Multi-inverter energy storage system control method based on double-layer multi-objective optimization

Abstract

The application discloses a multi-converter energy storage system control method based on double-layer multi-target optimization, which comprises a system layer and a unit layer, and specific steps are as follows: S1, initializing the energy storage system; S2, system layer optimization, based on power grid power instruction, generating initial power distribution through an upper-layer optimization model; S3, unit layer optimization, according to the initial power distribution, optimizing the output proportion of the battery unit through a lower-layer optimization model, and calculating a sensitivity matrix; S4, dynamic correction, according to the sensitivity matrix and a power grid fluctuation coefficient, dynamically adjusting target weights by the system layer; and S5, outputting an optimal solution. The control method adopts a double-layer optimization architecture of the system layer and the unit layer, the system layer is mainly responsible for dynamic power distribution among multiple converters, and the unit layer focuses on collaborative control between a single converter and a battery unit. Through a cross-layer collaborative engine, dynamic coupling is realized, information interaction and iterative optimization are carried out, and the performance and efficiency of the energy storage system are maximized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy storage system control, in particular to a multi-inverter energy storage system control method based on double-layer multi-objective optimization. BACKGROUND

[0002] At present, the scale of renewable energy power generation is increasing, and energy storage has become an important supporting technology for new power systems. Under the stimulation of the electricity market, non-pumped storage energy storage technologies such as electrochemical energy storage, compressed air heat storage, and flywheel energy storage have entered industrialization, and independent energy storage power stations and shared energy storage models have developed rapidly. Among them, electrochemical energy storage is particularly prominent in the application of non-pumped storage energy storage technology, with lithium-ion batteries as the main force, from power batteries to large-scale energy storage batteries. Therefore, it is of great significance to analyze the factors affecting the system efficiency and key improvement technologies of lithium battery energy storage.

[0003] The existing control method adopts a single-objective optimization method, which makes it difficult to simultaneously consider the efficiency, service life, and stability of the energy storage system, and has defects. For example, when only pursuing the maximization of system operation efficiency, the battery may be in a high-rate charging and discharging state for a long time, accelerating the aging of the battery, and may also cause system stability problems such as circulating current.

[0004] At the same time, in the existing control method, the system layer and the unit layer lack bidirectional information interaction, forming an "optimization island". The system layer cannot obtain real-time state information of the battery at the unit layer, such as SOC, health state, SOH, temperature, etc., and the power command issued may exceed the actual bearing capacity of the battery. The unit layer also cannot dynamically adjust its control strategy according to the overall scheduling demand of the system layer, resulting in the entire energy storage system being unable to achieve optimal performance.

[0005] Based on the above two problems, the existing multi-inverter energy storage system has the problems of low efficiency, short equipment life, poor stability, and insufficient scene adaptability in operation, and cannot maximize the performance and efficiency of the energy storage system. SUMMARY

[0006] To overcome the aforementioned problems, the present invention aims to provide a control method for multi-converter energy storage systems based on dual-layer multi-objective optimization. This control method adopts a dual-layer optimization architecture of system layer and unit layer, and incorporates a dynamic coupling dual-loop optimization mechanism. The system layer is mainly responsible for the dynamic power distribution among multiple converters, forming a dynamic response loop to ensure efficient power interaction between the entire energy storage system and the grid, and to maintain stable system operation. The unit layer focuses on the collaborative control between a single converter and a battery unit, forming a lifespan protection loop, dedicated to optimizing the battery's operating state and extending its lifespan. Through a cross-level collaborative engine, dynamic coupling between the system layer and the unit layer is achieved, enabling information interaction and iterative optimization to realize global optimization of the entire energy storage system, ultimately maximizing the performance and efficiency of the energy storage system.

[0007] The technical solution adopted in this invention is:

[0008] A control method for a multi-converter energy storage system based on dual-layer multi-objective optimization includes a system layer and a unit layer. The system layer is optimized through an upper-layer optimization model to achieve dynamic power distribution among multiple converters. The unit layer is optimized through a lower-layer optimization model to achieve coordinated control of a single converter and battery unit. The system layer and the unit layer are dynamically coupled through a cross-level collaborative engine, including the following steps:

[0009] S1: Initialize the energy storage system, load system parameters, and set the iteration count for the system layer and unit layer;

[0010] S2: System-level optimization, based on the power grid command, generates the initial power allocation through the upper-level optimization model;

[0011] S3: Unit layer optimization. Each converter optimizes the output ratio of the battery unit through the lower-level optimization model based on the initial power allocation in S2, and calculates the sensitivity matrix.

[0012] S4: Dynamic correction. The system layer dynamically adjusts the target weights based on the sensitivity matrix and the power grid fluctuation coefficient, calculates the power correction factor, and transmits it to the unit layer to adjust the power allocation.

[0013] S5: Output the optimal solution. If the maximum number of iterations at both the system level and the unit level is satisfied, then output the optimal solution. If the maximum number of iterations at the system level is not satisfied, then return to S2. If the maximum number of iterations at the unit level is not satisfied, then return to S3.

[0014] As a further description of the present invention, the optimization objectives of the system layer are system efficiency, circulating current suppression index, and power point tracking accuracy, and the output power vector of each converter is defined as follows: , The number of converters in the energy storage system. For the first The output power of the converter, and its objective function is:

[0015] .

[0016] in, The optimal state of the system layer is represented as All values ​​are at their minimum.

[0017] It represents the reciprocal of the overall efficiency of the energy storage system.

[0018] This indicates the circulation suppression index of the energy storage system.

[0019] This indicates the power tracking deviation of the energy storage system.

[0020] The system-level constraints include energy storage system demand power constraints, power change rate constraints, unit-level feedback constraints, and power tracking sensitivity and consistency constraints.

[0021] As a further description of the present invention, the optimization objectives of the unit layer are battery life, SOC equalization, and charge / discharge utilization rate. The output ratio vector of the M battery units connected to a single converter is defined as follows: , For the first The output ratio of each battery cell, that is, the proportion of its power to the total power of the converter, is defined by the objective function:

[0022] .

[0023] in, The optimal state of the unit layer is represented as All values ​​are at their minimum.

[0024] This indicates battery life indicators.

[0025] This indicates the battery's SOC balance.

[0026] This indicates the battery charge / discharge utilization rate.

[0027] The constraints of the unit layer include SOC range constraints, output ratio normalization constraints, and current safety constraints.

[0028] As a further description of the present invention, the The calculation formulas are as follows:

[0029] ,

[0030] .

[0031] wherein, is the battery efficiency of the energy storage system, is the battery loss power of the energy storage system, is the battery actual output power of the energy storage system.

[0032] .

[0033] wherein, is the circulating current of the kth converter and the average value, is the number of converters.

[0034] .

[0035] wherein, is the power instruction of the grid-connected point, is the output power of the ith converter.

[0036] As a further description of the present application, the calculation formula of the is respectively:

[0037] ,

[0038] .

[0039] wherein, is the remaining cycle number of the battery life, is the rated life number when the discharge depth is equal to 50%, is the dynamic internal resistance ratio, the value of which is the ratio of the dynamic resistance to the initial ohmic resistance, is the internal resistance-life correlation intensity coefficient.

[0040] .

[0041] wherein, is the state of charge of the jth battery cluster, , is the average state of charge of the battery cluster belonging to the ith converter, and M represents the total number of battery clusters connected to the converter.

[0042] .

[0043] wherein, is the effective charge and discharge capacity, is the rated capacity.

[0044] As a further description of the present application, the calculation formula of the sensitivity matrix in S3 is:

[0045] .

[0046] in:

[0047] This represents the partial derivative of battery power loss with respect to power.

[0048] This is the partial derivative of the SOC with respect to power.

[0049] This is the partial derivative of power due to battery aging.

[0050] As a further description of the present invention, the power correction factor in S4 The calculation formula is:

[0051] .

[0052] in, This is the temperature adaptive coefficient.

[0053] , The weighting coefficients and .

[0054] Let be the power loss of the i-th battery cluster.

[0055] The maximum temperature of the i-th battery cluster is given.

[0056] This represents the average loss.

[0057] This represents the average temperature.

[0058] As a further description of the present invention, the upper-level optimization model adopts an improved CMOPSO-MSI algorithm, and its speed update formula is as follows:

[0059] .

[0060] in: Indicates the particle at the 1st The velocity vector of each iteration.

[0061] This represents the dynamic inertia weight.

[0062] Indicates the particle at the 1st The velocity vector of the next iteration.

[0063] , For learning factors.

[0064] 、 a random number with a value range in the interval .

[0065] represents the individual optimal position of the particle.

[0066] represents the global optimal position.

[0067] represents the current position of the particle.

[0068] The dynamic inertia weight is calculated by the formula:

[0069] .

[0070] wherein: represents the current iteration number, represents the maximum iteration number, a random number with a value range in the interval .

[0071] The improved CMOPSO-MSI algorithm adopts a multi-strategy mutation mode for optimization, and the process is represented as:

[0072] .

[0073] .

[0074] wherein: represents a new position of the i-th particle after mutation, represents an original position of the i-th particle before updating, represents a Gaussian distribution random number with a mean of 0 and a standard deviation of 1, represents a Cauchy distribution random number with a mean of 0 and a standard deviation of 1 represents a mutation probability. As a further description of the application, the lower optimization model adopts an improved MOPSO algorithm, and the adaptive local search step is represented as:

[0075]

[0076]

[0077]

[0078] .

[0079] wherein: ​​​​represents a battery health state correction coefficient, represents a temperature correction coefficient, represents a health state of a battery, represents a temperature.

[0080] Advantages of the present application:

[0081] The present application is based on a multi-converter energy storage system control method based on double-layer multi-objective optimization, including a system layer and a unit layer, S1: initializing the energy storage system; S2: system layer optimization, generating an initial power distribution through an upper layer optimization model based on a power grid power instruction; S3: unit layer optimization, each converter optimizing the output proportion of the battery unit through a lower layer optimization model according to the initial power distribution in S2, and calculating a sensitivity matrix; S4: dynamic correction, the system layer dynamically adjusting target weights according to the sensitivity matrix and a power grid fluctuation coefficient, calculating a power correction factor and transmitting it to the unit layer to adjust the power distribution; S5: outputting an optimal solution. The control method adopts a double-layer optimization architecture of the system layer and the unit layer, and integrates a dynamic coupling double-loop optimization mechanism. The system layer is mainly responsible for dynamic power distribution between multiple converters, constitutes a dynamic response loop, and ensures efficient power interaction between the entire energy storage system and the power grid and maintains stable operation of the system. The unit layer focuses on collaborative control between a single converter and a battery unit, constitutes a life protection loop, and strives to optimize the operating state of the battery and prolong the service life of the battery. Through a cross-layer collaborative engine, dynamic coupling between the system layer and the unit layer is realized, information interaction and iterative optimization are carried out, global optimization of the entire energy storage system is realized, and finally the performance and efficiency of the energy storage system are maximized.

[0082] The present application is based on a multi-converter energy storage system control method based on double-layer multi-objective optimization. The upper layer optimization model adopts an improved CMOPSO-MSI algorithm, increases a random number item on the basis of a traditional inertia weight, improves the convergence speed of the algorithm, and simultaneously adopts a multi-strategy mutation optimization mode, adaptively selects different mutation strategies according to particle aggregation degree, and finally ensures the stability of the system layer when approaching an optimal solution. The lower layer optimization model adopts an improved MOPSO algorithm, dynamically adjusts a local search step length, balances search precision and efficiency, can better adapt to changes in the state of the battery, and finally ensures that the unit layer obtains an optimal solution. BRIEF DESCRIPTION OF DRAWINGS

[0083] Figure 1 The present application is based on a multi-converter energy storage system control method based on double-layer multi-objective optimization. DETAILED DESCRIPTION

[0084] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application are described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work should fall within the protection scope of the present application.

[0085] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be practiced without the specific details. In other instances, well-known methods have not been described in detail in order not to unnecessarily obscure aspects of the present application.

[0086] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. The "in one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments.

[0087] The present application is described in detail in conjunction with the schematic drawings. In the detailed description of the embodiments of the present application, the sectional view of the device structure is partially enlarged without the general proportion for the convenience of description, and the schematic drawings are only examples, which should not limit the scope of protection of the present application. In addition, the three-dimensional spatial dimensions of length, width and depth should be included in the actual manufacture.

[0088] Meanwhile, in the description of the present application, it should be noted that the orientation or position relationship of the terms "up, down, inner and outer" is based on the orientation or position relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first, second or third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0089] Unless otherwise specifically defined and limited, the terms "mounting, connecting, connection" in the present application should be understood broadly, for example: it can be fixed connection, detachable connection or integral connection; it can also be mechanical connection, electrical connection or direct connection, it can also be indirectly connected through intermediate medium, or it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0090] As shown in the drawings, the specific embodiments of the present application are shown: Figure 1

[0091] Embodiment one​

[0092] The control method of the multi-converter energy storage system based on double-layer multi-objective optimization includes a system layer and a unit layer. The system layer is optimized by an upper-layer optimization model to realize dynamic power distribution among the multiple converters. The unit layer is optimized by a lower-layer optimization model to realize collaborative control between a single converter and a battery unit. Dynamic coupling is realized between the system layer and the unit layer through a cross-layer collaborative engine. The method includes the following steps:

[0093] S1: initializing the energy storage system, loading system parameters, and setting the iteration times of the system layer and the unit layer;

[0094] S2: system layer optimization, generating an initial power distribution based on a power instruction of the power grid through the upper-layer optimization model;

[0095] S3: unit layer optimization, each converter optimizing the output proportion of the battery unit according to the initial power distribution in S2 through the lower-layer optimization model and calculating a sensitivity matrix;

[0096] S4: dynamic correction, dynamically adjusting target weights according to the sensitivity matrix and a power grid fluctuation coefficient, calculating a power correction factor, and transmitting the power correction factor to the unit layer to adjust the power distribution;

[0097] S5: outputting an optimal solution, if the maximum iteration times of the system layer and the unit layer are simultaneously satisfied, outputting the optimal solution, if the maximum iteration times of the system layer are not satisfied, returning to S2, and if the maximum iteration times of the unit layer are not satisfied, returning to S3.

[0098] In this embodiment, as shown in FIG. 1, the control method adopts a double-layer optimization architecture of the system layer and the unit layer and integrates a dynamic coupling double-loop optimization mechanism. Figure 1 The system layer is mainly responsible for dynamic power distribution among the multiple converters and constitutes a dynamic response loop to ensure efficient power interaction between the entire energy storage system and the power grid and maintain stable operation of the system. The unit layer focuses on collaborative control between a single converter and a battery unit and constitutes a life protection loop to optimize the operating state of the battery and prolong the service life of the battery. Dynamic coupling between the system layer and the unit layer is realized through a cross-layer collaborative engine to realize information interaction and iterative optimization and achieve global optimization of the entire energy storage system, thereby maximizing the performance and efficiency of the energy storage system.

[0099] Embodiment Two

[0100] Specifically, the optimization target of the system layer is system efficiency, circulating current suppression index, and power tracking accuracy. The output power vector of each converter is defined as , n is the number of converters in the energy storage system, is the number of battery units connected to the i-th converter, and The objective function of the output power of the converter is:

[0101] .

[0102] wherein, The optimal state of the system layer is The state in which the sum of the two is the minimum.

[0103] The reciprocal of the total efficiency of the energy storage system.

[0104] The circulating current suppression index of the energy storage system.

[0105] The power tracking deviation of the energy storage system.

[0106] Specifically, the calculation formula of the is respectively:

[0107] ,

[0108] .

[0109] wherein, The battery efficiency of the energy storage system, The battery loss power of the energy storage system, The actual output power of the battery of the energy storage system.

[0110] .

[0111] wherein, The circulating current of the kth converter and the average value, The number of converters.

[0112] .

[0113] wherein, The power instruction of the grid-connected point, The output power of the ith converter.

[0114] The constraint conditions of the system layer include the energy storage system demand power constraint, the power change rate constraint, the unit layer feedback constraint, and the power tracking sensitivity consistency constraint.

[0115] The system demand power constraint is:

[0116] ,

[0117] .

[0118] in, Let i be the minimum output power of the i-th converter. Let be the maximum output power of the i-th converter.

[0119] The power change rate constraint is:

[0120] .

[0121] in, For the first Each converter at the current moment 'output power' For the first The output power of each converter at the previous moment, with the interval being the control cycle. This represents the maximum power change rate of the converter.

[0122] The unit layer feedback constraint is:

[0123] .

[0124] in, This refers to the battery cell voltage. This represents the maximum permissible current based on the current SOC and temperature.

[0125] The power tracking sensitivity consistency constraint is:

[0126] .

[0127] in, For the power grid tracking deviation to the first The partial derivative of the power of the i-th converter represents the... Sensitivity to the impact of power variations of individual converters on the system's accuracy in tracking grid commands.

[0128] For the power grid tracking deviation to the first Partial derivatives of the power of the converter, , This represents the total number of converters.

[0129] The threshold value is 0.02, which represents the maximum allowable difference in power tracking sensitivity between different converters. The purpose is to ensure that the difference in sensitivity between different converters when tracking grid power is ≤2%, so as to avoid unbalanced power distribution in the system due to excessive differences in sensitivity of individual converters.

[0130] Example 3

[0131] Specifically, the optimization target of the unit layer is battery life, SOC balance, and charge-discharge utilization rate, and the output proportion vector of M battery units connected by a single converter is defined as , is the output proportion of the i-th battery unit, i.e., the proportion of the power it bears in the total power of the converter, and its objective function is:

[0132] .

[0133] wherein, represents the optimal state of the unit layer is the state in which all the above indicators are minimum.

[0134] represents the battery life index.

[0135] represents the battery SOC balance degree.

[0136] represents the battery charge-discharge utilization rate.

[0137] Specifically, the calculation formula of the above-mentioned is respectively:

[0138] ,

[0139] .

[0140] wherein, is the remaining cycle number of the battery life, is the rated life number when the discharge depth is equal to 50%, is the dynamic internal resistance ratio, whose value is the ratio of the dynamic resistance to the initial ohmic resistance, is the internal resistance-life correlation intensity coefficient.

[0141] .

[0142] wherein, is the state of charge of the j-th battery cluster, , is the average state of charge of the battery cluster belonging to the i-th converter, and M represents the total number of battery clusters connected by the converter.

[0143] .

[0144] wherein, is the effective charge-discharge capacity, is the rated capacity.

[0145] ​The constraints of the unit layer include SOC range constraints, output proportion normalization constraints, and current safety constraints.

[0146] The SOC range constraints are:

[0147] .

[0148] The output proportion normalization constraints are:

[0149] .

[0150] The current safety constraints are:

[0151] .

[0152] wherein, is the maximum allowable current based on SOC and temperature.

[0153] Embodiment Four

[0154] Specifically, the sensitivity matrix in S3 is calculated according to the following formula:

[0155] .

[0156] wherein,

[0157] is the partial derivative of the battery power loss with respect to power.

[0158] is the partial derivative of the SOC with respect to power.

[0159] is the partial derivative of the battery aging with respect to power.

[0160] In this embodiment, the sensitivity matrix is the feedback mode from the unit layer to the system layer, and , , is used to quantify the degree of influence of the power change on the unit layer target.

[0161] Specifically, the power correction factor in S4 is calculated according to the following formula:

[0162] .

[0163] wherein, is a temperature adaptive coefficient.

[0164] , is a weight coefficient and .

[0165] Ploss(i) is the power loss of the ith battery cluster.

[0166] Tmax(i) is the maximum temperature of the ith battery cluster.

[0167] Ploss_avg is the average loss.

[0168] Tmax_avg is the average temperature.

[0169] In this embodiment, the power correction factor is the feedback mode from the system layer to the unit layer, which is used to dynamically correct the converter power instruction.

[0170] Embodiment Five

[0171] Specifically, the upper layer optimization model adopts an improved CMOPSO-MSI algorithm, and the speed update formula is as follows:

[0172] .

[0173] Wherein: V(k) represents the velocity vector of the particle in the kth iteration, and the change rate of the power adjustment amount determines the search direction and step length of the particle in the next step.

[0174]

[0175] V(k) represents the velocity vector of the particle in the kth iteration, and the change rate of the power adjustment amount determines the search direction and step length of the particle in the next step.

[0176] , is a learning factor, which respectively controls the strength of the particle learning to the individual optimal solution and the global optimal solution.

[0177] , is a random number with a value range in the interval [0, 1], which increases the search randomness and avoids the algorithm from falling into a fixed track.

[0178]

[0179] ​​​​​​represents the global optimal position, i.e. the current power allocation scheme.

[0180] represents the current position of the particle, i.e. the optimal power allocation scheme found by the entire population history.

[0181] The dynamic inertia weight is calculated as follows:

[0182] .

[0183] wherein: represents the current iteration number, represents the maximum iteration number, is a random number with a value range of , increasing the randomness of the weight.

[0184] In this embodiment, the upper optimization model introduces a random number item 0.1rand() between 0 and 0.1 on the basis of the traditional inertia weight which linearly decreases with the iteration number, which makes the inertia weight not only monotonically change during the iteration, but also increases randomness to some extent. In the early stage of iteration, the larger inertia weight combined with random disturbance can make the particle explore more widely in the search space, avoiding the algorithm falling into local optimal solution too early; in the later stage of iteration, the inertia weight gradually decreases, reducing the influence of random disturbance, so that the particle can more accurately develop locally towards the optimal solution, balancing the global exploration and local development ability, thereby improving the convergence speed of the algorithm.

[0185] The improved CMOPSO-MSI algorithm adopts a multi-strategy mutation mode for optimization, and its process is represented as:

[0186] .

[0187] .

[0188] wherein: represents the new position of the th particle after mutation, representing the adjusted power allocation strategy.

[0189] represents the original position of the th particle before updating, which is the power allocation strategy currently being optimized.

[0190] represents a Gaussian distribution random number with a mean of 0 and a standard deviation of 1.

[0191] represents a Cauchy distribution random number with a mean of 0 and a standard deviation of 1.

[0192] represents the mutation probability.

[0193] In the embodiment, the aggregation degree is an index for measuring the dispersion or concentration degree of the particle group in the solution space. The value is closer to 1, the particles are more concentrated; the value is closer to 0, the particles are more dispersed. If the threshold is lower than 0.6, the Cauchy disturbance is started too early, the population diversity is insufficient, and it is easy to fall into local optimum; if the threshold is higher than 0.6, the Gaussian disturbance lasts too long, the convergence speed of the algorithm is reduced, and it is difficult to meet the control cycle requirement of 50 ms of the energy storage system. Therefore, the threshold of 0.6 is selected to ensure the sufficiency of global exploration and the efficiency of local development, and to adapt to the dynamic optimization requirements of the multi-inverter energy storage system.

[0194] At the same time, different mutation strategies are adaptively selected according to the particle aggregation degree. When the aggregation degree is low, that is, less than 0.6, the Gaussian disturbance is adopted, a relatively fine disturbance is performed, the particle position is fine-tuned to a certain extent, and it is helpful to search more carefully in the local area; when the aggregation degree is high, that is, greater than or equal to 0.6, the Cauchy disturbance is adopted, which has a thicker tail and can produce a larger disturbance, so that the particle has a greater probability to jump out of the local optimal area. Moreover, the mutation probability is dynamically adjusted with the iteration number. The mutation probability is larger at the early iteration stage, encourages the particle to perform more mutation operations, and enhances the exploration ability of the algorithm; the mutation probability is reduced at the late iteration stage, ensures the stability of the algorithm when approaching the optimal solution, and reduces unnecessary mutations.

[0195] Specifically, the lower layer optimization model adopts an improved MOPSO algorithm, and the adaptive local search step is represented as:

[0196]

[0197]

[0198]

[0199] wherein: represents a battery state of health correction coefficient, represents a temperature correction coefficient, represents a state of health of the battery, represents a temperature.

[0200] In the embodiment, the local search step is dynamically adjusted according to the state of health SOH and the temperature T of the battery. When the battery SOH is low, it means that the aging degree of the battery is high, at this time is small, the step ​​​Also correspondingly reduced, so that the algorithm in the search process more cautious, focus on fine search, avoid because of the larger step on the aging battery caused by excessive charge and discharge operation; when the battery temperature deviates from the optimal temperature 25℃ far, Reduced, step also decreases, balance the search accuracy and efficiency, can better adapt to the change of battery state.

[0201] The preferred embodiments of the application are described in detail above with reference to the accompanying drawings, but the application is not limited to the above-described embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the spirit of the application.

[0202] Many other changes and modifications can be made without departing from the spirit and scope of the application. It should be understood that the application is not limited to a particular embodiment, and the scope of the application is defined by the appended claims.

Claims

1. A multi-inverter energy storage system control method based on double-layer multi-objective optimization, characterized in that, The system layer is optimized by an upper layer optimization model to realize dynamic power distribution among multiple converters, and the unit layer is optimized by a lower layer optimization model to realize collaborative control of a single converter and a battery unit, dynamic coupling is realized between the system layer and the unit layer through a cross-layer collaborative engine, and the method comprises the following steps: S1: initializing the energy storage system, loading system parameters, and setting the iteration number of the system layer and the unit layer; S2: System-level optimization, based on grid power commands, generates initial power allocation through an upper-level optimization model; the optimization objectives of the system layer are system efficiency, circulating current suppression index, and power point tracking accuracy, defining the output power vector of each converter as follows. , The number of converters in the energy storage system. For the first The output power of the converter, and its objective function is: ， wherein, represents the optimal state of the system layer is state that both are minimum values; represents the inverse of the total efficiency of the energy storage system; a circulating current suppression index indicative of the energy storage system; represents the power tracking deviation of the energy storage system; S3: unit layer optimization, each converter optimizes the output proportion of the battery unit through the lower layer optimization model according to the initial power distribution in S2, and calculates the sensitivity matrix; S4: dynamic correction, the system layer dynamically adjusts the target weight according to the sensitivity matrix and the grid fluctuation coefficient, calculates the power correction factor and transmits it to the unit layer to adjust the power distribution; S5: output the optimal solution, if the maximum iteration number requirements of the system layer and the unit layer are met at the same time, output the optimal solution, if the maximum iteration number requirement of the system layer is not met, return to S2, if the maximum iteration number requirement of the unit layer is not met, return to S3.

2. The multi-converter energy storage system control method based on double-layer multi-objective optimization according to claim 1, characterized in that, The constraint conditions of the system layer include energy storage system demand power constraint, power change rate constraint, unit layer feedback constraint and power tracking sensitivity consistency constraint.

3. The multi-converter energy storage system control method based on double-layer multi-objective optimization according to claim 1, characterized in that, The optimization target of the unit layer is battery life, SOC balance, and charge-discharge capacity utilization rate, and the output proportion vector of M battery units connected by a single converter is defined as , is the output proportion of the first battery unit, that is, the proportion of the power it bears to the total power of the converter, and the objective function is: ， wherein, represents the optimal state of the cell layer is a state in which both are minimum values; represents an indicator of battery life; represents the battery SOC equalization degree; represents the battery charge-discharge utilization rate; The constraint conditions of the unit layer include SOC range constraint, output proportion normalization constraint and current safety constraint.

4. The multi-converter energy storage system control method based on double-layer multi-objective optimization according to claim 2, characterized in that, The The calculation formulas are respectively: ， ， wherein, Pbat is the battery power of the energy storage system, Pbatloss is the battery loss power of the energy storage system, Pbatactual is the battery actual output power of the energy storage system; ， wherein, is the circulating current of the kth converter with the average value, is the number of converters; ， wherein, is the power instruction for the point of interconnection, is the output power of the i-th converter.

5. The multi-converter energy storage system control method based on double-layer multi-objective optimization according to claim 3, characterized in that, The The calculation formulas are respectively: ， ， wherein, is the number of cycles remaining for the battery life, is the number of cycles for the rated life at a discharge depth equal to 50%, is the dynamic internal resistance ratio, which is the ratio of the dynamic resistance to the initial ohmic resistance, is the internal resistance-life correlation strength coefficient; ， wherein, Socj is the state of charge of the jth battery cluster, , Socj is the state of charge of the jth battery cluster, M represents the total number of battery clusters connected to the converter. ， wherein, is the effective charge-discharge capacity, is the rated capacity.

6. The dual-layer multi-objective optimization based multi-converter energy storage system control method of claim 1, wherein, The sensitivity matrix in S3 The formula for calculating is: ， Wherein: a partial derivative of the battery power loss with respect to power, To steer the power for the SOC, To bias the power for battery aging.

7. The dual-layer multi-objective optimization based multi-converter energy storage system control method of claim 1, wherein, The power correction factor in S4 The calculation formula is: ， wherein is a temperature adaptation coefficient, , are weight coefficients and , Ploss(i) = Ploss(i-1) + Ploss(i) The maximum temperature of the i-th battery cluster is... for average loss, is the average temperature.

8. The dual-layer multi-objective optimization based multi-converter energy storage system control method of claim 1, wherein, The upper layer optimization model adopts an improved CMOPSO-MSI algorithm, and the speed update formula is: ， wherein: represents the velocity vector of the particle at the th iteration, denotes the dynamic inertia weight, a velocity vector of the particle at the first iteration, , learning factor, , is a random number with values in the interval , This represents the individual optimal position of a particle. represents a globally optimal position, represents the current position of the particle; The dynamic inertia weight The calculation formula is: ， wherein: denotes the current iteration number, denotes the maximum iteration number, is a random number with values in the interval [0, 1]. The improved CMOPSO-MSI algorithm adopts a multi-strategy mutation mode for optimization, and the process is represented as: ， ， wherein: represents the new position of the th particle after mutation, represents the original position of the th particle before update, represents a Gaussian-distributed random number with mean 0 and standard deviation 1, represents a Cauchy-distributed random number with mean 0 and standard deviation 1, represents the mutation probability.

9. The dual-layer multi-objective optimization based multi-converter energy storage system control method of claim 1, wherein, The lower layer optimization model adopts an improved MOPSO algorithm, and the adaptive local search step is represented as: ， ， ， wherein: represents a battery health state correction coefficient, represents a temperature correction coefficient, represents a health state of the battery, represents a temperature.

Images