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

Abstract

The application 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 the control strategy under the low-load working condition is not fully considered in the prior art, and comprises the following steps: calculating the secondary frequency modulation power that needs to be compensated according to power grid operation data; identifying the current load interval and correcting the control parameters of the units located in the low-load interval; calculating a total output power reference value; calculating the optimal charge-discharge instructions of each energy storage unit by using a self-adaptive optimization algorithm for fine power distribution; and adjusting the output power in real time through double closed-loop control. The application further discloses a secondary frequency modulation optimization control device under a low-load working condition, an electronic device and a computer storage medium. According to the application, the control parameters of low-load thermal power units are dynamically corrected, and the mixed energy storage power is cooperatively distributed by combining a self-adaptive optimization algorithm, so that the safety and economy of the energy storage system are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power grid resource optimization, and in particular to a thermal power and hybrid energy storage collaborative secondary frequency modulation optimization control method, device, equipment and medium for low-load operation conditions. BACKGROUND

[0002] With the large-scale grid connection of wind power, photovoltaic and other new energy sources, the load fluctuation of the power system is intensified, and the problem of frequency instability is prone to occur. Especially when participating in deep peak shaving, thermal power units are often in low-load operation conditions, at which time the operating characteristics of the main equipment such as the boiler and the steam turbine of the unit change significantly, showing that the regulation inertia increases, the control gain decreases, and the output regulation capability deteriorates seriously. Long-term operation in this condition not only aggravates the wear and tear of the unit equipment, but also makes it difficult to meet the requirements of the power grid for the rapidity and accuracy of secondary frequency modulation. In order to make up for the frequency modulation defects of thermal power units, electrochemical energy storage systems are widely considered as ideal auxiliary frequency modulation resources due to their millisecond-level response speed and flexible bidirectional power regulation capability. In recent years, "thermal power-energy storage" collaborative frequency modulation has become a research hotspot in the industry. Existing technical solutions are mostly focused on: 1) adopting a double-layer optimization architecture, with the upper layer performing economic scheduling and the lower layer performing real-time power distribution; 2) designing an output strategy for participating in primary or secondary frequency modulation for a single type of energy storage.

[0003] However, most existing methods focus on the output distribution of primary frequency modulation or single energy storage participating in secondary frequency modulation, and have not fully considered the influence of the response lag of thermal power units under low-load conditions, the low economy and the SOC state of single energy storage on the control strategy.

[0004] Therefore, there is an urgent need for a secondary frequency modulation optimization control method that is specifically oriented to low-load operation conditions of thermal power units, can deeply integrate dynamic characteristic correction of thermal power units and collaborative optimization of internal SOC states of hybrid energy storage systems, so as to comprehensively improve the frequency modulation performance, operating economy and equipment safety of the system. SUMMARY

[0005] In order to overcome the shortcomings of the prior art, one of the purposes of the present application is to provide a secondary frequency modulation optimization control method under low-load conditions, which dynamically corrects the control parameters of low-load thermal power units and collaboratively distributes the power of hybrid energy storage by combining an adaptive optimization algorithm, thereby improving the accuracy and speed of secondary frequency modulation while ensuring the safety and economy of the energy storage system.

[0006] One of the purposes of the present application is achieved by adopting the following technical solutions:

[0007] A secondary frequency modulation optimization control method under low-load conditions, comprising the following steps:

[0008] Obtain power grid operation data, and calculate the secondary frequency modulation power currently required to be compensated according to the power grid operation data;

[0009] identifying the current load range of the thermal power unit in the power grid, and performing control parameter correction on the unit located in the low load range;

[0010] taking the minimum area control error of the thermal power unit operation cost as the optimization target, and calculating the total output power reference value;

[0011] based on the total output power reference value, using an adaptive optimization algorithm to calculate and perform fine power distribution to obtain optimal charging and discharging instructions of each energy storage unit;

[0012] According to the optimal charging and discharging instructions, the output power is adjusted in real time through double closed-loop control.

[0013] Further, the power grid operation data is obtained, and the secondary frequency modulation power currently required for compensation is calculated according to the power grid operation data, including:

[0014] obtaining the actual frequency of the power grid, the frequency deviation, the current active power of each thermal power unit, and the state of charge of each unit of the energy storage system;

[0015] According to the frequency deviation, the secondary frequency modulation power currently required for compensation is calculated , which satisfies:

[0016] , wherein, and are proportional and integral gains, represents the frequency deviation; when is positive, it means that the power output needs to be increased, is negative, it means that the power output needs to be reduced.

[0017] Further, the current load range of the thermal power unit in the power grid is identified, including: calculating the current load rate of the unit through real-time data of the thermal power unit , which satisfies: , wherein, is the real-time output power of the unit, is the rated maximum power of the unit, when <40%, it is identified as a low load condition.

[0018] Further, control parameter correction is performed on the unit located in the low load range, including introducing a dynamic correction weight in the cost function, and introducing a load state correction in the power ramp rate constraint;

[0019] wherein the cost function after introducing the dynamic correction weight is: , represents the cost function of the th unit after correction, is a performance correction factor, is the i-th thermal power unit, a standard cost function of the i-th thermal power unit;

[0020] the power ramp rate constraint is satisfied: wherein, is the modified upper limit of the ramp rate, is the i-th thermal power unit, is the current frequency modulation power of the i-th thermal power unit, represents the current time.

[0021] Further, the total output power reference value is calculated with the optimization objective of minimizing the regional control error in the region with the minimum operation cost of the thermal power unit, including:

[0022] A global optimization model is constructed with the joint objectives of minimizing the total operation cost and optimizing the frequency stability of the system;

[0023] The global optimization model is solved based on the secondary frequency modulation power and the operation constraints of the thermal power unit and the energy storage system, and the total reference power allocated to the thermal power unit and the total reference power allocated to the energy storage group are output.

[0024] Further, the calculation function of the global optimization model satisfies: ,

[0025] wherein, is the i-th thermal power unit, is the current frequency modulation power of the i-th thermal power unit, and represents the charging power and discharging power of the j-th energy storage system in the energy storage system, represents the total cost objective function, represents the regional control error, is a penalty coefficient, represents the standard cost function of the i-th thermal power unit, ACE represents the total number of units; The calculation function and constraints of the global optimization model are input into a convex optimization solver for solving, and the total reference power allocated to the thermal power unit and the total reference power allocated to the energy storage group are output. Further, based on the total output power reference value, an adaptive optimization algorithm is used for fine power distribution to obtain the optimal charging and discharging instructions of each energy storage unit, including:

[0026]

[0027] Further, based on the total output power reference value, an adaptive optimization algorithm is used for fine power distribution to obtain the optimal charging and discharging instructions of each energy storage unit, including:

[0028] ​​According to the rated capacity of the lithium iron phosphate battery and the lithium titanate battery and the preset response weight, the initial expected power of each battery unit is calculated; wherein the response weight is configured so that the lithium titanate battery with better dynamic response capability obtains a higher allocation ratio;

[0029] A distribution target function is constructed with the common optimization objectives of minimizing the tracking error of the actual output power of each battery unit and the expected power, and minimizing the imbalance degree of all battery unit state of charge (SOC);

[0030] An ADAM adaptive optimization algorithm is used to iteratively solve the distribution target function, the ADAM adaptive optimization algorithm adjusts the learning step of each energy storage unit power instruction adaptively by maintaining the first and second moment estimates of the historical gradient and correcting the deviation, and quickly converges to the optimal charging and discharging power instruction of each battery unit under the constraint of total power conservation;

[0031] Wherein, during the iteration process and before outputting the final instruction, the power instruction of all battery units is subjected to upper and lower limit truncation processing to ensure that it does not exceed the maximum allowed charging and discharging power.

[0032] The second purpose of the present application is to provide a secondary frequency modulation optimization control device under low load working condition.

[0033] The second purpose of the present application is achieved by the following technical solutions:

[0034] A secondary frequency modulation optimization control device under low load working condition, comprising:

[0035] A power calculation module is configured to obtain grid operation data and calculate the secondary frequency modulation power currently required for compensation according to the grid operation data;

[0036] A reference value calculation module is configured to identify the current load interval of the thermal power unit in the grid and correct the control parameters of the unit in the low load interval; and calculate the total output power reference value with the optimization objectives of minimizing the region control error of the thermal power unit operation cost.

[0037] An adjustment module is configured to calculate the optimal charging and discharging instructions of each energy storage unit by using an adaptive optimization algorithm based on the total output power reference value for fine power distribution; and adjust the output power in real time through double closed-loop control according to the optimal charging and discharging instructions.

[0038] The third purpose of the present application is to provide an electronic device for implementing the first purpose of the application, which comprises a processor, a storage medium and a computer program, the computer program is stored in the storage medium, and the computer program is executed by the processor to realize the secondary frequency modulation optimization control method under low load working condition.

[0039] The fourth object of the present application is to provide a computer readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned secondary frequency modulation optimization control method under low load conditions.

[0040] Compared with the prior art, the present application has the following advantages:

[0041] The present application effectively compensates for the nonlinear characteristics of the thermal power unit under low load, such as gain reduction and increased inertia, fundamentally reduces the regulation delay and overshoot caused by unit response lag, and uses the millisecond-level fast response capability of the hybrid energy storage system to make up for the shortcomings of the thermal power unit, so that the system can quickly track the frequency modulation instruction, thereby significantly improving the speed and accuracy of frequency regulation and reducing the regional control error.

[0042] In addition, the present application globally optimizes the operation cost of the thermal power unit and the frequency deviation penalty in one framework, automatically seeks the best balance point of economy and stability, avoids excessive dependence on high-cost unit frequency modulation, and reduces the overall frequency modulation cost; by promoting the balance of the SOC of the energy storage unit, it avoids the premature retirement of some batteries due to overcharging or overdischarging, prolongs the overall service life of the energy storage system, and improves the economy of the whole life cycle;

[0043] The dynamic correction mechanism automatically reduces the frequency modulation task allocation of the low-load unit, reduces the operating pressure of the unit, and is beneficial to improving the operation safety and prolonging the service life of the equipment; by setting the SOC, charging and discharging power, climbing rate and other constraints, and performing boundary processing in the algorithm, it ensures that the energy storage system always works in a safe interval, prevents overcharging and overdischarging, and greatly improves the reliability of the system;

[0044] The present application proposes an allocation strategy based on response weight, which fully utilizes the respective advantages of fast response of lithium titanate battery and large capacity of lithium iron phosphate battery, and realizes intelligent division of labor of frequency modulation tasks;

[0045] The present application also uses the ADAM adaptive optimization algorithm for power allocation, which not only has fast convergence speed, but also can optimize the power tracking error and SOC balance degree at the same time, realizing efficient and adaptive collaborative management of the hybrid energy storage system. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 is a secondary frequency modulation optimization control method system block diagram of example one;

[0047] Figure 2 is a flowchart of the secondary frequency modulation optimization control method under low load conditions of example one;

[0048] Figure 3 is a structure block diagram of the secondary frequency modulation optimization control device under low load working condition of embodiment two;

[0049] Figure 4 is a structure block diagram of the electronic device of embodiment three. DETAILED DESCRIPTION

[0050] The application will be described in more detail below with reference to the drawings, it should be noted that the following description of the application with reference to the drawings is only illustrative and not limiting. The various different embodiments can be combined with each other to constitute other embodiments not shown in the following description.

[0051] Embodiment one

[0052] Embodiment one provides a secondary frequency modulation optimization control method under low load working condition, aiming to solve the problem of low load frequency modulation of thermal power units, which can significantly improve the speed and accuracy of grid frequency regulation, while ensuring the safety of energy storage system and reducing the cost of frequency modulation.

[0053] The embodiment proposes a secondary frequency modulation optimization control method specially for low load operation condition, taking into account the SOC state of thermal lithium iron phosphate battery-lithium iron titanate hybrid battery energy storage, by fully considering the response lag and output constraint of thermal power units, combining the real-time SOC state of hybrid energy storage unit, comprehensively balancing the regulation capacity and state of charge, dynamically adjusting the power distribution ratio and control parameters of each resource in the frequency modulation process, effectively improving the secondary frequency modulation accuracy, response speed and frequency stability of the system under low load operation, while ensuring the safe and efficient operation of the energy storage system, prolonging the service life, and meeting the high reliability demand of grid for flexible frequency modulation resources.

[0054] Please refer to the system block diagram of the thermal power and hybrid energy storage collaborative secondary frequency modulation optimization control method system for low load operation condition of the embodiment shown in Figure 1 The embodiment dynamically identifies and corrects the control parameters of low load thermal power units, combines the upper global optimization and lower adaptive algorithm, and realizes the collaborative power distribution of thermal power and hybrid energy storage.

[0055] Please refer to Figure 2 The embodiment provides a secondary frequency modulation optimization control method under low load working condition, which includes the following steps:

[0056] S1, obtaining grid operation data, and calculating the secondary frequency modulation power currently needed to compensate according to the grid operation data;

[0057] S1 includes:

[0058] Obtaining the actual frequency of the grid, the frequency deviation, the current active power of each thermal power unit, and the state of charge of each unit of the energy storage system;

[0059] According to the frequency deviation, the secondary frequency modulation power required to be compensated is calculated , which satisfies:

[0060] , wherein, and are proportional and integral gains, respectively. The control algorithm can quickly convert the frequency deviation signal into the total power command required to be compensated, and the generated secondary frequency modulation power command is positive, indicating that the power generation needs to be increased, and is negative, indicating that the power generation needs to be reduced. The power command serves as the overall target for subsequent distribution. represents the frequency deviation; when is positive, indicating that the power generation needs to be increased, is negative, indicating that the power generation needs to be reduced. It should be noted that the above and refer to the control gain of the dispatch AGC (in EMS), which is set by the dispatch center according to system inertia, load fluctuation characteristics, etc. It can be set according to actual conditions and is an engineering setting parameter.

[0061] Specifically, system state acquisition is performed by the energy management system (EMS) of the power grid dispatch center, and real-time acquisition of power grid operation data is performed through remote devices (RTU) and supervisory control and data acquisition systems (SCADA). The collected information includes: the actual frequency and its deviation , wherein , is the rated frequency, the current active power of each thermal power unit , and the state of charge (SOC) of each unit of the energy storage system.

[0062] S2, identify the current load interval of the thermal power unit in the power grid, and perform control parameter correction on the unit located in the low load interval;

[0063] The load interval identification in S2 includes: calculating the current load rate of the thermal power unit through real-time data of the unit , satisfying: , wherein, is the real-time output power of the unit, is the rated maximum power of the unit, and when < 40%, it is determined as a low load condition.

[0064] The current load rate can be calculated by real-time data obtained through the electronic management system EMS. The low load operation of the thermal power unit is usually defined as: , and under the condition of partial ultra-supercritical unit or deep peak regulation, the low load interval 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.

[0065] 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.

[0066] 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;

[0067] 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.

[0068] The first of the above cost functions The standard cost function for the Taiwanese unit is:

[0069] ,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:

[0070] ,

[0071] 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: Thus, the inhibitory effect of low-load conditions on the economic efficiency of unit frequency modulation is directly reflected in the global optimization objective function.

[0072] Under this mechanism, even under the same total power demand, the low-load unit will have a lower power allocation ratio in the optimization result due to the increased weight of the corrected cost function, thereby avoiding its dynamic frequency fluctuation and over-regulation.

[0073] In addition, to reflect the increased output inertia under low load, the corrected ramp-up limit is defined as:

[0074] ,

[0075] is a proportionality coefficient (usually <1). At low load, due to the increase in time constant and the decrease in gain, the correction coefficient is less than 1, further compressing the ramp rate constraint and reducing the output variation. This correction makes the optimization model reflect the response lag characteristic of low-load units and the inertia constraint at the power dynamic constraint level.

[0076] S3, taking the minimum regional control error as the optimization objective, calculate the total output power reference value;

[0077] The S3 step first constructs an optimization objective function for the combined participation of thermal power units and hybrid energy storage systems (including lithium iron phosphate batteries and lithium titanate batteries) in frequency modulation, considering the system operating cost and frequency control error (ACE). The corresponding operating cost function has been defined previously.

[0078] Specifically, taking the minimum regional control error as the optimization objective, calculate the total output power reference value, including:

[0079] Taking the minimum system total operating cost and the optimal frequency stability as the joint objective, a global optimization model is constructed;

[0080] Based on the secondary frequency modulation power and the operating constraints of thermal power units and energy storage systems, the global optimization model is solved to output the total reference power allocated to thermal power units and the total reference power allocated to energy storage groups.

[0081] The discharge power and charge power of the first energy storage system in the energy storage system are represented as and :

[0082] .

[0083] To measure the system frequency regulation performance, the regional control error (ACE) is defined as follows:

[0084] ,

[0085] wherein represents the current predicted load demand power (MW), is a known quantity, and ACE is the current power balance error, used to reflect the size of the frequency deviation.

[0086] In summary, the calculation function of the global optimization model satisfies: ,

[0087] wherein, is the current frequency regulation power of the i-th unit, is the current frequency regulation power of the i-th unit, and represents the charging power and discharging power of the i-th power station in the energy storage system, represents the total cost objective function, represents the regional control error, is a penalty coefficient used to adjust the trade-off between cost and frequency recovery, represents the standard cost function of the i-th unit, ACE represents the total number of units; the first part of this objective function is the total operating cost of all units, and the second part is a square penalty term for frequency error, the larger the ACE value, the worse the frequency regulation performance. The calculation function of the global optimization model and the constraint input are solved by a convex optimization solver, and the total reference power allocated to thermal power units and the total reference power allocated to energy storage groups are output. The model should satisfy the physical and safe operation constraints of the power system, including:

[0088]

[0089] ,

[0090] ,

[0091] ,

[0092] ,

[0093] ,

[0094] ,

[0095] ,

[0096] In addition, it also includes the power ramping constraint in the foregoing steps, wherein ​​For the total power command of the secondary frequency modulation obtained in the foregoing step, the left side of the equation is the total output of the thermal power unit and the total output of the energy storage, which are variables to be optimized and allocated, respectively represent the minimum and maximum output power (MW) of the unit; For the first The maximum charging power and the maximum discharging power of the energy storage unit; 、 respectively represent the SOC of the energy storage in the current period and the next period, respectively represent the charging and discharging efficiencies (constants between 0 and 1), is the control period time (hours), is the total capacity of the energy storage unit (MWh).

[0097] Since there are both quadratic cost functions and non-convex constraints such as energy storage charging and discharging exclusion in the above optimization model, convex relaxation technology is needed to make the model solvable, and the following measures are taken:

[0098] Linearization of the quadratic cost function: the quadratic term in the power generation cost is replaced by a piecewise second-order cone relaxation, which is converted into a second-order cone constraint in the convex optimization problem. The method is to introduce auxiliary variables , and define the constraint:

[0099] ,

[0100] At this time, the in the objective function term is replaced by a second-order cone relaxation, so that the optimization problem remains convex:

[0101] .

[0102] Relaxation of charging and discharging exclusion: ideally, the energy storage unit should satisfy the exclusion condition:

[0103]

[0104] This constraint is a non-convex product constraint, which cannot be directly processed by conventional methods, so the constraint is relaxed by introducing a penalty term in the objective function:

[0105]

[0106] where is a large positive weight coefficient to suppress both being non-zero at the same time.

[0107] Processing of ACE square term: the ACE square term in the objective function is a convex function, but if it is converted into a linear constraint form, a new variable is introduced, and ACE is replaced by z.

[0108] Through the above steps, the final optimization problem constructed is:

[0109] ,

[0110] The above objective function and constraints are input into a convex optimization solver to automatically search for a global optimal solution. During the solving process, the optimal solution automatically considers both the lowest cost and the smallest ACE (the target weight setting balances them). The output results include: the reference power of each thermal power unit , the charge and discharge reference power of the jth energy storage power station , and the obtained reference output power value of the mixed energy storage group in the energy storage power station which will be used as the target value for the next step of mixed energy storage power distribution.

[0111] S4, based on the total output power reference value, an adaptive optimization algorithm is used to calculate and perform power fine distribution to obtain the optimal charge and discharge instructions of each energy storage unit;

[0112] S4 includes:

[0113] According to the rated capacity of the lithium iron phosphate battery and the lithium titanate battery and the preset response weight, the initial expected power of each battery unit is calculated; wherein the response weight is configured to make the lithium titanate battery with better dynamic response capability obtain a higher distribution proportion;

[0114] The tracking error of the actual output power of each battery unit and the expected power, and the imbalance degree of all battery unit state of charge (SOC) are minimized as common optimization objectives to construct a distribution objective function;

[0115] An ADAM adaptive optimization algorithm is used to iteratively solve the distribution objective function. The ADAM adaptive optimization algorithm maintains the first and second moment estimates of the historical gradient and corrects the deviation to adaptively adjust the learning step of each energy storage unit power instruction. Under the constraint of total power conservation, the optimal charge and discharge power instructions of each battery unit are quickly converged;

[0116] Wherein, during the iteration process and before outputting the final instructions, the power instructions of all battery units are subjected to upper and lower limit truncation processing to ensure that they do not exceed the maximum allowed charge and discharge power.

[0117] Specifically, the upper layer power instruction calculation outputs the total energy storage power reference value 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.

[0118] The expected power is determined by both capacity and response weight, and is defined by the following formula:

[0119] ,

[0120] ,

[0121] 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.

[0122] Define the average current SOC of all batteries:

[0123] ,

[0124] 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.

[0125] Comprehensive objective function It can be written as:

[0126]

[0127] where , is the power tracking error and each subunit power is close to the desired value, is the SOC balancing weight coefficient, the objective function contains two parts: power tracking error (the first two terms) and SOC balancing error (the last two terms). , By this objective, the total output of the system is equal to the upper-level instruction, and each battery SOC is encouraged to move towards the balanced direction.

[0128] To ensure that the power target is met, the power conservation constraint must also be met:

[0129]

[0130] That is, all battery outputs must meet the total energy storage power reference value obtained in step 3, where represents the actual reference output power of the lithium iron phosphate battery unit in the current control period, represents the actual reference output power of the lithium titanate battery unit in the current control period.

[0131] After determining the objective function and constraints, use the ADAM adaptive optimization algorithm to iteratively solve it. The ADAM algorithm has adaptive learning rate and momentum term, which can accelerate convergence and reduce oscillation. The specific iteration steps are as follows:

[0132] Initialization: Set the initial value of iteration and , set the reference output of each unit and , and set the first-order momentum vector , the second-order momentum vector , the learning rate , the first-order moment decay rate , the second-order moment decay rate , and a small constant for numerical stability.

[0133] Gradient calculation: in the th iteration, first calculate the gradient of the objective function with respect to each subunit output power, where is an example, its gradient can be represented as:

[0134]

[0135] where the consistency of the total power with is considered by adding a penalty term, is the penalty coefficient, and is similar.

[0136] First moment update: update the first moment estimate using an exponential weighting method:

[0137]

[0138] Here is the dimension of the decision variable , is the first order decay rate, where the gradient vector contains the partial derivatives of all subunits. This momentum preserves the historical first moment set, which is equivalent to gradient descent with momentum. Second moment update: similarly update the second moment estimate:

[0139]

[0140]

[0141] where denotes the component-wise square, is the second order decay rate, preserves the square information of the gradient, which is used to adaptively adjust the step size.

[0142] Bias correction: since the initial momentum is zero, in order to eliminate the bias of the momentum estimate in the first few iterations, introduce the corrected first and second moments:

[0143]

[0144] After correction, the unbiased first and second moment estimates can be obtained.

[0145] According to the corrected momentum, update the power:

[0146]

[0147] where is the learning rate, which is controlled in step 1 in low load state, is a very small constant, and the update is and are performed simultaneously, and the power of each subunit is corrected according to the corresponding dimension of the momentum and the gradient.

[0148] Boundary constraint processing: after updating, perform saturation processing on the output power of each unit to ensure that it is within the allowed range, and perform upper and lower limit truncation on the power of each energy storage unit to ensure that the technical limit is met:

[0149]

[0150] ​Stop after reaching the set number of iterations, due to the fast convergence characteristics of the ADAM algorithm, usually only a few tens to hundreds of iterations are required to achieve stable allocation.

[0151] The above ADAM iteration process gradually adjusts the output of each hybrid energy storage subunit to optimal allocation. In the optimization process, the objective function simultaneously reduces power tracking error and SOC imbalance, and finally realizes the convergence of the whole system power to the upper target and keeps the SOC of each subunit consistent. Since the output is truncated after each iteration, it always ensures that the charging and discharging power does not exceed the technical range. At the same time, it also highlights the high response characteristics of lithium iron phosphate batteries, automatically allocating more dynamic power, and lithium iron phosphate batteries bearing the smooth part, ensuring system fast response and energy storage life.

[0152] S5, according to the optimal charging and discharging instructions, real-time adjustment of output power through double closed loop control.

[0153] S5 through the power conversion device (PCS, Power Conversion System) of the energy storage system, according to the reference power instruction of each energy storage unit calculated by the lower optimization allocation, real-time adjustment of output power through double closed loop control, to ensure consistency with the frequency modulation target. Specifically, 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; then, through the outer ring power controller, the power error is converted into active current instruction, and the inner ring current controller further adjusts the inverter PWM duty cycle, so that the active power between the DC bus and the AC bus can track the reference value in real time. At the same time, PCS judges the feasibility of the instruction according to the state information of the energy storage battery such as SOC, voltage and temperature, when the power instruction exceeds the allowed range or there is the risk of overcharging and overdischarging, automatically limiting or adjusting the output to ensure safe operation.

[0154] In summary, the method of the embodiment unifies the power distribution of the energy storage system and the thermal power unit into the same optimization framework by establishing a convex relaxation optimization model containing the operation cost of the thermal power unit and the frequency control error. The model uses the thermal power unit cost function correction, dynamic ramping constraint and ACE weighting to achieve global optimal solution of power distribution, and can automatically generate power instructions with optimal economy and in line with operation safety according to real-time system load and unit state. Compared with the existing fixed ratio or hierarchical experience distribution scheme, the distribution accuracy is significantly improved, the frequency modulation cost is reduced, and the system response is more coordinated. The ADAM adaptive optimization algorithm is used to iteratively distribute the power output of the lithium iron phosphate battery and the lithium titanate battery energy storage unit, fully considering the differences in dynamic response performance of different battery chemical systems, and introducing a weighting strategy for the response characteristics of different batteries in the power distribution objective function. The ADAM algorithm automatically allocates power proportion according to the instantaneous SOC, capacity and response time constant of the battery in the iteration process, preferentially allocates fast frequency modulation tasks to lithium titanate batteries, and allocates medium-speed and long-time frequency modulation tasks to lithium iron phosphate batteries, so as to match the optimal working interval of each energy storage unit, and then realize the dynamic optimization of energy storage output between power tracking accuracy and SOC balance, significantly improve the frequency modulation response speed, shorten the convergence time of power distribution, and reduce the frequency deviation caused by response delay. Under the condition of mixed operation of various batteries, the response speed can be layered, the power distribution can be intelligent, and the service life of the energy storage can be prolonged. Real-time calculation of the load rate of the thermal power unit identifies whether the unit is in the low load interval, and automatically generates gain correction coefficients and time constant correction coefficients for dynamically adjusting the unit cost weight and the upper limit of the ramp rate. This mechanism takes effect in the optimization model in real time, so that the low-load unit frequency modulation task is automatically reduced, thereby significantly reducing the frequency modulation performance decline caused by lag or over-regulation under low-load conditions, improving the operation stability of the thermal power unit, and expanding the available load interval of secondary frequency modulation.

[0155] Embodiment Two

[0156] Embodiment Two discloses a device corresponding to the secondary frequency modulation optimization control method under the low-load working condition of the above-mentioned embodiment. For the virtual device structure of the above-mentioned embodiment, please refer to Figure 3 as shown, comprising:

[0157] The power calculation module 310 is configured to obtain grid operation data and calculate the secondary frequency modulation power to be compensated according to the grid operation data.

[0158] The reference value calculation module 320 is configured to identify the current load interval of the thermal power unit in the grid and correct the control parameters of the unit in the low load interval; and calculate the total output power reference value with the minimum region of the operation cost of the thermal power unit and the minimum error as the optimization target.

[0159] 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.

[0160] Example 3

[0161] 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.

[0162] 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.

[0163] 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.

[0164] 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.

[0165] Example 4

[0166] The embodiment four of the present application also provides a storage medium containing computer executable instructions, which can be used for a computer to execute a secondary frequency modulation optimization control method under a low load condition, the method comprising:

[0167] obtaining power grid operation data, and calculating a secondary frequency modulation power currently required to be compensated according to the power grid operation data;

[0168] identifying a current load interval of a thermal power unit in the power grid, and performing control parameter correction on a unit located in a low load interval;

[0169] taking a minimum region of a thermal power unit operation cost and a minimum control error as optimization objectives, and calculating a total output power reference value;

[0170] based on the total output power reference value, performing power fine distribution by using an adaptive optimization algorithm to obtain optimal charge and discharge instructions of each energy storage unit;

[0171] adjusting output power in real time by using double closed loop control according to the optimal charge and discharge instructions.

[0172] Of course, the computer executable instructions of the storage medium containing computer executable instructions provided by the embodiment of the present application are not limited to the method operations described above, and can also perform related operations in the secondary frequency modulation optimization control method under a low load condition provided by any embodiment of the present application.

[0173] Through the above description of the embodiments, those skilled in the art can clearly understand that the present application can be realized by means of software and necessary general hardware, and of course can also be realized by hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a floppy disk, a read-only memory (ROM), a random access memory (RAM), a FLASH, a hard disk or an optical disk, etc., including a plurality of instructions to make an electronic device (which can be a mobile phone, a personal computer, a server or a network device, etc.) execute the method described in each embodiment of the present application.

[0174] It is worth noting that in the above embodiment of the secondary frequency modulation optimization control method and device under a low load condition, each unit and module included is only divided according to the function logic, but is not limited to the above division, as long as the corresponding function can be realized; in addition, the specific name of each functional unit is only for easy distinction, and does not limit the protection scope of the present application.

[0175] As can be seen from the above description, various modifications and changes can be made to the technical solutions and concepts described above by those skilled in the art, and all these modifications and changes shall fall within the scope of protection of the present application.

Claims

1. A method for optimizing secondary frequency modulation under low load conditions, characterized in that, The method comprises the following steps: obtaining power grid operation data and calculating the secondary frequency modulation power currently required for compensation according to the power grid operation data; identifying the current load interval of the thermal power unit in the power grid and performing control parameter correction on the unit located in the low load interval; wherein the control parameter correction on the unit located in the low load interval comprises introducing a dynamic correction weight in the cost function and introducing a load state correction in the power ramp 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 of the unit satisfies: ,in, This is the cost coefficient for the unit; For the first Current frequency regulation power of the unit; The power ramp rate constraint is satisfied as follows: wherein, is the revised upper ramp limit, is the first is the current frequency modulation power of the generator unit, denotes the current time; taking the minimum regional control error of the thermal power unit operation cost as the optimization target, calculating the total output power reference value; based on the total output power reference value, using an adaptive optimization algorithm to calculate and perform power fine distribution to obtain the optimal charging and discharging instructions of each energy storage unit; adjusting the output power in real time through double closed-loop control according to the optimal charging and discharging instructions.

2. The method of claim 1, wherein, obtaining power grid operation data and calculating the secondary frequency modulation power currently required for compensation according to the power grid operation data, comprising: obtaining the actual frequency of the power grid, the frequency deviation, the current active power of each thermal power unit and the state of charge of each unit of the energy storage system; calculating a frequency deviation from the current frequency which fulfils: wherein, and are proportional and integral gains, respectively, denotes a frequency deviation; when is positive, it indicates that the power generation output needs to be increased, is negative, it indicates that the power generation output needs to be decreased.

3. The method of claim 1, wherein, Identify the current load interval of thermal power unit in power grid, comprising: calculating the current load rate of thermal power unit through real-time data of the unit , satisfying: Wherein, is the real-time output power of the unit, is the rated maximum power of the unit, and when <40%, it is identified as low load condition.

4. The method of claim 1, wherein, taking the minimum regional control error of the thermal power unit operation cost as the optimization target, calculating the total output power reference value, comprising: taking the lowest total operation cost of the system and the optimal frequency stability as the joint target to construct a global optimization model; based on the secondary frequency modulation power and the operation constraints of the thermal power unit and the energy storage system, solving the global optimization model to output the total reference power allocated to the thermal power unit and the total reference power allocated to the energy storage group.

5. The method of claim 4, wherein, The calculation 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 the constraint of the global optimization model are input into a convex optimization solver for solving to output the total reference power allocated to the thermal power unit and the total reference power allocated to the energy storage group.

6. The method according to claim 1 or 4, wherein, based on the total output power reference value, using an adaptive optimization algorithm to calculate and perform power fine distribution to obtain the optimal charging and discharging instructions of each energy storage unit, comprising: according to the rated capacity of the lithium iron phosphate battery and the lithium titanate battery and the preset response weight, the initial expected power of each battery unit is calculated; wherein the response weight is configured to enable the lithium titanate battery with better dynamic response capability to obtain a higher allocation ratio; taking the tracking error of the actual output power of each battery unit and the expected power and the imbalance degree of the state of charge SOC of all battery units as the common optimization target, constructing a distribution target function; using the ADAM adaptive optimization algorithm to iteratively solve the distribution target function, the ADAM adaptive optimization algorithm maintains the first and second moment estimates of the historical gradient and corrects the deviation to adaptively adjust the learning step of each energy storage unit power instruction; under the constraint of total power conservation, the optimal charging and discharging power instructions of each battery unit are quickly converged; wherein, during the iteration process and before outputting the final instruction, the power instructions of all battery units are subjected to upper and lower limit truncation processing to ensure that they do not exceed the maximum allowed charging and discharging power.

7. A secondary frequency modulation optimization control device under low load conditions, characterized in that, It comprises: a power calculation module for obtaining power grid operation data and calculating the secondary frequency modulation power currently required for compensation according to the power grid operation data; The reference value calculation module is configured to identify a current load interval of a thermal power unit in a power grid, and to correct control parameters of the unit in a low load interval; wherein the correction of the control parameters of the unit in the low load interval comprises introducing a dynamic correction weight in a cost function, and introducing a load state correction in a power ramp 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 of the unit satisfies: ,in, This is the cost coefficient for the unit; For the first Current frequency regulation power of the unit; The power ramp rate constraint is satisfied: wherein, is the modified ramp-up upper limit, is the first is the current frequency modulation power of the unit, denotes the current time; the total output power reference value is calculated with the minimum regional control error of the minimum operation cost of the thermal power unit as the optimization objective; The adjusting module is configured to calculate optimal charge and discharge instructions of each energy storage unit by using an adaptive optimization algorithm based on the total output power reference value, to perform fine power distribution, and to adjust output power in real time through double closed-loop control according to the optimal charge and discharge instructions.

8. An electronic device comprising a processor, a storage medium, and a computer program stored in the storage medium, characterized in that, The computer program is executed by a processor to implement the secondary frequency modulation optimization control method under a low load condition according to any one of claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by a processor to implement the secondary frequency modulation optimization control method under a low load condition according to any one of claims 1 to 6.

Images