Power distribution network multi-energy storage cooperative emergency energy management and rapid reconstruction method based on deep learning

Abstract

The invention provides a power distribution network multi-energy-storage cooperative emergency energy management and rapid reconstruction method based on deep learning, and the method achieves the rapid reconstruction of a power distribution network based on deep learning based on a multi-energy-storage cooperative emergency support optimization model and a power distribution network multi-target reconstruction optimization model. The multi-energy-storage cooperative emergency support optimization model takes the minimum system network loss as an optimization target; the power distribution network multi-target reconstruction optimization model considers fault scene information of line faults, distributed power supply faults and tie line closing and stable operation conditions of the reconstructed power distribution network; according to the power distribution network rapid reconstruction based on deep learning, a deep learning model is trained by using data obtained through simulation deduction, a fault system topological structure and load switch information are input online, and power flow parameters are rapidly output; and the optimal reconstruction scheme of the power distribution network is determined through multi-objective decision.

Description

Technical Field

[0001] The present invention relates to the technical fields such as redundant disaster prevention of distribution networks, and in particular to a multi-energy storage collaborative emergency energy management and rapid reconstruction method for distribution networks based on deep learning. Background Art

[0002] When disasters such as typhoons come, the distribution network is extremely prone to failures. Most of the systems after the topological structure changes operate in an abnormal state, and emergency energy management is required to ensure continuous and stable power supply for critical loads after the disaster and maintain the stability and security of the system. At present, through methods such as energy storage support and system reconstruction, the stability and security of the distribution network after the disaster can be rapidly improved. However, the current reconstruction methods involve a large amount of online power flow calculations, making the reconstruction speed unable to match the requirements of the rapid restoration and operation optimization of the distribution network after the disaster. Therefore, there is an urgent need to improve the emergency response ability and reconstruction speed of the post-disaster system. Summary of the Invention

[0003] Aiming at the defects and deficiencies existing in the prior art, the present invention proposes a multi-energy storage collaborative emergency energy management and rapid reconstruction method for distribution networks based on deep learning, which can not only effectively solve the problem of system emergency energy management, but also solve the problem of too long reconstruction time of the distribution network, and meet the requirements of the rapid restoration and operation of the distribution network after the disaster.

[0004] In the solution provided by the present invention, in the disaster occurrence stage, aiming at the problem of system energy imbalance, a multi-energy storage collaborative emergency energy management method is proposed. First, with the goal of minimizing the system line loss, a multi-energy storage output optimization model is established to achieve the emergency energy support of the system. However, the energy storage capacity is limited. To ensure the safe operation of the distribution network, a rapid reconstruction method based on deep learning is proposed. This method uses the data simulated and deduced by the disaster prediction model to train the deep learning model, realizes the online input of fault topology information and load switch information, and quickly outputs power flow parameters such as energy storage output, network loss, and voltage, effectively avoiding repeated power flow calculations. Secondly, a multi-objective decision-making evaluation function with decision-making preferences is constructed to achieve diversified reconstruction schemes.

[0005] The present invention specifically adopts the following technical solutions:

[0006] A multi-energy storage collaborative emergency energy management and rapid reconfiguration method for distribution networks based on deep learning, an emergency support optimization model based on multi-energy storage collaboration, and a multi-objective reconfiguration optimization model for distribution networks are used to achieve rapid reconfiguration of distribution networks based on deep learning. The emergency support optimization model based on multi-energy storage collaboration aims to minimize the system network loss. The multi-objective reconfiguration optimization model for distribution networks considers fault scenario information such as line faults, distributed power source faults, and tie line closures, as well as the stable operation conditions of the distribution network after reconfiguration. The rapid reconfiguration of the distribution network based on deep learning uses the data obtained from simulation deduction to train a deep learning model, inputs the fault system topology structure and load switch information online, and quickly outputs the power flow parameters. And the optimal reconfiguration plan of the distribution network is determined through multi-objective decision-making.

[0007] Further, in the multi-energy storage coordinated emergency support optimization model, the total output value of the energy storage is determined by the power deficit of the system, and the Distflow power flow model is used to describe the power flow equation with unknown energy storage states. The objective function is to minimize the lost energy. And it includes the constraints that the energy storage in the distribution network should meet when it is put into operation.

[0008] Further, the optimization objectives of the multi-objective reconfiguration optimization model for distribution networks include minimizing the load shedding amount, minimizing the network loss, minimizing the voltage deviation, and minimizing the balance power deficit. And the stable operation of the distribution network after reconfiguration satisfies the distributed power source power constraint, the power balance constraint, and the voltage constraint.

[0009] Further, the energy storage output and power flow parameter data in each fault scenario are solved through the multi-energy storage collaborative emergency support optimization model and the multi-objective reconfiguration optimization model for distribution networks, and used as the training set and verification set of deep learning.

[0010] Further, taking the power supply reliability as the division criterion, several loads with low reliability requirements are selected as Class III loads, and the Class III load switch states are used as the reconfiguration objects. If the i-th Class III load is P Lci , then the load shedding amount P cut is as follows:

[0011] P cut =∑ i∈I a i P Lci

[0012] where I is the set of Class III loads, and a i is 1 indicating normal power supply to the load, otherwise 0;

[0013] The N groups of switch decision schemes of M Class III loads are combined to form a matrix A. The operating states of the lines, tie lines, and generators in the fault scenario are described by b, c, and d respectively. When b i , c j , d tWhen the value is 1, it respectively represents that line i is in closed operation, tie line j is in closed operation, and distributed power source t is in normal operation. If the above components cannot operate normally, the value is 0. The status information of lines, tie lines, and distributed power sources is combined to obtain the system topology matrix S. The load information and topology information in the fault scenario are formed by combining the switch decision scheme matrix A and the system topology matrix S to determine the characteristics of the system in this scenario, which are used as the input information of the neural network model in deep learning.

[0014] Furthermore, the input information of the input layer of the neural network model for training the energy storage output is a multi-dimensional binary vector, which respectively represents the opening and closing states corresponding to lines, tie lines, distributed power sources, and class-III load switches. The output layer contains 6 nodes, which are associated with the system power loss, load shedding amount, maximum node voltage, minimum node voltage, and the output of two energy storages. In the offline training stage, after preprocessing the training data, the parameters of the CNN model are adjusted to complete the learning of the multiple complex mapping relationships between power flow data and line status, tie line status, distributed power source status, and load switch status.

[0015] Furthermore, set the load shedding amount P of class-III load cut Limit the power deficit P before the energy storage is put into operation e The magnification of to screen the reconstruction plan; input the decision after hierarchical screening into the CNN to obtain the reconstruction parameters; in the alternative reconstruction plans, preset the upper limits of the maximum voltage deviation and the power deficit of the system; determine the weight coefficients w1, w2, w3, w4 through the analytic hierarchy process combined with the influencing factors of post-disaster demand arrangement (the analytic hierarchy process is used to establish a weight calculation framework, and the post-disaster demand reflects the importance ranking of influencing factors), and then determine the optimal reconstruction decision of the distribution network through the multi-objective decision evaluation function;

[0016]

[0017] In the formula, P k cut0 respectively represent the normalized power loss, voltage deviation, power deficit, and load shedding amount under decision scheme k;

[0018] Select the switch combination with the smallest Ass and meeting the constraint conditions as the optimal reconstruction decision.

[0019] Furthermore, when all switch combinations in the load switch alternative solution set cannot meet the constraint conditions in the reconstruction optimization model, it is judged as reconstruction abnormality, and an over-limit evaluation function is used to evaluate the over-limit situation:

[0020]

[0021] In the formula, α, β ∈ (0,1) are weight coefficients and satisfy α + β = 1, U 0max 、U0min are the upper and lower limits of the node voltage, U jmax and U jmin are the upper and lower limits of the node voltage in the j-th decision; P emax and P emin are the upper and lower limits of the power deficit P ej respectively; r u and r e being 1 indicates that the voltage and the balanced power are out of limits, otherwise 0;

[0022] Take the load combination with the smallest out-of-limit value as the distribution network reconstruction plan when reconstruction is abnormal.

[0023] In addition, an electronic device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. The characteristic is that when the processor executes the program, it implements the steps of the above-mentioned distribution network multi-energy storage collaborative emergency energy management and rapid reconstruction method based on deep learning.

[0024] A non-transitory computer-readable storage medium stores a computer program. The characteristic is that when the computer program is executed by a processor, it implements the steps of the above-mentioned distribution network multi-energy storage collaborative emergency energy management and rapid reconstruction method based on deep learning.

[0025] Compared with the prior art, the present invention and its preferred solutions obtain the same reconstruction plan, but the reconstruction time is only 1% of the former. Compared with the system before reconstruction, the voltage distribution of the system is improved, and parameters such as system line loss are optimized, effectively improving the stability and economy of the distribution network operation. Description of the Drawings

[0026] The following further describes the present invention in detail with reference to the drawings and specific embodiments:

[0027] Figure 1 is the flowchart of the solution of the embodiment of the present invention;

[0028] Figure 2 is the deep learning method diagram of the embodiment of the present invention. Detailed Embodiments

[0029] In the following, specific embodiments of the present application will be described in detail with reference to the drawings. According to these detailed descriptions, those skilled in the art can clearly understand the present application and can implement the present application. Without departing from the principle of the present application, the features in different embodiments can be combined to obtain new implementation manners, or some features in certain embodiments can be replaced to obtain other preferred implementation manners.

[0030] To make the features and advantages of the present invention more obvious and understandable, specific embodiments are hereinafter given and described in detail in conjunction with the accompanying drawings as follows:

[0031] As Figure 1 shown, the implementation process of a multi-energy storage collaborative emergency energy management and rapid reconstruction method for a distribution network based on deep learning provided by an embodiment of the present invention is as follows:

[0032] (1) Construct an emergency support optimization model for multi-energy storage collaboration

[0033] The multi-energy storage coordinated emergency support optimization model takes the minimum system network loss as the optimization goal

[0034]

[0035] In the formula, for the line with endpoints i and j, I ij is the line current, r ij is the line resistance, and L is the set of lines.

[0036] The total output value of the energy storage is determined by the power deficit of the system, and the constraints are as follows:

[0037] P e = ∑ i∈E C i (2)

[0038] C imin ≤ C i ≤ C imax , i ∈ E (3)

[0039] In the formula, P e represents the power deficit, and E is the set of energy storage; for energy storage i, C i is the output, and C jmax , C jmin represent the upper and lower limits of the energy storage capacity respectively.

[0040] For the power flow equation describing the unknown state of the energy storage, the Distflow power flow model is adopted, and the objective function is the minimum loss of power:

[0041] F D = min ∑ i∈B P i (4)

[0042] The constraint conditions are as follows:

[0043]

[0044] U imin ≤ U i ≤ U imax (8)

[0045] where B is the set of power - off nodes, U i , U j are the terminal voltages at endpoints i and j, r ij , x ij are the line impedances, P ij , Q ij represent active and reactive powers respectively; for node j, h(j) is the set pointing from the end - point position in the power - flow direction to this node, t(j) is the set pointing from the starting - point position in the power - flow direction to this node, P Gj , Q Gj are the total active and reactive outputs of the connected distributed generation (DG), P Lj , Q Lj represent the active and reactive powers of the node load; U imax and U imin represent the upper and lower limits of the voltage of node i respectively. Equation (5) represents the relationship between terminal voltages, Equation (6) and Equation (7) represent the relationship constraints between power - flow parameters, and Equation (8) represents the voltage constraint.

[0046] In addition, when the energy storage in the distribution network is put into operation, the following constraints are satisfied:

[0047]

[0048] For energy storage i, represents the state of charge (SOC) at time t, being 1 indicates that the energy storage is in the charging or discharging state at time t, η cha , η dis are the charging and discharging efficiencies. Equation (9) is the SOC change expression, Equation (10) is the energy - storage output constraint, and Equation (11) is the energy - storage state constraint.

[0049] (2) Construct a multi - objective reconfiguration optimization model for the distribution network

[0050] In order to effectively solve the system's energy management problem and improve the system operation performance, the proposed distribution - network reconfiguration model considers the fault - scenario information of line faults, DG faults, and tie - line closures. The optimization objectives include minimizing the load shedding amount, minimizing the network loss, minimizing the voltage deviation, and minimizing the balance power deficit:

[0051] F1 = min∑ i∈B P i (12)

[0052]

[0053] F3 = |1 - U min |+|U max-1| (14)

[0054] F4 = min|P e | (15)

[0055] In the formula, U min , U max are the global minimum and maximum voltages respectively. Equation (12) represents the minimum load shedding, Equation (13) represents the minimum system power loss, Equation (14) represents the minimum voltage deviation, and Equation (15) represents the minimum power deficit.

[0056] The stable operation of the reconfigured distribution network satisfies DG power constraints, power balance constraints, and voltage constraints:

[0057] P Gimin ≤ P Gi ≤ P Gimax , i ∈ G (16)

[0058] ∑ i∈G P Gi + ∑ j∈C C j = ∑ i∈D P Li (17)

[0059] In the formula, G is the DG set, D represents the load set, P Gimax , P Gimin represent the upper and lower limits of the output P Gi of DG i. Equation (16) is the DG output constraint node, Equation (17) is the power balance constraint, and the voltage constraint equation is the same as Equation (8).

[0060] (3) Fast reconfiguration of distribution network based on deep learning

[0061] Introduce a deep learning model to realize online input of fault and load switch information, quickly output power flow parameters, and avoid repeated power flow calculations. Solve the energy storage output and power flow parameter data under each fault scenario through the emergency support optimization model with multi-energy storage coordination and the multi-objective reconfiguration optimization model, and use them as the training set and validation set of deep learning.

[0062] For reconfiguration, some loads with lower requirements for power supply reliability are selected as Class III loads, and their switch states are used as the reconfiguration objects. If the i-th Class III load is P Lci , then the load shedding amount P cut is as follows:

[0063] P cut = Σ i∈I a i P Lci (18)

[0064] Among them, I is the set of Class III loads, ai It is 1 indicating that the power supply for the load is normal, otherwise it is 0.

[0065] Combine the N - group switch decision schemes of M level - III loads to form matrix A. To simulate the post - disaster distribution network system, the operating states of lines, tie - lines, and generators in the fault scenario are described by b, c, and d respectively. When b i , c j , d t has a value of 1, it represents that line i is closed, tie - line j is closed, and DG t is operating normally, otherwise it is 0. Among them, the closing of the tie - line is to maintain the radial structure of the system and reduce the power outage area. Combining the line, tie - line, and DG status information, the system topology matrix S can be obtained. The CNN structure adopted in this embodiment is as Figure 2 shown:

[0066] The input information of the input layer of the neural network model for training the energy storage output is a multi - dimensional binary vector, representing the opening and closing states corresponding to lines, tie - lines, DGs, and level - III loads respectively. 1 represents that the component is operating normally, and 0 represents that the component is out of operation; the output layer contains 6 nodes, which are associated with the system power loss, load shedding amount, maximum node voltage, minimum node voltage, and two energy storage outputs. In the offline training stage, after pre - processing the training data, the CNN model parameters are adjusted to complete the learning of the multiple complex mapping relationships between power flow data and line status, tie - line status, DG status, and load switch status.

[0067] Limit P cut to 1.05 to 1.2 times of P e before the energy storage is put into use to screen the reconstruction schemes. Input the decision - making after hierarchical screening into the CNN to obtain the reconstruction parameters. Among the alternative reconstruction schemes, it is stipulated that the maximum voltage deviation does not exceed 0.1, and the power deficit of the system does not exceed 0.5 MW. To solve the energy deficit and meet the reconstruction requirements, the weight coefficients w1, w2, w3, w4 are determined by the analytic hierarchy process, and the optimal reconstruction decision of the distribution network is determined through the multi - objective decision - making evaluation function:

[0068]

[0069] In the formula, P k cut0 represents the normalized power loss, voltage deviation, power deficit, and load shedding amount under decision - making scheme k.

[0070] Among them, the involved weight coefficients are determined by the analytic hierarchy process. The reference implementation process includes:

[0071] (1) Establish a weight calculation hierarchical structure: Decompose the distribution network reconstruction goal into a hierarchical evaluation system containing elements related to post - disaster requirements;

[0072] (2) Element importance ranking: According to the actual importance of each influencing factor in the post-disaster scenario, prioritize the elements in the hierarchical structure;

[0073] (3) Weight calculation and verification: Based on the priority calibration results in step (2), calculate the weight coefficients through the built-in mathematical processing process of the analytic hierarchy process, and output the weight coefficients adapted to the current post-disaster scenario after completing the logical consistency verification.

[0074] Select the switch combination with the smallest Ass and meeting the constraint conditions as the optimal reconstruction decision. When all switch combinations in the load switch alternative solution set cannot meet the constraint conditions in the reconstruction optimization model, it is judged as a reconstruction anomaly. Construct an over-limit evaluation function to evaluate the over-limit situation:

[0075]

[0076] In the formula, α, β ∈ (0, 1) are weight coefficients and satisfy α + β = 1, U 0max , U 0min are the upper and lower limits of the node voltage respectively, and U jmax and U jmin are the upper and lower limits of the node voltage in the jth decision. P emax and P emin are the upper and lower limits of the power deficit P ej respectively. r u , r e being 1 indicates that the voltage and balance power are over-limit, otherwise it is 0.

[0077] Finally, take the load combination with the smallest over-limit value as the distribution network reconstruction plan in case of reconstruction anomaly.

[0078] Based on the same inventive concept, the present invention further provides a computer device, which includes: one or more processors, and a memory for storing one or more computer programs; the program includes program instructions, and the processor is configured to execute the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, and is used to implement one or more instructions. Specifically, it is used to load and execute one or more instructions in the computer storage medium to implement the above method.

[0079] It should be further noted that, based on the same inventive concept, the present invention further provides a computer storage medium, on which a computer program is stored, and the computer program, when run by a processor, executes the above method. The storage medium may be any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electrical, magnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a Random Access Memory (RAM), a Read-Only Memory (ROM), an Erasable Programmable Read-Only Memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or combined with an instruction execution system, apparatus, or device.

[0080] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0081] The above has shown and described the basic principles, main features and advantages of the present disclosure. Those skilled in the art should understand that the present disclosure is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present disclosure. Without departing from the spirit and scope of the present disclosure, the present disclosure will have various changes and improvements, and these changes and improvements all fall within the scope of the present disclosure claimed.

[0082] The present invention is not limited to the above best implementation manner. Anyone can obtain various other forms of the method for multi-energy storage collaborative emergency energy management and rapid reconstruction of a distribution network based on deep learning under the inspiration of the present invention. All equal changes and modifications made according to the scope of the patent application of the present invention shall fall within the scope covered by the present invention.

Claims

1. A method for emergency energy management and rapid reconstruction of multi-energy storage coordinated distribution network based on deep learning, characterized by: Based on the emergency support optimization model of multi-energy storage collaboration and the distribution network multi-objective reconstruction optimization model, the rapid reconstruction of the distribution network based on deep learning is realized; The multi-energy storage coordinated emergency support optimization model takes the minimum system network loss as the optimization goal; the distribution network multi-objective reconstruction optimization model considers the fault scenario information caused by line faults, distributed power supply faults and tie line closures, as well as the stable operation conditions of the distribution network after reconstruction; the deep learning-based distribution network rapid reconstruction uses the data obtained from simulation deduction to train the deep learning model, inputs the fault system topology and load switch information online, and outputs the power flow parameters; And determine the optimal reconstruction plan of the distribution network through multi-objective decision-making.

2. The method for emergency energy management and rapid reconstruction of distribution network multi-energy storage collaboration based on deep learning according to claim 1 is characterized in that: In the multi-energy storage coordinated emergency support optimization model, the total output value of the energy storage is determined by the power shortage of the system, and the Distflow flow model is used to describe the flow equation with unknown energy storage state. The objective function is to minimize the amount of power loss; and it includes the constraints that the energy storage in the distribution network should meet when it is put into operation.

3. The method for emergency energy management and rapid reconstruction of distribution network multi-energy storage collaboration based on deep learning according to claim 1 is characterized in that: The optimization objectives of the distribution network multi-objective reconstruction optimization model include minimizing load shedding, minimizing network loss, minimizing voltage deviation and minimizing balanced power shortage; and the stable operation of the distribution network after reconstruction satisfies the distributed power constraints, power balance constraints and voltage constraints.

4. The method for emergency energy management and rapid reconstruction of distribution network multi-energy storage collaboration based on deep learning according to claim 1 is characterized in that: The multi-energy storage coordinated emergency support optimization model and the distribution network multi-objective reconstruction optimization model are used to solve the energy storage output and power flow parameter data under various fault scenarios as training sets and verification sets for deep learning.

5. The method for emergency energy management and rapid reconstruction of distribution network multi-energy storage collaboration based on deep learning according to claim 1 is characterized in that: The loads with low power supply reliability requirements are classified as Class III loads, and the switch state of Class III loads is used as the reconstruction object; if the i-th Class III load is P Lci , then the load reduction amount P cut as follows: P cut =∑ i∈I a i P Lci Among them, I is the load set of level III, a i 1 indicates that the load power supply is normal, otherwise it is 0; Combine N groups of switch decision schemes of M level III loads to form a matrix A; describe the operating states of the line, tie line and generator in the fault scenario with b, c and d respectively; when b i 、c j ,d t When the value is 1, it means that line i is closed and running, tie line j is closed and running, and distributed power supply t is running normally. If the above components cannot run normally, it is 0. The line, tie line and distributed power supply status information are combined to obtain the system topology matrix S. The load information and topology information in the fault scenario are combined with the switch decision scheme matrix A and the system topology matrix S to determine the characteristics of the system in the scenario, which is used as the input information of the neural network model in deep learning.

6. The method for emergency energy management and rapid reconstruction of distribution network multi-energy storage collaboration based on deep learning according to claim 1 is characterized in that: The input information of the input layer of the neural network model for training energy storage output is a multi-dimensional binary vector, which represents the opening and closing states of the line, tie line, distributed power source, and level III switch respectively; the output layer contains 6 nodes, which are associated with system network loss, load removal, maximum node voltage, minimum node voltage and two energy storage outputs; In the offline training phase, the CNN model parameters are adjusted after preprocessing the training data to complete the learning of multiple complex mapping relationships between power flow data and line status, tie line status, distributed power supply status and load switch status.

7. The method for emergency energy management and rapid reconstruction of distribution network multi-energy storage collaboration based on deep learning according to claim 5 is characterized by: Set the load reduction amount P for level III load cut Limit the power shortage P before energy storage is put into use e The multiple of is used to screen the reconstruction scheme; The decision after the hierarchical screening is input into the CNN to obtain the reconstruction parameters; in the alternative reconstruction scheme, the upper limit of the maximum voltage deviation and the power shortage of the system is preset; The weight coefficients w1, w2, w3, w4 are determined by combining the influencing factors of post-disaster demand through the analytic hierarchy process, and the optimal reconstruction decision of the distribution network is determined based on the multi-objective decision evaluation function; Where L0 k 、U0 k , P e0 k , P k cut0 represents the normalized network loss, voltage deviation, power shortage, and load shedding under decision scheme k; The switch combination with the smallest Ass and satisfying the constraints is selected as the optimal reconstruction decision.

8. The method for emergency energy management and rapid reconstruction of distribution network multi-energy storage collaboration based on deep learning according to claim 7 is characterized in that: When all switch combinations in the load switch alternative solution set cannot meet the constraints in the reconstruction optimization model, it is judged as a reconstruction anomaly, and the over-limit evaluation function is used to evaluate the over-limit situation: In the formula, α, β∈(0,1) are weight coefficients and satisfy α+β=1, U 0max , U 0min are the upper and lower limits of node voltage, U jmax and U jmin is the upper and lower limits of the node voltage in the jth decision; P emax and P emin They are power deficit P ej The upper and lower limits of r u 、r e If it is 1, it means that the voltage or balanced power exceeds the limit, otherwise it is 0; The load combination with the smallest over-limit value is used as the distribution network reconstruction plan when reconstruction is abnormal.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements the steps of the distribution network multi-energy storage collaborative emergency energy management and rapid reconstruction method based on deep learning as described in any one of claims 1-8.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the distribution network multi-energy storage collaborative emergency energy management and rapid reconstruction method based on deep learning are implemented as described in any one of claims 1 to 8.

Images