Polar island micro-grid optimal scheduling method considering frequency constraint

Abstract

The invention discloses a polar region island micro-grid optimal scheduling method considering frequency constraint, and relates to the technical field of island micro-grid optimal scheduling, and the micro-grid optimal scheduling method comprises the following steps: building a system model; constructing and solving a multi-time scale joint optimization architecture; embedding a frequency transient constraint as a minute layer decision constraint in a linearization or convex approximation form; and according to a solving result, issuing a hierarchical instruction to various distributed resources and implementing online monitoring and safety guarantee measures. According to the method, through explicit embedding and linear convex approximation of frequency transient constraint, the defect that only steady-state frequency is considered or an empirical threshold value is adopted in traditional scheduling is overcome, uncontrollable frequency transient caused by short-time high-power disturbance is avoided, minute-level frequency safety guarantee is achieved, the frequency peak value and the minimum point are controlled, and the scheduling efficiency is improved. And load removal or equipment protection actions caused by transient overrun are reduced, so that the power supply continuity and the system reliability are improved.

Description

Technical Field

[0001] This invention relates to the field of islanded microgrid optimization scheduling technology, specifically to an optimization scheduling method for polar islanded microgrids that takes frequency constraints into account. Background Technology

[0002] The ongoing polar scientific research activities have placed unprecedented demands on the reliability and independence of energy supply. Limited by the difficulties, high costs, and environmental pollution associated with diesel refueling in polar regions, traditional diesel generator sets can no longer fully meet the long-term operational needs of polar scientific research stations. Therefore, developing and utilizing the abundant wind and solar energy resources in the polar regions to construct polar island microgrids primarily based on renewable energy has become an inevitable strategic direction to support the development of polar scientific research. However, the extreme polar climate conditions pose significant challenges to the utilization of renewable energy: wind turbines are subjected to strong winds, leading to drastic fluctuations in output power. Simultaneously, photovoltaic power generation is constrained by the alternating cycles of polar day and polar night, exhibiting diametrically opposed operating states. During the polar day, continuous photovoltaic power generation generates excess electricity, significantly exacerbating the system's regulation pressure; while during the long polar night, photovoltaic power generation drops to almost zero. This drastic and random fluctuation in power generation severely disrupts the power balance of polar island microgrids, directly triggering continuous system frequency oscillations.

[0003] To address these challenges, energy storage batteries, with their rapid response capabilities, have become a key technology for frequency regulation in isolated microgrids. However, existing energy storage frequency regulation methods have limitations in environmental adaptability. The extreme polar environment accelerates the aging of battery energy storage, significantly reducing the frequency stability and renewable energy absorption capacity of polar isolated microgrids. Simultaneously, the unique polar day and night phenomena and extremely strong winds further exacerbate the hourly, daily, and seasonal power imbalances in polar isolated microgrids, further increasing the difficulty of frequency optimization. Therefore, hybrid energy storage is urgently needed to provide multi-timescale regulation support. Existing research cannot meet the scheduling needs of polar isolated microgrids in terms of multi-timescale scheduling and hybrid energy storage synergy, resulting in scheduling schemes that fail to guarantee frequency stability in polar isolated microgrids.

[0004] Patent CN113708418B discloses a microgrid optimization scheduling method. The above patent improves the overall efficiency of the energy system and the reliability of energy supply, and increases the flexibility of absorbing renewable energy.

[0005] The aforementioned patents have stronger robustness, can converge quickly and reliably, and are highly reliable, practical, and accurate. However, they cannot meet the dispatching needs of polar island microgrids in terms of multi-timescale scheduling and hybrid energy storage coordination.

[0006] To this end, this application proposes an optimized scheduling method for polar island microgrids that takes into account frequency constraints, enabling hybrid energy storage to provide multi-timescale regulation support. Summary of the Invention

[0007] The purpose of this invention is to provide an optimized scheduling method for polar island microgrids that takes into account frequency constraints, in order to solve the technical problem mentioned in the background that existing research cannot meet the scheduling requirements of polar island microgrids in terms of multi-timescale scheduling and hybrid energy storage synergy, resulting in scheduling schemes that are difficult to guarantee the frequency stability of polar island microgrids.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for optimal scheduling of polar island microgrids considering frequency constraints, the microgrid optimal scheduling method comprising the following steps: Establish a system model, including (1) a power balance model and operating constraints at three time scales: hour, 15 minutes, and minute; (2) an equivalent pendulum frequency dynamic model, and associate RoCoF, the lowest frequency point nadir, the steady-state frequency deviation, and the power imbalance; (3) a hybrid energy storage dynamic decay model, which includes battery cycle decay and temperature-related decay terms. Construct and solve a multi-timescale joint optimization architecture: the hourly layer uses two-stage stochastic programming with scenario trees to generate the daily baseline plan; the 15-minute layer uses rolling model predictive control (MPC) to correct and fine-tune within the rolling window; the minute layer uses fast linear, quadratic programming, or analytical control to achieve transient frequency constraints and fast power injection. Frequency transient constraints are embedded as minute-level decision constraints through linearization or convex approximation, and energy storage degradation costs and battery thermal management actions are internalized as scheduling decision variables into the objective function. Based on the solution results, hierarchical instructions are issued to various distributed resources and online monitoring and security measures are implemented.

[0009] Preferably, the energy storage dynamic degradation model specifically includes: Cyclic decay function is used to represent the capacity loss related to the number of charge-discharge cycles and the depth of discharge; Temperature-dependent decay function is used to represent the effect of ambient temperature and thermal management power on capacity decay; Furthermore, the equivalent degradation cost is incorporated into the objective functions of the 15-minute and hourly layers to balance frequency support and battery life.

[0010] Preferably, the minute-level frequency constraint is established based on a linearized expression of the equivalent pendulum equation, including: RoCoF constraint, nadir constraint, steady-state deviation constraint; The linearized expression serves as an inequality constraint for minute-level optimization and is used for fast solution. Furthermore, a pre-computed conservative margin table is employed in the solution process to ensure real-time feasibility.

[0011] Preferably, the 15-minute layer is a rolling MPC controller, specifically including: At each rolling step, the wind and solar load forecasts are updated, and the SOC target trajectory is dynamically adjusted based on the updated forecasts, battery temperature, and available capacity. The charging and discharging power, SOC trajectory, and battery thermal management of lithium batteries are used as decision variables, and a dynamically updated decay cost term is added to the objective function to limit high DOD and low temperature operation. Consistent constraint is applied to ensure that the output of the 15-minute layer remains consistent with the baseline of the hourly layer.

[0012] Preferably, the hourly layer employs a two-stage stochastic programming approach to handle the uncertainty of renewable power output, specifically including: Construct a scenario tree containing several scenarios to represent the uncertainty of the scenery over 24 hours; The first phase determines the baseline output and hydrogen production and release plans. The second phase calculates the replenishment actions and expected replenishment costs for each scenario and incorporates the expected replenishment costs into the overall hourly target. The hourly targets include: fuel diesel costs, hydrogen operating costs, penalties for curtailment of electricity and hydrogen, and long-term lifetime guidance.

[0013] Preferably, the division of responsibilities between hybrid energy storage and power sources over a time scale includes: Hydrogen energy systems handle large-scale energy balancing and energy shifting on an hourly scale. Lithium-ion batteries provide energy buffering for 15 minutes and short-term frequency modulation support. Supercapacitors can handle high-frequency power injection on a timescale of minutes or less to cope with power disturbances; Furthermore, the optimized scheduling method includes an adaptive virtual inertia droop allocation strategy, which dynamically adjusts the inverter's virtual inertia and droop coefficient based on online estimated available energy storage capacity and temperature conditions, in order to balance frequency performance and battery life consumption under different operating conditions.

[0014] Preferably, the hierarchical solution employs a decomposition coordination strategy to improve computational efficiency. The decomposition coordination strategy is either Benders decomposition or the Alternating Multiplier Method (ADMM), wherein: The hourly layer serves as the primary problem, generating cut sets or Lagrange multiplier information. The 15-minute layer and the minute layer are treated as parallel subproblems, solved within their respective time scales, and return feasible cuts or price difference information to the main problem to gradually converge. During parallel solution, real-time executability constraints are preserved and conservative margins are applied to key frequency constraints to determine their feasibility.

[0015] Preferably, the thermal management and cryogenic protection include: The battery surface temperature is monitored online. When the battery surface temperature is lower than a preset threshold, the battery preheating power is introduced as a controllable decision variable into the optimization problem. The preheating power consumption is included in the objective functions of the 15-minute and hourly layers, and is linked with the decay function to calculate the equivalent lifetime loss caused by low-temperature operation; Within the optimization objective, lifetime loss is monetized in the form of equivalent energy unit price, thereby automatically balancing preheating energy consumption and accelerated decay during scheduling.

[0016] Preferably, the online security measures include: Real-time frequency and power imbalance monitoring; if the frequency is found to be approaching the warning threshold, a rapid response at the minute level is triggered and the virtual inertia and droop distribution are adjusted. Set up an emergency disconnection strategy. When multiple failures or extreme weather make dispatching infeasible, implement minimum load cut-off and start / stop standby diesel engines and fuel cells according to priority. A pre-computed frequency margin conservative table is used at the minute level to ensure the feasibility and safety boundaries of real-time control.

[0017] Preferably, the scheduling control device includes: The data acquisition unit is used to collect load, wind and solar power output, energy storage SOC, energy storage surface temperature, inverter status and frequency signals in real time across multiple time scales. The hourly scheduling module is used to solve hourly stochastic programming with two-stage decision-making based on scenario trees and output a baseline plan; The 15-minute layer MPC module is used to solve MPC problems with decay costs and thermal management decisions within a rolling window and output the SOC trajectory and preheating command. The minute-level fast frequency control module is used to quickly solve and output instantaneous power commands, virtual inertia and droop adjustment for supercapacitors and inverters based on linearized frequency constraints. The coordination communication unit is used to exchange cut sets, Lagrange multipliers, and baseline correction information between modules and to issue control commands to the field execution equipment; The scheduling control device implements the optimized scheduling method in the form of computer-executable instructions.

[0018] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention compensates for the shortcomings of traditional scheduling, which only considers steady-state frequency or uses empirical thresholds, by explicitly embedding frequency transient constraints and using linear convex approximation. It avoids uncontrollable frequency transients caused by short-term high-power disturbances, achieves minute-level frequency safety assurance, controls frequency peak and minimum points, reduces load shedding or equipment protection actions caused by transient over-limits, and thus improves power supply continuity and system reliability. 2. This invention uses dynamic energy storage decay and thermal management as scheduling decision variables to solve the problem that traditional scheduling ignores the accelerated decay caused by low temperature and the impact of cycle depth on lifespan, which leads to increased long-term operating costs and frequent energy storage degradation. It reduces harmful operation under high DoD and low temperature conditions and improves the system sustainability in polar environments. 3. This invention achieves optimal resource allocation based on capability and lifespan through multi-timescale responsibility division and adaptive virtual inertia droop allocation. It solves the problem that static or artificially fixed inertia droop settings cannot maintain optimal frequency performance when equipment availability or temperature changes, reduces high-frequency and high-amplitude cycling of batteries, and improves the overall system dynamic response capability and equipment health management. 4. This invention achieves parallel solving of the main and sub-problems and information interaction through parallel solving, solving the problem that traditional solvers cannot meet the requirements of parallel solving in real time, greatly improving the solving efficiency and real-time executability, enabling the system to make economic plans for future uncertainties and respond to frequent sudden events in a short time. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the multi-time-scale optimized scheduling of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 The present invention provides an embodiment of an optimized scheduling method for polar island microgrids that takes frequency constraints into account, and establishes a dynamic capacity decay model for polar energy storage: Energy storage systems play a crucial role in polar island microgrids, fulfilling multiple important functions such as mitigating power fluctuations, providing frequency support, and enabling energy time-shifting. The extreme low temperatures of the polar environment pose challenges to the performance and lifespan of energy storage, necessitating the development of accurate degradation models to guide optimized operation. This model comprehensively considers the impact of polar temperatures, cycle count, and depth of discharge on lithium-ion battery capacity degradation, quantifying capacity degradation under polar day, polar night, and stormy weather conditions, and embedding degradation costs into the polar dispatch optimization model.

[0022] (1) (2) (3) Equation (1) is the dynamic capacity decay model of lithium battery, where: For lithium battery capacity in Capacity of time, This refers to the depth of discharge of the battery. This refers to the number of battery cycles. Where α is the battery surface temperature, and β and γ are empirical parameters obtained by fitting the battery's physical properties. This refers to the battery capacity degradation caused by temperature. Equation (2-3) is... The specific expression is: and These are the ambient temperature and the battery surface temperature, respectively. and These represent the frequency factors for solid electrolyte phase interface formation and lithium deposition, respectively; and represents the activation energy of the two side reactions; R is the molar gas constant; b is the ratio coefficient of temperature to power.

[0023] (4) (5) (6) Equation (4) is the charge / discharge count model, where: and They are respectively The charging and discharging state at any given time; and They are respectively and The number of charge-discharge cycles at time t. Equation (5-6) is the model for calculating the depth of discharge, where: These are the three stages of the state of charge of a lithium battery; and , respectively, represent the depth of discharge at times t and t-1; Establish a dynamic capacity decay model for energy storage.

[0024] Please see Figure 1 One embodiment of this invention provides: an optimized scheduling method for polar island microgrids considering frequency constraints, which constructs a multi-time-scale polar hybrid energy storage model. This model incorporates supercapacitors, lithium batteries, and hydrogen energy, with the hydrogen energy system including an electrolyzer, hydrogen storage tank, and fuel cell. Dynamic response models of the hybrid energy storage model are established at different time scales, forming a multi-time-scale hybrid energy storage collaborative model.

[0025] Polar wind and solar resources are subject to hourly fluctuations that are prolonged and have large amplitudes. Hydrogen energy systems can produce hydrogen through electrolyzers to store excess electrical energy, or supplement the power shortage through fuel cells, adapting to hourly energy regulation cycles. By balancing hourly energy fluctuations, hydrogen energy systems achieve hourly timescale frequency adjustment, providing an important guarantee for frequency stability in extreme polar environments.

[0026] Modeling of hourly timescale hydrogen energy systems in polar hybrid energy storage: (7) (8) (9) (10) (11) (12) Equation (7-12) is the hydrogen model on an hourly timescale. Equation (7-8) is the hydrogen storage tank model, where: and These represent the hydrogen content in the hydrogen storage tank at times t and t-1, respectively. This indicates the amount of hydrogen produced by the electrolyzer; This indicates the amount of hydrogen consumed by the fuel cell; , These represent the minimum and maximum hydrogen storage capacities of the hydrogen storage tank, respectively. Equation (9-10) is the model for the electrolyzer, where: This represents the power of the electrolytic cell equipment. This indicates its operating efficiency. This indicates its hydrogen-to-electric conversion coefficient. Indicates the runtime. and respectively The minimum and maximum values. Equation (11-12) is the fuel cell model, where: This represents the output power of the fuel cell at time t. This indicates its maximum output power. Indicates the working efficiency of the fuel cell. This indicates its hydrogen-to-electric conversion coefficient.

[0027] Hydrogen energy systems focus on regulating fluctuations at the hourly level, while lithium batteries fill the gap in regulation at the minute-level. Polar wind and solar resources experience extremely frequent short-term fluctuations, which are characterized by short durations and rapid changes. Lithium batteries, acting as a 15-minute timescale regulation unit, can balance these fluctuations: when wind and solar power output surges, lithium batteries charge rapidly; when output drops sharply, they quickly discharge to replenish energy. By balancing short-term energy fluctuations in polar regions, lithium battery energy storage systems achieve 15-minute timescale frequency adjustment, providing crucial support for frequency stability in extreme polar environments.

[0028] Modeling of lithium-ion battery systems on a 15-minute timescale in polar hybrid energy storage: (13) (14) (15) (16) (17) Equations (13-17) represent a lithium battery model on a 15-minute timescale, where: For the power of lithium batteries, , These are the charging and discharging power of the lithium battery, respectively. This is the maximum power of the lithium battery; , This indicates the state of charge of the lithium battery at times t+1 and t; , The charging and discharging efficiencies of the lithium battery at time t are respectively. Initial lifespan; , These represent the maximum and minimum states of charge of a lithium battery, respectively.

[0029] In addition to hourly and 15-minute fluctuations, the polar regions also experience minute-level high-frequency power fluctuations. If these fluctuations are not precisely mitigated, they can gradually lead to significant frequency fluctuations, threatening the power supply security of isolated polar power grids. Supercapacitors, as regulating units operating on a minute-scale, directly participate in frequency regulation. By balancing minute-level high-frequency energy fluctuations in the polar regions, supercapacitors achieve minute-scale frequency adjustment, providing crucial support for frequency stability in the extreme polar environment.

[0030] Modeling of minute-scale supercapacitor systems in polar hybrid energy storage: (18) (19) (20) Equation (18-20) is a supercapacitor model on a minute timescale, where: , These represent the charging and discharging power of the supercapacitor, respectively. This refers to the rated power of the supercapacitor. , This represents the supercapacitor charge at times t+1 and t; , The charging and discharging efficiencies of the supercapacitor at time t are respectively.

[0031] Please see Figure 1 The present invention provides an embodiment of an optimal scheduling method for polar island microgrids that takes frequency constraints into account, comprising a multi-time-scale optimal scheduling model: To address the strong hourly fluctuations in polar wind and solar resources, an hourly-scale scheduling model can optimize the output plans of diesel generators, electrolyzers, and fuel cells in advance based on hourly forecast data of wind and solar power output and load. This avoids power imbalances caused by wind and solar fluctuations and ensures continuous power supply to isolated polar island power grids. An hourly-scale scheduling model is established, aiming to minimize the daily operating cost of the system. Its optimized mathematical expression is as follows: (twenty one) (twenty two) (twenty three) (twenty four) (25) (26) (27) (28) (29) (30) Equation (21) is the objective function for hourly optimized scheduling, where: [Equation missing]. This represents the total operating cost of the system over 24 hours; T represents 24 hours. This indicates the diesel fuel cost of the diesel generator set; This represents the overall operation and maintenance cost of the system; This reflects the penalty costs incurred due to restrictions on wind and solar power generation. Equations (22-24) provide specific expressions for each cost, where: This represents the penalty coefficient for wind and solar power curtailment; This represents the curtailed power of wind and solar power at time t; Δt is the time interval, set to one hour for hourly optimization. This indicates the price of fuel consumed by the diesel generator; This refers to the output power of the diesel generator. The unit operating cost of the i-th device is included in this layer of optimization, which includes: diesel generator, wind turbine, photovoltaic panel, hydrogen fuel cell and electrolyzer; For the first The output power of the equipment. Equation (25) is the power balance constraint when the power is low, where: This indicates the output power of the diesel generator. Indicates the output power of wind power generation. Indicates the output power of photovoltaic power generation. This indicates the load at this time. Equation (26-27) is for a diesel generator, where: This indicates the minimum output power of the diesel generator. This indicates the maximum output power of the diesel generator. and These represent the maximum power limits of the diesel generator during ramp-up and ramp-down, respectively. Equations (28-29) represent the new energy equipment model, where: and These represent the minimum and maximum output values ​​of wind power generation, respectively. and Let represent the minimum and maximum values ​​of the photovoltaic output, respectively. Equation (30) is the hydrogen storage capacity constraint of the hydrogen storage tank, and the hydrogen storage capacity remains equal on the first and seventh days of the week.

[0032] Hourly scheduling models rely on predicted wind and solar power output and load data. However, polar weather conditions are extremely variable, and directly executing hourly plans would lead to power imbalances. A 15-minute timescale model uses real-time wind and solar power output and load data as input. Within the framework of hourly planning, it dynamically optimizes lithium battery output, quickly mitigating prediction errors and power fluctuations caused by short-term power fluctuations, ensuring power supply reliability. A 15-minute timescale scheduling model is established, incorporating energy storage aging costs into the sum of operating costs. With the objective of minimizing the system's 15-minute operating cost, its optimization model mathematical expression is: (31) (32) (33) (34) (35) (36) (37) Equation (31) is the objective function for 15-minute optimized scheduling, where: This represents the total operating cost of the system over 15 minutes; equations (32-33) are the specific expressions for each cost. This indicates the cost incurred by lithium batteries due to aging; This indicates the cost of replacing lithium batteries; Let be the unit operating cost of the i-th device. The devices in this layer of scheduling include: wind turbine, photovoltaic panel, and lithium battery. Equation (34) is the 15-minute power balance constraint. Equations (35-36) are the new energy output constraints. Equation (37) is the state of charge constraint for lithium battery, where the state of charge is equal in the first hour and the 24th hour of the day.

[0033] A minute-scale timescale scheduling model is established to address the minute-level high-frequency power fluctuations in polar regions. This model leverages the rapid response characteristics of supercapacitors to mitigate these fluctuations, aiming to minimize the system's minute operating cost. It accepts hourly scheduling for diesel generators, electrolyzers, and fuel cells, and 15-minute scheduling for lithium battery output. Frequency safety constraints caused by minute-level polar fluctuations are incorporated into the optimization model. The supercapacitor output is optimized based on real-time load and renewable energy output to obtain the supercapacitor output per minute. The mathematical expression of the optimization model is as follows: (38) (39) (40) (41) (42) (43) (44) Equation (38) is the objective function for minute-optimized scheduling, where: This represents the operating cost of the system's minute-by-minute scheduling. The unit operating cost of the kth device is given. The devices in this layer of scheduling include: wind turbines, photovoltaic panels, and supercapacitors; Equation (40) represents the minute power balance constraint. This is the minute power imbalance at this time; Equation (41-43) is the system frequency constraint; The frequency fluctuation representing the power imbalance Represents the system's unit regulating power; frequency value at time t. It can be calculated using equation (42), where The frequency value at time t-1; and These represent the minimum and maximum allowable frequencies of the system, respectively. Equation (44) is the supercapacitor charge constraint, where the charge in the first minute and the 60th minute of an hour remains equal.

[0034] Please see Figure 1 The present invention provides an embodiment of an optimized scheduling method for polar island microgrids that takes frequency constraints into account, which is verified by simulation at a polar research station. System Configuration: Two wind turbines, each 1.0MW, with rated variation characteristics set according to historical data; 0.8MW of photovoltaic power; a lithium-ion battery pack with a rated capacity of 1.5MWh, a maximum charge / discharge power of 750kW, a rated energy efficiency of 0.95, and an initial SOC of 0.6; a supercapacitor (SC) with a total capacity of 200kWh, a maximum power of 1.0MW, and a response time of <1s; a hydrogen energy system: a 500kW electrolyzer with an oxygen storage capacity equivalent to approximately 50kg, and a 200kW fuel cell; a diesel engine: one 500kW unit capable of ramping up to rated power within 5 minutes, with a minimum output of 100kW; frequency safety threshold: RoCoF. max =0.5Hz / s, lowest frequency f min =49.0Hz, steady-state deviation Battery low-temperature preheating threshold: T preheat =-20℃, preheating power limit 50kW.

[0035] Offline preprocessing (linearization and margin table): Offline detailed transient model for different equivalent inertia H eff Simulate the system damping D with disturbance ΔP (0.1-1.0MW) and fit a linear approximation of nadir with ΔP (fitting coefficients for given H and D): Example linearized form: nadir≈f0-(α1ΔP+α2), and generate a conservative margin table for real-time control (e.g., when H...). eff When =3s and D=0.8, the nadir linear approximation corresponding to ΔP=0.5MW produces nadir≥49.35Hz, and a conservative margin of 0.05Hz is added, setting 49.40Hz as the lower bound of the constraint. These offline data are used for parameterization and rapid evaluation of minute-level constraints.

[0036] Hourly level (24 hours, two-stage stochastic programming): The time step is 1 hour, the prediction time domain is 24 hours, and the number of scenarios is 10 (scenario trees are constructed by sampling historical wind and solar data); the decision variable is the hourly output P of the diesel engine. dg,t Hydrogen production power P elec,t Fuel cell discharge P fc,t Hydrogen tank terminal SOC, etc.; Example of objective function (hour level): ; in Lifetime cost guiding weights (example taken) =20 / kWh-ep), diesel fuel and operation and maintenance costs are given according to the common fuel model. The first phase gives the baseline plan, and the second phase calculates the expected cost of the replenishment action under each scenario and feeds it back to the first phase.

[0037] 15-minute layer (rolling MPC): Step size 15 minutes, rolling window 4 hours; Decision variable: Lithium battery charging and discharging power P b,k SOC trajectory within 15 minutes, preheating power P heat,k 1. Make minor adjustments to the backup diesel fuel; Objective function: ; Among them, attenuation cost The specific formulas used are: α1 = 0.8 yuan / (kWh / cdotpcycle), α2 = 1.2 yuan / (kWh / cdotp℃). Constraints include: upper and lower bounds of SOC, charge / discharge power limits, and consistency constraints with the hourly baseline.

[0038] Minute level: Refresh cycle 60s; Decision variable: Supercapacitor power P sc,t Inverter instantaneous response allocation, virtual inertia parameter H v (t), droop coefficient k drop (t); Constraint form (linearization): ; nadir: Using offline linearization and a conservative margin, nadir≥f min +ϵ nadir ; steady-state frequency ≤Δf max .

[0039] Real-time solution: Construct the above constraints and linear cost into an LP, solve it within 60 seconds, and issue control commands. If the LP has no feasible solution, trigger the safety policy.

[0040] Adaptive virtual inertia and droop strategy: Example of a rule: If the usable capacity of the lithium battery is C avail >0.3C rated And T > -10℃, let H v =H v,base +1.0s, droop coefficient k drop =0.05 pu / Hz.

[0041] If T < -20℃ or Cavail <0.15C rated Reduce the virtual inertia of the supercapacitor and increase the preheating weight to avoid deep battery degradation.

[0042] Offline and online testing metrics: The simulation scenarios include three conditions: "polar day with strong winds," "polar night with low light," and "sudden load surge + wind stoppage." Example results summary: The curtailment rate is approximately 30% lower than that of hydrogen-free absorption control; The maximum frequency fluctuation amplitude at the minute level decreased by approximately 40%; The battery's equivalent lifespan is extended by approximately 20%.

[0043] Please see Figure 1 This invention provides an embodiment of an optimized scheduling method for polar island microgrids considering frequency constraints, with a description of the field controller configuration and operation process: On-site hardware and communication architecture: Controller: 1 main control server and 1 edge controller for rapid control at the minute level; IED / RTU: Used to collect inverter, battery temperature, frequency, and power data. Sampling frequency: 1Hz for minute-level data, and aggregated data from 15-minute and hour-level data in 1-minute and 5-minute windows. Communication protocol: IEC61850 / Modbus TCP; Control delay requirements: Total delay of minute-level end commands ≤2s.

[0044] Online operation steps: Every hour on the hour: The master server runs hourly random optimization, generates the next 24-hour baseline plan, and distributes the baseline to the 15-minute level; Every 15 minutes: The 15-minute layer MPC performs rolling optimization with the latest load, wind and solar forecasts and SOC temperature data, generates SOC trajectory, preheating plan and reserve power allocation, and sends it down to the minute layer and updates several soft constraints in the hour layer. Every minute: Read the real-time frequency and power deviation signals, calculate the instantaneous output of the supercapacitor and inverter and adjust the virtual inertia based on the minute-level LP, and implement closed-loop control. If there is no feasible solution for LP or the frequency is critical (frequency < 49.1 Hz and RoCoF > 0.6 Hz / s), trigger the emergency strategy (immediately start / stop the diesel engine or perform load shedding); Time logging and online learning: Whenever a significant frequency event occurs, the event data is recorded and used for offline regression to update nadir, linearization coefficients, and conservative margin tables.

[0045] Safety and Redundancy: Two-level redundancy control: Both the main control server and the edge controller have watchdogs. When communication is lost, the edge controller enters an independent safety mode to perform minute-level frequency protection. Data verification: Cross-comparison verification and anomaly filtering are performed on key sensor data to prevent erroneous data from causing malfunctions.

[0046] Please see Figure 1 The present invention provides an embodiment of an optimized scheduling method for polar island microgrids considering frequency constraints, illustrated under extreme fault and emergency grid disconnection conditions: Scenario: Sudden wind turbine shutdown + major diesel engine failure: Initial operating conditions: High wind day, system equivalent inertia H eff =3s; Sudden event: Instantaneous available wind power decreases by 0.8MW, and diesel engine 1 becomes unavailable after 2 minutes.

[0047] Real-time response process: Second-level monitoring: A sudden drop in frequency triggers a minute-level strategy. The edge controller immediately calculates the maximum injection of the supercapacitor, Psc,max, according to LP and adjusts the inverter droop to increase the first response. Concurrent actions: Simultaneously, the minute layer sends an alert to the 15-minute layer, the 15-minute layer shortens the rolling window and adjusts the lithium battery discharge and fuel cell output; if the current SOC of the lithium battery is >0.4, short-term high power output is allowed (constrained by degradation cost), and a battery thermal management alert is initiated (if the temperature is low, high-rate discharge is prioritized after preheating to protect battery life). Hourly standby: After the hourly layer detects a diesel engine failure, it restarts the two-stage plan and updates the hydrogen production and release strategies to compensate for the long-term energy gap.

[0048] Emergency load shedding logic: Priority settings: Non-critical loads → Secondary loads → Critical life support loads; Example of triggering conditions: If the frequency is not controlled to ≥49.2Hz within 60s and RoCoF exceeds the threshold, then 10% of the total load will be cut off according to priority. Recovery process: If the frequency is stable after the cut-off, the scheduling module (15-minute layer) will gradually restore the cut-off load and correct the SOC and fuel strategy in the next rolling step.

[0049] Working principle: This method operates under a multi-timescale architecture consisting of hourly, 15-minute, and minute layers. The hourly layer uses two-stage stochastic programming with scenario trees to generate the daily baseline energy plan. The 15-minute layer uses a rolling model to predict and fine-tune the SOC trajectory and thermal management strategy of the lithium battery. The minute layer uses real-time frequency measurement as input to perform fast linear, analytical optimization, or closed-loop control to ensure that the transient frequency index meets the safety boundary. At the same time, the energy storage degradation cost and preheating and temperature maintenance energy consumption are internalized into the scheduling decision to balance economy and lifespan. At the minute level, the frequency transient response is linearized or approximated based on the equivalent pendulum equation, forming the upper bound of RoCoF, the lower bound of nadir, and the steady-state deviation constraint, which are then incorporated into the optimization problem. At the same time, frequency responsibility is allocated according to the time scale: supercapacitors and high-power energy storage are responsible for transient power injection, lithium batteries bear short-term energy and part of the frequency support, hydrogen energy systems bear hourly large-capacity energy shifting, and diesel engines and fuel cells perform long-term power maintenance. To adapt to the operating conditions, the system adopts an adaptive virtual inertia and droop allocation mechanism, dynamically adjusting the inverter inertia contribution and droop coefficient according to the available capacity and temperature. The scheduling uses hierarchical decomposition to accelerate parallel computing: the hourly layer generates cut and multiplier information as the main problem, and the 15-minute and minute layers are used as sub-problems to perform local feasibility checks and fine-tuning and send correction signals back to the main problem; during online implementation, the system continuously monitors temperature and frequency indicators. Low temperature triggers preheating decisions and is included in the target. If minute-level optimization is not feasible or multiple failures occur, an emergency strategy is triggered to ensure frequency safety and gradually roll back to the economic operating point through rolling optimization after recovery.

[0050] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for optimal scheduling of polar island microgrids considering frequency constraints, characterized in that: The microgrid optimization scheduling method includes the following steps: Establish a system model, including (1) a power balance model and operating constraints at three time scales: hour, 15 minutes, and minute; (2) an equivalent pendulum frequency dynamic model, and associate RoCoF, the lowest frequency point nadir, the steady-state frequency deviation, and the power imbalance; (3) a hybrid energy storage dynamic decay model, which includes battery cycle decay and temperature-related decay terms. Construct and solve a multi-timescale joint optimization architecture: the hourly layer uses two-stage stochastic programming with scenario trees to generate the daily baseline plan; the 15-minute layer uses rolling model predictive control (MPC) to correct and fine-tune within the rolling window; the minute layer uses fast linear, quadratic programming, or analytical control to achieve transient frequency constraints and fast power injection. Frequency transient constraints are embedded as minute-level decision constraints through linearization or convex approximation, and energy storage degradation costs and battery thermal management actions are internalized as scheduling decision variables into the objective function. Based on the solution results, hierarchical instructions are issued to various distributed resources and online monitoring and security measures are implemented.

2. The optimal scheduling method for polar island microgrids considering frequency constraints according to claim 1, characterized in that: The energy storage dynamic decay model specifically includes: Cyclic decay function is used to represent the capacity loss related to the number of charge-discharge cycles and the depth of discharge; Temperature-dependent decay function is used to represent the effect of ambient temperature and thermal management power on capacity decay; Furthermore, the equivalent degradation cost is incorporated into the objective functions of the 15-minute and hourly layers to balance frequency support and battery life.

3. The optimal scheduling method for polar islanded microgrids considering frequency constraints according to claim 1, characterized in that: The minute-level frequency constraint is established based on a linearized expression of the equivalent pendulum equation, including: RoCoF constraint, nadir constraint, steady-state deviation constraint; The linearized expression serves as an inequality constraint for minute-level optimization and is used for fast solution. Furthermore, a pre-computed conservative margin table is employed in the solution process to ensure real-time feasibility.

4. The optimal scheduling method for polar islanded microgrids considering frequency constraints according to claim 1, characterized in that: The 15-minute layer is a rolling MPC controller, specifically including: At each rolling step, the wind and solar load forecasts are updated, and the SOC target trajectory is dynamically adjusted based on the updated forecasts, battery temperature, and available capacity. The charging and discharging power, SOC trajectory, and battery thermal management of lithium batteries are used as decision variables, and a dynamically updated decay cost term is added to the objective function to limit high DOD and low temperature operation. Consistent constraint is applied to ensure that the output of the 15-minute layer remains consistent with the baseline of the hourly layer.

5. The optimal scheduling method for polar islanded microgrids considering frequency constraints according to claim 1, characterized in that: The hourly layer employs a two-stage stochastic programming approach to handle the uncertainty of renewable power output, specifically including: Construct a scenario tree containing several scenarios to represent the uncertainty of the scenery over 24 hours; The first phase determines the baseline output and hydrogen production and release plans. The second phase calculates the replenishment actions and expected replenishment costs for each scenario and incorporates the expected replenishment costs into the overall hourly target. The hourly targets include: fuel diesel costs, hydrogen operating costs, penalties for curtailment of electricity and hydrogen, and long-term lifetime guidance.

6. The optimal scheduling method for polar islanded microgrids considering frequency constraints according to claim 1, characterized in that: The division of responsibilities between hybrid energy storage and power sources over time includes: Hydrogen energy systems handle large-scale energy balancing and energy shifting on an hourly scale. Lithium-ion batteries provide energy buffering for 15 minutes and short-term frequency modulation support. Supercapacitors can handle high-frequency power injection on a timescale of minutes or less to cope with power disturbances; Furthermore, the optimized scheduling method includes an adaptive virtual inertia droop allocation strategy, which dynamically adjusts the inverter's virtual inertia and droop coefficient based on online estimated available energy storage capacity and temperature conditions, in order to balance frequency performance and battery life consumption under different operating conditions.

7. The optimal scheduling method for polar islanded microgrids considering frequency constraints according to claim 1, characterized in that: The hierarchical solution employs a decomposition coordination strategy to improve computational efficiency. The decomposition coordination strategy is either Benders decomposition or the alternating method multiplier method (ADMM), where: The hourly layer serves as the primary problem, generating cut sets or Lagrange multiplier information. The 15-minute layer and the minute layer are treated as parallel subproblems, solved within their respective time scales, and return feasible cuts or price difference information to the main problem to gradually converge. During parallel solution, real-time executability constraints are preserved and conservative margins are applied to key frequency constraints to determine their feasibility.

8. The optimal scheduling method for polar islanded microgrids considering frequency constraints according to claim 1, characterized in that: The thermal management and cryogenic protection include: The battery surface temperature is monitored online. When the battery surface temperature is lower than a preset threshold, the battery preheating power is introduced as a controllable decision variable into the optimization problem. The preheating power consumption is included in the objective functions of the 15-minute and hourly layers, and is linked with the decay function to calculate the equivalent lifetime loss caused by low-temperature operation; Within the optimization objective, lifetime loss is monetized in the form of equivalent energy unit price, thereby automatically balancing preheating energy consumption and accelerated decay during scheduling.

9. The optimal scheduling method for polar islanded microgrids considering frequency constraints according to claim 1, characterized in that: The online security measures include: Real-time frequency and power imbalance monitoring; if the frequency is found to be approaching the warning threshold, a rapid response at the minute level is triggered and the virtual inertia and droop distribution are adjusted. Set up an emergency disconnection strategy. When multiple failures or extreme weather make dispatching infeasible, implement minimum load cut-off and start / stop standby diesel engines and fuel cells according to priority. A pre-computed frequency margin conservative table is used at the minute level to ensure the feasibility and safety boundaries of real-time control.

10. A scheduling control device for a polar island microgrid optimization scheduling method considering frequency constraints according to any one of claims 1-9, characterized in that: The scheduling and control device includes: The data acquisition unit is used to collect load, wind and solar power output, energy storage SOC, energy storage surface temperature, inverter status and frequency signals in real time across multiple time scales. The hourly scheduling module is used to solve hourly stochastic programming with two-stage decision-making based on scenario trees and output a baseline plan; The 15-minute layer MPC module is used to solve MPC problems with decay costs and thermal management decisions within a rolling window and output the SOC trajectory and preheating command. The minute-level fast frequency control module is used to quickly solve and output instantaneous power commands, virtual inertia and droop adjustment for supercapacitors and inverters based on linearized frequency constraints. The coordination communication unit is used to exchange cut sets, Lagrange multipliers, and baseline correction information between modules and to issue control commands to the field execution equipment; The scheduling control device implements the optimized scheduling method in the form of computer-executable instructions.

Images