Active power distribution network cooperative operation optimization method considering'source-load-storage 'coordination

Abstract

The invention discloses a source-load-storage coordinated active power distribution network cooperative operation optimization method, and belongs to the technical field of active power distribution network optimization. Comprising the following steps: partitioning the power distribution network by using an electrical coupling degree and a power matching degree to obtain an optimal power distribution network partitioning result; taking the minimum comprehensive operation cost of the active power distribution network as a target function, introducing a demand response constraint, and constructing an active power distribution network partition collaborative optimization model; and constructing a regional agent for each regional partition based on the optimal power distribution network partition result, modeling an objective function into a distributed partial observable Markov decision process, and performing iterative optimization on the active power distribution network partition collaborative optimization model through the regional agent to obtain the real-time electricity price in each region. Distributed management of flexible resources of each region can be realized, and the problem that the operation regulation and control difficulty is increased due to multivariate uncertainty factors and ever-increasing structure complexity in the existing ADN is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of active distribution network optimization, and in particular to an active distribution network collaborative operation optimization method considering "source-load-storage" coordination. Background Art

[0002] The statements in this section merely mention background art related to the present invention and do not necessarily constitute prior art.

[0003] With the advancement of energy structure transformation and the proposal of carbon peak and carbon neutrality strategic goals, the penetration rate of distributed renewable energy generation (DRG) represented by photovoltaic and wind power in distribution networks is rapidly increasing. In addition, the penetration rate of distributed resources (DER) such as energy storage and flexible loads in distribution networks is also gradually increasing. The passive characteristics, unidirectional flow distribution, and single-point radial grid structure of traditional distribution networks have been replaced by active characteristics, bidirectional flow distribution, and complex network structures with multiple power sources, which greatly increases the difficulty of operation and regulation of distribution networks.

[0004] In order to adapt to the development needs of the distribution network, the concept of "Active Distribution Network" (ADN) came into being. That is, based on a flexible network structure, through the application of advanced technologies such as power electronics, information communication, and intelligent control, flexible resources are actively controlled and managed, to promote the rational allocation of resources and the efficient use of equipment, and to achieve synergy and win-win results for different stakeholders.

[0005] Building ADN will become the medium- and long-term goal of the development of distribution networks under the background of new power systems. In distribution networks, the output power of high-proportion renewable energy resources such as photovoltaics and wind power is difficult to adjust and control; the load level supported by renewable energy resources is limited, and photovoltaic output is difficult to meet the demand for peak power consumption at night. Under extreme weather conditions such as extremely hot and windless weather, it is even more difficult to guarantee power supply requirements by relying solely on distributed wind and photovoltaic power generation.

[0006] Therefore, under the new situation and environment, it is very necessary to apply deep reinforcement learning to the coordinated operation optimization of sources, loads and storage in active distribution networks. Summary of the invention

[0007] In order to address the deficiencies of the prior art, the present invention provides a method, system, electronic device, computer-readable storage medium and computer program product for optimizing the coordinated operation of an active distribution network considering "source-load-storage", which can realize multi-region coordinated control of a distribution network without relying on inter-regional communication; at the same time, a method for optimizing the coordinated operation of partitioned active distribution networks with "source-load-storage" coordination is proposed, and a partitioned intelligent agent is constructed with the goal of reducing the comprehensive operating cost of the distribution network. The Markov game model of this problem is modeled as Dec-POMDPs to realize the distributed management of flexibility resources in each region; electric vehicle loads are added and the demand response of flexible loads is considered, thereby exploring the flexible potential of the "source-load-storage" coordination of the distribution network.

[0008] In a first aspect, the present invention provides a method for optimizing the coordinated operation of an active distribution network taking into account “source-load-storage” coordination; An active distribution network coordinated operation optimization method considering "source-load-storage" coordination includes: Use the electrical coupling degree and power matching degree to partition the distribution network and obtain the optimal distribution network partition result; Taking the minimum comprehensive operation cost of active distribution network as the objective function, and taking power flow constraint, node voltage constraint, line transmission capacity constraint, substation transmission power constraint and demand response constraint as the constraint conditions, a partitioned collaborative optimization model of active distribution network is constructed. Based on the optimal distribution network partitioning results, regional intelligent agents are constructed for each region. The objective function is modeled as a decentralized partially observable Markov decision process. The active distribution network partitioning collaborative optimization model is iteratively optimized through deep reinforcement learning to obtain the real-time electricity price in each region.

[0009] In some implementations, partitioning the distribution network using the electrical coupling degree and the power matching degree to obtain an optimal distribution network partitioning result includes: The modularity index of each node in the distribution network is calculated using the degree of electrical coupling; the active power balance capability index and reactive power balance capability index of each node are calculated using the power matching degree; Generate a partition comprehensive index for each node based on the modularity index, active power balance capability index and reactive power balance capability index; Nodes are merged according to the partition comprehensive index until the partition comprehensive index is maximized to obtain the optimal distribution network partition result.

[0010] In some embodiments, the use of electrical coupling degree to calculate the modularity index of each node in the distribution network is specifically: constructing an electrical distance based on a voltage-active sensitivity matrix and a voltage-reactive sensitivity matrix, and constructing a modularity index using the electrical distance as an edge weight.

[0011] In some embodiments, the power matching degree is used to calculate the active balance capability index and reactive balance capability index of each node, specifically: obtaining the regional net power within a preset time period, and calculating the active balance degree corresponding to the region; obtaining the maximum reactive power supply of the region within the preset time period, and calculating the reactive balance degree corresponding to the region.

[0012] In some embodiments, the demand response constraints include node injection constraints, demand response load constraints, and demand response electricity price constraints.

[0013] In some implementations, a reward function is constructed by combining the optimization objective and the penalty function to reward or punish each agent.

[0014] In a second aspect, the present invention provides an active distribution network collaborative operation optimization system considering “source-load-storage” coordination; An active distribution network coordinated operation optimization system considering "source-load-storage" coordination includes: The distribution network partitioning module is configured to: partition the distribution network using the electrical coupling degree and the power matching degree to obtain the optimal distribution network partitioning result; The model building module is configured to: take the minimization of the comprehensive operation cost of the active distribution network as the objective function, and take the power flow constraint, node voltage constraint, line transmission capacity constraint, substation transmission power constraint and demand response constraint as the constraint conditions to build the active distribution network partition collaborative optimization model; The collaborative optimization module is configured to: construct a regional intelligent agent for each region based on the optimal distribution network partitioning result, model the objective function as a decentralized partially observable Markov decision process, iteratively optimize the active distribution network partitioning collaborative optimization model through deep reinforcement learning, and obtain the real-time electricity price in each region.

[0015] In a third aspect, the present invention provides an electronic device; An electronic device comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above-mentioned active distribution network collaborative operation optimization method considering "source-load-storage" coordination.

[0016] In a fourth aspect, the present invention provides a computer-readable storage medium; A computer-readable storage medium stores a computer program / instruction thereon, which, when executed by a processor, implements the steps of the above-mentioned active distribution network collaborative operation optimization method considering "source-load-storage" coordination.

[0017] In a fifth aspect, the present invention provides a computer program product; A computer program product includes a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned active distribution network collaborative operation optimization method considering "source-load-storage" coordination.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1. The technical solution provided by the present invention is not only conducive to coping with the uncertainties brought about by the access of massive distributed resources, but also conducive to quickly generating adaptive strategies for the coordinated operation of distributed resources. It is of great significance for strengthening the cooperation and mutual assistance between sources, loads and storage, giving full play to the flexibility potential of distributed resources, and improving the stability and economy of the distribution network.

[0019] 2. The technical solution provided by the present invention can realize multi-region coordinated control of distribution networks without relying on inter-regional communication; a "source-load-storage" coordinated active distribution network partition coordinated operation optimization method is proposed, and a partition intelligent body is constructed with the goal of reducing the comprehensive operation cost of the distribution network. The Markov game model of this problem is modeled as Dec-POMDPs to realize the distributed management of flexibility resources in each region, add electric vehicle loads and consider the demand response of flexible loads, thereby exploring the flexible potential of "source-load-storage" coordination of distribution networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0021] Figure 1 A flow chart of an active distribution network collaborative operation optimization method considering “source-load-storage” coordination provided by an embodiment of the present invention; Figure 2 A schematic diagram of the system framework of an active distribution network collaborative operation optimization system considering “source-load-storage” coordination provided by an embodiment of the present invention; Figure 3 An example diagram of optimization scheduling results provided by an embodiment of the present invention; FIG4(a) is an example diagram of the optimization scheduling results of region 1, FIG4(b) is an example diagram of the optimization scheduling results of region 2, and FIG4(c) is an example diagram of the optimization scheduling results of region 3. DETAILED DESCRIPTION

[0022] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0023] It should be noted that the terms used herein are only for describing specific embodiments, and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that the terms "include" and "have" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0025] Embodiment 1 Next, combine Figure 1 - Figure 4 (c) describes in detail an active distribution network coordinated operation optimization method considering "source-load-storage" coordination disclosed in this embodiment. The active distribution network coordinated operation optimization method considering "source-load-storage" coordination includes the following steps: S1. Use the electrical coupling degree and power matching degree to partition the distribution network and obtain the optimal distribution network partition result.

[0026] In this embodiment, the distribution network is partitioned by comprehensively considering the degree of electrical coupling and the degree of power matching. In terms of structure, the electrical distance is defined based on the voltage-active and reactive sensitivity matrix, and the modularity index is constructed with the electrical distance between nodes as the edge weight to describe the degree of electrical coupling between nodes in the region, thereby forming a partition pattern with strong coupling within the region and weak coupling between regions; in terms of power matching, the source-load complementary characteristics within the region are considered through the active balance capability index, and the reactive balance capability index is used to prepare sufficient reactive reserves in the region to cope with voltage fluctuations caused by uncertain disturbances.

[0027] As an implementation mode, S1 specifically includes: S101, initializing partitioning. At this point, each node in the distribution network is regarded as an independent area.

[0028] S102, calculate the partition comprehensive index of each node, and merge nodes pairwise based on the partition comprehensive index, that is, if the comprehensive index value is higher than before merging, select the scheme with the largest comprehensive index value to continue merging. Repeat the above process until the comprehensive index value reaches the maximum and stop merging, and the optimal distribution network partition result is obtained.

[0029] Specifically, the modularity index of each node in the distribution network is calculated using the degree of electrical coupling; the active power balance capability index and reactive power balance capability index of each node are calculated using the power matching degree; and the partition comprehensive index of each node is generated based on the modularity index, the active power balance capability index and the reactive power balance capability index.

[0030] For example, the specific calculation process of the modularity index is as follows: Since the large amount of active power brought by the access of distributed resources affects the voltage of the distribution network nodes, the relationship between active power and voltage needs to be considered when partitioning. By taking the inverse of the Jacobi matrix in the power flow calculation, the following equation can be obtained: (1) In the formula, , They represent the influence of unit amount of active power and reactive power injection on the change of node voltage phase angle respectively; , They respectively represent the impact of unit amount of active and reactive power injection on the node voltage amplitude.

[0031] Construct the electrical distance based on the voltage-active sensitivity matrix and the voltage-reactive sensitivity matrix, that is: (2) (3) In the formula, , They are the electrical distances based on voltage-active sensitivity and voltage-reactive sensitivity, and the coefficients are introduced To balance the two; Representation Node The voltage change of the node Sensitivity of active power changes; Representation Node The voltage change of the node The sensitivity of reactive power change. The smaller the value of electrical distance, the closer the electrical coupling between the two nodes.

[0032] The modularity function concept extended to weighted networks can be used to measure the structural coupling strength of complex networks. The modularity index is constructed with electrical distance as the edge weight as follows: (4) In the formula, To connect nodes and nodes The edge weight (abbreviated as edge weight) of ; If the node and nodes In the same partition, , otherwise 0.

[0033] The calculation process of active power balance capability index is as follows: Autonomous regions should give full play to their self-consumption capacity, optimize the source-load imbalance within the sub-region, realize the source-load complementarity between distributed power sources and between distributed power sources and loads, improve the independence of power supply within the region, and reduce the exchange power required for inter-regional coordination and optimization. Therefore, the active balance capacity indicators are set as follows: (5) In the formula, For Region Active power balance; for Time zone of net power.

[0034] The calculation process of reactive power balance capability index is as follows: The autonomous region should have a certain voltage regulation capability to reduce the voltage over-limit caused by the increase of renewable energy output. The reactive power supply capacity within the autonomous region should meet the needs of reactive power local balance as much as possible to reduce the cross-regional transmission of reactive power. The reactive power balance capability indicators are set as follows: (6) (7) In the formula, For Region Reactive power balance; For Region Reactive power demand within For Region The maximum value of reactive power supply within.

[0035] In summary, the partition comprehensive index proposed in this embodiment is expressed as: (8) In the formula, , , is the weight coefficient.

[0036] S2. Taking the minimization of the comprehensive operating cost of the active distribution network as the objective function, and taking the power flow constraint, node voltage constraint, line transmission capacity constraint, substation transmission power constraint and demand response constraint as the constraint conditions, a partitioned collaborative optimization model of the active distribution network is constructed.

[0037] The optimization goal is to minimize the comprehensive operating cost of the system. Since the RTP demand response in this embodiment mainly plays a guiding role and the user decides on the electricity adjustment by himself, the corresponding cost of demand response is not taken into account in the optimization goal. The objective function includes the main grid power purchase cost, the operating cost of energy storage and CDG, and the network loss penalty cost, as shown in formula (9): (9) The constraints are expressed as: (1) Node injection equation Changes in electricity load due to demand response: (10) In the formula, Indicates that the electric vehicle group is unified from the node Access to the distribution network.

[0038] (2) Demand response load constraints In this embodiment, it is hoped that the real-time electricity price will guide users to achieve load transfer rather than load reduction, so the total electricity consumption of users should remain unchanged: (11) (12) In the formula, , They are the upper and lower limits of the active power of the response load respectively.

[0039] (3) Demand response electricity price constraints (13) In the formula, and They are the upper and lower limits of the real-time electricity price respectively.

[0040] In addition, the distribution network power flow equation, node voltage constraints, line transmission capacity constraints, substation transmission power constraints, equipment operation constraints, etc. are consistent with the conventional constraints and will not be repeated here.

[0041] In summary, the active distribution network partition collaborative optimization model considering the “source-load-storage” coordination can be expressed as: Objective function: (14) Constraints: (15) S3. Based on the optimal distribution network partitioning results, a regional intelligent agent is constructed for each region, and the objective function is modeled as a decentralized partially observable Markov decision process. The active distribution network partitioning collaborative optimization model is iteratively optimized through the regional intelligent agent to obtain the real-time electricity price in each region.

[0042] The composition of Dec-POMDPs can be expressed as: in this process, the regional agent observes the environment and generates an observation set, and then generates the action at that moment based on the observation and accumulated data mining experience, forming an action set. The actions taken will have an impact on the environment and generate new observation sets. This process will continue until enough data features are extracted. Finally, multiple agents work together to find an optimal joint strategy. The details are as follows: (1) State set and observation set Both the state set and the observation set have added demand response load and electric vehicle load. When learning price-based DR decision-making experience, the agent needs to make judgments based on the load response of the previous period and the load forecast data before the current period response, and give reasonable real-time electricity price actions. In order to enable the model to cope with the uncertainty of DRG output and load, random Gaussian noise is superimposed on the DRG and load power data. The system environment information contained in the time period is expressed as: (16) In the formula, for The active power of electric vehicle loads at each node of the distribution network during the period is obtained by calling the EVCC module by the distribution network environment; , They are The response load active power and reactive power at each node of the distribution network during the time period.

[0043] Regional Agent exist The observation information of the time period is expressed as: (17) In the formula, for Time zone autonomous area Active power of electric vehicle load at each node; , They are Time zone autonomous area The response load active power and reactive power at each node within the system.

[0044] (2) Action set The elements in the action set are based on those in Chapter 2 and include real-time electricity prices related to demand response, namely: (18) In the formula, for Time Autonomous Region Real-time electricity prices.

[0045] Output of the DRL algorithm strategy network The range is (-1,1), which needs to be mapped to the real range through linear transformation and then acted on the distribution network environment.

[0046] (3) State transition probability function and state transition process The model-free DRL method does not require the specific expression of the state transition probability function to be known in advance. It only needs to use the empirical samples obtained from the interaction with the environment at each state transition. Train to maximize the shared expected reward. During the state transfer process, when After the multi-agent joint action acts on the distribution network environment, the EVCC module is immediately called. That is, the observed variable of the EVCC module , the total charging load obtained after the EVCC module is running is Electric vehicle load during the period Then, the state set and other states of the observation set are updated through power flow calculation. In order to approximate the environment to be stable in the short term, the model still assumes .

[0047] (4) Reward Function Reward Function Including running optimization goals and penalty function .when The smaller the comprehensive operating cost and penalty item of a period, the greater the reward given to each agent in that period.

[0048] (19) (20) In order to punish the agent when it violates constraints (12)-(13) during the learning process, we set The second term is shown in formula (20). For the case of violating constraint formula (12), a barrier function is set as formula (22), and for the case of violating constraint formula (13), the barrier function formula (21) is still used.

[0049] (twenty one) (twenty two) In the formula, , , is the penalty coefficient; It is a very small positive number, indicating the acceptable range of total power consumption.

[0050] (5) Special treatment of constraints in the DRL method In this embodiment, after punishing the violation of the constraint conditions in the reward function, it is expected that the over-limit variables will be constrained to a reasonable range through correction. Regarding the constraints on the decision variables, they can be mapped into the constraint range through linear transformation. Regarding the constraints on the state variables, the over-limit situation of constraint (11) can only be known in the last scheduling period. If a large number of adjustments are made at this time, it may affect the node voltage constraints, line transmission capacity constraints and other important constraints related to the safe and stable operation of the distribution network. Therefore, it is not appropriate to use forced correction measures; if constraint (12) exceeds the limit, it will be Restricted to the boundary as in formula (23), the current period is derived The real-time electricity price correction value is as shown in formula (24). If the corrected electricity price does not satisfy formula (13), then exit the current round of exploration and start a new round of training.

[0051] (twenty three) (twenty four) In order to verify the effectiveness of the proposed method, the IEEE 33-node power distribution system is used as the environment.

[0052] Use the trained model to perform real-time optimization scheduling. Select the scheduling results obtained on a test day. Figure 3 As shown. When the output of renewable energy is greater than the load demand, the surplus output will be stored through energy storage charging, and part of it will be sold to the main grid to earn revenue. When the output of renewable energy is less than the load demand, the system will first meet the load through DRG output, and make up for the power shortage through gas turbine output, energy storage discharge and purchase of electricity from the main grid.

[0053] The optimization results of each region are shown in Figure 4. The load before optimization is the sum of the power load before DR adjustment and the EV load under disordered charging. On that day, the photovoltaic output in region 1 was relatively large. During the peak photovoltaic output at noon, the gas turbine output was basically maintained at the lower power limit, and the electric energy storage was charged to store surplus power; during the period when the photovoltaic output was limited, the gas turbine gave priority to release energy for supply. Region 2 is closest to the main grid, and the gas turbine is close to full power, which can not only make up for the limited photovoltaic output in the early morning and at night, but also sell more electricity to the main grid to obtain revenue; the energy storage releases energy in the early morning to make up for the power demand, stores energy at noon to fully absorb the photovoltaic output, and releases electricity during the peak electricity price period to reduce the purchase of electricity from the main grid, reducing the operating cost of the distribution network. The renewable energy output in region 3 is relatively small, and the power shortage is mainly made up by the gas turbine output.

[0054] Embodiment 2 Combination Figure 2 This embodiment discloses an active distribution network coordinated operation optimization system considering “source-load-storage” coordination, including: The distribution network partitioning module is configured to: partition the distribution network using the electrical coupling degree and the power matching degree to obtain the optimal distribution network partitioning result; The model building module is configured to: take the minimization of the comprehensive operation cost of the active distribution network as the objective function, and take the power flow constraint, node voltage constraint, line transmission capacity constraint, substation transmission power constraint and demand response constraint as the constraint conditions to build the active distribution network partition collaborative optimization model; The collaborative optimization module is configured to: construct a regional intelligent agent for each region based on the optimal distribution network partitioning result, model the objective function as a decentralized partially observable Markov decision process, iteratively optimize the active distribution network partitioning collaborative optimization model through the regional intelligent agent, and obtain the real-time electricity price in each region.

[0055] It should be noted that the above-mentioned distribution network partitioning module, model building module and collaborative optimization module correspond to the steps in Embodiment 1, and the examples and application scenarios implemented by the above-mentioned modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above-mentioned Embodiment 1. It should be noted that the above-mentioned modules as part of the system can be executed in a computer system such as a set of computer executable instructions.

[0056] Embodiment 3 Embodiment 3 of the present invention provides an electronic device, including a memory and a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps of the above-mentioned active distribution network collaborative operation optimization method considering "source-load-storage" coordination are completed.

[0057] Embodiment 4 Embodiment 4 of the present invention provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the steps of the above-mentioned active distribution network collaborative operation optimization method considering "source-load-storage" coordination are completed.

[0058] Embodiment 5 Embodiment 5 of the present invention provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned active distribution network collaborative operation optimization method considering "source-load-storage" coordination.

[0059] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0060] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0061] These computer program instructions can also be loaded onto a computer or other programmable data processing device, and a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0062] The description of each embodiment in the above embodiments has different emphases. For parts not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0063] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An active distribution network coordinated operation optimization method considering "source-load-storage" coordination, characterized in that: include: Use the electrical coupling degree and power matching degree to partition the distribution network and obtain the optimal distribution network partition result; Taking the minimum comprehensive operation cost of active distribution network as the objective function, and taking power flow constraint, node voltage constraint, line transmission capacity constraint, substation transmission power constraint and demand response constraint as the constraint conditions, a partitioned collaborative optimization model of active distribution network is constructed. Based on the optimal distribution network partitioning results, regional intelligent agents are constructed for each region. The objective function is modeled as a decentralized partially observable Markov decision process. The active distribution network partitioning collaborative optimization model is iteratively optimized through deep reinforcement learning to obtain the real-time electricity price in each region.

2. The active distribution network coordinated operation optimization method considering "source-load-storage" coordination as claimed in claim 1 is characterized in that: The method of partitioning the distribution network by using the electrical coupling degree and the power matching degree to obtain the optimal distribution network partition result includes: The modularity index of each node in the distribution network is calculated using the degree of electrical coupling; the active power balance capability index and reactive power balance capability index of each node are calculated using the power matching degree; Generate a partition comprehensive index for each node based on the modularity index, active power balance capability index and reactive power balance capability index; Nodes are merged according to the partition comprehensive index until the partition comprehensive index is maximized to obtain the optimal distribution network partition result.

3. The active distribution network coordinated operation optimization method considering "source-load-storage" coordination as claimed in claim 2 is characterized in that: The method of calculating the modularity index of each node in the distribution network by utilizing the electrical coupling degree is as follows: constructing an electrical distance based on a voltage-active sensitivity matrix and a voltage-reactive sensitivity matrix, and constructing a modularity index using the electrical distance as an edge weight.

4. The active distribution network coordinated operation optimization method considering "source-load-storage" coordination as claimed in claim 2 is characterized in that: The method of using the power matching degree to calculate the active balance capability index and reactive balance capability index of each node is as follows: obtaining the regional net power within a preset time period and calculating the active balance degree corresponding to the region; obtaining the maximum reactive power supply of the region within the preset time period and calculating the reactive balance degree corresponding to the region.

5. The active distribution network coordinated operation optimization method considering "source-load-storage" coordination as claimed in claim 1 is characterized in that: The demand response constraints include node injection constraints, demand response load constraints and demand response electricity price constraints.

6. The active distribution network coordinated operation optimization method considering "source-load-storage" coordination as claimed in claim 1 is characterized in that: By combining the optimization objective and penalty function, a reward function is constructed to reward and punish each agent.

7. Active distribution network coordinated operation optimization system considering "source-load-storage" coordination, characterized by: include: The distribution network partitioning module is configured to: partition the distribution network using the electrical coupling degree and the power matching degree to obtain the optimal distribution network partitioning result; The model building module is configured to: take the minimization of the comprehensive operation cost of the active distribution network as the objective function, and take the power flow constraint, node voltage constraint, line transmission capacity constraint, substation transmission power constraint and demand response constraint as the constraint conditions to build the active distribution network partition collaborative optimization model; The collaborative optimization module is configured to: construct a regional intelligent agent for each region based on the optimal distribution network partitioning result, model the objective function as a decentralized partially observable Markov decision process, iteratively optimize the active distribution network partitioning collaborative optimization model through deep reinforcement learning, and obtain the real-time electricity price in each region.

8. An electronic device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the active distribution network collaborative operation optimization method considering "source-load-storage" coordination as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the steps of the method for optimizing the coordinated operation of an active distribution network considering "source-load-storage" coordination as described in any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the steps of the method for optimizing the coordinated operation of an active distribution network considering "source-load-storage" coordination as described in any one of claims 1 to 6 are implemented.

Images