Dispatching method, device and electronic equipment for distributed grid-connected electric energy storage system group

Abstract

The present invention relates to a scheduling method, device and electronic device for distributed grid connection of an electric energy storage system group, belonging to the technical field of electric power grid, wherein the method comprises: constructing an objective function and constraint conditions for each electric energy storage system in the electric energy storage system group, the objective function comprising a first objective function constructed with maximum benefit of the electric energy storage system, a second objective function constructed with minimum peak-to-valley difference, a third objective function constructed with minimum load standard deviation and a fourth objective function constructed with minimum voltage amplitude deviation; selecting a global search operator in a global search operator pool, selecting a local search operator in a local search operator pool; obtaining an update operator based on the global search operator and the local search operator; selecting an operator in the update operator based on a preset deep reinforcement learning algorithm to solve the objective function under constraint conditions, and obtaining a scheduling result for distributed grid connection of the electric energy storage system group. The present invention improves the efficiency of scheduling of a comprehensive electric energy storage system.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power grids, and in particular to a method, device and electronic equipment for dispatching a distributed grid-connected electric energy storage system group. Background Art

[0002] Integrated energy storage systems usually integrate multiple energy forms (such as electricity, hydrogen energy, wind energy, solar energy, etc.), which can achieve complementary utilization of different energy sources. When the integrated energy storage system absorbs / releases electricity to the distribution network, if the amount of electricity absorbed or released is too large, the energy storage system will consume other energy to meet the electricity demand, or convert the excess electricity into other forms of energy.

[0003] In the prior art, in order to achieve the collaborative optimization scheduling of the above-mentioned integrated electric energy storage system group and solve the high-dimensional multi-objective optimization strategy, it is necessary to rely on a high-dimensional multi-objective optimization algorithm. The core of the high-dimensional multi-objective optimization algorithm lies in the simultaneous optimization of multiple competing objective functions. This process requires not only extensive exploration of the solution space, but also fine-tuning of local solutions. High-dimensional multi-objective optimization algorithms usually rely on update operators composed of global search operators and local search operators. The global search operator is responsible for exploring the entire solution space and discovering potential high-quality solutions; while the local search operator focuses on the local area and improves the accuracy of the solution by refining the existing solution. The effective collaboration of update operators is crucial to improving algorithm performance.

[0004] Although many high-dimensional multi-objective optimization algorithms use a fixed-structure update operator combination (such as a fixed pairing of global search operators and local search operators), this design, although simple, will lead to insufficient population diversity, thereby weakening the exploration ability, affecting the quality of the final solution and the global convergence of the algorithm. In this context, some algorithms try to improve flexibility by randomly selecting update operators. This method adjusts the update strategy by randomly selecting between global and local search operators. Although this method improves the adaptability of the algorithm to a certain extent, it can also lead to population instability and affect the consistency of the results.

[0005] In summary, the prior art lacks a scheduling method to improve the diversity and stability of the population in the multi-objective function model in the scheduling of the comprehensive electric energy storage system, thereby improving the efficiency of the scheduling of the comprehensive electric energy storage system. Summary of the invention

[0006] In view of this, it is necessary to provide a distributed grid-connected scheduling method, device and electronic equipment for an electric energy storage system group to solve the problem of insufficient population diversity and stability in a multi-objective function model in the scheduling of an integrated electric energy storage system.

[0007] In order to solve the above problems, the present invention provides a method for dispatching a distributed grid-connected electric energy storage system group, comprising:

[0008] Constructing objective functions and constraints for each electric energy storage system in the electric energy storage system group, wherein the objective functions include a first objective function constructed by maximizing the benefit of the electric energy storage system, a second objective function constructed by minimizing the peak-to-valley difference, a third objective function constructed by minimizing the load standard deviation, and a fourth objective function constructed by minimizing the voltage amplitude deviation, and the constraints include battery energy storage system constraints, hydrogen energy storage system constraints, and distribution network power balance constraints;

[0009] Select a global search operator from a preset global search operator pool, and select a local search operator from a preset local search operator pool;

[0010] An update operator is obtained based on the global search operator and the local search operator;

[0011] Based on the preset deep reinforcement learning algorithm, the operator is selected in the update operator to solve the objective function under the constraints, and the scheduling result of the distributed grid-connected electric energy storage system group is obtained.

[0012] In a possible implementation, the operator selected in the update operator based on the preset deep reinforcement learning algorithm solves the objective function under the constraint condition to obtain the dispatch result of the distributed grid-connected electric energy storage system group, including:

[0013] S1: Obtaining the initial value of the objective function and determining the initialization parameters of the preset deep reinforcement learning algorithm, wherein the initialization parameters include value network parameters and action network parameters;

[0014] S2: taking the power absorbed and released by the electric energy storage system group as the initial population, calculating the IGD+ and HV of each individual in the initial population based on the initial value of the objective function, and taking the IGD+ and HV as the initial state of the initial population;

[0015] S3: Generate an action strategy and store action probability based on action network parameters according to the initial state;

[0016] S4: the initial population generates a first population of offspring in the function space of the objective function based on the update operator;

[0017] S5: merging the initial population and the first population to obtain a second population, applying a non-dominated sorting and a reference point-based hierarchical strategy to the second population to obtain a third population;

[0018] S6: Calculate the state of the third population, and obtain a reward value based on the state of the third population and the initial state;

[0019] S7: Obtaining experience based on the initial state, the stored action probability, the reward value, and the state of the third population, and putting the experience into an experience pool, wherein the experience pool obtains target value network parameters and target action network parameters based on preset deep reinforcement learning training value network parameters and action network parameters;

[0020] S8: Use the third population as the initial population, the state of the third population as the initial state, the target value network parameters as the value network parameters, and the target action network parameters as the action network parameters, repeat steps S3-S7 iterations, and determine the scheduling results of the distributed grid-connected electric energy storage system group based on the iterated population.

[0021] In a possible implementation, the obtaining the initial value of the objective function includes: setting a first initial value of the SOC of the battery energy storage system, setting a second initial value of the HSL of the hydrogen energy system, setting a third initial value of the photovoltaic power generation power, setting a fourth initial value of the electric vehicle flow data, and setting a fifth initial value of the load of the electric energy storage system;

[0022] The first initial value, the second initial value, the third initial value, the fourth initial value and the fifth initial value are input into the first objective function, the second objective function, the third objective function and the fourth objective function to obtain the first target initial value, the second target initial value, the third target initial value and the fourth target initial value.

[0023] In a possible implementation, the expression of the first objective function is:

[0024]

[0025] In the formula, represents the profit of the integrated energy storage system from selling electricity to electric vehicles; It represents the profit of mutual feedback between the integrated electric energy storage system and the distribution network; represents the profit obtained by the hydrogen energy storage system; m represents the number of electric energy storage systems; and are the discharge power and charging power of the distribution network in period t respectively; and are the prices of electricity sold to and purchased from the grid during period t, respectively; Power for charging electric vehicles; the price of charging electric vehicles from integrated electrical energy storage systems; and They are the capacity and price of hydrogen energy sold; and They are the capacity and price of hydrogen purchased;

[0026] The expression of the second objective function is:

[0027]

[0028] In the formula, Indicates time period t Load power; T=24; The peak-to-valley difference of the day;

[0029] The expression of the third objective function is:

[0030]

[0031] In the formula, Indicates the average load power in a day;

[0032] The expression of the fourth objective function is:

[0033] In the formula, Is a node The voltage amplitude; is the target voltage; N is the set of all nodes in the distribution network.

[0034] In a possible implementation, the battery energy storage system constraint includes upper and lower limit constraints of battery energy storage system capacity and upper and lower limit constraints of charge and discharge power. The expression of the upper and lower limit constraints of battery energy storage system capacity is:

[0035]

[0036] In the formula, is the minimum capacity of the battery subsystem; is the capacity of the battery subsystem at each moment; is the maximum capacity of the battery subsystem;

[0037] The expression of the upper and lower limits of the charge and discharge power is:

[0038]

[0039] In the formula, The charging power per hour for the battery subsystem; is the discharge power of the battery subsystem per hour; is the maximum discharge power of the battery subsystem; The maximum charging power for the battery subsystem;

[0040] The hydrogen energy storage system constraint includes a hydrogen energy storage system capacity upper limit constraint and a hydrogen charging and discharging constraint. The expression of the hydrogen energy storage system capacity upper limit constraint is:

[0041]

[0042] In the formula, is the minimum capacity of the hydrogen energy storage device; is the hydrogen storage level of the hydrogen energy storage device at each moment; is the maximum capacity of the hydrogen energy storage device;

[0043] The expression of the hydrogen charging and discharging constraint is:

[0044]

[0045] In the formula, one represents the first case, that is, C stores hydrogen energy and D releases hydrogen energy at the same time; two represents the second case, that is, D stores hydrogen energy and C releases hydrogen energy at the same time;

[0046] The expression of the power constraint of the distribution network is:

[0047]

[0048] In the formula, The power generated by the generator per hour; Release power to the integrated energy storage system group for the distribution network; absorbing power from the integrated electrical energy storage system for the distribution grid; For each hour load power; is the power loss of the grid line every hour; The value of is 1 or 0, indicating that the grid cannot release and absorb power at the same time.

[0049] In a possible implementation, it also includes adjusting the absorption and release of power of the battery energy storage system and the hydrogen energy storage system through preset adaptive balancing parameters.

[0050] In a possible implementation, when the battery energy storage system includes a first battery energy storage subsystem and a second battery energy storage subsystem, and the hydrogen energy storage system includes a first hydrogen energy storage subsystem and a second hydrogen energy storage subsystem, the preset adaptive balancing parameters include a first adaptive balancing parameter and a second adaptive balancing parameter; when the output power of the distribution network or the photovoltaic system is not greater than the power absorption upper limit or the available power space of the first battery energy storage subsystem, the first adaptive balancing parameter satisfies a first preset condition, and the formula of the first preset condition is:

[0051]

[0052] in, represents a first adaptive equalization parameter; The value of is 1 or 0, indicating that the grid cannot release and absorb power at the same time; fort At the moment the first battery energy storage subsystem absorbs power; for t The grid releases power at all times; The photovoltaic power released at time t; for t The fan releases power at all times; for t At time 1, the first hydrogen energy storage subsystem C absorbs power; is the total capacity of the battery subsystem; for t The state of charge of the BESS at the moment; is the maximum absorbed power of the battery subsystem; is the maximum state of charge of the battery subsystem; is the available power space; The remaining state of charge space of the first battery energy storage subsystem;

[0053] When the output power of the distribution network or the photovoltaic system is greater than the power absorption upper limit or the available power space of the first battery energy storage subsystem, the first adaptive balancing parameter meets the second preset condition. The formula of the second preset condition is:

[0054]

[0055]

[0056] in, Indicates the power absorbed by the first battery energy storage subsystem at time t; Indicates the power that cannot be absorbed by the first battery energy storage subsystem at time t and is transferred to the first hydrogen energy storage subsystem; Represents the power absorbed by the first hydrogen energy storage subsystem at time t.

[0057] In a possible implementation, when the power absorbed by the distribution network or the electric vehicle EV is not greater than the maximum released power or the remaining power of the second battery energy storage subsystem, the second adaptive balancing parameter satisfies a third preset condition, and the formula of the third preset condition is:

[0058]

[0059] In the formula, represents a second adaptive equalization parameter; for t The grid absorbs power at all times; For electric vehicles t Power is required at all times; for t At this moment, the second battery energy storage subsystem releases power; is the remaining capacity of the second battery energy storage subsystem; The remaining power of the second battery energy storage subsystem; Release the power ceiling for the battery subsystem; is the minimum state of charge of the battery subsystem; is the total capacity of the battery subsystem; Represented as the remaining state of charge space of the first battery energy storage subsystem; express t The absorbed power of the electric vehicle at any moment; The value of is 1 or 0, indicating that the grid cannot release and absorb power at the same time;

[0060] When the power absorbed by the distribution network or the electric vehicle EV is not greater than the maximum released power or the remaining power of the second battery energy storage subsystem, the second adaptive balancing parameter meets the fourth preset condition. The formula of the fourth preset condition is:

[0061]

[0062]

[0063]

[0064] In the formula, represents the output power of the fuel cell at time t; represents the absorbed power of the electric vehicle at time t.

[0065] On the other hand, the present invention also provides a dispatching device for distributed grid-connected power storage system group, comprising:

[0066] Construct an objective function module, which is used to construct the objective function and constraint conditions of each electric energy storage system in the electric energy storage system group. The objective function includes a first objective function constructed by maximizing the benefit of the electric energy storage system, a second objective function constructed by minimizing the peak-to-valley difference, a third objective function constructed by minimizing the load standard deviation, and a fourth objective function constructed by minimizing the voltage amplitude deviation. The constraint conditions include battery energy storage system constraints, hydrogen energy storage system constraints, and distribution network power balance constraints;

[0067] An operator selection module, used to select a global search operator from a preset global search operator pool and a local search operator from a preset local search operator pool;

[0068] An update operator acquisition module, used to obtain an update operator based on a global search operator and a local search operator;

[0069] The scheduling module is used to select operators in the update operators based on a preset deep reinforcement learning algorithm to solve the objective function under constraints and obtain the scheduling results of the distributed grid-connected electric energy storage system group.

[0070] On the other hand, the present invention also provides an electronic device, including a memory and a processor, wherein:

[0071] The memory is used to store programs;

[0072] The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the scheduling method for distributed grid-connected power storage system group described in any one of the above implementations.

[0073] The beneficial effect of the present invention is as follows: a scheduling method for distributed grid-connected energy storage system groups provided by the present invention first constructs the objective function and constraints of each energy storage system in the energy storage system group, the objective function includes a first objective function constructed with the maximum benefit of the energy storage system, a second objective function constructed with the minimum peak-to-valley difference, a third objective function constructed with the minimum load standard deviation, and a fourth objective function constructed with the minimum voltage amplitude deviation, and the constraints include battery energy storage system constraints, hydrogen energy storage system constraints, and distribution network power balance constraints, and then selects a global search operator from a preset global search operator pool, selects a local search operator from a preset local search operator pool, and obtains an update operator based on the global search operator and the local search operator, so that the operator in the deep reinforcement learning algorithm is replaced by the update operator, and selects an operator from the update operator based on the preset deep reinforcement learning algorithm to solve the objective function under the constraints to obtain the scheduling result of the distributed grid-connected energy storage system group. When the existing global or local operators are difficult to improve the diversity of solutions at a certain stage, the present invention reselects the most suitable search operator based on the deep reinforcement learning algorithm, so as to better balance exploration and utilization, improve the diversity and stability of the population, optimize the overall performance, and thus improve the efficiency of the integrated electric energy storage system scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A method flow chart of an embodiment of a method for dispatching a distributed grid-connected electric energy storage system group provided by the present invention;

[0075] Figure 2 A diagram of integrating an integrated electric energy storage system group into a distribution network system in a method for distributing an electric energy storage system group in a distributed grid-connected dispatching method provided by the present invention;

[0076] Figure 3 For the present invention Figure 1 A schematic flow chart of an embodiment of S104;

[0077] Figure 4 A schematic diagram of a node of a distribution network 30 where the integrated electric energy storage system provided by the present invention is connected to the distribution network under normal conditions;

[0078] Figure 5A schematic diagram of a comprehensive electric energy storage system provided by the present invention connected to a distribution network 30 node under extreme conditions;

[0079] Figure 6 A diagram showing the output of the electric vehicle, photovoltaic power generation system and wind turbine system of the integrated electric energy storage system 1 in the integrated electric energy storage system group provided by the present invention within 24 hours;

[0080] Figure 7 A diagram showing the output of the electric vehicle, photovoltaic power generation system and wind turbine system of the integrated electric energy storage system 2 in the integrated electric energy storage system group provided by the present invention within 24 hours;

[0081] Figure 8 A diagram showing the output of the electric vehicle, photovoltaic power generation system and wind turbine system of the integrated electric energy storage system 8 in the integrated electric energy storage system group provided by the present invention within 24 hours;

[0082] Fig. 9 A daily load curve diagram of the comprehensive electric energy storage system of Case 1 provided by the present invention;

[0083] Fig.10 A voltage distribution diagram of the integrated electric energy storage system of Case 1 provided by the present invention;

[0084] Fig.11 A daily load curve diagram of the comprehensive electric energy storage system of Case 2 provided by the present invention;

[0085] Fig.12 A voltage distribution diagram of the integrated electric energy storage system of Case 2 provided by the present invention;

[0086] Fig.13 A daily load curve diagram of the comprehensive electric energy storage system of Case 3 provided by the present invention;

[0087] Fig.14 The voltage distribution diagram of the integrated electric energy storage system of Case 3 provided by the present invention;

[0088] Fig.15 A power diagram of the integrated electric energy storage system of Case 1 provided by the present invention;

[0089] Fig.16 A power diagram of the integrated electric energy storage system of Case 2 provided by the present invention;

[0090] Fig.17 A power diagram of the integrated electric energy storage system of Case 3 provided by the present invention;

[0091] Fig.18 A profit diagram of the integrated electric energy storage system of Case 1 provided by the present invention;

[0092] Fig.19 A profit diagram of the integrated electric energy storage system of Case 2 provided by the present invention;

[0093] Fig. 20 A profit diagram of the integrated electric energy storage system of Case 3 provided by the present invention;

[0094] Fig.21 A schematic flow chart of an embodiment of a dispatching device for distributed grid-connected power storage system group provided by the present invention;

[0095] Fig. 22 A schematic structural diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0096] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0097] The present invention provides a method, device and electronic equipment for dispatching a distributed grid-connected electric energy storage system group, which are described below respectively.

[0098] Figure 1 A schematic flow chart of an embodiment of a method for dispatching a distributed grid-connected electric energy storage system group provided by the present invention is shown in FIG. Figure 1 As shown, the dispatching method for distributed grid-connected power storage system group includes:

[0099] S101, constructing the objective function and constraint conditions of each electric energy storage system in the electric energy storage system group, wherein the objective function includes a first objective function constructed by maximizing the benefit of the electric energy storage system, a second objective function constructed by minimizing the peak-to-valley difference, a third objective function constructed by minimizing the load standard deviation, and a fourth objective function constructed by minimizing the voltage amplitude deviation, and the constraint conditions include battery energy storage system constraints, hydrogen energy storage system constraints, and distribution network power balance constraints;

[0100] S102, selecting a global search operator from a preset global search operator pool, and selecting a local search operator from a preset local search operator pool;

[0101] S103, obtaining an update operator based on the global search operator and the local search operator;

[0102] S104. Based on a preset deep reinforcement learning algorithm, an operator is selected from the update operators to solve the objective function under constraints, and a dispatch result of the distributed grid-connected electric energy storage system group is obtained.

[0103] Compared with the prior art, the present embodiment provides a method for dispatching a distributed grid-connected energy storage system group. First, the objective function and constraint conditions of each energy storage system in the energy storage system group are constructed. The objective function includes a first objective function constructed with the maximum benefit of the energy storage system, a second objective function constructed with the minimum peak-to-valley difference, a third objective function constructed with the minimum load standard deviation, and a fourth objective function constructed with the minimum voltage amplitude deviation. The constraint conditions include battery energy storage system constraints, hydrogen energy storage system constraints, and distribution network power balance constraints. Then, a global search operator is selected from a preset global search operator pool, and a local search operator is selected from a preset local search operator pool. An update operator is obtained based on the global search operator and the local search operator, so that the operator in the deep reinforcement learning algorithm is replaced with the update operator. Based on the preset deep reinforcement learning algorithm, an operator is selected from the update operator to solve the objective function under the constraint conditions to obtain the dispatching result of the distributed grid-connected energy storage system group. When the existing global or local operators are difficult to improve the diversity of solutions at a certain stage, the present invention reselects the most suitable search operator based on the deep reinforcement learning algorithm, so as to better balance exploration and utilization, improve the diversity and stability of the population, optimize the overall performance, and thus improve the efficiency of the integrated electric energy storage system scheduling.

[0104] In a specific embodiment of the present invention, Figure 2 The figure shows the system diagram of the integrated energy storage system group incorporated into the distribution network. The integrated energy storage system group is composed of multiple integrated energy storage systems (IES). Each integrated energy storage system is integrated by electric vehicles (EV), wind turbines (WT), photovoltaic power generation (PV), hydrogen production / storage systems (HPSS), battery energy storage systems (BESS), fuel cells (FC), etc. Each integrated energy storage system is distributedly connected to different nodes of the distribution network. When the distribution network is operating normally, these integrated energy storage systems can coordinately adjust load fluctuations and realize the function of peak shaving and valley filling; when the distribution network encounters extreme natural disasters, some power sources may be paralyzed, but the integrated energy storage system group can improve the resilience of the distribution network through multi-energy complementarity while taking into account its own economic benefits.

[0105] The following introduces the battery energy storage system model, wind turbine model, photovoltaic model and hydrogen energy system model in the comprehensive electric energy storage system. It should be understood that the objective function is obtained based on the battery energy storage system model, wind turbine model, photovoltaic model and hydrogen energy system model.

[0106] Battery Energy Storage Model:

[0107] In order to simultaneously provide electric energy to electric vehicles, interact with the distribution network, and maximize the power generation of the wind and solar systems, the battery energy storage system model in this embodiment is composed of two subsystems. The two subsystems are energy storage subsystem A and energy storage subsystem B. According to the remaining capacity of A and B, the respective connection objects are flexibly adjusted to optimize the performance of the entire system. In order to introduce the principle of the model in detail, it is described through the following content.

[0108] During initialization, it is first assumed that energy storage subsystem A absorbs power and energy storage system B releases power. The formula is as follows:

[0109]

[0110] In the formula, For in time t Input power of energy storage subsystem A at the moment; is the output power of energy storage subsystem B at time t; and They are charging efficiency and discharging efficiency, respectively, and are coefficients between 0 and 1.

[0111] Energy storage subsystem A will be connected to the wind and solar system. At the same time, if the distribution network is releasing power, it will also be connected to energy storage subsystem A. The formula is as follows:

[0112]

[0113] In the formula The time of the photovoltaic system t of power generation; The power is released to the distribution grid.

[0114] If the distribution network is t When is absorbing electrical energy, then The formula is as follows:

[0115]

[0116] Energy storage subsystem B will be connected to the electric vehicle. At the same time, if the distribution network absorbs electric energy, it will also be connected to energy storage subsystem B. The formula is as follows:

[0117]

[0118] In the formula is the power absorbed by the electric vehicle at time t; is the power absorbed by the distribution network at time t.

[0119] If the distribution network releases power at time t, then The formula is as follows:

[0120]

[0121] According to the SOC status of energy storage subsystems A and B, decide whether to swap the roles of absorbing and releasing power. Assuming that one of the following conditions is met at time t, the connection object is swapped:

[0122] (1) If , then the energy storage subsystems A and B exchange connection objects.

[0123] (2) If , then the energy storage subsystem B and A exchange connection objects.

[0124] Wind Turbine Model:

[0125] Wind speed and tower height directly affect the power generated by a wind turbine. A wind turbine generator model is used to estimate the power generation of a wind turbine with an integrated energy storage system:

[0126]

[0127] in , , , and are the actual wind speed at the wind turbine hub height (m / s), the rated power of the wind turbine (kW), the rated wind speed (m / s) and the cut-in and cut-out wind speeds (m / s).

[0128] Photovoltaic model:

[0129] The amount of electricity generated by photovoltaics is determined by several parameters, including the solar load on the surface of the photovoltaic panel ( ) in W / ㎡, the working temperature of the photovoltaic panel ( ) in °C, inverter efficiency ( ) and loss( ) is in %, and the power generation of photovoltaic modules can be expressed as:

[0130]

[0131] in The reference temperature is 25℃( ) under the benchmark efficiency of photovoltaic modules, in %, is the temperature coefficient, is the planned area of ​​the PV array in the hybrid system (㎡). Most of the data in the formula are provided by the manufacturer, as shown in Table 1.

[0132] Table 1: Photovoltaic panel parameters

[0133]

[0134] The hydrogen energy system consists of three parts: electrolyzer (for producing hydrogen), compressor, hydrogen storage tank and fuel cell. The four module models are described as follows:

[0135] Hydrogen is obtained by electrolyzing water. The polymer electrolyte membrane (PEM) electrolyzer has the advantages of low operating temperature and simple structure. When the pressure reaches 200 bar, the PEM electrolyzer can produce high-purity hydrogen. The calculation model formula for the electrolysis conversion of water in the PEM electrolyzer is:

[0136]

[0137] in is the electrolyzer efficiency, is the mass flow rate of hydrogen at the output of the electrolyser, is the lower calorific value of hydrogen, yes t The power consumed by the electrolytic cell during the period and the specific parameter values ​​in the electrolytic cell are shown in Table 2.

[0138] Table 2: Electrolyzer parameters

[0139]

[0140] Compressor Model

[0141] When water electrolysis is in progress to produce hydrogen, the compressor also runs at the same time. The power of the compressor comes from the PV module, WT module and PG. The function of the compressor is to compress hydrogen from the low-pressure storage tank to the high-pressure storage tank. Assuming that there is no hydrogen loss during the compression process, the compressor model formula is:

[0142]

[0143] in, represents the specific heat of hydrogen at constant pressure, Indicates at time t The power consumed by the compressor motor is and are the inlet pressure and outlet pressure of the compressor, r represents the specific heat ratio of hydrogen, represents the hydrogen temperature at the compressor inlet, is the compressor isentropic efficiency, is the mechanical efficiency. The specific parameter values ​​of the compressor are shown in Table 3.

[0144] Table 3: Compressor parameters

[0145]

[0146] Hydrogen storage model

[0147] HT is used to store hydrogen from the compressor and then supply it to the distributor to meet the hydrogen demand of HVs in each period. This paper converts the change form of hydrogen content in HT into the change form of electricity in BESS. Therefore, the change of hydrogen content in HT is as follows:

[0148]

[0149] in, and Respectively in time t and the hydrogen storage content in HT at t+1, in %. is the hydrogen supplied to the HVs at time t. is the hydrogen loss coefficient occurring in HT. is the total capacity of HT. The specific parameter values ​​of the hydrogen storage tank are shown in Table 4.

[0150] Table 4: Hydrogen storage tank parameters

[0151]

[0152] Fuel cell model

[0153] Fuel cell output power The formula is:

[0154]

[0155] In the formula is the number of cells connected in series; is the open circuit voltage; is the reversible open circuit voltage; R is the universal gas constant; is the battery stack temperature; , and are the partial pressures of hydrogen, oxygen and water vapor respectively; F is the Faraday constant; It's entropy change.

[0156] It should be noted that, from the perspective of the operation of the integrated electric energy storage system group, the comprehensive benefit of the electric energy storage system group is a basic goal. Energy management of electric vehicles and power grids based on time-of-use electricity prices can maximize the operating benefits of the integrated electric energy storage system group. In some embodiments of the present invention, the first objective function is constructed by maximizing the benefit of the electric energy storage system, and the expression of the first objective function is:

[0157]

[0158] In the formula, represents the profit of the integrated energy storage system from selling electricity to electric vehicles; It represents the profit of mutual feedback between the integrated electric energy storage system and the distribution network; represents the profit obtained by the hydrogen energy storage system; m represents the number of electric energy storage systems; and are the discharge power and charging power of the distribution network in period t respectively; and are the prices of electricity sold to and purchased from the grid during period t, respectively; Power for charging electric vehicles; the price of charging electric vehicles from integrated electrical energy storage systems; and They are the capacity and price of hydrogen energy sold; and They are the capacity and price of hydrogen purchased;

[0159] It should be noted that the power demand of the load nodes at each moment of the distribution network is obtained through flow calculation. In order to improve the stability of the distribution network and reduce the difference between the maximum and minimum values ​​of the distribution network load, peak shaving and valley filling can be effectively achieved, thereby maintaining the balance and reliable operation of the power system. The second objective function constructed with the minimum peak-to-valley difference is expressed as:

[0160]

[0161] In the formula, Indicates time period t Load power; T=24; The peak-to-valley difference of the day;

[0162] It should be noted that by minimizing the load standard deviation of the distribution network, load smoothing and balancing can be achieved, thereby improving the stability and operation efficiency of the distribution network. The third objective function constructed with the minimum load standard deviation is expressed as follows:

[0163]

[0164] In the formula, Indicates the average load power in a day;

[0165] It should be noted that the minimum voltage amplitude deviation of the distribution network usually refers to the difference between all node voltages and the target voltage. This indicator can be used to measure the voltage stability of the distribution network. The fourth objective function constructed with the minimum voltage amplitude deviation is expressed as:

[0166] In the formula, Is a node The voltage amplitude; is the target voltage, usually the rated voltage, such as 1.0.pu; N is the set of all nodes in the distribution network.

[0167] It should be noted that the objective function of maximizing the revenue of the above-mentioned electric energy storage system group, minimizing the peak-to-valley difference of the distribution network and minimizing the load standard deviation uses three constraints, namely, battery energy storage system constraints, hydrogen energy storage device constraints and distribution network constraints. In some embodiments of the present invention, the battery energy storage system constraints include upper and lower limit constraints on the capacity of the battery energy storage system and upper and lower limit constraints on the charge and discharge power. The expression of the upper and lower limit constraints on the capacity of the battery energy storage system is:

[0168]

[0169] In the formula, is the minimum capacity of the battery subsystem; is the capacity of the battery subsystem at each moment; is the maximum capacity of the battery subsystem;

[0170] The expression of the upper and lower limits of the charge and discharge power is:

[0171]

[0172] In the formula, The charging power per hour for the battery subsystem; is the discharge power of the battery subsystem per hour; is the maximum discharge power of the battery subsystem; The maximum charging power for the battery subsystem;

[0173] The hydrogen energy storage system constraint includes a hydrogen energy storage system capacity upper limit constraint and a hydrogen charging and discharging constraint. The expression of the hydrogen energy storage system capacity upper limit constraint is:

[0174]

[0175] In the formula, is the minimum capacity of the hydrogen energy storage device; is the hydrogen storage level of the hydrogen energy storage device at each moment; is the maximum capacity of the hydrogen energy storage device;

[0176] The expression of the hydrogen charging and discharging constraint is:

[0177]

[0178] In the formula, one represents the first case, that is, C stores hydrogen energy and D releases hydrogen energy at the same time; two represents the second case, that is, D stores hydrogen energy and C releases hydrogen energy at the same time;

[0179] The expression of the power constraint of the distribution network is:

[0180]

[0181] In the formula, The power generated by the generator per hour; Release power to the integrated energy storage system group for the distribution network; absorbing power from the integrated electrical energy storage system for the distribution grid; For each hour load power; is the power loss of the grid line every hour; The value of is 1 or 0, indicating that the power grid cannot release and absorb power at the same time. If the power grid is releasing power at this time, then it is 1, otherwise it is 0.

[0182] It should be noted that in order to maximize the complementary advantages of the battery energy storage system and the hydrogen energy system, two battery subsystems (A and B) and two hydrogen energy subsystems (C and D) are designed. Through the adjustment of adaptive balancing parameters, the system can dynamically adjust the power absorption and release process of the battery energy storage and hydrogen energy systems, thereby optimizing the overall performance.

[0183] In some embodiments of the present invention, the power absorption and release of the battery energy storage system and the hydrogen energy storage system are adjusted by preset adaptive balancing parameters. The preset adaptive balancing parameters include a first adaptive balancing parameter and a second adaptive balancing parameter.

[0184] In some embodiments of the present invention, when the battery energy storage system includes a first battery energy storage subsystem A and a second battery energy storage subsystem B, and the hydrogen energy storage system includes a first hydrogen energy storage subsystem C and a second hydrogen energy storage subsystem D, the preset adaptive balancing parameters include a first adaptive balancing parameter and a second adaptive balancing parameter.

[0185] In order to maximize the complementary advantages of battery energy storage systems and hydrogen energy systems, this paper designs two battery subsystems (A and B) and two hydrogen energy subsystems (C and D). According to their respective energy storage and release characteristics, the battery and hydrogen energy systems achieve cascade coupling absorption and release. Cascade coupling absorption means that the battery energy storage system absorbs the power output of the distribution network and photovoltaics, while the hydrogen energy system mainly absorbs wind energy; when the battery energy storage system reaches the absorption limit or capacity limit , the hydrogen energy system absorbs it alternately. The cascade coupling release is that the battery energy storage system releases energy to the grid first; when the battery release power reaches the upper limit Or the battery capacity reaches the lower limit When the hydrogen energy is converted into electricity from the hydrogen energy system, the fuel cell takes over the release of energy to the power grid. In addition, in order to ensure the adaptive balance of the system, the first battery energy storage subsystem A and the second battery energy storage subsystem B can dynamically switch the connection objects according to their respective SOC states. The hydrogen energy system is the same. The specific process is as follows:

[0186] First, the power output by the distribution network or photovoltaic system is preferentially released to the first battery energy storage subsystem A, and the wind energy is released to the first hydrogen energy storage subsystem C. There are two cases, and the formula is as follows:

[0187] The first case is that the power output of the distribution network or photovoltaic system is not greater than the power absorption limit of the first battery energy storage subsystem A. or available power headroom When the first adaptive equalization parameter satisfies the following formula, that is, the formula of the first preset condition is:

[0188]

[0189] in, Indicates the first adaptive equalization parameter, select and The minimum power between The value of is 1 or 0, indicating that the grid cannot release and absorb power at the same time; for t At the moment, the first battery energy storage subsystem A absorbs power; The power released by the power grid at time t; for t Photovoltaic releases power at all times; The power released by the fan at time t; is the power absorbed by the first hydrogen energy storage subsystem C at time t; is the total capacity of the battery subsystem; is the charge state of BESS at time t; is the maximum absorbed power of the battery subsystem; is the maximum state of charge of the battery subsystem; is the available power space; The remaining state of charge space of the first battery energy storage subsystem A;

[0190] In the second case, when the output power of the distribution network or the photovoltaic system is greater than the power absorption upper limit or the available power space of the first battery energy storage subsystem A, the first adaptive balancing parameter satisfies the following formula, that is, the formula of the second preset condition is:

[0191]

[0192]

[0193] in, Indicates the power absorbed by the first battery energy storage subsystem at time t; represents the power that cannot be absorbed by the first battery energy storage subsystem A at time t and is transferred to the first hydrogen energy storage subsystem C; Represents the power absorbed by the first hydrogen energy storage subsystem C at time t.

[0194] If the first hydrogen energy storage subsystem C reaches the absorption upper limit , the excess power is converted into hydrogen energy for sale, the formula is as follows: Substitute the following formula:

[0195]

[0196]

[0197]

[0198]

[0199] Get power and convert it into hydrogen energy , the specific formula is as follows:

[0200]

[0201] In the formula for t Hydrogen energy is always on sale; Provides an upper limit for the absorption of the hydrogen energy subsystem; for t Hydrogen energy state at all times.

[0202] In addition, the power grid or electric vehicle absorbs power, and the battery subsystem C preferentially transmits power to the distribution network or electric vehicle EV. When the battery subsystem reaches the upper limit of the released power or capacity limit When the battery subsystem is unable to compensate for the power distribution grid or EV, the power is assisted by the fuel cell, and the hydrogen of the fuel cell comes from the hydrogen energy subsystem D. There are two specific cases.

[0203] In the first case, in some embodiments of the present invention, when the power absorbed by the distribution network or the electric vehicle EV is not greater than the maximum released power or the remaining power of the second battery energy storage subsystem B, the second adaptive balancing parameter satisfies the following formula, that is, the formula of the third preset condition is:

[0204]

[0205] In the formula, represents a second adaptive equalization parameter; is the power absorbed by the grid at time t; For electric vehicles t Power is required at all times; The power released by the second battery energy storage subsystem B at time t; is the remaining capacity of the second battery energy storage subsystem B; The remaining power of the second battery energy storage subsystem; Release the power ceiling for the battery subsystem; is the minimum state of charge of the battery subsystem; is the total capacity of the battery subsystem; Represented as the remaining state of charge space of the first battery energy storage subsystem; represents the absorbed power of the electric vehicle at time t; The value of is 1 or 0, indicating that the grid cannot release and absorb power at the same time;

[0206] In the second case, when the power absorbed by the distribution network or the electric vehicle EV is not greater than the maximum released power or the remaining power of the second battery energy storage subsystem, the second adaptive balancing parameter satisfies the following formula, that is, the formula of the fourth preset condition is:

[0207]

[0208]

[0209]

[0210] In the formula, express t The output power of the fuel cell at any moment; express t The absorbed power of the electric vehicle at each moment.

[0211] If the second hydrogen energy subsystem D reaches the release lower limit , hydrogen will be purchased for fuel cells to convert into power, the formula is as follows:

[0212] Will Substitute it into the following formula, the following formula is:

[0213]

[0214]

[0215] Obtain hydrogen energy needed to supply fuel cells capacity.

[0216]

[0217] In the formula for t Hydrogen energy state of the second hydrogen energy subsystem D at time t; is the minimum hydrogen energy state of the hydrogen energy subsystem; for t The hydrogen energy capacity that needs to be provided to the fuel cell at all times; The hydrogen capacity that needs to be purchased.

[0218] Finally, if or , then the first battery energy storage subsystem A and the second battery energy storage subsystem B exchange connection objects. Similarly, or Then the first hydrogen energy subsystem C and the second hydrogen energy subsystem D will exchange connection objects.

[0219] The cascade utilization optimization strategy based on batteries and hydrogen energy systems provided in this embodiment is designed to maximize the complementary advantages of multiple batteries and hydrogen energy systems and ensure that the system can fully exert its performance and effectively respond to challenges when power grid failures or energy supply shortages occur.

[0220] It should be noted that in order to solve the optimal strategy for the mutual feedback relationship between the integrated electric energy storage system group and the distribution network, this paper proposes an adaptive selection mechanism based on deep reinforcement learning, which aims to dynamically adjust the use of global and local search operators according to the different solution stages of the same multi-objective optimization problem. When the existing global or local operators are difficult to improve the diversity of solutions at a certain stage, the algorithm will reselect the most suitable search operator to better balance exploration and utilization and optimize the overall performance.

[0221] Specifically, the proximal policy optimization algorithm 2 (PPO2) was adopted as the core deep reinforcement learning algorithm, and two algorithm pools were designed: the local search operator pool (LSOP) and the global search operator pool (GSOP).

[0222] 1) Local Search Operator Pool (LSOP): Contains a variety of local search operators, such as polynomial mutation, Gaussian mutation, and differential mutation operators, that is, LSOP = {L1, L2, ..., Li}, where Li is a local search operator.

[0223] 2) Global Search Operator Pool (GSOP): Contains a variety of global search operators, such as two-point crossover, uniform crossover, differential crossover and simulated binary crossover (SBX) operators, that is, GSOP = {G1, G2, ..., Gi}, where Gi is a global search operator.

[0224] During the execution of the algorithm, a local search operator and a global search operator are selected from LSOP and GSOP respectively to form the action of the PPO2 agent, that is, action = {Li, Gi}. This newly composed hybrid operator replaces the update operator of the multi-objective optimization algorithm framework, thus forming a new multi-objective optimization algorithm.

[0225] In order to evaluate the status of the population, we introduced two indicators: IGD+ and HV:

[0226] 1) IGD+: reflects the convergence and diversity of the population.

[0227] 2) HV (Hypervolume): can reflect the overall quality and distribution of the solution set.

[0228] Based on these indicators, the state of the population is defined as S = {IGD+, HV}. After each iteration, the change in the population state is used to calculate the reward R, which is calculated as: R = {IGD+ - IGD+`, HV` - HV}, where IGD+ and HV` are calculated for the state of the new population in the next iteration.

[0229] Based on the above definitions, a constrained multi-objective optimization model and pseudocode based on deep reinforcement learning adaptive operator hybrid strategy are proposed. The algorithm model has four steps:

[0230] 1) Evolution Procedure: During the evolution process, the MOEA algorithm (evolutionary algorithm) generates offspring through the selected global and local search operators (determined by PPO2), and survives to the next generation through non-dominated sorting.

[0231] 2) Interaction Procedure: During the interaction process, the agent obtains a reward value for taking actions by interacting in the environment. This reward value can affect the agent's next action choice.

[0232] 3) Learning Procedure: After the agent receives feedback from the environment, it learns and improves the action network 2 and the value network in the learning process, and updates the action network 1 according to the action network 2. First, the experience [s, a, r, s_, p] collected during the interaction process is added to the experience pool, where s is the state, a is the action, r is the reward, s_ is the next state, and p is the probability of the action. Secondly, the Value network is trained based on the experience data in the experience pool, and finally the advantage function value is obtained, and the advantage function value (adv) is used as the reward value. Finally, the Actor network 2 is jointly updated based on the experience pool and the updated Value network.

[0233] 4) Decision Procedure: After PPO2 is trained, the agent estimates the reward values ​​of all actions based on the population state and selects the action with the highest reward.

[0234] The pseudo code is as follows:

[0235]

[0236] In some embodiments of the present invention, Figure 3 As shown, the preset deep reinforcement learning algorithm is used to select an operator in the update operator to solve the objective function under the constraint condition, and obtain the dispatch result of the distributed grid-connected electric energy storage system group, including:

[0237] S301: Obtaining an initial value of the objective function and determining initialization parameters of a preset deep reinforcement learning algorithm, wherein the initialization parameters include value network parameters and action network parameters;

[0238] S302: taking the power absorbed and released by the electric energy storage system group as the initial population, calculating the IGD+ and HV of each individual in the initial population based on the initial value of the objective function, and taking the IGD+ and HV as the initial state of the initial population;

[0239] S303: Generate an action strategy and store action probability based on action network parameters according to the initial state;

[0240] S304: The initial population generates a first population of offspring in the function space of the objective function based on the update operator;

[0241] S305: merging the initial population and the first population to obtain a second population, applying a non-dominated sorting and a reference point-based stratification strategy to the second population to obtain a third population;

[0242] S306: Calculate the state of the third population, and obtain a reward value based on the state of the third population and the initial state;

[0243] S307: Obtaining experience based on the initial state, the stored action probability, the reward value, and the state of the third population, and putting the experience into an experience pool, wherein the experience pool obtains target value network parameters and target action network parameters based on preset deep reinforcement learning training value network parameters and action network parameters;

[0244] S308: Use the third population as the initial population, the state of the third population as the initial state, the target value network parameters as the value network parameters, and the target action network parameters as the action network parameters, repeat steps S303-S307, and determine the scheduling results of the distributed grid-connected electric energy storage system group based on the iterated population.

[0245] In some embodiments of the present invention, the obtaining of the initial value of the objective function includes: setting a first initial value of the SOC of the battery energy storage system, setting a second initial value of the HSL of the hydrogen energy system, setting a third initial value of the photovoltaic power generation power, setting a fourth initial value of the electric vehicle flow data, and setting a fifth initial value of the load of the electric energy storage system;

[0246] The first initial value, the second initial value, the third initial value, the fourth initial value and the fifth initial value are input into the first objective function, the second objective function, the third objective function and the fourth objective function to obtain the first target initial value, the second target initial value, the third target initial value and the fourth target initial value.

[0247] In a specific embodiment of the present invention, an adaptive operator hybrid strategy based on proximal strategy optimization algorithm 2 is used to solve the objective function of the integrated electric energy storage system. Specifically, the algorithm parameters are initialized: the population size is set to 100, and the maximum number of iterations is 500. Each algorithm in this article is run 30 times. The model parameters are initialized as follows: the battery capacity of a single integrated electric energy storage system is 1000 kWh, and the number of batteries is 50.

[0248] Figure 4 The 30-node test system of the distribution network under normal conditions is shown, in which the integrated electric energy storage system access points are represented by black solid circles, the integrated electric energy storage system groups are connected separately, and the power supply is represented by red solid circles. Figure 5 The 30-node test system of the distribution network under extreme conditions is shown, which is disconnected from the main grid and has only one power supply to maintain power supply. Table 5 lists the load level and power purchase price of the distribution network for 24 hours. Figures 6 to 8The output of PV and WT of 3 IES within 24 hours and the 24-hour load demand of EV are shown. In the actual application, the output of PV and WT of 15 IES within 24 hours and the 24-hour load demand of EV are selected. The photovoltaic and wind speed data in PV and WT are provided by Weatherbit weather data service provider. The selected locations include Hohhot, Lanzhou, Urumqi, Xining, Zhongwei, Weinan, Jinan, Zhangjiakou, Aba, Haikou, Baotou, Dingxi, Yushu, Guiyang and Bijie in China. The time is September 2, 2024. The 24-hour load demand of EV is calculated by the Monte Carlo load forecasting method. In order to be closer to the actual working conditions, 20% random disturbance is further applied to the photovoltaic and wind speed data and the traffic flow data in solving the objective function to simulate the uncertainty in the real environment.

[0249] Table 5: 24-hour load levels and electricity prices

[0250]

[0251] In order to verify that under extreme conditions (17 to 21 hours of peak load, the distribution network is disconnected from the main grid and 5 power sources fail, leaving only 1 power node to supply power, such as Figure 5 As shown, the rest of the time is normal operation), the impact of 15 integrated energy stations connected to different nodes of the distribution network, this embodiment uniformly sets the photovoltaic (PV) and wind power (WT) outputs and electric vehicle (EV) demand loads of these 15 stations.

[0252] The performance of the adaptive search operator selection mechanism based on the proximal strategy optimization algorithm 2 proposed in the present invention, namely the ASOS-DRL algorithm and other algorithms on different objective functions is summarized as follows:

[0253] The first objective function: As shown in Table 6, the ASOS-DRL algorithm outperforms the other five algorithms in terms of both Min value and Mean value, showing its advantage in optimizing the overall performance.

[0254]

[0255] Second objective function: As shown in Table 7, Spea2 and Ctaea algorithms outperform ASOS-DRL in terms of Min value. ASOS-DRL algorithm outperforms other algorithms in terms of Max value and Mean value, especially in the optimization effect of Max value.

[0256]

[0257] The third objective function: As shown in Table 8, the ASOS-DRL algorithm performs best in terms of Min value and Mean value, showing its advantage in average performance.

[0258]

[0259] The fourth objective function: As shown in Table 9, the ASOS-DRL algorithm performs best in terms of Min value, Max value and Mean value, indicating that it has significant advantages in comprehensively optimizing this objective function.

[0260]

[0261] In summary, the ASOS-DRL algorithm performs well in terms of the Min and Mean values ​​of most objective functions, showing its advantages in overall performance optimization.

[0262] The extreme conditions (17 to 21 hours of peak load period, the distribution network is disconnected from the main network and 5 power supplies fail, leaving only 1 power supply node for power supply, such as Figure 5 As shown, the rest of the time is normal operation) and under normal conditions, the impact of the integrated energy storage system group connected to the distribution network on load fluctuations, peak-to-valley differences and voltage amplitude is analyzed. We analyzed the following three scenarios:

[0263] 1) Case 1: The overall benefit of the integrated electric energy storage system group is maximized.

[0264] 2) Case 2: Balanced benefits among sites in the integrated electric energy storage system group.

[0265] The number of targets in Case 2 is 18, which are the distribution network's response to load fluctuation, peak-to-valley difference and voltage amplitude, as well as the economic benefits of 15 IES.

[0266] Under normal conditions (1 to 24 hours when all power nodes are operating normally and the distribution network is connected to the main network, such as Figure 4 As shown in the figure, the following scenario case 3 is analyzed: the overall benefit maximization of the integrated electric energy storage system group.

[0267] Through the comparative analysis of these three cases, we can clearly understand the different effects of the integrated energy storage system group connected to the distribution network on load fluctuations, peak-to-valley differences and voltage amplitudes under extreme conditions and normal conditions. In addition, it is further analyzed that in extreme cases, whether to focus on the overall economic benefits of the integrated energy storage system group or to optimize the benefits of each site separately will have a significant difference in the stability and performance of the distribution network.

[0268] Cases 1-3 all randomly select scheduling solutions in the objective function using the ASOS-DRL algorithm. Figures 9 to 14The 24-hour daily load curves and 30-node voltage distributions for Cases 1-3 are shown. Fig. 9 and Fig.10 The daily load curve and voltage distribution diagram of Case 1 are shown respectively. After connecting to IESG, the load volatility of the distribution network is reduced by 47.6%, the peak-to-valley difference is reduced by 39.3%, and the voltage amplitude deviation is reduced by 12.7% compared with the original. Fig.11 and Fig.12 The daily load curve and voltage distribution diagram of Case 2 are shown respectively. After connecting to IESG, the load volatility of the distribution network is reduced by 44.3%, the peak-to-valley difference is reduced by 38.2%, and the voltage amplitude deviation is reduced by 9.9% compared with the original. Fig.13 and Fig.14 The daily load curve and voltage distribution diagram of Case 3 are shown respectively. After connecting to IESG, the load volatility of the distribution network is reduced by 44.6%, the peak-to-valley difference is reduced by 36.8%, and the voltage amplitude deviation is reduced by 6.1% compared with the original.

[0269] Figures 15 to 20 The system power and IES benefits for Cases 1-3 are shown. Fig.15 , Fig.16 as well as Fig.17 The power release of the distribution network, photovoltaic panels and wind turbines, and the power absorption of electric vehicles are shown in the 1-10 hour period. During this period, the A energy storage subsystem absorbs the power of the distribution network and the wind and solar system. When the upper limit is reached, the excess power is converted into hydrogen energy through the hydrogen production system, while the B energy storage subsystem releases power to the electric vehicle.

[0270] During the 11-22 hour period, electric vehicles and distribution networks absorb power, while wind turbines and photovoltaic panels release power. The A energy storage subsystem provides power to electric vehicles and the grid. If the release limit is reached, the fuel cell provides additional power. At the same time, the B energy storage subsystem absorbs photovoltaic power and converts wind energy into hydrogen energy.

[0271] In the 23-24 hour period, electric vehicles and distribution networks absorb power again, and wind turbines and photovoltaic panels release power. The B energy storage subsystem releases power to electric vehicles and the grid. If the release limit is reached, the fuel cell provides additional power. The A energy storage subsystem absorbs photovoltaic power and converts wind energy into hydrogen energy.

[0272] Fig.18 , Fig.19 as well as Fig. 20 The revenues of the integrated energy storage systems of different cases are shown respectively: the average revenue of the integrated energy storage system in Case 1 is 1.1591 million yuan, the average revenue of the integrated energy storage system in Case 2 is 741,900 yuan, and the average revenue of the integrated energy storage system in Case 3 is 814,400 yuan.

[0273] The above analysis shows that the integrated electric energy storage system group has a significant impact on the load fluctuation, peak-to-valley difference and voltage amplitude deviation of the distribution network under different conditions (extreme and normal). Specifically:

[0274] Effects under extreme conditions: Case 1 (overall profit maximization strategy) is more significant in improving load volatility, peak-to-valley difference and voltage amplitude than Case 2 (site profit balancing strategy), indicating that the overall optimization strategy is more effective in improving the performance of the distribution network.

[0275] Effect under normal conditions: Case 3 (overall profit maximization strategy) performs better in load fluctuation optimization than Case 1, but the improvement of peak-to-valley difference is slightly inferior. Although the voltage amplitude deviation optimization effect of Case 3 is not as significant as that of Case 1, it is still improved, indicating its optimization effect in voltage stability.

[0276] In summary, under extreme conditions, the overall benefit maximization strategy (Case 1) can significantly improve the performance and economic benefits of the distribution network. In contrast, although the balanced benefit strategy (Case 2) also has a certain effect, it is not as good as the overall optimization strategy in terms of system stability and economic benefits. The overall benefit maximization strategy can also achieve corresponding performance improvements under normal conditions, proving its feasibility and effectiveness under different conditions.

[0277] In order to better implement a method for dispatching a distributed grid-connected electric energy storage system group in an embodiment of the present invention, based on a method for dispatching a distributed grid-connected electric energy storage system group, correspondingly, as follows: Fig.21 As shown, the embodiment of the present invention further provides a distributed grid-connected dispatching device for an electric energy storage system group. A distributed grid-connected dispatching device 2100 for an electric energy storage system group includes:

[0278] Constructing an objective function module 2101, for constructing an objective function and constraint conditions for each electric energy storage system in the electric energy storage system group, wherein the objective function includes a first objective function constructed by maximizing the benefit of the electric energy storage system, a second objective function constructed by minimizing the peak-to-valley difference, a third objective function constructed by minimizing the load standard deviation, and a fourth objective function constructed by minimizing the voltage amplitude deviation, and the constraint conditions include a battery energy storage system constraint, a hydrogen energy storage system constraint, and a distribution network power balance constraint;

[0279] An operator selection module 2102, used to select a global search operator from a preset global search operator pool and a local search operator from a preset local search operator pool;

[0280] An update operator acquisition module 2103, used to obtain an update operator based on a global search operator and a local search operator;

[0281] The scheduling module 2104 is used to select an operator in the update operator based on a preset deep reinforcement learning algorithm to solve the objective function under constraints, and obtain the scheduling result of the distributed grid-connected electric energy storage system group.

[0282] The above-mentioned embodiment provides a distributed grid-connected dispatching device 2100 for a group of electric energy storage systems, which can implement the technical solution described in the above-mentioned embodiment of a method for dispatching a distributed grid-connected group of electric energy storage systems. The specific implementation principles of the above-mentioned modules or units can refer to the corresponding contents in the above-mentioned embodiment of a method for dispatching a distributed grid-connected group of electric energy storage systems, which will not be repeated here.

[0283] like Fig. 22 As shown, the present invention also provides an electronic device 2200. The electronic device 2200 includes a processor 2201, a memory 2202 and a display 2203. Fig. 22 Only some components of the electronic device 2200 are shown, but it should be understood that it is not required to implement all of the shown components, and more or fewer components may be implemented instead.

[0284] In some embodiments, the processor 2201 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 2202, such as a distributed grid-connected scheduling method for a group of electric energy storage systems in the present invention.

[0285] In some embodiments, the processor 2201 may be a single server or a server group. The server group may be centralized or distributed. In some embodiments, the processor 2201 may be local or remote. In some embodiments, the processor 2201 may be implemented in a cloud platform. In some embodiments, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-cloud, etc., or any combination thereof.

[0286] In some embodiments, the memory 2202 may be an internal storage unit of the electronic device 2200, such as a hard disk or memory of the electronic device 300. In other embodiments, the memory 2202 may also be an external storage device of the electronic device 2200, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., equipped on the electronic device 2200.

[0287] Furthermore, the memory 2202 may include both an internal storage unit of the electronic device 2200 and an external storage device. The memory 2202 is used to store application software installed in the electronic device 2200 and various data.

[0288] In some embodiments, the display 2203 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 2203 is used to display information on the electronic device 2200 and to display a visual user interface. The components 2201-2203 of the electronic device 2200 communicate with each other via a system bus.

[0289] In one embodiment, when the processor 2201 executes a scheduling program for distributed grid connection of a group of electric energy storage systems in the memory 2202, the following steps may be implemented:

[0290] Constructing objective functions and constraints for each electric energy storage system in the electric energy storage system group, wherein the objective functions include a first objective function constructed by maximizing the benefit of the electric energy storage system, a second objective function constructed by minimizing the peak-to-valley difference, a third objective function constructed by minimizing the load standard deviation, and a fourth objective function constructed by minimizing the voltage amplitude deviation, and the constraints include battery energy storage system constraints, hydrogen energy storage system constraints, and distribution network power balance constraints;

[0291] Select a global search operator from a preset global search operator pool, and select a local search operator from a preset local search operator pool;

[0292] An update operator is obtained based on the global search operator and the local search operator;

[0293] Based on the preset deep reinforcement learning algorithm, the operator is selected in the update operator to solve the objective function under the constraints, and the scheduling result of the distributed grid-connected electric energy storage system group is obtained.

[0294] It should be understood that: when the processor 2201 executes a distributed grid-connected scheduling program for a group of electric energy storage systems in the memory 2202, in addition to the above functions, other functions can also be implemented. For details, please refer to the description of the corresponding method embodiment above.

[0295] Furthermore, the embodiment of the present invention does not specifically limit the type of the electronic device 2200 mentioned, and the electronic device 2200 may be a portable electronic device such as a mobile phone, a tablet computer, a personal digital assistant (PDA), a wearable device, a laptop computer, etc. Exemplary embodiments of portable electronic devices include but are not limited to portable electronic devices equipped with IOS, Android, Microsoft or other operating systems. The above-mentioned portable electronic devices may also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, the electronic device 2200 may not be a portable electronic device, but a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0296] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0297] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for dispatching a distributed grid-connected electric energy storage system group, characterized in that: include: Constructing objective functions and constraints for each electric energy storage system in the electric energy storage system group, wherein the objective functions include a first objective function constructed by maximizing the benefit of the electric energy storage system, a second objective function constructed by minimizing the peak-to-valley difference, a third objective function constructed by minimizing the load standard deviation, and a fourth objective function constructed by minimizing the voltage amplitude deviation, and the constraints include battery energy storage system constraints, hydrogen energy storage system constraints, and distribution network power balance constraints; Select a global search operator from a preset global search operator pool, and select a local search operator from a preset local search operator pool; An update operator is obtained based on the global search operator and the local search operator; Based on the preset deep reinforcement learning algorithm, an operator is selected in the update operator to solve the objective function under the constraint condition, and the dispatch result of the distributed grid-connected electric energy storage system group is obtained; The preset deep reinforcement learning algorithm is based on selecting an operator in the update operator to solve the objective function under the constraint condition to obtain the dispatch result of the distributed grid-connected power storage system group, including: S1: Obtaining the initial value of the objective function and determining the initialization parameters of the preset deep reinforcement learning algorithm, wherein the initialization parameters include value network parameters and action network parameters; S2: taking the power absorbed and released by the electric energy storage system group as the initial population, calculating the IGD+ and HV of each individual in the initial population based on the initial value of the objective function, and taking the IGD+ and HV as the initial state of the initial population; S3: Generate an action strategy and store action probability based on action network parameters according to the initial state; S4: the initial population generates a first population of offspring in the function space of the objective function based on the update operator; S5: merging the initial population and the first population to obtain a second population, applying a non-dominated sorting and a reference point-based hierarchical strategy to the second population to obtain a third population; S6: Calculate the state of the third population, and obtain a reward value based on the state of the third population and the initial state; S7: Obtaining experience based on the initial state, the stored action probability, the reward value, and the state of the third population, and putting the experience into an experience pool, wherein the experience pool obtains target value network parameters and target action network parameters based on preset deep reinforcement learning training value network parameters and action network parameters; S8: Use the third population as the initial population, the state of the third population as the initial state, the target value network parameters as the value network parameters, and the target action network parameters as the action network parameters, repeat steps S3-S7 iterations, and determine the scheduling results of the distributed grid-connected electric energy storage system group based on the iterated population.

2. The method for dispatching a distributed grid-connected electric energy storage system group according to claim 1, characterized in that: The obtaining of the initial value of the objective function includes: setting a first initial value of the SOC of the battery energy storage system, setting a second initial value of the HSL of the hydrogen energy system, setting a third initial value of the photovoltaic power generation power, setting a fourth initial value of the electric vehicle flow data, and setting a fifth initial value of the load of the electric energy storage system; The first initial value, the second initial value, the third initial value, the fourth initial value and the fifth initial value are input into the first objective function, the second objective function, the third objective function and the fourth objective function to obtain the first target initial value, the second target initial value, the third target initial value and the fourth target initial value.

3. The method for dispatching a distributed grid-connected electric energy storage system group according to claim 1, characterized in that: The expression of the first objective function is: In the formula, represents the profit of the integrated energy storage system from selling electricity to electric vehicles; It represents the profit of mutual feedback between the integrated electric energy storage system and the distribution network; represents the profit obtained by the hydrogen energy storage system; m represents the number of electric energy storage systems; and They are t Discharge power and charging power of the distribution network during the time period; and are the prices of electricity sold to and purchased from the grid during period t, respectively; Power for charging electric vehicles; the price of charging electric vehicles from integrated electrical energy storage systems; and They are the capacity and price of hydrogen energy sold; and They are the capacity and price of hydrogen purchased; The expression of the second objective function is: In the formula, Indicates time period t Load power; T=24; The peak-to-valley difference of the day; The expression of the third objective function is: In the formula, Indicates the average load power in a day; The expression of the fourth objective function is: In the formula, Is a node The voltage amplitude of is the target voltage; N is the set of all nodes in the distribution network.

4. The method for dispatching a distributed grid-connected electric energy storage system group according to claim 1, characterized in that: The battery energy storage system constraints include upper and lower limit constraints of battery energy storage system capacity and upper and lower limit constraints of charge and discharge power. The expression of the upper and lower limit constraints of battery energy storage system capacity is: In the formula, is the minimum capacity of the battery subsystem; is the capacity of the battery subsystem at each moment; is the maximum capacity of the battery subsystem; The expression of the upper and lower limits of the charge and discharge power is: In the formula, The charging power per hour for the battery subsystem; is the discharge power of the battery subsystem per hour; is the maximum discharge power of the battery subsystem; The maximum charging power for the battery subsystem; The hydrogen energy storage system constraint includes a hydrogen energy storage system capacity upper limit constraint and a hydrogen charging and discharging constraint. The expression of the hydrogen energy storage system capacity upper limit constraint is: In the formula, is the minimum capacity of the hydrogen energy storage device; is the hydrogen storage level of the hydrogen energy storage device at each moment; is the maximum capacity of the hydrogen energy storage device; The expression of the hydrogen charging and discharging constraint is: In the formula, one represents the first case, that is, C stores hydrogen energy and D releases hydrogen energy at the same time; two represents the second case, that is, D stores hydrogen energy and C releases hydrogen energy at the same time; The expression of the power balance constraint of the distribution network is: In the formula, The power generated by the generator per hour; Release power to the integrated energy storage system group for the distribution network; absorbing power from the integrated electrical energy storage system for the distribution grid; For each hour load power; is the power loss of the grid line every hour; The value of is 1 or 0, indicating that the grid cannot release and absorb power at the same time.

5. The method for dispatching a distributed grid-connected electric energy storage system group according to claim 4, characterized in that: It also includes adjusting the absorption and release of power of the battery energy storage system and the hydrogen energy storage system through preset adaptive balancing parameters.

6. The method for dispatching a distributed grid-connected electric energy storage system group according to claim 5, characterized in that: When the battery energy storage system includes a first battery energy storage subsystem and a second battery energy storage subsystem, and the hydrogen energy storage system includes a first hydrogen energy storage subsystem and a second hydrogen energy storage subsystem, the preset adaptive balancing parameters include a first adaptive balancing parameter and a second adaptive balancing parameter; when the output power of the distribution network or the photovoltaic system is not greater than the power absorption upper limit or the available power space of the first battery energy storage subsystem, the first adaptive balancing parameter satisfies the first preset condition, and the formula of the first preset condition is: in, represents a first adaptive equalization parameter; The value of is 1 or 0, indicating that the grid cannot release and absorb power at the same time; for t At the moment the first battery energy storage subsystem absorbs power; for t The grid releases power at all times; for t Photovoltaic releases power at all times; for t The fan releases power at all times; for t At time 1, the first hydrogen energy storage subsystem C absorbs power; is the total capacity of the battery subsystem; for t The state of charge of the BESS at the moment; is the maximum absorbed power of the battery subsystem; is the maximum state of charge of the battery subsystem; is the available power space; The remaining state of charge space of the first battery energy storage subsystem; When the output power of the distribution network or the photovoltaic system is greater than the power absorption upper limit or the available power space of the first battery energy storage subsystem, the first adaptive balancing parameter meets the second preset condition. The formula of the second preset condition is: in, express t At the moment the first battery energy storage subsystem absorbs power; express t The power that cannot be absorbed by the first battery energy storage subsystem and is transferred to the first hydrogen energy storage subsystem at the moment; express t The power absorbed by the first hydrogen energy storage subsystem at the moment.

7. The method for dispatching a distributed grid-connected electric energy storage system group according to claim 6, characterized in that: When the power absorbed by the distribution network or the electric vehicle EV is not greater than the maximum released power or the remaining power of the second battery energy storage subsystem, the second adaptive balancing parameter meets the third preset condition, and the formula of the third preset condition is: In the formula, represents a second adaptive equalization parameter; for t The grid absorbs power at all times; For electric vehicles t Power is required at all times; The power released by the second battery energy storage subsystem at time t; is the remaining capacity of the second battery energy storage subsystem; The remaining power of the second battery energy storage subsystem; Release the power ceiling for the battery subsystem; is the minimum state of charge of the battery subsystem; is the total capacity of the battery subsystem; Represented as the remaining state of charge space of the first battery energy storage subsystem; express t The absorbed power of the electric vehicle at any moment; The value of is 1 or 0, indicating that the grid cannot release and absorb power at the same time; When the power absorbed by the distribution network or the electric vehicle EV is not greater than the maximum released power or the remaining power of the second battery energy storage subsystem, the second adaptive balancing parameter meets the fourth preset condition. The formula of the fourth preset condition is: In the formula, express t The output power of the fuel cell at any moment; express t The absorbed power of the electric vehicle at each moment.

8. A dispatching device for distributed grid-connected power storage system group, characterized in that: include: Construct an objective function module, which is used to construct the objective function and constraint conditions of each electric energy storage system in the electric energy storage system group. The objective function includes a first objective function constructed by maximizing the benefit of the electric energy storage system, a second objective function constructed by minimizing the peak-to-valley difference, a third objective function constructed by minimizing the load standard deviation, and a fourth objective function constructed by minimizing the voltage amplitude deviation. The constraint conditions include battery energy storage system constraints, hydrogen energy storage system constraints, and distribution network power balance constraints. An operator selection module, used to select a global search operator from a preset global search operator pool and a local search operator from a preset local search operator pool; An update operator acquisition module, used to obtain an update operator based on a global search operator and a local search operator; A scheduling module is used to select an operator from the update operator based on a preset deep reinforcement learning algorithm to solve the objective function under constraints, and obtain a scheduling result for the distributed grid connection of the electric energy storage system group; The preset deep reinforcement learning algorithm is based on selecting an operator in the update operator to solve the objective function under the constraint condition to obtain the dispatch result of the distributed grid-connected power storage system group, including: S1: Obtaining the initial value of the objective function and determining the initialization parameters of the preset deep reinforcement learning algorithm, wherein the initialization parameters include value network parameters and action network parameters; S2: taking the power absorbed and released by the electric energy storage system group as the initial population, calculating the IGD+ and HV of each individual in the initial population based on the initial value of the objective function, and taking the IGD+ and HV as the initial state of the initial population; S3: Generate an action strategy and store action probability based on action network parameters according to the initial state; S4: the initial population generates a first population of offspring in the function space of the objective function based on the update operator; S5: merging the initial population and the first population to obtain a second population, applying a non-dominated sorting and a reference point-based hierarchical strategy to the second population to obtain a third population; S6: Calculate the state of the third population, and obtain a reward value based on the state of the third population and the initial state; S7: Obtaining experience based on the initial state, the stored action probability, the reward value, and the state of the third population, and putting the experience into an experience pool, wherein the experience pool obtains target value network parameters and target action network parameters based on preset deep reinforcement learning training value network parameters and action network parameters; S8: Use the third population as the initial population, the state of the third population as the initial state, the target value network parameters as the value network parameters, and the target action network parameters as the action network parameters, repeat steps S3-S7 iterations, and determine the scheduling results of the distributed grid-connected electric energy storage system group based on the iterated population.

9. An electronic device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the distributed grid-connected scheduling method of an electric energy storage system group as described in any one of claims 1 to 7.

Images