Secondary frequency modulation optimization control method, device and equipment under low-load working condition and medium

Abstract

The invention discloses a secondary frequency modulation optimization control method under a low-load working condition, relates to the technical field of power grid resource optimization, and is used for solving the problem that a control strategy under the low-load working condition is not fully considered in the prior art. The current load interval is recognized, and control parameter correction is conducted on the unit located in the low load interval; calculating a total output power reference value; carrying out calculation by adopting a self-adaptive optimization algorithm, and carrying out power fine distribution to obtain an optimal charging and discharging instruction of each energy storage unit; and the output power is adjusted in real time through double-closed-loop control. The invention further discloses a secondary frequency modulation optimization control device under the low-load working condition, electronic equipment and a computer storage medium. By dynamically correcting the control parameters of the low-load thermal power generating unit and cooperatively distributing the hybrid energy storage power in combination with the adaptive optimization algorithm, the safety and economical efficiency of the energy storage system are guaranteed.

Description

Technical Field

[0001] This invention relates to the field of power grid resource optimization technology, and in particular to a method, device, equipment and medium for coordinated secondary frequency regulation optimization control of thermal power and hybrid energy storage for low load operation conditions. Background Technology

[0002] With the large-scale grid connection of new energy sources such as wind power and photovoltaics, the load volatility of the power system has intensified, easily leading to frequency instability. Especially when participating in deep peak shaving, thermal power units often operate under low load conditions. At this time, the operating characteristics of the main equipment such as boilers and turbines change significantly, manifesting as increased regulation inertia, decreased control gain, and severely deteriorated output regulation capability. Long-term operation under these conditions not only exacerbates equipment losses but also makes it difficult to meet the grid's requirements for the speed and accuracy of secondary frequency regulation. To compensate for the frequency regulation deficiencies of thermal power units, electrochemical energy storage systems, due to their millisecond-level response speed and flexible bidirectional power regulation capabilities, are widely considered ideal auxiliary frequency regulation resources. In recent years, "thermal power-energy storage" coordinated frequency regulation has become a research hotspot in the industry. Existing technical solutions mainly focus on: 1) adopting a two-layer optimization architecture, with the upper layer performing economic dispatch and the lower layer performing real-time power allocation; 2) designing output strategies for a single type of energy storage to participate in primary or secondary frequency regulation.

[0003] However, most existing methods focus on the output distribution of primary frequency regulation or single energy storage participating in secondary frequency regulation, and have not fully considered the impact of thermal power unit response lag, low economic efficiency and the SOC state of individual energy storage units on control strategies under low load conditions.

[0004] Therefore, there is an urgent need for a secondary frequency regulation optimization control method specifically designed for low-load operation of thermal power units, which can deeply integrate the dynamic characteristics correction of thermal power units and collaboratively optimize the internal SOC state of hybrid energy storage systems, so as to comprehensively improve the frequency regulation performance, operating economy and equipment safety of the system. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, one of the objectives of this invention is to provide a secondary frequency regulation optimization control method under low load conditions. This method dynamically corrects the control parameters of low-load thermal power units and combines them with an adaptive optimization algorithm to coordinate the allocation of hybrid energy storage power, thereby improving the accuracy and speed of secondary frequency regulation while ensuring the safety and economy of the energy storage system.

[0006] One of the objectives of this invention is achieved through the following technical solution: A secondary frequency modulation optimization control method under low load conditions includes the following steps: Acquire power grid operation data and calculate the secondary frequency regulation power that needs to be compensated based on the power grid operation data; Identify the current load range of thermal power units in the power grid and correct the control parameters of units located in the low load range; With the goal of minimizing the regional control error and minimizing the operating cost of thermal power units, a reference value for total output power is calculated. Based on the total output power reference value, an adaptive optimization algorithm is used to calculate and perform fine-grained power allocation to obtain the optimal charging and discharging commands for each energy storage unit. The output power is adjusted in real time through dual closed-loop control according to the optimal charge and discharge command.

[0007] Further, acquiring power grid operation data and calculating the secondary frequency regulation power that currently needs compensation based on the power grid operation data includes: Obtain the actual frequency of the power grid, frequency deviation, current active power output of each thermal power unit, and state of charge of each unit of the energy storage system; Calculate the secondary frequency modulation power that needs to be compensated based on the frequency deviation. Its calculation satisfies: ,in, and These are the proportional and integral gains, respectively. Indicates frequency deviation; when A positive value indicates that power generation output needs to be increased. A negative value indicates that power generation output needs to be reduced.

[0008] Furthermore, identifying the current load range of thermal power units in the power grid includes: using real-time data from thermal power units to calculate the current load rate. ,satisfy: ,in, For the real-time output power of the unit, When the rated maximum power of the unit is... When the load is less than 40%, it is considered a low-load condition.

[0009] Furthermore, control parameters are modified for units located in the low load range, including introducing dynamic correction weights into the cost function and introducing load state corrections into the power ramp-up rate constraint. The cost function after introducing dynamically adjusted weights is as follows: , Indicates the corrected number The cost function of the unit, As a performance correction factor, For the first The standard cost function for a Taiwanese generator set; The power ramp rate constraint is satisfied: ,in, This is the revised upper limit of the climb. For the first The current frequency regulation power of the Taiwanese unit, Indicates the current time.

[0010] Furthermore, with the optimization objective of minimizing the control error in the area with the lowest operating cost of thermal power units, the reference value for total output power is calculated, including: A global optimization model is constructed with the joint objectives of minimizing the total system operating cost and optimizing frequency stability. Based on the secondary frequency regulation power and the operating constraints of the thermal power units and energy storage systems, the global optimization model is solved to output the total reference power allocated to the thermal power units and the total reference power allocated to the energy storage group.

[0011] Furthermore, the computation function of the global optimization model satisfies: , in, For the first The current frequency regulation power of the Taiwanese unit, and Indicating the first in the energy storage system The charging and discharging power of the power station This represents the total cost objective function. Indicates the area control error. for ACE Penalty coefficient, Indicates the first The standard cost function of the Taiwanese unit, Indicates the total number of generating units; The calculation function and constraints of the global optimization model are input into the convex optimization solver to solve the problem, and the output is the total reference power allocated to the thermal power unit and the total reference power allocated to the energy storage group.

[0012] Furthermore, based on the total output power reference value, an adaptive optimization algorithm is used to calculate and perform fine-grained power allocation to obtain the optimal charging and discharging commands for each energy storage unit, including: Based on the rated capacity of lithium iron phosphate batteries and lithium iron titanate batteries and the preset response weights, the initial expected power of each battery cell is calculated; wherein, the response weights are configured to give lithium iron titanate batteries with better dynamic response capabilities a higher allocation ratio. With the common optimization objectives of minimizing the tracking error between the actual output power of each battery cell and the desired power, and minimizing the imbalance of the state of charge (SOC) of all battery cells, an allocation objective function is constructed. The ADAM adaptive optimization algorithm is used to iteratively solve the allocation objective function. The ADAM adaptive optimization algorithm maintains the first and second moments of the historical gradient and corrects for their deviations to adaptively adjust the learning step size of the power command for each energy storage unit. Under the constraint of total power conservation, it quickly converges to obtain the optimal charge and discharge power command for each battery unit. During the iteration process and before the final command is output, the power commands of all battery cells are subject to upper and lower limit truncation to ensure that they do not exceed the maximum allowable charge and discharge power.

[0013] The second objective of this invention is to provide a secondary frequency modulation optimization control device under low load conditions.

[0014] The second objective of this invention is achieved by the following technical solution: A secondary frequency modulation optimization control device for low-load conditions, comprising: The power calculation module is used to acquire power grid operation data and calculate the secondary frequency regulation power that needs to be compensated based on the power grid operation data. The reference value calculation module is used to identify the current load range of thermal power units in the power grid and to correct the control parameters of units located in the low load range; with the optimization objective of minimizing the control error in the area with the lowest operating cost of thermal power units, the total output power reference value is calculated. The adjustment module is used to calculate and perform fine-grained power allocation based on the total output power reference value using an adaptive optimization algorithm to obtain the optimal charging and discharging command for each energy storage unit; and to adjust the output power in real time through dual closed-loop control according to the optimal charging and discharging command.

[0015] A third objective of this invention is to provide an electronic device that performs one of the objectives of the invention, comprising a processor, a storage medium, and a computer program, wherein the computer program is stored in the storage medium, and when the computer program is executed by the processor, it implements the above-mentioned secondary frequency modulation optimization control method under low load conditions.

[0016] A fourth objective of this invention is to provide a computer-readable storage medium storing one of the objectives of the invention, wherein a computer program is stored thereon, and when the computer program is executed by a processor, it implements the above-described secondary frequency modulation optimization control method under low load conditions.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention effectively compensates for the nonlinear characteristics of thermal power units, such as decreased gain and increased inertia under low load, by introducing a load range identification and dynamic parameter correction mechanism. This fundamentally reduces the regulation delay and overshoot problems caused by unit response lag. By utilizing the millisecond-level fast response capability of the hybrid energy storage system, it makes up for the shortcomings of thermal power units, enabling the system to quickly track frequency regulation commands, thereby significantly improving the speed and accuracy of frequency regulation and reducing regional control errors. Furthermore, this application optimizes the operating cost of thermal power units and frequency deviation penalties within a unified framework, automatically seeking the optimal balance between economy and stability. This avoids over-reliance on high-cost unit frequency regulation and reduces overall frequency regulation costs. By promoting the balance of the SOC of energy storage units, it prevents some batteries from prematurely retiring due to overcharging or over-discharging, extending the overall lifespan of the energy storage system and improving the economy throughout its entire life cycle. The dynamic correction mechanism automatically reduces the frequency regulation task allocation of low-load units, alleviating the operating pressure of the units and helping to improve their operational safety and extend equipment life. By setting constraints such as SOC, charging and discharging power, and ramp rate, and performing boundary processing in the algorithm, it ensures that the energy storage system always operates within the safe range, preventing risks such as overcharging and over-discharging, and greatly improving the reliability of the system. This invention proposes a response weight-based allocation strategy, which fully leverages the respective advantages of lithium iron titanate batteries (fast response) and lithium iron phosphate batteries (large capacity), and realizes intelligent division of labor for frequency modulation tasks. This invention also employs the ADAM adaptive optimization algorithm for power allocation, which not only has a fast convergence speed but also simultaneously optimizes two objectives: power tracking error and SOC balance, thereby achieving efficient and adaptive collaborative management within the hybrid energy storage system. Attached Figure Description

[0018] Figure 1 This is a system block diagram of the secondary frequency modulation optimization control method in Embodiment 1; Figure 2 This is a flowchart of the secondary frequency modulation optimization control method under low load conditions in Example 1; Figure 3 This is a structural block diagram of the secondary frequency modulation optimization control device under low load conditions in Embodiment 2; Figure 4 This is a structural block diagram of the electronic device in Embodiment 3. Detailed Implementation

[0019] The present invention will now be described in more detail with reference to the accompanying drawings. It should be noted that the following description of the present invention with reference to the accompanying drawings is merely illustrative and not restrictive. Various embodiments can be combined with each other to form other embodiments not shown in the following description.

[0020] Example 1 Example 1 provides a secondary frequency regulation optimization control method under low load conditions, which aims to solve the problem of low load frequency regulation of thermal power units, significantly improve the speed and accuracy of grid frequency regulation, and at the same time ensure the safety of energy storage system and reduce frequency regulation costs.

[0021] This embodiment proposes a secondary frequency regulation optimization control method specifically designed for low-load operation conditions, taking into account the SOC state of the lithium iron phosphate battery-lithium iron titanate hybrid battery energy storage. By fully considering the response lag and output constraints of thermal power units, and combining the real-time SOC state of the hybrid energy storage unit, the method comprehensively balances regulation capabilities and state of charge, dynamically adjusting the power allocation ratio and control parameters of each resource during the frequency regulation process. This effectively improves the secondary frequency regulation accuracy, response speed, and frequency stability of the system during low-load operation, while ensuring the safe and efficient operation of the energy storage system, extending its service life, and meeting the grid's high reliability requirements for flexible frequency regulation resources.

[0022] Please refer to Figure 1 The diagram shown is a system block diagram of the coordinated secondary frequency regulation optimization control method for thermal power and hybrid energy storage under low-load operating conditions in this embodiment. This embodiment achieves coordinated power allocation between thermal power and hybrid energy storage by dynamically identifying and correcting the control parameters of low-load thermal power units and combining upper-level global optimization and lower-level adaptive algorithms.

[0023] Please refer to Figure 2 As shown, this embodiment provides a secondary frequency modulation optimization control method under low load conditions, including the following steps: S1. Obtain power grid operation data and calculate the secondary frequency regulation power that needs to be compensated based on the power grid operation data; S1 includes: Obtain the actual frequency of the power grid, frequency deviation, current active power output of each thermal power unit, and state of charge of each unit of the energy storage system; Calculate the secondary frequency modulation power that needs to be compensated based on the frequency deviation. Its calculation satisfies: ,in, and These are proportional and integral gains, respectively. This control algorithm can quickly convert the frequency deviation signal into a total power command that needs compensation, generating a secondary frequency-modulated power command. A positive value indicates that power generation output needs to be increased, while a negative value indicates that power generation output needs to be reduced. This power command serves as the overall target for subsequent allocations. Indicates frequency deviation; when A positive value indicates that power generation output needs to be increased. A negative value indicates that power generation output needs to be reduced. It should be noted that the above... and This refers to the control gain of the dispatch-side AGC (within the EMS), which is set by the dispatch center based on system inertia, load fluctuation characteristics, etc. It can be set according to actual conditions and belongs to engineering tuning parameters.

[0024] Specifically, system status data acquisition is performed by the Energy Management System (EMS) of the power grid dispatch center, which collects power grid operation data in real time through Remote Terminal Units (RTUs) and Supervisory Control and Data Acquisition (SCADA) systems. The collected information includes: the actual frequency of the power grid. and its deviation ,in , At the rated frequency, the current active power output of each thermal power unit And the state of charge (SOC) of each unit in the energy storage system.

[0025] S2. Identify the current load range of thermal power units in the power grid and correct the control parameters of units located in the low load range; Load range identification in S2 includes: using real-time data from the thermal power unit computer to determine the current load rate of the unit. ,satisfy: ,in, For the real-time output power of the unit, When the rated maximum power of the unit is... When the load is less than 40%, it is considered a low-load condition.

[0026] The current load factor can be calculated by obtaining real-time data from the Electronic Management System (EMS). Low-load operation of thermal power units is typically defined as: For some ultra-supercritical units or under deep peak-shaving conditions, the low-load range can be extended to: This invention takes a load rate of less than 40% as a low-load condition. Under this condition, the heat transfer inertia of the boiler, turbine and reheat system of the thermal power unit is significantly enhanced, and there is a significant time delay between the regulation command and the actual power change. The control system exhibits gain nonlinearity and increased inertia under low load, which leads to an increase in the probability of regulation overshoot.

[0027] After identifying a low-load area, the control parameter correction mechanism is automatically triggered, including introducing dynamic correction weights into the cost function and introducing load state corrections into the power ramp-up rate constraint. The cost function after introducing dynamically adjusted weights is as follows: , Indicates the corrected number The cost function of the unit, As a performance correction factor, For the first The standard cost function for a Taiwanese generator set; The power ramp rate constraint is satisfied: ,in, This is the revised upper limit of the climb. For the first The current frequency regulation power of the Taiwanese unit, Indicates the current time.

[0028] The first of the above cost functions The standard cost function for the Taiwanese unit is: ,in, These are the cost coefficients for the unit, all of which are known quantities; For the first The current frequency regulation power of the generator unit is a variable to be optimized. When the unit is identified as being under low load, the system dynamically adjusts the cost weighting coefficient based on real-time correction parameters and introduces a performance correction factor. Defined as: , in, and These are the standard time constant and standard gain (known quantities) of the unit under nominal load, respectively. and The correction values ​​obtained for low-load identification are all known quantities. Performance correction factor An overall value greater than 1 indicates that the unit responds slowly under low load conditions, requiring the system to increase its frequency regulation allocation cost. Therefore, the cost function after introducing dynamically adjusted weights is: Therefore, the effect of low-load conditions on the economic efficiency of unit frequency regulation is directly reflected in the global optimization objective function.

[0029] Under this mechanism, even under the same total power demand, low-load units will obtain a lower power allocation ratio in the optimization results due to the increased weight of the modified cost function, thereby avoiding frequency fluctuations and overshooting in dynamics.

[0030] In addition, to reflect the characteristic of increased output inertia under low load, a modified ramp upper limit is defined: , This is a proportionality coefficient (usually <1). Under low load conditions, due to the increase in time constant and decrease in gain, the correction coefficient... A value less than 1 further compresses the ramp rate constraint, reducing the magnitude of output changes. This correction allows the optimization model to simultaneously reflect the response lag characteristics and inertia constraints of low-load units at the power dynamic constraint level.

[0031] S3. Calculate the reference value of total output power with the optimization objective of minimizing the control error in the area with the lowest operating cost of thermal power units; Step S3 first constructs an optimization objective function for frequency regulation jointly participated in by thermal power units and hybrid energy storage systems (including lithium iron phosphate batteries and lithium iron titanate batteries), taking into account both system operating costs and frequency control error (ACE). The corresponding operating cost function has already been defined previously.

[0032] Specifically, with the optimization objective of minimizing the control error in the area with the lowest operating cost of thermal power units, the reference value for total output power is calculated, including: A global optimization model is constructed with the joint objectives of minimizing the total system operating cost and optimizing frequency stability. Based on the secondary frequency regulation power and the operating constraints of the thermal power units and energy storage systems, the global optimization model is solved to output the total reference power allocated to the thermal power units and the total reference power allocated to the energy storage group.

[0033] The first in the energy storage system The discharge power and charging power of the power station are expressed as follows: and : .

[0034] To measure the system's frequency regulation performance, the range control error (ACE) is defined as follows: , in The current predicted load demand (MW) is a known quantity, and ACE is the current power balance error, used to reflect the magnitude of the frequency deviation.

[0035] In summary, the computation function of the global optimization model satisfies: , in, For the first The current frequency regulation power of the Taiwanese unit, and Indicating the first in the energy storage system The charging and discharging power of the power station This represents the total cost objective function. Indicates the area control error. for ACE The penalty factor is used to adjust the trade-off between cost and frequency recovery. Indicates the first The standard cost function of the Taiwanese unit, This represents the total number of generating units. The first part of this objective function is the sum of the operating costs of all generating units, and the second part is the squared penalty term for the frequency error. The larger the ACE value, the worse the frequency regulation performance.

[0036] The calculation function and constraints of the global optimization model are input into the convex optimization solver to solve the problem, and the output is the total reference power allocated to the thermal power unit and the total reference power allocated to the energy storage group.

[0037] The model should meet the physical and safety operation constraints of the power system, including: , , , , , , In addition, it also includes the power ramping constraint in the aforementioned steps, where The total power command for secondary frequency regulation obtained in the preceding steps represents the total output of thermal power units and the total output of energy storage on the left side of the equation, both of which are variables to be optimized. These represent the minimum and maximum output power of the unit, respectively (MW). For the first Maximum charging power and maximum discharging power of the energy storage unit; , These are the SOC of energy storage for the current cycle and the next cycle, respectively. These are the charging and discharging efficiencies (constants between 0 and 1), respectively. To control the cycle time (hours). This represents the total capacity of the energy storage unit (MWh).

[0038] Since the above optimization model contains both a quadratic cost function and non-convex constraints such as mutual exclusion between energy storage charging and discharging, convex relaxation techniques are needed to make the model solvable. Therefore, the following measures are taken: Linearization of the quadratic cost function: The quadratic term in the power generation cost is relaxed using piecewise second-order cone relaxation, transforming it into a second-order cone constraint in a convex optimization problem. This is achieved by introducing auxiliary variables. and define constraints: , At this point, the objective function term will be... Replace it with a second-order cone relaxation to maintain the convexity of the optimization problem: .

[0039] Relaxation of charge-discharge mutual exclusion: Ideally, energy storage units should satisfy mutual exclusion conditions.

[0040] This constraint is a non-convex product constraint, which cannot be handled directly using conventional methods. Therefore, to relax this constraint, a penalty term can be introduced into the objective function:

[0041] in The positive weighting coefficient is large enough to suppress the simultaneous non-zero values ​​of both.

[0042] Handling the ACE squared term: The ACE squared term in the objective function is a convex function, but to convert it into a linear constraint form, we can introduce a new variable 𝑧 and replace ACE with z.

[0043] Through the above steps, the final optimization problem form is as follows: , The objective function and constraints are input into the convex optimization solver, which automatically searches for the global optimum. During the solution process, the optimum automatically balances minimum cost and minimum ACE (objective weights are set to balance these). The output includes: reference power for each thermal power unit. and the charging and discharging reference power of the j-th energy storage power station. The obtained reference value of output power of the hybrid energy storage group in the energy storage power station This will serve as the target value for the next step in the allocation of hybrid energy storage power.

[0044] S4. Based on the total output power reference value, an adaptive optimization algorithm is used to calculate and perform fine-grained power allocation to obtain the optimal charging and discharging commands for each energy storage unit. S4 includes: Based on the rated capacity of lithium iron phosphate batteries and lithium iron titanate batteries and the preset response weights, the initial expected power of each battery cell is calculated; wherein, the response weights are configured to give lithium iron titanate batteries with better dynamic response capabilities a higher allocation ratio. With the common optimization objectives of minimizing the tracking error between the actual output power of each battery cell and the desired power, and minimizing the imbalance of the state of charge (SOC) of all battery cells, an allocation objective function is constructed. The ADAM adaptive optimization algorithm is used to iteratively solve the allocation objective function. The ADAM adaptive optimization algorithm maintains the first and second moments of the historical gradient and corrects for their deviations to adaptively adjust the learning step size of the power command for each energy storage unit. Under the constraint of total power conservation, it quickly converges to obtain the optimal charge and discharge power command for each battery unit. During the iteration process and before the final command is output, the power commands of all battery cells are subject to upper and lower limit truncation to ensure that they do not exceed the maximum allowable charge and discharge power.

[0045] Specifically, the upper-level power command calculates and outputs a reference value for the total energy storage power. Building upon this foundation, the total power target is further rationally allocated to each energy storage sub-unit, including lithium iron phosphate (LFP) and lithium iron titanate (LTO) batteries. This allocation scheme fully considers the differences in response speed among different batteries. LTO batteries have a faster dynamic response capability for power regulation and are prioritized for high-frequency and large-amplitude frequency regulation requirements; LFP batteries have a relatively slower response and are more suitable for smoothing power output and maintaining long-term power support. By dynamically adjusting the power allocation weights, tracking error is minimized and SOC is balanced.

[0046] The expected power is determined by both capacity and response weight, and is defined by the following formula: , , in No. The expected power distribution (MW) of each lithium iron phosphate battery. No. The expected power distribution (MW) of each lithium iron titanate battery. This corresponds to the battery's rated capacity (MWh). The response weight of lithium iron phosphate batteries (smaller values) indicates that their dynamic response capability is slow. A higher response weight (value) for lithium iron phosphate batteries indicates a faster dynamic response. This formula ensures that lithium iron phosphate batteries achieve a higher power distribution ratio for the same capacity. Indicates the first The rated capacity of each lithium iron phosphate battery cell, Indicates the first The rated capacity of each lithium iron titanate battery cell.

[0047] Define the average current SOC of all batteries: , in The average state of charge (%) of all batteries is used to measure the degree of balance. For the first SOC (%) of each LFP battery For the first SOC (%) of each LTO battery This indicates the total number of LFP and LTO batteries.

[0048] Comprehensive objective function It can be written as:

[0049] in , To minimize power tracking error, efforts are made to ensure that the power of each sub-unit is close to the desired value. The SOC equalization weight coefficients are known quantities. The objective function consists of two parts: power tracking error (the first two terms) and SOC equalization error (…). , This objective aims to ensure that the total system output is as close as possible to the upper-level instructions, while simultaneously encouraging each battery's SOC to move towards a more balanced state.

[0050] To ensure the power target is met, the power conservation constraint must also be satisfied:

[0051] That is, the output of all batteries must meet the reference value of total energy storage power obtained in step 3, where, This indicates the actual reference output power of the lithium iron phosphate battery cell during the current control cycle. This indicates the actual reference output power of the lithium iron phosphate battery cell during the current control cycle.

[0052] After determining the objective function and constraints, the ADAM adaptive optimization algorithm is used to iteratively solve it. The ADAM algorithm features an adaptive learning rate and momentum term, which can accelerate convergence and reduce oscillations. The specific iterative steps are as follows: Initialization: Set the initial values ​​for iteration. and Set as reference output for each unit and and the first-order momentum vector Second-order momentum vector Set the learning rate First-order moment decay rate Second-order moment decay rate and small constants Used for numerical stability.

[0053] Gradient calculation: in the first... In this iteration, the gradient of the objective function with respect to the output power of each sub-unit is first calculated, where... For example, its gradient can be expressed as:

[0054] This involves considering the relationship between total power and... Consistency As the penalty coefficient, for Similar to.

[0055] Moment Update: Update the first-order momentum estimate using an exponentially weighted method:

[0056] here and Dimensions and decision variables same, It is the first-order decay rate, where the gradient vector It contains the partial derivatives of all subunits. This momentum preserves the historical set of first-order momentum, which is equivalent to gradient descent of the momentum.

[0057] Second moment update: Similarly, update the second moment estimate:

[0058] in Indicates the square of each component. It is the second-order decay rate. The squared gradient information is preserved for adaptive step size adjustment.

[0059] Bias Correction: Since the initial momentum is zero, to eliminate the bias in momentum estimation from the previous iterations, corrected first and second moments are introduced:

[0060] After correction, unbiased first and second moment estimates can be obtained.

[0061] Power is updated based on the corrected momentum:

[0062] in The learning rate, under low load conditions, is adjusted as described in step 1. For a minimal constant, update the pair and Simultaneously, the power of each sub-unit is corrected based on the momentum and gradient of the corresponding dimension.

[0063] Boundary constraint handling: After the update, saturation processing is performed on the output power of each unit to ensure it is within the allowable range. Upper and lower limits are truncated for the power of each energy storage unit to ensure that technical limits are met.

[0064] The algorithm stops after reaching the set number of iterations. Because the ADAM algorithm has a fast convergence characteristic, it usually only takes tens to hundreds of iterations to achieve a stable allocation.

[0065] The ADAM iterative process described above gradually adjusts the output of each hybrid energy storage sub-unit towards an optimal allocation. During the optimization process, the objective function simultaneously reduces power point tracking error and State of Charge (SOC) imbalance, ultimately achieving power convergence of the entire system to the upper-level objective. It also maintains a consistent State of Charge (SOC) across all sub-units. By trunculating the output after each iteration, the charging and discharging power is consistently kept within technical limits. Furthermore, it emphasizes the high response characteristics of lithium iron phosphate batteries, automatically allocating more dynamic power while lithium iron phosphate handles the smoothing portion, ensuring rapid system response and long energy storage life.

[0066] S5. Based on the optimal charge / discharge command, adjust the output power in real time through dual closed-loop control.

[0067] The S5, through the Power Conversion System (PCS) of the energy storage system, adjusts the output power in real time via dual closed-loop control based on the reference power commands for each energy storage unit calculated by the lower-level optimization allocation, ensuring consistency with the frequency regulation target. Specifically, the PCS first receives the target power value of each energy storage unit and compares it with the actual output power measured in real time to calculate the power error. Subsequently, the outer-loop power controller converts the power error into an active current command, and the inner-loop current controller further adjusts the inverter's PWM duty cycle to ensure that the active power between the DC bus and AC bus tracks the reference value in real time. Simultaneously, the PCS judges the feasibility of the command based on the SOC, voltage, temperature, and other status information of the energy storage battery. When the power command exceeds the allowable range or there is a risk of overcharging or over-discharging, it automatically limits or adjusts the output to ensure safe operation.

[0068] In summary, the method in this embodiment unifies the power allocation problem between energy storage systems and thermal power units within the same optimization framework by establishing a convex relaxation optimization model that includes the operating cost of thermal power units and frequency control errors. This model utilizes methods such as thermal power unit cost function correction, dynamic ramping constraints, and ACE weighting to achieve a globally optimal solution for power allocation. It can automatically generate economically optimal and safe power commands based on real-time system load and unit status. Compared to existing fixed-ratio or hierarchical experience-based allocation schemes, this method significantly improves allocation accuracy, reduces frequency regulation costs, and enhances system response coordination. The ADAM adaptive optimization algorithm is employed to iteratively allocate the power output of lithium iron phosphate (LFP) and lithium iron titanate (LTI) battery energy storage units. Taking into full account the differences in dynamic response performance between different battery chemistry systems, a weighted strategy for the response characteristics of different batteries is introduced into the power allocation objective function. During the iteration process, the ADAM algorithm automatically allocates power ratios based on the instantaneous state of charge (SOC), capacity, and response time constant of the batteries. It prioritizes assigning fast frequency regulation tasks to LTI batteries and medium-speed and long-term frequency regulation tasks to LFP batteries, thereby matching the optimal operating range of each energy storage unit. This achieves dynamic optimization of energy storage output between power point tracking accuracy and SOC balance, significantly improving the fast response capability of frequency regulation, shortening the allocation convergence time, and reducing frequency deviation caused by response delay. Under mixed battery operation conditions, it achieves significant advantages such as tiered response speed, intelligent power allocation, and extended energy storage life. Real-time calculation of the thermal power unit load rate identifies whether the unit is in a low-load range and automatically generates gain correction coefficients and time constant correction coefficients for dynamically adjusting the unit cost weight and ramp rate upper limit. This mechanism takes effect in real time in the optimization model, automatically reducing the frequency regulation tasks of low-load units, thereby significantly reducing the frequency regulation performance degradation caused by lag or over-regulation under low-load conditions, improving the operational stability of thermal power units, and expanding the available load range for secondary frequency regulation.

[0069] Example 2 Example 2 discloses a device corresponding to the low-load condition secondary frequency modulation optimization control method of the above embodiments. It is a virtual device structure of the above embodiments. Please refer to... Figure 3 As shown, it includes: The power calculation module 310 is used to acquire power grid operation data and calculate the secondary frequency regulation power that needs to be compensated based on the power grid operation data. The reference value calculation module 320 is used to identify the current load range of thermal power units in the power grid and to correct the control parameters of units located in the low load range; with the optimization objective of minimizing the control error in the area with the lowest operating cost of thermal power units, it calculates the reference value of total output power. The adjustment module 330 is used to calculate and perform fine-grained power allocation based on the total output power reference value using an adaptive optimization algorithm to obtain the optimal charging and discharging command for each energy storage unit; and to adjust the output power in real time through dual closed-loop control according to the optimal charging and discharging command.

[0070] Example 3 Figure 4 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention, as shown below. Figure 4 As shown, the electronic device includes a processor 410, a memory 420, an input device 430, and an output device 440; the number of processors 410 in the computer device can be one or more. Figure 4 Taking a processor 410 as an example; the processor 410, memory 420, input device 430, and output device 440 in the electronic device can be connected via a bus or other means. Figure 4 Taking the example of a connection between China and Israel via a bus.

[0071] The memory 420, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the low-load operating condition secondary frequency modulation optimization control method in this embodiment of the invention. The processor 410 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory 420, thereby implementing the low-load operating condition secondary frequency modulation optimization control method of the above embodiment.

[0072] The memory 420 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, the memory 420 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 420 may further include memory remotely located relative to the processor 410, which can be connected to the electronic device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0073] Input device 430 can be used to receive input user identity information, power grid operation data, etc. Output device 440 may include display devices such as a display screen.

[0074] Example 4 Embodiment 4 of the present invention also provides a storage medium containing computer-executable instructions, which can be used by a computer to execute a secondary frequency modulation optimization control method under low-load conditions, the method comprising: Acquire power grid operation data and calculate the secondary frequency regulation power that needs to be compensated based on the power grid operation data; Identify the current load range of thermal power units in the power grid and correct the control parameters of units located in the low load range; With the goal of minimizing the regional control error and minimizing the operating cost of thermal power units, a reference value for total output power is calculated. Based on the total output power reference value, an adaptive optimization algorithm is used to calculate and perform fine-grained power allocation to obtain the optimal charging and discharging commands for each energy storage unit. The output power is adjusted in real time through dual closed-loop control according to the optimal charge and discharge command.

[0075] Of course, the computer-executable instructions provided in the embodiments of the present invention are not limited to the method operations described above, but can also execute related operations in the secondary frequency modulation optimization control method under low load conditions provided in any embodiment of the present invention.

[0076] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause an electronic device (which may be a mobile phone, personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0077] It is worth noting that in the embodiments of the above-mentioned device based on the secondary frequency modulation optimization control method under low load conditions, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.

[0078] For those skilled in the art, various other corresponding changes and modifications can be made based on the technical solutions and concepts described above, and all such changes and modifications should fall within the protection scope of the claims of this invention.

Claims

1. A secondary frequency modulation optimization control method under low load conditions, characterized in that, Includes the following steps: Acquire power grid operation data and calculate the secondary frequency regulation power that needs to be compensated based on the power grid operation data; Identify the current load range of thermal power units in the power grid and correct the control parameters of units located in the low load range; With the goal of minimizing the regional control error and minimizing the operating cost of thermal power units, a reference value for total output power is calculated. Based on the total output power reference value, an adaptive optimization algorithm is used to calculate and perform fine-grained power allocation to obtain the optimal charging and discharging commands for each energy storage unit. The output power is adjusted in real time through dual closed-loop control according to the optimal charge and discharge command.

2. The secondary frequency modulation optimization control method under low load conditions as described in claim 1, characterized in that, Acquire power grid operation data and calculate the secondary frequency regulation power that needs to be compensated based on the power grid operation data, including: Obtain the actual frequency of the power grid, frequency deviation, current active power output of each thermal power unit, and state of charge of each unit of the energy storage system; Calculate the secondary frequency modulation power that needs to be compensated based on the frequency deviation. Its calculation satisfies: ,in, and These are the proportional and integral gains, respectively. Indicates frequency deviation; when A positive value indicates that power generation output needs to be increased. A negative value indicates that power generation output needs to be reduced.

3. The secondary frequency modulation optimization control method under low load conditions as described in claim 1, characterized in that, Identifying the current load range of thermal power units in the power grid includes: using real-time data from the thermal power units to calculate the current load rate of the units. ,satisfy: ,in, For the real-time output power of the unit, When the rated maximum power of the unit is... When the load is less than 40%, it is considered a low-load condition.

4. The secondary frequency modulation optimization control method under low load conditions as described in claim 1 or 3, characterized in that, For units located in the low load range, control parameters are adjusted, including introducing dynamic adjustment weights into the cost function and introducing load state adjustments into the power ramp-up rate constraint. The cost function after introducing dynamically adjusted weights is as follows: , Indicates the corrected number The cost function of the unit, As a performance correction factor, For the first The standard cost function for a Taiwanese generator set; The power ramp rate constraint is satisfied: ,in, This is the revised upper limit of the climb. For the first The current frequency regulation power of the Taiwanese unit, Indicates the current time.

5. The secondary frequency modulation optimization control method under low load conditions as described in claim 1, characterized in that, With the optimization objective of minimizing the regional control error and minimizing the operating cost of thermal power units, the reference value for total output power is calculated, including: A global optimization model is constructed with the joint objectives of minimizing the total system operating cost and optimizing frequency stability. Based on the secondary frequency regulation power and the operating constraints of the thermal power units and energy storage systems, the global optimization model is solved to output the total reference power allocated to the thermal power units and the total reference power allocated to the energy storage group.

6. The secondary frequency modulation optimization control method under low load conditions as described in claim 5, characterized in that, The computation function of the global optimization model satisfies: , in, For the first The current frequency regulation power of the Taiwanese unit, and Indicating the first in the energy storage system The charging and discharging power of the power station This represents the total cost objective function. Indicates the area control error. for ACE Penalty coefficient, Indicates the first The standard cost function of the Taiwanese unit, Indicates the total number of generating units; The calculation function and constraints of the global optimization model are input into the convex optimization solver to solve the problem, and the output is the total reference power allocated to the thermal power unit and the total reference power allocated to the energy storage group.

7. The secondary frequency modulation optimization control method under low load conditions as described in claim 1 or 5, characterized in that, Based on the total output power reference value, an adaptive optimization algorithm is used to calculate and perform fine-grained power allocation to obtain the optimal charging and discharging commands for each energy storage unit, including: Based on the rated capacity of lithium iron phosphate batteries and lithium iron titanate batteries and the preset response weights, the initial expected power of each battery cell is calculated; wherein, the response weights are configured to give lithium iron titanate batteries with better dynamic response capabilities a higher allocation ratio. With the common optimization objectives of minimizing the tracking error between the actual output power of each battery cell and the desired power, and minimizing the imbalance of the state of charge (SOC) of all battery cells, an allocation objective function is constructed. The ADAM adaptive optimization algorithm is used to iteratively solve the allocation objective function. The ADAM adaptive optimization algorithm maintains the first and second moments of the historical gradient and corrects for their deviations to adaptively adjust the learning step size of the power command for each energy storage unit. Under the constraint of total power conservation, it quickly converges to obtain the optimal charge and discharge power command for each battery unit. During the iteration process and before the final command is output, the power commands of all battery cells are subject to upper and lower limit truncation to ensure that they do not exceed the maximum allowable charge and discharge power.

8. A secondary frequency modulation optimization control device for low-load conditions, characterized in that, It includes: The power calculation module is used to acquire power grid operation data and calculate the secondary frequency regulation power that needs to be compensated based on the power grid operation data. The reference value calculation module is used to identify the current load range of thermal power units in the power grid and to correct the control parameters of units located in the low load range; with the optimization objective of minimizing the control error in the area with the lowest operating cost of thermal power units, the total output power reference value is calculated. The adjustment module is used to calculate and perform fine-grained power allocation based on the total output power reference value using an adaptive optimization algorithm to obtain the optimal charging and discharging command for each energy storage unit; and to adjust the output power in real time through dual closed-loop control according to the optimal charging and discharging command.

9. An electronic device comprising a processor, a storage medium, and a computer program, wherein the computer program is stored in the storage medium, characterized in that, When the computer program is executed by the processor, it implements the secondary frequency modulation optimization control method under low load conditions as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the secondary frequency modulation optimization control method under low load conditions as described in any one of claims 1 to 7.

Images