Power distribution network-micro-grid collaborative dual-time scale scheduling method in combination with topological optimization

Abstract

The invention discloses a power distribution network-micro-grid collaborative dual-time scale scheduling method in combination with topological optimization. According to the method, a long-time-scale scheduling period is set at the hierarchy of the power distribution network, a scheduling model combining topology reconstruction and active optimization is adopted, and unified regulation and control of global operation characteristics such as grid power flow, line heavy load and node voltage are achieved; an ultra-short time scale real-time scheduling period is set in a micro-grid level, and new energy fluctuation is rapidly tracked and compensated based on rapid adjustable resources; and through a multi-level information interaction mechanism, power distribution network scheduling constraint downloading and micro-grid feedback uploading are realized, and friendly interaction and cooperative operation between the power distribution network and the micro-grid are realized. According to the method, the flexibility and the new energy consumption capability of the power distribution system are remarkably improved through a topology flexible regulation and control means, meanwhile, the rapid response capability of the system to the time-varying uncertainty is enhanced through a double-level scheduling mode, the operation loss and the safety risk of the power distribution network are reduced, and the method has good engineering application value.

Description

Technical Field

[0001] This invention belongs to the field of distribution microgrid optimization scheduling technology, specifically involving a distribution network-microgrid collaborative dual-time-scale scheduling method that combines topology optimization. Background Technology

[0002] With the rapid penetration of distributed photovoltaic and wind power and other new energy sources into the distribution network, the operation of the distribution system exhibits characteristics such as strong multi-source randomness, rapid power flow fluctuations, and easy voltage overruns. The power flow distribution of the distribution system shows significant randomness and time-varying characteristics. After the large-scale integration of distributed power sources, the distribution network is gradually shifting from the traditional "source follows load" mode to a "source and load dual random" mode. Node voltage, current, and power flow direction fluctuate frequently, and local lines are prone to operational risks such as reverse power flow and node voltage overruns. At the same time, the strong volatility of new energy sources significantly increases the difficulty of distribution network scheduling. The traditional slow-time-scale scheduling method, which is mainly based on thermal power, is unable to meet the dynamic adjustment needs under a high proportion of new energy sources.

[0003] Traditional distribution network dispatching methods primarily rely on optimal power flow calculations under fixed topologies, making it difficult to simultaneously address network security, equipment capacity constraints, and renewable energy integration needs within a short timeframe. On the other hand, while microgrids possess rapid adjustment capabilities such as energy storage and inverters, they lack coordination mechanisms with the upper-level distribution network, often failing to fully leverage their rapid adjustment advantages in actual operation. Furthermore, due to the strong nonlinearity of AC power flow equations and the complexity of topological variable discretization, existing dispatching methods struggle to achieve stable and feasible optimization results within actual operating time windows.

[0004] In actual operation, distribution networks and microgrids coexist in a multi-level structure: the distribution network bears the overall responsibility for wide-area stability, voltage maintenance, and power flow security; microgrids typically include energy storage, inverter-type distributed power sources, and flexible loads, possessing rapid adjustment capabilities and able to compensate for local renewable energy fluctuations within seconds. Therefore, how to achieve coordinated control between the distribution network and microgrids in a multi-level structure has become an important issue for the current distributed energy consumption and safe operation of the distribution network.

[0005] Furthermore, there is a significant time scale difference between distribution networks and microgrids: distribution networks require topology and active power optimization at the minute level, while microgrids need to rapidly control renewable energy fluctuations within seconds. The lack of a coordinated scheduling method that takes into account multiple time scales and multiple levels makes it easy for problems such as inconsistent objectives, conflicting boundary conditions, and delayed regulation response to arise between distribution networks and microgrids.

[0006] On the other hand, most existing distribution network dispatching methods are based on fixed topology optimization. Faced with power flow shifts caused by the time-varying nature of new energy sources, a single topology may be insufficient to simultaneously address operational performance aspects such as line load balancing, voltage compliance, and network losses. Although topology reconfiguration can achieve power flow redistribution by changing switch states, traditional topology optimization is an NP-hard problem involving a large number of discrete variables, resulting in high solution complexity and making it difficult to solve efficiently within dispatching timescales such as fifteen minutes.

[0007] For real-time control of microgrids, conventional methods often rely on classic optimization algorithms such as Newton's method and interior-point method. These algorithms are sensitive to the initial point and are prone to convergence failure under highly nonlinear AC power flow constraints. Furthermore, they cannot guarantee continuous and stable provision of feasible scheduling commands within a strict second-level time window, affecting the timely absorption of renewable energy and operational safety.

[0008] Overall, with a high proportion of renewable energy connected to the grid, the following technical challenges exist: (1) The operating status of the distribution network fluctuates greatly, and it is necessary to use topology reconfiguration to dynamically adjust the power flow, but it is difficult to complete the large-scale mixed integer AC-OPF solution in a short period of time.

[0009] (2) Microgrid regulation resources have a fast response speed but strict calculation time limit, and the optimization solution that satisfies the AC power flow constraint needs to be completed within 10 seconds.

[0010] (3) There is a lack of effective coupling mechanism between multi-time scale scheduling, making it difficult to achieve coordination between distribution network (minute level) and microgrid (second level).

[0011] Traditional solution methods lack robustness and are prone to optimization failure or convergence failure under complex constraints.

[0012] Therefore, there is an urgent need for a new scheduling method that can take into account topology optimization efficiency, AC power flow feasibility, second-level fast solution capability, and multi-timescale coordinated control, so as to improve the adaptability of the power distribution system to the fluctuation of new energy sources and enhance the system flexibility and operational safety. Summary of the Invention

[0013] The purpose of this invention is to overcome the shortcomings of the prior art and provide a dual-time-scale scheduling method for distribution network-microgrid collaboration that combines topology optimization. This method significantly improves the flexibility of the distribution system and the capacity for renewable energy absorption through topology flexible control, while enhancing the system's rapid response to time-varying uncertainties through a dual-level scheduling mode. This reduces the operating losses and safety risks of the distribution network and has good engineering application value.

[0014] The technical solution to achieve the above objectives is: a dual-time-scale scheduling method for distribution network-microgrid collaborative scheduling combining topology optimization, comprising the following steps: S1, Data Acquisition and Initialization Steps: Acquire power grid operation status data and microgrid operation status data; S2, Candidate Topology Filtering Steps: Filter the switchable topology and output the candidate topology set; S3, 15-minute dispatch steps for distribution networks: Combining topology optimization and active power optimization, global operation optimization is achieved on a 15-minute time scale, and boundary conditions are issued to microgrids; S4, Microgrid ultra-short-term (10-second) real-time scheduling steps: Construct a real-time optimization model for the microgrid and output the real-time scheduling results; S5, Rolling Coordinated Scheduling and Status Feedback Mechanism Steps: Realize the linkage and mutual correction between the distribution network and microgrid to form a complete two-level scheduling cycle.

[0015] The aforementioned method for coordinated dual-time-scale scheduling of distribution networks and microgrids, which combines topology optimization, includes the following in step S1: the distribution network operating status data includes node voltage, line current, power flow distribution, switch status, location of adjustable switches, actual output of renewable energy and 15-minute predicted values, energy storage charge status, and adjustable range of controllable loads; the microgrid operating status data includes microgrid bus voltage, current, adjustable capability of distributed power sources / energy storage / inverters, ramp rate, actual output of renewable energy and 10-second predicted values.

[0016] The aforementioned method for coordinated dual-time-scale scheduling of distribution networks and microgrids, which combines topology optimization, includes the following specific steps in step S2: S2.1, Switch Feasibility Screening: Screening switchable switches based on distribution network protection configuration and operation procedures; S2.2, Offline feasibility pre-screening reduces the search space dimension: Topologies that exceed the power flow limit are screened out through power flow calculation verification, and obviously non-compliant topologies are excluded by combining historical experience rules, and a candidate topology set is output.

[0017] The aforementioned distribution network-microgrid collaborative dual-time-scale scheduling method combining topology optimization includes the following specific steps in step S3: S3.1, Construct an optimal AC power flow optimization model for a mixed-integer distribution network, and determine the objective function and constraints of the optimal AC power flow optimization model for the distribution network; S3.2, a two-stage solution method based on AC feasibility recovery is used to solve the AC optimal power flow optimization model of the distribution network. The first stage is to use a convex approximation relaxation model for fast solution; the second stage is to transform the feasibility recovery problem into a search for a conventional stable equilibrium point through a negative gradient second-order dynamic system. S3.3, Distribution network long-term scale dispatch output: 15-minute cycle optimal topology, including generator active and reactive power output settings, energy storage active and reactive power output settings, node voltage reference values ​​and microgrid exchange power boundaries; S3.4, Issue boundary conditions to the microgrid: Boundary conditions include the upper / lower limit of the microgrid bus switching power, the expected range of voltage support for the microgrid, the reference power for energy storage charging and discharging, and the maximum adjustable range of the DG / inverter.

[0018] In the aforementioned distribution network-microgrid coordinated dual-time-scale scheduling method combining topology optimization, the objective function of the distribution network AC optimal power flow optimization model in step S3.1 is: (1) In formula (1): These are the weight coefficients of the objective function; Operating costs; For network loss; This refers to the amount of wind and solar power that has been curtailed. Operating costs The formula is: (2) In equation (2), The number of thermal power units involved in operation; NESS represents the active power output of the generator at node i; NESS represents the number of energy storage units. , , For thermal power units Cost coefficient; , , For energy storage units Cost coefficient; Network loss The formula is: (3) In equation (3), This represents the active power output of the generator at node i. Let J be the active power consumption of the load at node j. Wind and solar curtailment The formula is: (4) In equation (4): This represents the actual active power output of the new energy generating units at node i. NRES represents the on-grid active power of the renewable energy generating units at node i, and NRES represents the number of renewable energy generating units.

[0019] In the aforementioned distribution network-microgrid coordinated dual-time-scale scheduling method combining topology optimization, step S3.1 includes the following constraints on the distribution network AC optimal power flow optimization model: 1) Node communication flow constraints: (5) In equation (5), For the reactive power output of thermal power generator i, For the reactive power output of energy storage unit i, The reactive power consumption of the load. Let be the voltage amplitude at node i; They are nodes and nodes The electrical conductance and susceptance between them; Let be the voltage phase angle between node j and node i, and NB be the number of nodes; This is a line open / closed variable, representing the state of the line between node i and node j. When it is 1, it means the line is connected, and when it is 0, it means the line is disconnected. 2) Voltage amplitude limit: (6) In equation (6), and Let be the lower and upper voltage limits for node i; 3) Line thermal limit: (7) In equation (7), Let l be the apparent power of branch l; NL represents the rated capacity of branch i, and NL represents the number of lines. 4) Generator output limit: (8) In equation (8), , They are thermal power generators The lower and upper limits of active power output; 5) Restrictions on the output of new energy sources: (9) In equation (9), The active power of the new energy unit i. This is the upper limit of the active power of the new energy unit i; 6) Energy storage device operation constraints: (10) In equation (10), , These are the lower and upper limits of the output power of energy storage unit i, respectively; For energy storage units i in The state of charge at any given moment; For the charging and discharging efficiency of energy storage units; , These represent the state-of-charge and upper limit of energy storage unit i, respectively; The duration of the charging and discharging period; The above AC optimal power flow optimization model for distribution networks considers both the AC optimal power flow model and discrete line disconnection variables. This is a mixed integer programming problem; by introducing a slack variable S over all inequality constraints, it can be simplified to the following form: (11) In equation (11), This represents the set of all constraints in the optimal power flow optimization model for a distribution network. These are all the equality constraints of the optimal power flow optimization model for the distribution network. Let x be the set of all inequality constraints in the optimal power flow optimization model of the distribution network; let x be the set of all continuous variables and Z be the set of all discrete variables.

[0020] The above-mentioned distribution network-microgrid collaborative dual-time-scale scheduling method combining topology optimization, in step S3.2, the solution of stage one specifically involves: performing a convex approximation on the node AC power flow constraint equation (5), and simultaneously converting the discrete line disconnection variables... Relaxation as a continuous variable To solve quickly; the approximate nodal AC power flow constraint equation (5) is shown below: (12) And increase the relaxation of continuous variables Due to the limitations, we obtain: (13) Using commercial solvers such as IPOPT to solve the approximate constrained optimization problem using equations (1)-(4), (6)-(10), and (12)-(13), the obtained solution is the relaxed approximate solution, denoted as . ; Relaxed continuous variables for all relaxed approximate solutions Discretization is performed using a threshold method when... When the line ij is in the connected state, otherwise, when At this time, the ij state of the line is set to disconnected. It is usually set to 0.5; According to the distribution network operation rules, the 0-1 results of the lines are adjusted to obtain the final topology state vector that satisfies the network structure constraints. .

[0021] The aforementioned distribution network-microgrid coordinated dual-time-scale scheduling method combining topology optimization, in step S3.2, specifically the second stage solution involves: [The solution is performed under a fixed topology.] In the case of continuous variable x, the feasibility of communication recovery is as follows: The continuous part of the relaxation solution As initial values, combined with the final topology state vector To obtain the initial state of the system ; In fixed topology In the case of x, construct an AC optimal power flow optimization model for the distribution network with only x as the variable, as shown in equation (11), and denote the constraint set as... ; according to Constructing a second-order dynamical system with negative gradient: (14) In equation (14), express The Jacobian matrix, where T denotes the transpose of the matrix; The feasibility recovery problem is transformed into a search for the conventional stable equilibrium point of the dynamic system by using a second-order dynamic system with negative gradients; by Starting from the negative gradient second-order dynamic system equation (14), we find a feasible AC solution that satisfies the topological requirements by integrating the negative gradient until a normal stable equilibrium point is reached.

[0022] The aforementioned method for coordinated dual-time-scale scheduling of distribution networks and microgrids, which combines topology optimization, includes the following specific steps in step S4: S4.1, Constructing a real-time optimization model for the microgrid: The objective function of the real-time optimization model for the microgrid is: (15) In equation (15): These are the weight coefficients of the objective function. This represents the actual exchange power between the microgrid and the distribution network. This is the reference switching power of the microgrid provided by the upper layer; The constraints of the real-time optimization model for microgrids are as follows: Power balance constraints: (16) In equation (16), For microgrid network losses, NGW, NESSW, NRESW, and NBW represent the number of thermal power units (diesel engines, etc.) in the microgrid, the number of energy storage units in the microgrid, the number of new energy units in the microgrid, and the number of nodes in the microgrid, respectively. 2) Node communication flow constraints: (17) In equation (17), For the reactive power output of thermal power generator i, For the reactive power output of energy storage unit i, The reactive power consumption of the load. Let be the voltage amplitude at node i; They are nodes and nodes The electrical conductance and susceptance between them; Let be the voltage phase angle between node j and node i, and NB be the number of nodes; This is a line open / closed variable, representing the state of the line between node i and node j. When it is 1, it means the line is connected, and when it is 0, it means the line is disconnected. 3) Microgrid voltage amplitude limitation: (18) In equation (18), and Let be the lower and upper voltage limits for node i; 4) Generator output limit: (19) In equation (19), , Generators The lower and upper limits of active power output; 5) Restrictions on the output of new energy sources: (20) In equation (20), The active power of the new energy unit i. This is the upper limit of the active power of the new energy unit i; 6) Energy storage device operation constraints: (twenty one) In equation (21), , These are the lower and upper limits of the output power of energy storage unit i, respectively; Let i be the state of charge of energy storage unit i at time t; Let i be the state of charge of energy storage unit i at time t-1; For the charging and discharging efficiency of energy storage units; , These represent the state-of-charge and upper limit of energy storage unit i, respectively; The duration of the charging and discharging period; 7) Interface constraints with the upper-level distribution network: (twenty two) In equation (22), These are the lower and upper limits of the microgrid switching power given by the upper layer, respectively; S4.2 Real-time solution method based on dynamic system aims to find the approximate optimal solution of the real-time optimization model of microgrid within 10 seconds using a set of numerically robust dynamic system methods, avoiding possible divergence of the traditional Newton / interior point method, and ensuring the feasibility and safety of the solution. Construct a feasible region projection dynamics system such that the optimized intermediate trajectory lies around the feasible region, and output a safe approximate optimal solution during the 10s output phase; The real-time optimization model of microgrid (15)-(22) can be simplified as follows: (twenty three) In equation (23), The constraint set for the real-time optimization model of the microgrid includes equations (16) to (22). The feasible region Ψ of this problem is constructed as follows: (twenty four) The feasible region is appropriately relaxed to an approximate feasible region. : (25) In equation (25), The approximate relaxation parameter is set to 0.01; The feasible region projection dynamics system is constructed as follows: (26) In equation (26), The objective function is... The Jacobian matrix; K is the integrated gain matrix; For the approximate feasible region The projection operator; The projection operator is defined as:

[0023] Each iteration first updates the data using gradient descent, then projects it, ensuring that all iteration points x(k) always fall within the approximate feasible region. Within, the optimized intermediate trajectory always "stays close to the safe and feasible domain" during movement; The solution outputs the real-time scheduling results of the microgrid.

[0024] The aforementioned method for coordinated dual-time-scale scheduling of distribution networks and microgrids, which combines topology optimization, includes the following specific steps in step S5: S5.1, Distribution Network 15-Minute Cycle Rolling Optimization: The distribution network dispatch center performs S3-stage topology + active power dual-level optimization every 15 minutes to obtain: the current optimal topology, voltage and power flow reference values ​​of each node, reference power of each distributed power source and energy storage, and the switching power boundary of the microgrid interface; then, the above parameters are sent down to the microgrid level as hard boundary conditions for subsequent 10-second dispatch. S5.2, Real-time Optimization Control of Microgrid in 10-Second Cycles: After receiving information from the distribution network, the microgrid controller performs S4-stage projection dynamics system scheduling every 10 seconds. Using the latest measured value as the initial point, it evolves under constraints such as switching power, equipment capacity, and voltage, always ensuring that the trajectory is near the safe and feasible region. At the end of 10 seconds, it outputs the current point as a "safe approximate optimal solution". The output includes: energy storage charging and discharging commands, active and reactive power regulation values ​​of DG / inverter, actual switching power of the microgrid, and output commands of thermal power generator units. S5.3, Two-way information interaction: The microgrid updates and uploads prediction deviation, execution results, energy storage status, and actual switching power every 15 minutes; enabling the distribution network to receive feedback and update its internal model parameters; The distribution network issues instructions every 15 minutes: current topology, voltage and power flow reference values ​​of each node, reference power of each distributed power source and energy storage, and switching power boundary of the microgrid interface.

[0025] The distribution network-microgrid collaborative dual-time-scale scheduling method combining topology optimization of the present invention has the following beneficial effects: (1) Improve the solution efficiency and reliability of topology optimization: This invention divides topology optimization into two stages. By first performing an approximate solution of the communication feasibility and then performing a recovery of the communication feasibility, the discrete-continuous hybrid optimization problem can be solved quickly within a 15-minute scheduling cycle, significantly reducing the computational complexity. (2) Enhance the robustness of AC power flow solution and avoid infeasible solutions: By introducing the AC feasibility recovery model and the dynamic system solution framework, the optimization process can effectively avoid the problems of numerical divergence and infeasible solutions that are common in traditional methods, and ensure that the output solution always meets physical constraints such as voltage, current and power balance. (3) Microgrid achieves second-level, interruptible safety control: The microgrid adopts a feasible region projection dynamics system, which keeps the optimized intermediate trajectory near the safe feasible region. Even if the iteration is terminated early at the end of the 10-second cycle, it can directly output safe and executable near-optimal control commands, thereby improving real-time performance and operational stability. (4) Achieve coordinated operation of distribution and microgrids under multiple time scales: The distribution network layer provides topology, operating benchmark and exchange power constraints, and the microgrid layer makes rapid adjustments within the boundary to form a coordination mechanism of "minute-level global optimization + second-level local control", which effectively improves the capacity for new energy consumption;

[0026] (5) Improve the overall economy and operating performance of the system: By coordinating and optimizing the topology, power flow distribution and local fast adjustment resources, network losses can be reduced, operating costs can be reduced, voltage quality can be improved, and the overall operating performance of the system can be significantly improved. (6) It is feasible for engineering implementation: The proposed two-stage solution method and dynamic scheduling framework can be implemented on existing scheduling platforms and microgrid energy management systems, without relying on high-performance servers, and are suitable for actual power distribution system operation. Attached Figure Description

[0027] Figure 1 The flowchart shows the distribution network-microgrid collaborative dual-time-scale scheduling method combining topology optimization according to the present invention. Detailed Implementation

[0028] To enable those skilled in the art to better understand the technical solution of the present invention, its specific embodiments are described in detail below with reference to the accompanying drawings: Please see Figure 1 A method for coordinated dual-time-scale scheduling of distribution networks and microgrids combining topology optimization includes the following steps: S1, System Data Acquisition Steps, specifically include: S1.1, Distribution network operation status data acquisition: The acquisition content includes: node voltage, line current, power flow distribution, switch status, location of adjustable switches, actual output and 15-minute forecast value of new energy (photovoltaic, wind power), state of charge (SOC) of energy storage, and adjustable range of controllable load; S1.2, Microgrid Operation Status Data Acquisition: The acquisition content includes: microgrid bus voltage, current, adjustable capability of distributed power sources / energy storage / inverters, ramp rate, actual output of new energy sources and 10-second predicted values.

[0029] S2, Feasible Topology Set Generation Step. The goal of this step is to reduce the topology optimization search space and improve solution efficiency. It includes the following steps: S2.1 Switch Feasibility Screening: Based on the distribution network protection configuration and operation procedures, screen switchable switches, including but not limited to: ① It will not result in isolated areas; ② Do not damage the ring network or zoned power supply structure; ③ It does not affect the coordination of protection settings; S2.2, Offline feasibility pre-screening reduces the search space dimension: Topologies that exceed the power flow limit are screened out by power flow calculation verification, and obviously non-compliant topologies are excluded by combining historical experience rules, and the candidate topology set Ω_0 is output (significantly reduced in size, which is conducive to 15min optimization).

[0030] S3, Solution steps for long-term dispatching of distribution networks using topology optimization and active power dispatching: Combining topology optimization and active power optimization, global operational optimization is achieved on a 15-minute timescale. Specifically, the process includes the following steps: S3.1, Construct an AC optimal power flow optimization model for a mixed-integer distribution network. The objective function of this AC optimal power flow optimization model is: (1) In formula (1): These are the weight coefficients of the objective function; Operating costs; For network loss; This refers to the amount of wind and solar power that has been curtailed.

[0031] Operating costs The formula is: (2) In equation (2), The number of thermal power units involved in operation; NESS represents the active power output of the generator at node i; NESS represents the number of energy storage units. , , For thermal power units Cost coefficient; , , For energy storage units The cost coefficient.

[0032] Network loss The formula is: (3) In equation (3), This represents the active power output of the generator at node i. Let J be the active power consumption of the load at node j.

[0033] Wind and solar curtailment The formula is: (4) In equation (4): This represents the actual active power output of the new energy generating units at node i. NRES represents the on-grid active power of the renewable energy generating units at node i, and NRES represents the number of renewable energy generating units.

[0034] The constraints of the AC optimal power flow optimization model for distribution networks include: 1) Node communication flow constraints: (5) In equation (5), For the reactive power output of thermal power generator i, For the reactive power output of energy storage unit i, The reactive power consumption of the load. Let be the voltage amplitude at node i; They are nodes and nodes The electrical conductance and susceptance between them; Let be the voltage phase angle between node j and node i, and NB be the number of nodes. This is a line open / closed variable, representing the state of the line between node i and node j. A value of 1 indicates that the line is connected, and a value of 0 indicates that the line is disconnected.

[0035] 2) Voltage amplitude limit: (6) In equation (6), and Let be the lower and upper voltage limits for node i.

[0036] 3) Line thermal limit: (7) In equation (7), Let l be the apparent power of branch l; Let NL be the rated capacity of branch i, and NL be the number of lines.

[0037] 4) Generator output limit: (8) In equation (8), , They are thermal power generators The lower and upper limits of the active power output.

[0038] 5) Restrictions on the output of new energy sources: (9) In equation (9), The active power of the new energy unit i. This is the upper limit of the active power of the new energy unit i.

[0039] 6) Energy storage device operation constraints: (10) In equation (10), , These are the lower and upper limits of the output power of energy storage unit i, respectively; For energy storage units i in The state of charge at any given moment; For the charging and discharging efficiency of energy storage units; , These represent the state-of-charge and upper limit of energy storage unit i, respectively; This refers to the duration of the charging and discharging period.

[0040] The above AC optimal power flow optimization model for distribution networks considers both the AC optimal power flow model and discrete line disconnection variables. This is a mixed integer programming problem. By introducing a slack variable S over all inequality constraints, it can be simplified to the following form: (11) In equation (11), This represents the set of all constraints in the optimal power flow optimization model for a distribution network. These are all the equality constraints of the optimal power flow optimization model for the distribution network. Let x be the set of all continuous variables and Z be the set of all discrete variables in the optimal power flow optimization model for the distribution network.

[0041] S3.2, a two-stage solution method based on communication feasibility recovery: S3.2.1 Stage 1: Fast solution of convex approximation relaxation model: The nodal AC power flow constraint equation (5) is approximated by convex approximation, and the discrete line disconnection variables are also approximated. Relaxation as a continuous variable To achieve a fast solution, the approximate nodal AC power flow constraint equation (5) is shown below: (12) And increase the relaxation of continuous variables Due to the limitations, we obtain: (13) Using commercial solvers such as IPOPT to solve the approximate constrained optimization problem using equations (1)-(4), (6)-(10), and (12)-(13), the obtained solution is the relaxed approximate solution, denoted as . .

[0042] Relaxed continuous variables for all relaxed approximate solutions Discretization is performed using a threshold method when... When the line ij is in the connected state, otherwise, when At this time, the ij state of the line is set to disconnected. It is usually set to 0.5.

[0043] Based on the distribution network operation rules (such as avoiding island formation and maintaining a radial structure), the 0-1 results of the lines are adjusted to obtain the final topology state vector that satisfies the network structure constraints. .

[0044] S3.2.2 Phase Two: Recovery of the feasibility of communication based on the second-order dynamical system with negative gradient, specifically in a fixed topology. In the case of continuous variable x, the feasibility of communication recovery is as follows: The continuous part of the relaxation solution As initial values, combined with the final topology state vector To obtain the initial state of the system ; In fixed topology In the case of x, construct an AC optimal power flow optimization model for the distribution network with only x as the variable, as shown in equation (11), and denote the constraint set as... ; according to Constructing a second-order dynamical system with negative gradient: (14) In equation (14), express The Jacobian matrix, where T denotes the transpose of the matrix.

[0045] The feasibility recovery problem is transformed into a search for the regular stable equilibrium point of the second-order dynamic system with negative gradient by using a negative gradient second-order dynamic system.

[0046] by Starting from point A, the second-order dynamical system with the negative gradient is integrated according to equation (14) until a normal stable equilibrium point is reached, at which point a feasible AC solution that satisfies the topological requirements is found.

[0047] S3.3, Distribution network long-term scale dispatch output: 15-minute cycle optimal topology, generator active and reactive power output settings, energy storage active and reactive power output settings, node voltage reference value, microgrid switching power boundary.

[0048] S3.4, Issue boundary conditions to the microgrid: This step is the coupling layer between the distribution network and the microgrid. It includes: upper / lower limits of the microgrid bus switching power, the expected range of voltage support for the microgrid, reference power for energy storage charging and discharging, and the maximum adjustable range of the DG / inverter.

[0049] S4, Microgrid Ultra-Short-Term Real-Time Scheduling Steps: This is the third core innovation of this invention, used to address the second-level fluctuations in new energy sources. Specifically, it includes the following process: S4.1, Constructing a real-time optimization model for microgrids: The objective function of the real-time optimization model for microgrids considers minimizing the amount of wind and solar power curtailment and minimizing the deviation from the upper-level power setting, and is modeled as follows: (15) In equation (15): These are the weight coefficients of the objective function. This represents the actual exchange power between the microgrid and the distribution network. This is the reference switching power of the microgrid provided by the upper layer.

[0050] The constraints of the real-time optimization model for microgrids are as follows: Power balance constraints: (16) In equation (16), The network loss is represented by NGW, NESSW, NRESW, and NBW, which represent the number of thermal power units (diesel engines, etc.) in the microgrid, the number of energy storage units in the microgrid, the number of new energy units in the microgrid, and the number of nodes in the microgrid, respectively.

[0051] 2) Node communication flow constraints: (17) In equation (17), For the reactive power output of thermal power generator i, For the reactive power output of energy storage unit i, The reactive power consumption of the load. Let be the voltage amplitude at node i; They are nodes and nodes The electrical conductance and susceptance between them; Let be the voltage phase angle between node j and node i, and NB be the number of nodes. This is a line open / closed variable, representing the state of the line between node i and node j. A value of 1 indicates that the line is connected, and a value of 0 indicates that the line is disconnected.

[0052] 3) Microgrid voltage amplitude limitation: (18) In equation (18), and Let be the lower and upper voltage limits for node i.

[0053] 4) Generator output limit: (19) In equation (19), , Generators The lower and upper limits of the active power output.

[0054] 5) Restrictions on the output of new energy sources: (20) In equation (20), The active power of the new energy unit i. This is the upper limit of the active power of the new energy unit i.

[0055] 6) Energy storage device operation constraints: (twenty one) In equation (21), , These are the lower and upper limits of the output power of energy storage unit i, respectively; Let i be the state of charge of energy storage unit i at time t; Let i be the state of charge of energy storage unit i at time t-1; For the charging and discharging efficiency of energy storage units; , These represent the state-of-charge and upper limit of energy storage unit i, respectively; This refers to the duration of the charging and discharging period.

[0056] 7) Interface constraints with the upper-level distribution network: (twenty two) In equation (22), These are the lower and upper limits of the microgrid switching power given by the upper layer, respectively.

[0057] S4.2 Real-time solution method based on dynamic system. Objective: To find the approximate optimal solution of the real-time optimization model of the microgrid within 10 seconds using a numerically robust dynamic system method, avoiding possible divergence of the traditional Newton / interior point method, and ensuring the feasibility and safety of the solution.

[0058] Construct a feasible region projection dynamics system such that the optimized intermediate trajectory lies around the feasible region, and output a safe approximate optimal solution during the 10s output phase.

[0059] The real-time optimization model of microgrid (15)-(22) can be simplified as follows: (twenty three) In equation (23), The constraint set for the real-time optimization model of the microgrid includes equations (16) to (22).

[0060] The feasible region Ψ of this problem is constructed as follows: (twenty four) The feasible region is appropriately relaxed to an approximate feasible region. : (25) In equation (25), To approximate the relaxation parameter, it is usually taken as 0.01.

[0061] The feasible region projection dynamics system is constructed as follows: (26) In equation (26), The objective function is... The Jacobian matrix; K is the integrated gain matrix; For the approximate feasible region The projection operator.

[0062] The projection operator is defined as:

[0063] Each iteration first updates the data using gradient descent, then projects it, thus ensuring that all iteration points x(k) always fall within the approximate feasible region. The internal principle is to optimize the intermediate trajectory so that it always moves "close to the safe and feasible domain".

[0064] The solution process is as follows: ① Initialization process: Use the current measurement value as the initial point x0, and perform projection calculations to obtain a safe initial point. : ; ②Construct the feasible domain projection dynamic system equation (26); ③ Starting from a safe initial point Starting from the integral feasible domain projection dynamics system equation (26); ④ Stop integrating when the time reaches 10s or the feasible region projected dynamics system finds a stable equilibrium point; ⑤ Output the real-time scheduling results of the microgrid.

[0065] Since each update step involves projection into the feasible region, the entire trajectory remains within the approximately feasible region. Therefore, at the end of 10 seconds, regardless of whether full convergence has been achieved, taking the current iteration point yields a feasible solution that satisfies all safety constraints. Furthermore, because the dynamic system continuously evolves along the descent direction of the objective function, it also represents an approximately optimal solution obtainable within the current time window.

[0066] S5, Rolling Coordinated Scheduling and Status Feedback Mechanism Steps: How the distribution network (15-minute level) and microgrid (10-second level) work together and correct each other to form a complete two-level scheduling loop. The specific process is as follows: S5.1, Distribution Network 15-Minute Cycle Rolling Optimization: Every 15 minutes, the distribution network dispatch center performs a topology + active power dual-level optimization in phase S3, obtaining: the current optimal topology, voltage and power flow reference values ​​for each node, baseline power for each distributed power source and energy storage, and the switching power boundary of the microgrid interface. Then, these parameters are distributed to the microgrid level as hard boundary conditions for subsequent 10-second dispatching.

[0067] S5.2, Real-time Optimization Control of Microgrid in 10-Second Cycle: After receiving information from the distribution network, the microgrid controller performs the S4 stage projection dynamics system scheduling once every 10 seconds: using the latest measured value as the initial point, it evolves under constraints such as switching power, equipment capacity, and voltage, always ensuring that the trajectory is near the safe and feasible region, and outputs the current point as the "safe approximate optimal solution" at the end of 10 seconds.

[0068] The output includes: energy storage charging and discharging commands, active and reactive power regulation values ​​of DG / inverter, actual exchange power of microgrid, and output commands of thermal power generator units.

[0069] S5.3, Two-way Information Interaction: The microgrid updates and uploads information every 15 minutes, including prediction deviations, execution results, energy storage status, and actual switching power. This allows the distribution network to receive feedback and update its internal model parameters. The distribution network also issues commands every 15 minutes, including the current topology, voltage and power flow reference values ​​for each node, reference power for each distributed power source and energy storage, and the switching power boundary of the microgrid interface.

[0070] The following example illustrates the specific implementation process of the method of the present invention by connecting a 10 kV distribution network of a city to a microgrid containing photovoltaic and energy storage.

[0071] System Structure and Operating Environment: The distribution network in this area adopts a radial structure, comprising 32 nodes, 34 lines, and several tie switches. The system is connected to a microgrid, which includes: 2 inverter-type photovoltaic power sources (total installed capacity 1.2MW), 1 energy storage system (rated 500kW / 1MWh), and a common coupling point (PCC) for connection to the distribution network.

[0072] The dispatch cycle for the distribution network is 15 minutes; the real-time dispatch cycle for the microgrid is 10 seconds.

[0073] High photovoltaic output leads to problems such as reverse power flow and node voltage deviation on some lines of the distribution network at noon on sunny days. Therefore, it is necessary to implement the distribution-micro coordinated scheduling method provided by this invention.

[0074] The steps of this embodiment are as follows: Step S1: Data Acquisition and Model Initialization: The distribution network dispatch center collects the predicted load and photovoltaic output data within the current 15-minute time window, including: voltage, current, and power flow measurements at each node, topology information such as line impedance and switch status, photovoltaic predicted power curve and uncertainty range, and energy storage SOC=68%.

[0075] Step S2: Candidate Topology Screening: Based on the protection configuration and operating procedures, the system allows operation of 7 switches. Using a fast power flow verification method, 4 feasible topologies are screened, forming a candidate set Ω1, including: τ1: Maintaining the original topology; τ2: Close the 11–14 interlocking switch; τ3: Close the 23–24 interlocking switch; τ4: Simultaneously close switches 11–14 and open switches 8–9; Topologies τ3 and τ4 may mitigate the overload trend of the main line and are considered key candidates for future applications.

[0076] Step S3: Solving the 15-minute distribution network topology optimization and joint active power scheduling problem, specifically: S3.1 Construction of Mixed Integer AC Optimal Power Flow Model: The dispatch center first establishes a mixed integer AC optimal power flow model based on the current cycle's operating data. The model's objective function includes distribution network operating costs, network losses, and renewable energy consumption. Constraints include node AC power flow equations, node voltage amplitude limits, line current limits, active and reactive power output limits for distributed generation, and operational limitations of energy storage devices. The model includes continuous variables and line on / off state variables, preparing for the subsequent two-stage solution.

[0077] S3.2 A two-stage solution method based on communication feasibility recovery, specifically including: S3.2.1 Stage 1: Fast solution of convex approximation relaxation model: Perform the following steps on candidate topologies τ1, τ2, τ3, and τ4: (1) By performing a convex approximation on the AC power flow equation, an approximate AC power flow equation is obtained; (2) Relax the line opening and closing variables from binary variables to continuous variables on the interval [0,1], and add corresponding relaxation constraints; (3) The IPOPT solver is used to solve the relaxed optimization model consisting of the objective function and convex approximation constraints to obtain the relaxed approximate solution. ; (4) Discretize the line state variables obtained by relaxation according to a fixed threshold. When the relaxation value is greater than the threshold, set it to 1; otherwise, set it to 0 to obtain the preliminary discrete topology Z0. (5) Perform structural correction on Z0 according to the distribution network operation rules, eliminate islanding, maintain radial structure, etc., and finally form a feasible topology structure vector. After completing the above steps, the initial continuous variables of each candidate topology can be obtained. and its corresponding discretized topology .

[0078] S3.2.2 Phase Two: Feasibility Recovery of the Communication Based on the Second-Order Dynamical System with Negative Gradient The topology obtained in Phase 1 Based on this, perform the following steps: (1) Fixed topology Construct an AC power flow and equipment constraint model containing only continuous variables; (2) Obtained in Phase One As the initial point of the dynamic system; (3) Construct a second-order dynamic system with negative gradient based on the feasibility recovery model, and form dynamic equations by constraining the Jacobian matrix through the constraint set; (4) Integrate the dynamic system so that the system state gradually approaches the stable equilibrium point that satisfies all AC constraints along the dynamic trajectory; (5) After reaching the equilibrium point, the AC feasible solution under this topology is obtained. .

[0079] In this embodiment, the overall performance of topology τ3 is better than that of other topologies, so τ3 is selected as the optimal topology for this scheduling cycle.

[0080] S3.3 Issue boundary conditions to the microgrid: After completing the topology optimization and active power dispatch of the distribution network layer, the dispatch center issues the operating boundary conditions for the current cycle to the microgrid based on the operating results of the optimal topology τ3. The issued conditions include: (1) The upper and lower limits of the switching power at the microgrid connection point are used to limit the range of power injection or absorption from the microgrid to the distribution network; (2) The allowable voltage fluctuation range at the grid connection point is used to guide the microgrid inverter in performing reactive power support; (3) The reference charging and discharging power and adjustable range of the energy storage system; (4) The range of active and reactive power regulation capabilities of distributed power sources and inverters; (5) The upper limit of the proportion of new energy power deviation undertaken by the microgrid in this cycle is used to guide real-time adjustment.

[0081] After receiving the above boundary conditions, the microgrid energy management system uses them as constraint inputs for subsequent real-time scheduling, providing a constraint basis for the next stage of second-level optimization process.

[0082] S4, Microgrid Ultra-Short-Term Real-Time Scheduling: Within a 10-second cycle, the microgrid performs ultra-short-term real-time scheduling based on boundary conditions issued by the distribution network and local real-time measurement data. The scheduling process includes two parts: constructing a real-time optimization model and solving the dynamic system.

[0083] S4.1 Construction of Real-time Optimization Model for Microgrids In this embodiment, the decision variables of the real-time optimization model include the active / reactive power of the energy storage system, the active / reactive power regulation of the distributed generation, and the voltage and phase angle of each node in the microgrid. The real-time optimization objective function comprehensively considers the node voltage deviation, the regulation cost of the energy storage system, and the regulation cost of flexible loads, and may include a deviation term of the power exchanged at the grid connection point from the reference value, thus constructing a complete real-time optimization model for the microgrid.

[0084] S4.2 Real-time solution of feasible region projection dynamics system (example) This embodiment employs a feasible region projection dynamics system to perform real-time scheduling calculations for the microgrid. The specific steps are as follows: (1) Use the real-time optimization result of the previous cycle as the starting point of the current cycle, and project the starting point to the safe and feasible region; (2) Construct a dynamic system based on the gradient of the objective function and the real-time constraint residuals, so that the system state evolves along the descent direction of the objective function; (3) After each iteration, the intermediate state is projected to the safe and feasible region so that the system state satisfies the node voltage, equipment capacity and grid connection point constraints throughout the entire scheduling process; (4) Repeat the dynamic iteration within the 10-second scheduling cycle to continuously reduce the objective function value and constraint residual; (5) When the 10-second cycle ends, the current system status is directly output as the real-time scheduling result of this cycle, including energy storage charging and discharging commands, inverter reactive power support commands and flexible load adjustment amounts.

[0085] Because of the feasible region projection mechanism, even if the iteration does not fully converge, a feasible scheduling solution that satisfies the safety constraints can be output at any time, ensuring the safe and stable operation of the microgrid under rapid dynamic conditions.

[0086] S5, Rolling Coordinated Scheduling and Status Feedback: In this embodiment, the distribution network and microgrid form a rolling coordinated scheduling closed loop with different time scales of 15 minutes and 10 seconds, as detailed below: (1) After the microgrid performs real-time scheduling every 10 seconds, it feeds back the actual exchange power, node voltage status, energy storage charge status and new energy prediction deviation for this cycle to the distribution network dispatch center. (2) The distribution network dispatch center updates the forecast of regional load and new energy output based on the feedback information and corrects the parameters of the dispatch model for the next cycle; (3) When the operating status fed back by the microgrid is close to the safety boundary, the distribution network dispatch center can adjust the grid connection point power limit, voltage reference value and other parameters in advance for this cycle, and send them to the microgrid for execution through the communication link; (4) At the start of the next 15-minute scheduling cycle, the distribution network re-executes topology optimization and active power scheduling, and generates a new topology and operating boundary by combining the feedback information from the previous cycle. (5) The microgrid continues to execute the 10-second real-time scheduling of the next cycle based on the updated boundary conditions, forming a closed loop of coordinated operation between the distribution network and the microgrid.

[0087] Through the aforementioned rolling collaborative scheduling mechanism, microgrids can quickly absorb local renewable energy fluctuations at the upper boundary, and distribution networks can dynamically adjust topology and power flow within a minute-level optimization cycle, making the overall system operation more flexible, secure, and economical.

[0088] Specific numerical examples I. Distribution Network Side (1) The predicted total active load within the 15-minute dispatch cycle is 5.0 MW, and the predicted total reactive load is 1.8 Mvar; (2) The voltage of the main transformer and the upstream power supply is 35kV, and the acceptable range of node voltage is 0.95 to 1.05 per unit value; (3) The upper limit of the load rate corresponding to the current limit of each line is 100%.

[0089] microgrid side (1) The microgrid contains two photovoltaic power sources with a total installed capacity of 1.2MW; (2) The energy storage system has a rated power of 500kW and a capacity of 1MWh. The current state of charge (SOC) is 60%, and the allowable SOC range is 20% to 90%. (3) The allowable deviation range of the grid connection point PCC node voltage is 0.98 to 1.04 per unit value.

[0090] II. 15-minute dispatch results of the distribution network Applying the mixed-integer AC optimal power flow and two-stage solution method described in S3 to the two candidate topologies τ1 and τ2 yields the following results: Topology τ3 (close the 23–24 interconnection switch): (1) The system network loss is 0.18MW; (2) All node voltages are in the range of 0.97 to 1.04 pu; (3) The maximum line load rate is 86%; (4) The photovoltaic power that can be absorbed has increased from 0.95MW to 1.10MW.

[0091] Based on a comprehensive comparison of objective functions, topology τ3 was selected as the optimal topology for this cycle. Under this topology, the operating boundaries issued to the microgrid include: a maximum outgoing power of 566.21 kW from the grid connection point; a maximum power absorbed from the distribution network of -300.89 kW from the grid connection point; and a voltage reference value of 1.02 pu. III. Numerical Results of 10-Second Real-Time Dispatch for Microgrids Suppose that at some point within this 15-minute period, the actual photovoltaic output deviates from the predicted value: The predicted photovoltaic output is 800 kW; The actual photovoltaic output is 950kW; Therefore, the power deviation of new energy sources in this cycle is 150kW. In the real-time optimization model, the main optimization objectives are node voltage deviation, energy storage power variation, and flexible load adjustment. Constraints include energy storage power range [−500, 500] kW, SOC range, flexible load adjustment range [−100, 200] kW, and grid connection point switching power constraint [−300, 500] kW.

[0092] The real-time scheduling results for this cycle are obtained by iteratively solving the feasible region projection dynamics system within a 10-second time window: The energy storage system has a charging power of 110kW (absorbing excess photovoltaic power), and its SOC increases from 60% to 71%. The active power exchange at the connection point between the microgrid and the distribution network has increased from 200kW to 240kW. The voltage at each node within the microgrid remains within the range of 0.99–1.03 pu; The grid connection point voltage is 1.021 pu, which meets the control requirements near the reference value of 1.02 pu; At the end of the 10-second scheduling cycle, the dynamic system reaches a state near stability, all constraints are met, and the current iteration point is directly used as the real-time optimization solution for this cycle and sent to the energy storage converter, inverter and flexible load control device for execution.

[0093] IV. Effects of Rolling Collaborative Operation Under the above numerical conditions, the key data fed back by the microgrid to the distribution network dispatch center includes: The actual grid connection point switching power is 240kW; Current SOC = 71%; The current photovoltaic deviation is 150kW; Neither the node voltage nor the line load rate exceeded the limit.

[0094] In the subsequent 15-minute cycle, the distribution network dispatch center updated the load and photovoltaic forecasts using feedback information, and performed topology optimization and active power dispatch again under the new boundary conditions. Through continuous rolling execution over multiple cycles, the system network loss stabilized at 0.21 MW from the initial operating condition, the average photovoltaic absorption capacity increased from 0.95 MW to approximately 1.08 MW, and the microgrid node voltage remained within the allowable range, verifying the numerical feasibility and effectiveness of the method of this invention.

[0095] The purpose of this invention is to: solve the problem of distribution network topology optimization being difficult to solve in a short time, thereby improving scheduling efficiency and solution feasibility; address the problem of unstable solutions caused by strong nonlinearity of AC power flow, ensuring that scheduling results are always feasible at the operational level; solve the problem of microgrids being unable to obtain stable real-time adjustment commands within seconds, enhancing the rapid response capability to renewable energy fluctuations; and address the lack of a unified coordination mechanism between the distribution network and microgrids, achieving multi-level, multi-timescale collaborative scheduling. Through these improvements, the global optimization of the distribution network and the local rapid control of the microgrid are unified, improving the renewable energy absorption capacity and system operational security.

[0096] The innovative aspects of the distribution network-microgrid collaborative dual-time-scale scheduling method combining topology optimization of this invention are mainly reflected in the following three aspects: (1) A two-stage solution method combining topology optimization and AC feasibility recovery is proposed: In view of the characteristics of large scale, many discrete variables and strong nonlinearity of distribution network topology optimization problem, a method is proposed to first use an approximate AC model to quickly obtain the relaxation solution, and then use an AC feasibility recovery model to refine the relaxation solution, so as to achieve fast and feasible topology optimization solution.

[0097] (2) A feasible region projection dynamic system suitable for second-level scheduling was constructed: In order to solve the problems of high solution speed and easy divergence in real-time scheduling of microgrids, a feasible region projection dynamic system was constructed so that the system state remains within the safe feasible region throughout the entire evolution process, and can output an executable approximate optimal solution at any time, which significantly improves the robustness and real-time performance of the solution.

[0098] (3) A multi-time-scale collaborative scheduling framework for distribution network and microgrid has been formed: a consistency constraint and feedback mechanism has been established between the minute-level global optimization of distribution network and the second-level rapid adjustment of microgrid, realizing multi-level and multi-time-scale collaborative operation, enabling distribution network to use the rapid adjustment capability of microgrid to cope with new energy fluctuations and improve the flexibility and reliability of the overall system.

[0099] In summary, the distribution network-microgrid collaborative dual-time-scale scheduling method of the present invention, which combines topology optimization, significantly improves the flexibility of the distribution system and the capacity for renewable energy absorption through topology flexible control. At the same time, it enhances the system's rapid response to time-varying uncertainties through a dual-level scheduling mode, reducing the operating losses and safety risks of the distribution network, and has good engineering application value.

[0100] Those skilled in the art should recognize that the above embodiments are merely illustrative of the present invention and are not intended to limit the present invention. Any variations or modifications to the above embodiments that are within the spirit and essence of the present invention will fall within the scope of the claims of the present invention.

Claims

1. A dual-time-scale scheduling method for distribution network-microgrid collaborative scheduling combining topology optimization, characterized in that, Includes the following steps: S1, Data Acquisition and Initialization Steps: Acquire power grid operation status data and microgrid operation status data; S2, Candidate Topology Filtering Steps: Filter the switchable topology and output the candidate topology set; S3, 15-minute dispatch steps for distribution networks: Combining topology optimization and active power optimization, global operation optimization is achieved on a 15-minute time scale, and boundary conditions are issued to microgrids; S4, Microgrid 10-second real-time scheduling steps: Construct a real-time optimization model for the microgrid and output the real-time scheduling results; S5, Rolling Coordinated Scheduling and Status Feedback Mechanism Steps: Realize the linkage and mutual correction between the distribution network and microgrid to form a complete two-level scheduling cycle.

2. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 1, characterized in that, In step S1, the power distribution network operation status data includes node voltage, line current, power flow distribution, switch status, position of adjustable switches, actual output of new energy sources and 15-minute predicted values, energy storage charge status, and adjustable range of controllable loads; the microgrid operation status data includes microgrid bus voltage, current, adjustable capability of distributed power sources / energy storage / inverters, ramp rate, actual output of new energy sources and 10-second predicted values.

3. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 1, characterized in that, Step S2 specifically includes the following process: S2.1, Switch Feasibility Screening: Screening switchable switches based on distribution network protection configuration and operation procedures; S2.2, Offline feasibility pre-screening reduces the search space dimension: Topologies that exceed the power flow limit are screened out through power flow calculation verification, and obviously non-compliant topologies are excluded by combining historical experience rules, and a candidate topology set is output.

4. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 1, characterized in that, Step S3 specifically includes the following process: S3.1, Construct an optimal AC power flow optimization model for a mixed-integer distribution network, and determine the objective function and constraints of the optimal AC power flow optimization model for the distribution network; S3.2, a two-stage solution method based on AC feasibility recovery is used to solve the AC optimal power flow optimization model of the distribution network. The first stage is to use a convex approximation relaxation model for fast solution; the second stage is to transform the feasibility recovery problem into a search for a conventional stable equilibrium point through a negative gradient second-order dynamic system. S3.3, Distribution network long-term scale dispatch output: 15-minute cycle optimal topology, including generator active and reactive power output settings, energy storage active and reactive power output settings, node voltage reference values ​​and microgrid exchange power boundaries; S3.4, Issue boundary conditions to the microgrid: Boundary conditions include the upper / lower limit of the microgrid bus switching power, the expected range of voltage support for the microgrid, the reference power for energy storage charging and discharging, and the maximum adjustable range of the DG / inverter.

5. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 4, characterized in that, In step S3.1, the objective function of the AC optimal power flow optimization model for the distribution network is: （1） In formula (1): These are the weight coefficients of the objective function; Operating costs; For network loss; This refers to the amount of wind and solar power that has been curtailed. Operating costs The formula is: （2） In equation (2), The number of thermal power units involved in operation; NESS represents the active power output of the generator at node i; NESS represents the number of energy storage units. , , For thermal power units Cost coefficient; , , For energy storage units Cost coefficient; Network loss The formula is: （3） In equation (3), This represents the active power output of the generator at node i. Let J be the active power consumption of the load at node j. Wind and solar curtailment The formula is: （4） In equation (4): This represents the actual active power output of the new energy generating units at node i. NRES represents the on-grid active power of the renewable energy generating units at node i, and NRES represents the number of renewable energy generating units.

6. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 5, characterized in that, In step S3.1, the constraints of the AC optimal power flow optimization model for the distribution network include: 1) Node communication flow constraints: （5） In equation (5), For the reactive power output of thermal power generator i, For the reactive power output of energy storage unit i, The reactive power consumption of the load. Let be the voltage amplitude at node i; They are nodes and nodes The electrical conductance and susceptance between them; Let be the voltage phase angle between node j and node i, and NB be the number of nodes; This is a line open / closed variable, representing the state of the line between node i and node j. When it is 1, it means the line is connected, and when it is 0, it means the line is disconnected. 2) Voltage amplitude limit: （6） In equation (6), and Let be the lower and upper voltage limits for node i; 3) Line thermal limit: （7） In equation (7), Let l be the apparent power of branch l; NL represents the rated capacity of branch i, and NL represents the number of lines. 4) Generator output limit: （8） In equation (8), , They are thermal power generators The lower and upper limits of active power output; 5) Restrictions on the output of new energy sources: （9） In equation (9), The active power of the new energy unit i. This is the upper limit of the active power of the new energy unit i; 6) Energy storage device operation constraints: （10） In equation (10), , These are the lower and upper limits of the output power of energy storage unit i, respectively; For energy storage units i in The state of charge at any given moment; For the charging and discharging efficiency of energy storage units; , These represent the state-of-charge and upper limit of energy storage unit i, respectively; The duration of the charging and discharging period; The above AC optimal power flow optimization model for distribution networks considers both the AC optimal power flow model and discrete line disconnection variables. This is a mixed integer programming problem; by introducing a slack variable S over all inequality constraints, it can be simplified to the following form: （11） In equation (11), This represents the set of all constraints in the optimal power flow optimization model for a distribution network. These are all the equality constraints of the optimal power flow optimization model for the distribution network. Let x be the set of all inequality constraints in the optimal power flow optimization model of the distribution network; let z be the set of all continuous variables and z be the set of all discrete variables.

7. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 6, characterized in that, In step S3.2, the solution of stage one specifically involves: performing a convex approximation on the node AC power flow constraint equation (5), and simultaneously converting the discrete line disconnection variables... Relaxation as a continuous variable To solve quickly; the approximate nodal AC power flow constraint equation (5) is shown below: （12） And increase the relaxation of continuous variables Due to the limitations, we obtain: （13） The approximate constrained optimization problem (Equations (1)-(4), (6)-(10), and (12)-(13) is solved using the commercial solver IPOPT. The resulting solution is the relaxed approximate solution, denoted as . ; Relaxed continuous variables for all relaxed approximate solutions Discretization is performed using a threshold method when... When the line ij is in the connected state, otherwise, when At this time, the ij state of the line is set to disconnected. It is usually set to 0.5; According to the distribution network operation rules, the 0-1 results of the lines are adjusted to obtain the final topology state vector that satisfies the network structure constraints. .

8. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 7, characterized in that, In step S3.2, the solution for stage two specifically involves: under a fixed topology... In the case of continuous variable x, the feasibility of communication recovery is as follows: The continuous part of the relaxation solution As initial values, combined with the final topology state vector To obtain the initial state of the system ; In fixed topology In the case of x, construct an AC optimal power flow optimization model for the distribution network with only x as the variable, as shown in equation (11), and denote the constraint set as... ; according to Constructing a second-order dynamical system with negative gradient: （14） In equation (14), express The Jacobian matrix, where T denotes the transpose of the matrix; The feasibility recovery problem is transformed into a search for the conventional stable equilibrium point of the dynamic system by using a second-order dynamic system with negative gradients; by Starting from the negative gradient second-order dynamic system equation (14), we find a feasible AC solution that satisfies the topological requirements by integrating the negative gradient until a normal stable equilibrium point is reached.

9. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 1, characterized in that, Step S4 specifically includes the following process: S4.1, Constructing a real-time optimization model for the microgrid: The objective function of the real-time optimization model for the microgrid is: （15） In equation (15): These are the weight coefficients of the objective function. This represents the actual exchange power between the microgrid and the distribution network; This is the reference switching power of the microgrid provided by the upper layer; The constraints of the real-time optimization model for microgrids are as follows: Power balance constraints: （16） In equation (16), For microgrid network losses, NGW, NESSW, NRESW, and NBW represent the number of thermal power units, energy storage units, new energy units, and nodes in the microgrid, respectively. 2) Node communication flow constraints: （17） In equation (17), For the reactive power output of thermal power generator i, For the reactive power output of energy storage unit i, The reactive power consumption of the load. Let be the voltage amplitude at node i; They are nodes and nodes The electrical conductance and susceptance between them; Let be the voltage phase angle between node j and node i, and NB be the number of nodes; This is a line open / closed variable, representing the state of the line between node i and node j. When it is 1, it means the line is connected, and when it is 0, it means the line is disconnected. 3) Microgrid voltage amplitude limitation: （18） In equation (18), and Let be the lower and upper voltage limits for node i; 4) Generator output limit: （19） In equation (19), , Generators The lower and upper limits of active power output; 5) Restrictions on the output of new energy sources: （20） In equation (20), The active power of the new energy unit i. This is the upper limit of the active power of the new energy unit i; 6) Energy storage device operation constraints: （21） In equation (21), , These are the lower and upper limits of the output power of energy storage unit i, respectively; Let i be the state of charge of energy storage unit i at time t; Let i be the state of charge of energy storage unit i at time t-1; For the charging and discharging efficiency of energy storage units; , These represent the state-of-charge and upper limit of energy storage unit i, respectively; The duration of the charging and discharging period; 7) Interface constraints with the upper-level distribution network: （22） In equation (22), These are the lower and upper limits of the microgrid switching power given by the upper layer, respectively; S4.2 Real-time solution method based on dynamic system aims to find the approximate optimal solution of the real-time optimization model of microgrid within 10 seconds using a set of numerically robust dynamic system methods, avoiding possible divergence of the traditional Newton / interior point method, and ensuring the feasibility and safety of the solution. Construct a feasible region projection dynamics system such that the optimized intermediate trajectory lies around the feasible region, and output a safe approximate optimal solution during the 10s output phase; The real-time optimization model of microgrid (15)-(22) can be simplified as follows: （23） In equation (23), The constraint set for the real-time optimization model of the microgrid includes equations (16) to (22). The feasible region Ψ for this problem is constructed as follows: （24） The feasible region is appropriately relaxed to an approximate feasible region. : （25） In equation (25), The approximate relaxation parameter is set to 0.01; The feasible region projection dynamics system is constructed as follows: （26） In equation (26), The objective function is... The Jacobian matrix; K is the integrated gain matrix; For the approximate feasible region The projection operator; The projection operator is defined as: Each iteration first updates the data using gradient descent, then projects it, ensuring that all iteration points x(k) always fall within the approximate feasible region. Within, the optimized intermediate trajectory always "stays close to the safe and feasible domain" during movement; The solution outputs the real-time scheduling results of the microgrid.

10. The method for coordinated dual-time-scale scheduling of distribution networks and microgrids combined with topology optimization according to claim 1, characterized in that, Step S5 specifically includes the following process: S5.1, Distribution Network 15-Minute Cycle Rolling Optimization: The distribution network dispatch center performs S3-stage topology + active power dual-level optimization every 15 minutes to obtain: the current optimal topology, voltage and power flow reference values ​​of each node, reference power of each distributed power source and energy storage, and the switching power boundary of the microgrid interface; then, the above parameters are sent down to the microgrid level as hard boundary conditions for subsequent 10-second dispatch. S5.2, Real-time Optimization Control of Microgrid in 10-Second Cycles: After receiving information from the distribution network, the microgrid controller performs S4-stage projection dynamics system scheduling every 10 seconds. Using the latest measured value as the initial point, it evolves under constraints of exchange power, equipment capacity, and voltage, always ensuring that the trajectory is near the safe and feasible region. At the end of 10 seconds, it outputs the current point as a "safe approximate optimal solution". The output includes: energy storage charging and discharging commands, active and reactive power regulation values ​​of DG / inverter, actual exchange power of the microgrid, and output commands of thermal power generator units. S5.3, Two-way information interaction: The microgrid updates and uploads prediction deviation, execution results, energy storage status, and actual switching power every 15 minutes; enabling the distribution network to receive feedback and update its internal model parameters; The distribution network issues instructions every 15 minutes: current topology, voltage and power flow reference values ​​of each node, reference power of each distributed power source and energy storage, and switching power boundary of the microgrid interface.

Images