the ramp rate constraint and the state of charge constraint; The expression for the gradeability constraint is: ； In the formula, This is the absolute value operator; For a moment Energy storage charging and discharging power (kW); For a moment Energy storage charging and discharging power (kW); The maximum ramp rate (kW / h) of the energy storage system. To control the time step length (h); The expression for the state-of-charge constraint is: ； In the formula, This is the upper limit value under charged state; For a moment The state of charge value; This represents the upper limit of the state of charge. S24. Solution and Output: The rolling optimization model is solved using a quadratic programming algorithm under the constraints of ramp rate and state of charge, yielding the optimal charging and discharging power sequence in the predicted time domain. The first component of this sequence is taken as the corrected real-time power command for the current moment. The command is then sent to the energy storage converter for execution. At the next control moment, the initial state is updated with the new measurement value, and steps S21 to S24 are repeated.

4. The multi-objective hierarchical scheduling method for smart energy storage systems oriented towards source-grid-load-storage coordination as described in claim 1, characterized in that, Step S3 includes the following sub-steps: S31. Status Acquisition: The energy storage converter receives the corrected real-time power command. The frequency detection module collects the real-time frequency of the power grid and calculates the frequency deviation. The battery management module obtains the current health status value of the energy storage battery pack. and the average health status value of the energy storage battery pack ; S32. Calculation of Adaptive Sag Coefficient: The adaptive sag coefficient is calculated based on the health status value. The calculation formula is as follows: ； In the formula, For a moment The adaptive droop factor (kW / Hz); The baseline droop factor (kW / Hz) is used. As a regulator of health status; For a moment Health status values ​​of the energy storage battery pack; This represents the average health status value of the energy storage battery pack. For a moment Frequency deviation penalty factor; The formula for calculating the frequency deviation penalty factor is: ； In the formula, For a moment Frequency deviation penalty factor; This is the penalty intensity coefficient; This is the absolute value operator; For a moment The power grid frequency deviation (Hz); The permissible frequency deviation limit (Hz); S33. Power Superposition Output: The primary frequency regulation response power is superimposed with the corrected real-time power command to calculate the final power of the energy storage converter. The calculation formula is as follows: ； In the formula, For a moment Final power (kW) of the energy storage converter; For a moment Revised real-time power command (kW); For a moment The adaptive droop factor (kW / Hz); For a moment Grid frequency deviation (Hz); control the energy storage converter according to the final power To output power.

5. The multi-objective hierarchical scheduling method for smart energy storage systems oriented towards source-grid-load-storage coordination as described in claim 4, characterized in that, In step S31, the health status value Obtain it through the following steps: The battery management module collects the terminal voltage, charging / discharging current, and temperature of the energy storage battery pack in real time; it calculates the state of charge (SOC) of the battery pack by integrating the charging / discharging current over time and correcting it with the open-circuit voltage; and it calculates the health status value based on the ratio of the current maximum release capacity to the rated capacity of the energy storage battery pack. The calculation formula is as follows: ； In the formula, For a moment Health status value; For a moment Maximum release capacity of the energy storage battery pack (Ah); This refers to the rated capacity (Ah) of the energy storage battery pack.

6. A multi-objective hierarchical scheduling system for intelligent energy storage systems oriented towards source-grid-load-storage coordination, used to implement the method of any one of claims 1 to 5, characterized in that, include: The source-load panoramic perception module communicates with photovoltaic inverters, smart meters and power grid dispatching systems to collect photovoltaic power generation prediction data, load power prediction data and time-of-use electricity price data. The cloud-based game scheduling platform communicates with the source-load panoramic perception module to receive photovoltaic power generation forecast data, load power forecast data, and time-of-use electricity price data. It constructs economic objective functions, environmental objective functions, and smoothing objective functions, and runs the Nash negotiation algorithm to output the day-ahead charging and discharging power plan. The edge-side collaborative controller communicates with the cloud-based game scheduling platform and the source-load panoramic perception module to receive day-ahead charging and discharging power plans and source-load ultra-short-term forecast data, and runs a rolling optimization model to output corrected real-time power commands. The intelligent converter execution unit is signal-connected to the edge-side co-controller and electrically connected to the power grid and energy storage battery pack. It is used to receive the corrected real-time power command, collect the power grid frequency deviation and the health status value of the energy storage battery pack, calculate the adaptive droop coefficient and perform power superposition, and output the final power of the energy storage converter.

7. The multi-objective hierarchical scheduling system for intelligent energy storage systems oriented towards source-grid-load-storage coordination as described in claim 6, characterized in that, The edge-side collaborative controller includes: The data receiving unit is connected to the cloud-based game scheduling platform and the source-load panoramic perception module to receive day-ahead charging and discharging power plans and source-load ultra-short-term forecast data. The MPC computing unit is connected to the data receiving unit and is used to build a rolling optimization model and set the ramp rate constraint and state of charge constraint. It solves the rolling optimization model to generate the corrected real-time power command. The instruction issuing unit is connected to the MPC calculation unit and the intelligent converter execution unit via signals, and is used to issue the corrected real-time power instruction to the intelligent converter execution unit.

8. The multi-objective hierarchical scheduling system for intelligent energy storage systems oriented towards source-grid-load-storage coordination as described in claim 6, characterized in that, The intelligent converter execution unit includes: The battery management module is connected to the energy storage battery pack and is used to collect the terminal voltage, charging and discharging current and temperature of the energy storage battery pack, calculate and output the health status value and average health status value of the energy storage battery pack. The frequency detection module is electrically connected to the power grid and is used to detect the power grid frequency and calculate the frequency deviation. The control calculation module is connected to the battery management module, frequency detection module and edge-side collaborative controller. It is used to receive the corrected real-time power command, health status value and frequency deviation, calculate the adaptive droop coefficient, and perform power superposition calculation to calculate the final power of the energy storage converter. The power conversion module is signal-connected to the control calculation module and electrically connected to the energy storage battery pack and the power grid. It is used to control the energy storage converter to perform power conversion based on the final power output of the control calculation module.

9. The multi-objective hierarchical scheduling system for intelligent energy storage systems oriented towards source-grid-load-storage coordination as described in claim 6, characterized in that: A two-way communication connection is established between the cloud-based game scheduling platform and the edge-side collaborative controller. The cloud-based game scheduling platform sends the day-ahead charging and discharging power plan to the edge-side collaborative controller, and the edge-side collaborative controller uploads the energy storage system operation status data to the cloud-based game scheduling platform. A bidirectional signal connection is formed between the edge-side collaborative controller and the intelligent converter execution unit. The edge-side collaborative controller sends a corrected real-time power command to the intelligent converter execution unit, and the intelligent converter execution unit feeds back the operating status of the energy storage converter to the edge-side collaborative controller.

10. The multi-objective hierarchical scheduling system for intelligent energy storage systems oriented towards source-grid-load-storage coordination as described in claim 8, characterized in that, The control calculation module is integrated into the DSP chip inside the energy storage converter. The control calculation module is connected to the IGBT drive circuit of the power conversion module. The control calculation module generates a PWM control signal based on the final power and outputs it to the IGBT drive circuit of the power conversion module.

Images