Multi-time scale coordinated optimal dispatching method for main and distribution micro-integrated power grid based on cloud-edge collaborative architecture

Abstract

The application discloses a kind of main distribution micro-integrated power grid multi-time scale coordinated optimization scheduling method based on cloud edge cooperation architecture, establishes cloud-edge-end three-level scheduling system, and cloud receives the operation data from edge and terminal and utilizes day-ahead scheduling model to calculate day-ahead scheduling plan;Edge receives the scheduling instruction issued by cloud and is coordinated to the distribution network control and feeds back the result to cloud, and also uses the scheduling instruction as initial condition, uses the rolling optimization model to calculate correction amount and issues correction instruction, while regularly calculating system operation risk value and issuing action signal;Terminal receives the scheduling instruction and correction instruction to calculate the correction result as control reference, uses real-time scheduling model to calculate optimal control amount and is coordinated to microgrid control and feeds back to cloud, when receiving action signal, corresponding strategy is executed to re-coordinate microgrid control.The application can improve new energy consumption rate and system safety margin, realize economic-safety-flexible collaborative optimization under multi-time scale.

Description

Technical Field

[0001] This invention relates to power system operation and dispatch optimization technology, specifically to a multi-time-scale coordinated optimization dispatch method for a main-distribution-micro integrated power grid based on a cloud-edge collaborative architecture. Background Technology

[0002] With the widespread integration of distributed renewable energy, energy storage, and electric vehicles, power grid operation is exhibiting characteristics of strong uncertainty, multi-scale, and strong coupling. Traditional centralized dispatching models face bottlenecks in terms of data scale and real-time performance, lack unified coordination among the primary, distribution, and micro-level objectives, have insufficient connections between multiple time scales, and suffer from inadequate operational risk assessment and rapid response mechanisms. Summary of the Invention

[0003] The technical problem to be solved by this invention is to provide a multi-time-scale coordinated optimization scheduling method for a main-distribution-micro integrated power grid based on a cloud-edge collaborative architecture, which combines cloud-edge-end collaborative architecture with multi-time-scale coordinated optimization to achieve a balance between economy, security and flexibility.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A multi-time-scale coordinated optimization scheduling method for a main-distribution-microgrid integrated power grid based on a cloud-edge collaborative architecture is proposed. The integrated power grid deploys a main grid control module on a cloud server or dispatch control center to establish a cloud control layer, sets regional edge nodes at the distribution network layer to construct an edge coordination layer, and installs terminal control units on the microgrid and user side to deploy a terminal execution layer. The method includes: The cloud control layer receives real-time operational data from the edge coordination layer and the terminal execution layer and uses the day-ahead scheduling model to calculate the day-ahead scheduling plan. Then, it periodically issues scheduling instructions for the day-ahead scheduling plan according to the first time scale. The edge coordination layer receives scheduling instructions from the cloud control layer to coordinate and control the distribution network and feeds back the results to the cloud control layer. It also uses the scheduling instructions as initial conditions, uses the intraday rolling optimization model to calculate the correction amount according to the second time scale which is less than the first time scale, and issues the corresponding correction instructions. At the same time, it periodically calculates the system operation risk value and issues the corresponding action signal based on the comparison result between the system operation risk value and the preset threshold. The terminal execution layer receives scheduling instructions from the cloud control layer and correction instructions from the edge coordination layer. It calculates the correction result based on the scheduling instructions and correction instructions, uses the correction result as a control reference, and uses the real-time scheduling model to calculate the optimal control quantity on a rolling basis according to a third time scale that is smaller than the second time scale. It uses the optimal control quantity to coordinate and control the microgrid and feeds back the result to the edge coordination layer. If it receives an action signal from the edge coordination layer, it executes the corresponding strategy to re-coordinate and control the microgrid and feeds back the result to the edge coordination layer.

[0005] Furthermore, when issuing corresponding action signals based on the comparison results between the system operation risk value and the preset threshold, specifically, a green light signal is issued when the system operation risk value is less than the lower threshold, a yellow light signal is issued when the system operation risk value is greater than the lower threshold but less than the upper threshold, and a red light signal is issued when the system operation risk value is greater than the upper threshold; when executing the corresponding strategy to re-coordinate and control the microgrid, this includes: If the action signal is a green light signal, the optimal control quantity calculated by the real-time scheduling model is obtained according to the original rolling cycle, and the optimal control quantity is used to coordinate the control of the microgrid. If the action signal is a yellow light signal, the moment the action signal is received will be taken as the start time of the new rolling cycle. The optimal control quantity will be calculated according to the new rolling cycle using the real-time scheduling model and then used to coordinate the control of the microgrid. If the action signal is a red light signal, emergency control is performed on the microgrid. The emergency control includes load reduction, energy storage discharge, and tie line current limiting.

[0006] Furthermore, the mathematical expression of the objective function of the day-ahead scheduling model is as follows:

[0007] in, The objective function for the total cost in the current day phase; This is the set of days prior to the scheduled time. Indicates the day-ahead scheduling period; For a set of dispatchable power sources, Indicates the first One power source; For power supply During the period Those who have made contributions , Time periods Purchased power and sold power, For time period Network power loss For power supply The unit power generation cost coefficient, and Time periods Electricity purchase and sales price coefficients, This is the network loss reduction factor.

[0008] Furthermore, the mathematical expression of the objective function of the intraday rolling optimization model is as follows:

[0009] in, The objective function for rolling optimization during the intraday phase; For the first Power correction amount for a given time period; , These represent transferable load and load that can be reduced, respectively. , These are the power adjustment and demand response cost coefficients, respectively.

[0010] Furthermore, the mathematical expression of the objective function of the real-time scheduling model is as follows:

[0011] in, A and B are the state transition matrix and the control influence matrix, respectively. For system state variables; For control variables; The desired reference state vector; To control the increment; This is the state deviation weight matrix; To control the cost weight matrix; To predict the length of the time domain.

[0012] Furthermore, the method also includes: A three-layer optimization model consisting of a main network, a distribution network, and a microgrid is constructed. The main network aims to minimize system cost, the distribution network aims to balance risk and network loss, and the microgrid aims to optimize user-side economy and comfort. The objectives are cascaded using an augmented Lagrangian function. When calculating the day-ahead scheduling plan using the day-ahead scheduling model, when calculating the correction amount using the intraday rolling optimization model at a second time scale smaller than the first time scale, and when calculating the optimal control amount using the real-time scheduling model at a third time scale smaller than the second time scale, the following steps are all included: The optimization model solves its respective target optimization problem in parallel at each layer. It achieves coordination by exchanging boundary power variables and Lagrange multipliers. When the boundary power difference satisfies the convergence condition, a joint scheduling scheme of the optimal solutions of each layer is obtained. The cloud control layer summarizes the optimal solutions of each layer to form a global scheduling instruction.

[0013] Furthermore, the day-ahead scheduling model, intraday rolling optimization model, and real-time scheduling model are coupled through information exchange via boundary power consistency constraints. The mathematical expression for the boundary power consistency constraints is as follows:

[0014] in, This is the rolling correction amount. , , These represent the boundary exchange power for the day-ahead, intraday, and real-time periods, respectively.

[0015] Furthermore, the mathematical expression of the objective function of the main network in the optimization model is as follows:

[0016] in, For power supply During the period Those who have made contributions , Time periods Purchased power and sold power, For time period Network power loss For power supply The unit power generation cost coefficient, and Time periods Electricity purchase and sales price coefficients, This is the network loss reduction factor; The mathematical expression for the objective function of the distribution network in the optimization model is as follows:

[0017] in, For time period The system operation risk value, For distribution network losses, For node voltage, This is the voltage reference value. , , These are the weighting coefficients; The mathematical expression for the objective function of the microgrid in the optimization model is as follows:

[0018] in, , Microgrid time periods Purchased power and sold power, For energy storage power, In order to reduce the load, This represents the offset of the transferable load. and These are the actual and preferred load curves, respectively. This is a comfort weighting coefficient; and This is the energy storage usage cost coefficient. In order to reduce the load penalty factor, This is the penalty factor for transferable load offset. This represents the absolute value operator.

[0019] Furthermore, the day-ahead scheduling model, intraday rolling optimization model, and real-time scheduling model all incorporate equipment availability to weighted adjust the output of energy storage and adjustable loads. Specifically, the upper limit of energy storage power, the upper limit of adjustable load response, and the upper limit of the feasible domain of the real-time control variable in the real-time scheduling model are multiplied by availability, thereby achieving dynamic adjustment of the upper limit of equipment output. The mathematical expression for availability is as follows:

[0020] in, For the first Equipment availability For equipment failure rate, This refers to the equipment repair rate.

[0021] Furthermore, the mathematical expression for the system's operational risk value is as follows:

[0022] in, As a risk weighting factor, This is a comprehensive system risk index; This is an indicator of voltage over-limit risk. This serves as an indicator of line overload risk. As a power deviation risk indicator; The mathematical expressions for each energy storage risk indicator are as follows:

[0023]

[0024]

[0025]

[0026] in, For the first Branch power; This is the upper limit of line power; Number of branch roads; This is the upper limit of line power; For the first Node voltage; Reference voltage; For permissible voltage deviation limits; The number of nodes participating in the evaluation; For the first Measured power of the line; For the first Thermal stability or operating limits of each line; and They are time points The predicted value and the actual power; As a normalized benchmark; To avoid extremely small positive numbers with a denominator of zero; Number of time points; The mean; For the first Each energy storage unit at time The state of charge; For the first The upper and lower limits of operation for each energy storage unit; This represents the number of energy storage units.

[0027] Compared with the prior art, the advantages of the present invention are as follows: This invention achieves cross-level goal consistency and information closure through cloud-edge-device layered collaboration; it achieves continuous connection between planning, correction and rapid control through day-to-day-real-time joint optimization; it achieves proactive perception and dynamic response to operational risks through a traffic light mechanism; and it improves computational scalability and real-time performance through goal cascading and augmented Lagrange distributed solution. Attached Figure Description

[0028] Figure 1 This is a flowchart of a method according to an embodiment of the present invention.

[0029] Figure 2 This is a schematic diagram of the integrated main-distribution-microgrid structure of the cloud-edge collaborative architecture according to an embodiment of the present invention.

[0030] Figure 3 This is a flowchart illustrating the multi-timescale coordination and scheduling process according to an embodiment of the present invention.

[0031] Figure 4 This is a risk response control logic diagram under the traffic light mechanism in an embodiment of the present invention.

[0032] Figure 5 This is a diagram of a three-level target cascaded distributed optimization framework of the main-component-micro level according to an embodiment of the present invention.

[0033] Figure 6 This is a timing diagram of the algorithm execution and information interaction for multi-scale coordinated optimization scheduling in an embodiment of the present invention. Detailed Implementation

[0034] The present invention will be further described below with reference to the accompanying drawings and specific preferred embodiments, but this does not limit the scope of protection of the present invention.

[0035] This embodiment proposes a multi-timescale coordinated optimization scheduling method for a main-distribution-microgrid integrated power grid based on a cloud-edge collaborative architecture, establishing a three-level scheduling system of cloud-edge-end. It performs plan optimization with uncertainties at the day-ahead level, rolling prediction and re-optimization at the intraday level, and rapid coordinated control using model predictive control and a traffic light mechanism at the real-time level. A risk index system and equipment availability model are introduced, and the three-level coupled solution of main-distribution-microgrid is achieved through target cascading and augmented Lagrange distributed optimization. This combines the cloud-edge-end collaborative architecture with multi-timescale coordinated optimization to achieve a balance between economy, security, and flexibility.

[0036] like Figure 1 As shown, the method includes the following steps: Step S1: Establish a three-level scheduling system of cloud-edge-device, forming a cloud control layer, an edge coordination layer, and a device execution layer; Step S2: Establish a multi-timescale scheduling model and constraint set for day-to-day and real-time scheduling. Step S3: Construct an operational risk indicator system and equipment availability model to complete uncertainty modeling; Step S4: Set up a traffic light mechanism and provide thresholds and action sets to achieve dynamic risk response; Step S5: Use the target concatenation (ATC) distributed optimization algorithm to complete the cross-level coupled solution and instruction closed-loop execution; Step S6: Based on the configuration results of steps S1 to S5, the cloud control layer, edge coordination layer, and terminal execution layer are used to realize the multi-time-scale coordinated optimization scheduling of the main distribution and microgrid using a multi-time-scale scheduling model, uncertainty modeling, dynamic risk response, and ATC algorithm.

[0037] The following is a detailed explanation of each step.

[0038] Step S1 of this embodiment aims to construct a three-tiered cloud-edge-device architecture that coordinates the main grid, distribution network, and microgrid, enabling data sharing, command transmission, and optimized collaboration. The integrated main grid, distribution, and microgrid power grid deploys a main grid control module on a cloud server or dispatch control center to establish a cloud control layer; sets regional edge nodes in the distribution network layer to construct an edge coordination layer; and installs terminal control units on the microgrid and user side to deploy a terminal execution layer. Specifically, it includes the following steps: S11: Establish a cloud-based control layer.

[0039] Deploy a mainnet-level control platform in the cloud, specifically by deploying the mainnet control module on a cloud server or in a scheduling and control center, including a global optimization scheduling unit, a system risk identification unit, and a multi-layer information fusion unit; The cloud-based control layer establishes a network power flow model based on the data uploaded from each layer, and performs unified optimization of power generation plans, reserve capacity, interconnected power and voltage security. This enables the cloud-based control layer to receive real-time operating data from the edge and terminal layers, perform unified optimization calculations on the main grid power flow, interconnected power and reserve capacity, and issue day-ahead operating plans and real-time dispatch instructions. The cloud layer ensures data timescale consistency between different nodes through the unified time synchronization protocol (NTP), thereby achieving unified timescale with lower-level nodes and ensuring scheduling cycle consistency.

[0040] S12: Construct the edge coordination layer.

[0041] Edge computing nodes are set up and deployed at the distribution network layer, and distribution network optimization scheduling module, rolling prediction module and state estimation module are configured. The edge coordination layer receives global instructions (including day-ahead operation plans and boundary constraints) from the cloud control layer, and combines local load forecasting and real-time monitoring information to achieve coordinated control of active and reactive power within the distribution network, and complete the power distribution and reactive voltage regulation of the distribution network layer. The edge coordination layer also enables high-speed bidirectional communication with the terminal execution layer via message queues or lightweight communication protocols (such as MQTT, IEC 61850 MMS).

[0042] S13: Deploy the terminal execution layer.

[0043] A local controller (LC) is installed on the microgrid and user side. This unit is responsible for local real-time monitoring and control of distributed photovoltaic, energy storage, electric vehicles and other equipment. The terminal execution layer collects power, voltage, SOC and load status information in real time and feeds back the operating status to the edge coordination layer, realizing fine-grained management of multi-source loads.

[0044] S14: Establish a cloud-edge-device data interaction and control link.

[0045] Using a unified data communication protocol (such as CIM or IEC 61970 / 61968 format), a closed-loop data flow of "reporting - analysis - decision-making - distribution - execution - feedback" is formed through a two-way communication link. The cloud performs operational status summary and risk prediction, the edge layer is responsible for intermediate forwarding and local correction, and the terminal layer performs specific control, realizing dynamic information interaction between different levels.

[0046] S15: Improve data security and redundancy mechanisms.

[0047] Layered encryption and access control strategies are introduced to ensure the secure transmission of scheduling instructions and operational data; A fault detection and redundancy node mechanism is set up so that when any communication link is interrupted, the edge node can temporarily take over the local scheduling task to maintain the local stable operation of the system.

[0048] S16: Form a three-level closed-loop collaborative operation mechanism.

[0049] The cloud periodically issues global planning and scheduling instructions. Edge nodes, based on real-time predictions, make corrections according to prediction deviations. The terminal execution layer adjusts power according to the correction instructions, and after implementing the adjustment instructions, it sends status information back to the cloud. The cloud then uses the returned information to perform model correction and risk reassessment, forming a closed-loop iteration to achieve [the desired result]. Figure 2 The cloud-edge-device closed-loop collaborative operation and optimized control and scheduling system is shown.

[0050] In step S2 of this embodiment, a hierarchical, rolling-updating multi-time-scale coordination model is established to achieve continuous scheduling optimization across multiple time scales. This includes the following steps: S21: Establish a day-ahead scheduling model.

[0051] definition This is the set of days prior to the scheduled time. Indicates the day-ahead scheduling period; definition For a set of dispatchable power sources, Indicates the first One power source; definition A collection of energy storage devices. Indicates the first One energy storage device.

[0052] definition For power supply During the period Those who have made contributions and efforts, and , These are its lower and upper limits of output, respectively; definition , Time periods The purchased power and the sold power, and , These are the upper limits for electricity purchase and electricity sales, respectively.

[0053] For any energy storage device , definition , They are respectively in the time period The charging power and discharging power, and , These are the upper limits for charging power and discharging power, respectively. definition For stored energy (or SOC equivalent), and , These are the lower and upper limits of energy storage capacity, respectively. definition , These are charging efficiency and discharging efficiency, respectively. definition Let the time step be ; define the initial conditions as . .

[0054] definition For time period Predicted load power, For time period Network power loss (or equivalent).

[0055] First, a multi-source optimization model including power generation units, energy storage devices, and purchased power is established. Meteorological data, load curves, and electricity price information are collected during the forecast period. A multi-scenario stochastic modeling method is used to construct a set of renewable energy output and load output scenarios. Then, with the goal of minimizing system operating costs, a day-ahead optimization model is established, comprehensively considering electricity prices, purchased and sold power, and network losses. The objective function can be expressed as: (1) in, For power supply The unit power generation cost coefficient, and Time periods Electricity purchase and sales price coefficients, This is the network loss reduction factor.

[0056] And satisfy the following constraints: (1) Power balance constraint: (2) (2) Output boundary constraints of power generation unit: for any ,satisfy (3) (3) Boundary constraints for purchased and sold power: satisfy (4) (4) Energy storage charging and discharging power boundary constraints: for any ,satisfy (5) (5) Dynamic constraints on energy storage: for any ,satisfy (6) (6) Energy storage boundary constraints: for any ,satisfy (7) And the initial conditions are met. When it is necessary to ensure the energy level at the end of the energy storage system, further terminal energy constraints can be set. .

[0057] Furthermore, when energy storage does not allow simultaneous charging and discharging, a mutual exclusion constraint between charging and discharging can be introduced: for any ,satisfy (8) S22: Establish an intraday rolling optimization model.

[0058] definition Define the rolling optimization objective function for the intraday phase; For the first Power correction amount for a given time period; definition , Representing transferable load and reducible load respectively; definition , These are the power adjustment and demand response cost coefficients, respectively.

[0059] Based on the day-ahead results and short-term forecast revisions, a rolling time window mechanism is used to dynamically adjust the power generation plan; Within each rolling window, energy storage capacity, transferable load, and boundary power flow are re-optimized to minimize forecast bias. The optimization objective is to adjust the weighted sum of costs and operational deviations to achieve local compensation for new energy output deviations and load fluctuations. The objective function can be expressed as: (9) S23: Establish a real-time scheduling model.

[0060] definition These are system state variables (such as bus voltage, branch power, etc.). definition The control variables include generator output, energy storage power, and load response; A and B are defined as the state transition matrix and control influence matrix, respectively, reflecting the dynamic characteristics of the system. definition The desired reference state vector; definition To control the increment, used to constrain the smoothness of the action; definition This is the state deviation weight matrix, which reflects the penalty intensity for state errors; definition The control cost weight matrix reflects the trade-off between control energy or economic costs. definition To predict the time domain length, i.e., the model in the future The time span for in-step scrolling optimization; objective function Characterizing the comprehensive performance index in the prediction time domain, by minimizing Achieve dynamic optimal control of the system.

[0061] Minimize the state deviation and control cost within the control prediction time domain H: (10) Building a rolling prediction model: (11) When the new moment When the optimal control value arrives, the scrolling optimization window slides forward one step and recalculates the optimal control value. This enables a real-time closed-loop control process of "prediction-optimization-execution-feedback".

[0062] S24: Establish multi-timescale coupling constraints.

[0063] definition This is the rolling correction amount; definition , , These represent the boundary exchange power for the day-ahead, intraday, and real-time periods, respectively. The power plan output by the daytime layer is used as the initial condition for the intraday layer, and the correction result of the intraday layer is used as the control reference for the real-time layer. The three layers exchange information through boundary power consistency constraints: (12) S25: Establish a time coordination mechanism.

[0064] Set up time synchronization and data inheritance mechanisms to ensure the continuity of scheduling at the day-ahead, intraday, and real-time levels; It adopts a closed-loop structure of "planning-correction-execution", and has the ability to adaptively correct the day-ahead plan in both intraday and real-time stages.

[0065] S26: Achieve multi-scale joint optimization scheduling.

[0066] The cloud is responsible for the overall planning for the day, edge nodes perform intraday rolling adjustments, and the terminal layer performs real-time control and response, forming a process like... Figure 3 The collaborative optimization system shown is a cross-timescale system.

[0067] In step S3 of this embodiment, to achieve quantifiable and controllable operational risks, a systematic model of risk indicators is performed, taking into account equipment availability and uncertainty. This includes the following steps: S31: Establish a risk indicator system.

[0068] definition This is an indicator of voltage over-limit risk. definition This serves as an indicator of line overload risk. definition As a power deviation risk indicator; definition As an indicator of energy storage risk; definition This is a comprehensive system risk index; definition For the first Branch power; definition This is the upper limit of line power; definition Number of branch roads; definition This is the upper limit of line power; definition For the first Node voltage; definition Reference voltage; definition For permissible voltage deviation limits; definition The number of nodes participating in the evaluation; definition For the first Measure / calculate the power of each line (or branch); definition Its thermal stability or operating limits; definition and They are time points The predicted value and the actual power; definition For normalized references (such as rated power or day-ahead plans); definition To avoid extremely small positive numbers with a denominator of zero; definition Number of time points; definition The mean; definition For the first Each energy storage unit at time The state of charge; definition , Set its upper and lower limits for operation; definition This refers to the number of energy storage units; Define the system operation risk indicator set ; The calculation methods for each indicator are as follows: (13) (14) (15) (16) (17) (18) S32: Establish an equipment availability model.

[0069] definition Equipment failure rate; definition For equipment repair rate; definition For the first Such devices at all times Availability.

[0070] Equipment availability is calculated using historical data and real-time monitoring information. Considering failure rate and repair rate, availability satisfies the following relationship: when the failure rate and repair rate are approximately constant within the statistical period, steady-state availability can be taken. (19) Availability when considering time-varying characteristics A sliding window approach can be used to estimate the status based on online status and historical fault maintenance records.

[0071] Furthermore, the equipment availability parameter, as a constraint upper limit correction coefficient, is applied simultaneously to the day-ahead scheduling model, the intraday rolling optimization model, and the real-time scheduling model in step S2. This means that the upper limit of energy storage power, the upper limit of adjustable load response, and the feasible domain of real-time control variables in step S2 are reduced according to availability, thereby achieving dynamic correction of the equipment output upper limit. Specifically: For energy storage devices Its charging and discharging power upper limit is revised to (20) For adjustable load Its transferable and reduceable capacity limits have been revised to (twenty one) Furthermore, for the control variables in the real-time scheduling model (model predictive control) (Including energy storage capacity, load response, etc.), its feasible region can be synchronously modified to (twenty two) in This is the availability vector corresponding to each control variable. This indicates element-wise multiplication.

[0072] With the above correction, the "equipment output boundary" of energy storage and adjustable load in step S2 is transformed from a fixed upper limit form to a dynamic upper limit form that varies with availability, so that the equipment availability parameters are explicitly reflected in the day-ahead, intraday, and real-time scheduling models.

[0073] S33: Perform uncertainty modeling.

[0074] Random samples are generated using the Monte Carlo simulation method to perform probabilistic modeling of wind power, photovoltaic power, and load. K-means or K-medoids algorithms are used for scene clustering, and representative scenes are selected for computational optimization to reduce scene size.

[0075] S34: Calculate the comprehensive risk index.

[0076] Risk values ​​for the integrated weighted calculation system: (twenty three) in The risk weighting factor can be obtained through sensitivity analysis, regression analysis of historical operating data, or calibration based on expert experience. To maintain consistency in the project, it is recommended to use a fixed value within the same assessment period.

[0077] S35: Establish a risk-control mapping relationship.

[0078] By backtracking historical data, a mapping rule between risk levels and scheduling actions is established, providing input for the subsequent traffic light mechanism.

[0079] In step S4 of this embodiment, a dynamic risk control mechanism for traffic lights is established to achieve real-time identification and rapid control response of system risks. For example... Figure 4 As shown, the traffic light mechanism enables dynamic perception and rapid response to system operational risks, including the following steps: S41: Set traffic light thresholds.

[0080] definition For a moment The system operation risk value; definition , , These are safety, early warning, and danger thresholds, respectively. Will run risk indicators Divided into three level ranges: (twenty four) Where the threshold , , Based on the system's operating characteristics and historical risk distribution, it is determined through experience and statistical analysis.

[0081] S42: Establish a risk assessment mechanism.

[0082] Edge nodes periodically calculate the current system risk value using formula (13) and determine the operating status in real time based on the threshold range; When the system's operating status approaches the risk boundary, an early warning will be automatically triggered.

[0083] S43: Define the action set.

[0084] Green light: The system is operating normally; maintain the original scheduling plan. Yellow light: Triggers rolling re-optimization, adjusting energy storage output and load distribution; Red light: Triggers emergency control, executing load reduction, energy storage discharge, and tie line current limiting.

[0085] S44: Build a dynamic response process.

[0086] Risk assessment → Action selection → Strategy execution → Status update → Risk reassessment; This process runs cyclically across the cloud, edge, and terminal layers, enabling dynamic risk tracking and rapid closed-loop response.

[0087] S45: Recording and backtracking mechanism.

[0088] The system automatically records each risk state transition and control execution result for subsequent optimization parameter tuning and threshold self-learning.

[0089] In step S5 of this embodiment, in order to ensure the consistency and coordination of the three-layer scheduling of primary, secondary, and micro-level systems, a target-cascaded distributed optimization method is adopted to achieve collaborative solution.

[0090] Furthermore, in the collaborative solution process of step S5, each layer still uses the optimization model established in step S2 as the basis at its own time scale. That is, the objective function of each layer remains in the form of the objective function corresponding to the day-ahead, intraday and real-time scheduling models in step S2, while satisfying the local constraint set such as power balance, equipment output boundary, energy storage energy constraint and multi-time scale coupling constraint in step S2. The layers are coupled and coordinated only through boundary exchange power consistency constraints, and coordination terms are introduced into the objective through augmented Lagrangian form to achieve distributed collaborative solution.

[0091] At any time scale For example, the first The local optimization problem of a layer can be solved by satisfying the constraints of that layer. , Under the given conditions, solve (25) in Let be the original objective function corresponding to the time scale in step S2. , This is the set of equality and inequality constraints corresponding to step S2. Exchange the power variables at the boundary of this layer. These are reference consistency variables from adjacent layers or the main coordination layer. Step S52 involves exchanging boundary power variables and multiplier information between layers and iteratively updating them according to the update rules until the convergence criterion described in step S53 is met. This yields a joint scheduling scheme that simultaneously satisfies the constraints of step S2 and achieves coordination and consistency among the objectives of the main network, distribution network, and microgrid.

[0092] like Figure 5 As shown, the target-cascaded distributed optimization method realizes multi-level distributed cooperative optimization scheduling of master, dispatcher, and micro-level systems, including the following steps: S51: Establish a target-cascaded distributed optimization framework.

[0093] definition For the first Layer optimization objective function; definition For Lagrange multipliers; definition As a penalty factor; definition This is the constraint matrix; definition For constraint vectors; Construct a three-tier optimization model consisting of the main network (TN), distribution network (DN), and microgrid (MG). The objective of the main network is to minimize system cost, the objective of the distribution network is to balance risk and network loss, and the objective of the microgrid is to optimize user-side economy and comfort. The objective functions for the three tiers can be constructed as follows (and can be expanded or tailored according to system scale and resource type): The main grid layer (TN) aims to minimize the total system operating cost. Its objective function may include the unit generation cost, the cost of purchasing and selling electricity, and the cost of network losses. For example, it can be expressed as: (26) in The unit power generation cost, and Time periods Electricity purchase and sales price coefficients, This is the network loss reduction factor.

[0094] The distribution network (DN) aims to achieve a comprehensive trade-off between risk and operational efficiency. Its objective function may include comprehensive risk indicators, distribution network losses, and voltage deviations, and can be exemplarily expressed as: (27) in The comprehensive risk index mentioned in step S3, i.e., the time period The system operation risk value, For distribution network losses, For node voltage, This is the voltage reference value. , , These are the weighting coefficients.

[0095] The microgrid layer (MG) aims to achieve a comprehensive balance between user-side economy and comfort. Its objective function may include terms such as electricity purchase and sale costs, energy storage usage costs, load response costs, and comfort deviation penalties, which can be exemplarily expressed as: (28) in , Microgrid time periods Purchased power and sold power, For energy storage power, In order to reduce the load, This represents the offset of the transferable load. and These are the actual and preferred load curves, respectively. This is a comfort weighting coefficient; and This is the energy storage usage cost coefficient. In order to reduce the load penalty factor, This is the penalty factor for transferable load offset. This represents the absolute value operator.

[0096] The target cascading is achieved using the augmented Lagrangian function form: Let the inter-layer coupling consistency constraint be written as... (in (where the vector consists of the power variables exchanged at each boundary level) then the augmented Lagrangian objective function can be expressed as: (29) The solution is obtained by satisfying the local constraints at each level (including power balance, equipment output boundary, energy storage energy constraint, and multi-time-scale coupling constraint at the corresponding time scale in step S2), thereby achieving distributed coordination of the three-layer optimization objectives of the main grid, distribution grid, and microgrid under boundary consistency constraints. The local constraints include at least the power balance, equipment output boundary, energy storage energy constraint, and multi-time-scale coupling constraint at the corresponding time scale in step S2, and the variables and parameters in the above objective function are consistent with the scheduling models at each time scale established in step S2 to ensure the traceability and consistency of the meaning of variables and constraints during the collaborative solution process.

[0097] S52: Distributed solution and information exchange.

[0098] Each layer solves its own optimization problem in parallel, and multi-layer coordinated updates are achieved by exchanging boundary power variables and Lagrange multipliers. The update rules are as follows: (30) (31) S53: Define convergence and coordination criteria.

[0099] When the boundary power difference satisfies When the system reaches coordinated convergence, a joint scheduling scheme is output.

[0100] S54: Execution result feedback and closed-loop update.

[0101] The cloud aggregates the optimal solutions from each layer, forms a global scheduling command, and sends it to the edge and terminal execution layers. It also performs rolling optimization updates based on the real-time status feedback. After the terminal layer executes the control, it sends back the real-time status, and the cloud updates the parameters accordingly, thus realizing a rolling optimization closed loop.

[0102] The updated parameters include at least the following: (1) Update of state and initial conditions: including system state variables (such as bus voltage, branch power, etc.), boundary exchange power measurement, and initial value of energy storage energy / charge state, which are used as initial conditions for the next rolling cycle or the next prediction time domain. (2) Forecast and external input update: including load forecast, new energy output forecast and scenario probability, electricity price curve, etc., used to correct the input data for the next cycle optimization calculation; (3) Equipment and constraint parameter updates: including equipment availability (obtained by online statistics or monitoring estimates), and dynamic operating limits such as the upper limit of energy storage and adjustable load output after discounting availability, and line thermal limits and transformer capacity when necessary; (4) Collaborative Iterative Parameter Update: When using the target cascade / augmented Lagrange distributed collaborative solution, the Lagrange multipliers and penalty factors can also be updated periodically as hot-start information to improve the efficiency and stability of rolling solution.

[0103] By updating the above parameters, a rolling closed-loop optimized scheduling process of "planning-correction-execution-feedback" is achieved.

[0104] S55: Algorithm parallelism and fault tolerance mechanisms.

[0105] A multi-threaded, parallel, and partitioned update strategy is employed to perform distributed solving, ensuring real-time performance. If a node in a certain layer fails to compute, other layers can retain the previous valid solution to maintain the system's operational capability.

[0106] This embodiment, through step S6, coordinates the execution of steps S1 to S5, enabling multi-level collaborative optimization and risk-aware scheduling among the main grid, distribution network, and microgrid, significantly improving the renewable energy absorption rate, system flexibility, and safety margin. Figure 6 As shown, it includes the following steps: The optimization objective functions of the main network layer, distribution network layer and microgrid layer, as shown in formulas (26) to (28), are constructed for the cloud control layer, edge coordination layer and terminal execution layer respectively. The objective cascade is achieved by using the augmented Lagrangian function form according to formula (29) to obtain the three-layer optimization model. The boundary power consistency constraint is set according to formula (12) to couple the day-ahead, day-in, and real-time scheduling models. The local constraints of each layer are set according to formulas (2) to (8). The cloud control layer receives real-time operation data from the edge coordination layer and the terminal execution layer and calculates the day-ahead scheduling plan using the day-ahead scheduling model of formula (1). Then, it periodically issues scheduling instructions for the day-ahead scheduling plan according to the first time scale. The edge coordination layer receives the scheduling instructions issued by the cloud control layer to coordinate and control the distribution network and feeds back the results to the cloud control layer. It also uses the scheduling instructions as initial conditions, uses the intraday rolling optimization model of formula (9) to calculate the correction amount according to the second time scale which is less than the first time scale and issues the corresponding correction instructions. At the same time, it periodically calculates the system operation risk value according to formula (13) to formula (18) and issues the corresponding action signal according to the comparison result between the system operation risk value and the preset threshold. Specifically, when the system operation risk value is less than the lower threshold, a green light signal is issued; when the system operation risk value is greater than the lower threshold and less than the upper threshold, a yellow light signal is issued; and when the system operation risk value is greater than the upper threshold, a red light signal is issued. The terminal execution layer receives the scheduling instructions issued by the cloud control layer and the correction instructions issued by the edge coordination layer. It calculates the correction result based on the scheduling instructions and the correction instructions, uses the correction result as the control reference, and uses the real-time scheduling model of formula (10) to calculate the optimal control quantity in a rolling manner according to the third time scale which is smaller than the second time scale. It uses the optimal control quantity to coordinate and control the microgrid and feeds back the result to the edge coordination layer. If the terminal execution layer receives an action signal from the edge coordination layer, it executes the corresponding strategy to re-coordinate and control the microgrid and feeds back the result to the edge coordination layer, including: If the action signal is a green light signal, the original scheduling plan is maintained, the optimal control quantity calculated by the real-time scheduling model is obtained according to the original rolling cycle, and the optimal control quantity is used to coordinate the control of the microgrid. If the action signal is a yellow light signal, rolling re-optimization is triggered. After adjusting the energy storage output and load distribution, the moment the action signal is received is taken as the start time of the new rolling cycle. The real-time scheduling model is used to calculate the optimal control quantity according to the new rolling cycle and use the optimal control quantity to coordinate the control of the microgrid. If the action signal is a red light signal, emergency control is triggered to perform emergency control on the microgrid. The emergency control includes load reduction, energy storage discharge, and tie line current limiting.

[0107] When calculating the day-ahead scheduling plan using the day-ahead scheduling model in the first time scale, the correction amount using the intraday rolling optimization model in the second time scale, and the optimal control amount using the real-time scheduling model in the third time scale, the cloud control layer, the edge coordination layer, and the terminal execution layer all solve their respective target optimization problems in parallel according to the optimization model. Cross-level coordination is achieved by exchanging boundary power variables and Lagrange multipliers. When the boundary power difference satisfies the convergence condition, a joint scheduling scheme of the optimal solutions of each layer is obtained. The cloud control layer summarizes the optimal solutions of each layer after correction by the augmented Lagrange coordination term in formula (29) to form a globally consistent scheduling instruction.

[0108] Compared with traditional centralized scheduling methods, the above steps have significant advantages in terms of communication load, risk response speed, and global convergence.

[0109] Furthermore, this embodiment proposes a multi-time-scale coordinated optimization scheduling system for a main-distribution-micro integrated power grid based on a cloud-edge collaborative architecture. The system includes a cloud control layer, an edge coordination layer, and a terminal execution layer. The three layers achieve information interaction and collaborative optimization through a unified data interface and communication protocol. The system is configured to execute the steps of the multi-time-scale coordinated optimization scheduling method for a main-distribution-micro integrated power grid based on a cloud-edge collaborative architecture of this embodiment.

[0110] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0111] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A multi-time-scale coordinated optimization scheduling method for a master-distribution-microgrid integrated power grid based on a cloud-edge collaborative architecture, characterized in that, The integrated main grid, distribution, and microgrid network deploys a main grid control module on a cloud server or dispatch control center to establish a cloud control layer, sets regional edge nodes in the distribution network layer to construct an edge coordination layer, and installs terminal control units on the microgrid and user side to deploy a terminal execution layer. The method includes: The cloud control layer receives real-time operational data from the edge coordination layer and the terminal execution layer and uses the day-ahead scheduling model to calculate the day-ahead scheduling plan. Then, it periodically issues scheduling instructions for the day-ahead scheduling plan according to the first time scale. The edge coordination layer receives scheduling instructions from the cloud control layer to coordinate and control the distribution network and feeds back the results to the cloud control layer. It also uses the scheduling instructions as initial conditions, uses the intraday rolling optimization model to calculate the correction amount according to the second time scale which is less than the first time scale, and issues the corresponding correction instructions. At the same time, it periodically calculates the system operation risk value and issues the corresponding action signal based on the comparison result between the system operation risk value and the preset threshold. The terminal execution layer receives scheduling instructions from the cloud control layer and correction instructions from the edge coordination layer. It calculates the correction result based on the scheduling instructions and correction instructions, uses the correction result as a control reference, and uses the real-time scheduling model to calculate the optimal control quantity on a rolling basis according to a third time scale that is smaller than the second time scale. It uses the optimal control quantity to coordinate and control the microgrid and feeds back the result to the edge coordination layer. If it receives an action signal from the edge coordination layer, it executes the corresponding strategy to re-coordinate and control the microgrid and feeds back the result to the edge coordination layer.

2. The multi-time-scale coordinated optimization scheduling method for a master-distribution-micro integrated power grid based on a cloud-edge collaborative architecture as described in claim 1, characterized in that, When issuing corresponding action signals based on the comparison results between the system operation risk value and the preset threshold, specifically, a green light signal is issued when the system operation risk value is less than the lower threshold, a yellow light signal is issued when the system operation risk value is greater than the lower threshold but less than the upper threshold, and a red light signal is issued when the system operation risk value is greater than the upper threshold. When implementing the corresponding strategy to re-coordinate control of the microgrid, the following is included: If the action signal is a green light signal, the optimal control quantity calculated by the real-time scheduling model is obtained according to the original rolling cycle, and the optimal control quantity is used to coordinate the control of the microgrid. If the action signal is a yellow light signal, the moment the action signal is received will be taken as the start time of the new rolling cycle. The optimal control quantity will be calculated according to the new rolling cycle using the real-time scheduling model and then used to coordinate the control of the microgrid. If the action signal is a red light signal, emergency control is performed on the microgrid. The emergency control includes load reduction, energy storage discharge, and tie line current limiting.

3. The multi-time-scale coordinated optimization scheduling method for a master-distribution-microgrid integrated power grid based on a cloud-edge collaborative architecture as described in claim 1, characterized in that, The mathematical expression for the objective function of the day-ahead scheduling model is as follows: in, The objective function for the total cost in the current day phase; For the day-ahead scheduling time set, Indicates the day-ahead scheduling period; For a set of dispatchable power sources, Indicates the first One power source; For power supply During the period Those who have made contributions , Time periods Purchased power and sold power, For time period Network power loss For power supply The unit power generation cost coefficient, and Time periods Electricity purchase and sales price coefficients, This is the network loss reduction factor.

4. The multi-time-scale coordinated optimization scheduling method for a master-distribution-micro integrated power grid based on a cloud-edge collaborative architecture as described in claim 1, characterized in that, The mathematical expression for the objective function of the intraday rolling optimization model is as follows: in, The objective function for rolling optimization during the intraday phase; For the first Power correction amount for a given time period; , These represent transferable load and load that can be reduced, respectively. , These are the power adjustment and demand response cost coefficients, respectively.

5. The multi-time-scale coordinated optimization scheduling method for a master-distribution-microgrid integrated power grid based on a cloud-edge collaborative architecture as described in claim 1, characterized in that, The mathematical expression for the objective function of the real-time scheduling model is as follows: in, A and B are the state transition matrix and the control influence matrix, respectively. For system state variables; For control variables; The desired reference state vector; To control the increment; This is the state deviation weight matrix; To control the cost weight matrix; To predict the length of the time domain.

6. The multi-time-scale coordinated optimization scheduling method for a master-distribution-microgrid integrated power grid based on a cloud-edge collaborative architecture as described in claim 1, characterized in that, The method further includes: A three-layer optimization model consisting of a main network, a distribution network, and a microgrid is constructed. The main network aims to minimize system cost, the distribution network aims to balance risk and network loss, and the microgrid aims to optimize user-side economy and comfort. The objectives are cascaded using an augmented Lagrangian function. When calculating the day-ahead scheduling plan using the day-ahead scheduling model, when calculating the correction amount using the intraday rolling optimization model at a second time scale smaller than the first time scale, and when calculating the optimal control amount using the real-time scheduling model at a third time scale smaller than the second time scale, the following steps are all included: The optimization model solves the optimization problem of its respective objectives in parallel at each layer. It achieves coordination by exchanging boundary power variables and Lagrange multipliers. When the boundary power difference satisfies the convergence condition, a joint scheduling scheme of the optimal solutions of each layer is obtained. The cloud control layer summarizes the optimal solutions of each layer to form a global scheduling instruction, which is then sent to the edge coordination layer and the terminal execution layer. After the edge coordination layer and the terminal execution layer execute control according to the global scheduling instructions, they send the real-time status back to the cloud control layer to update the model parameters.

7. The multi-time-scale coordinated optimization scheduling method for a master-distribution-micro integrated power grid based on a cloud-edge collaborative architecture as described in claim 6, is characterized in that... The day-ahead scheduling model, intraday rolling optimization model, and real-time scheduling model are coupled through information exchange via boundary power consistency constraints. The mathematical expression for the boundary power consistency constraints is as follows: in, This is the rolling correction amount. , , These represent the boundary exchange power for the day-ahead, intraday, and real-time periods, respectively.

8. The multi-time-scale coordinated optimization scheduling method for a master-distribution-micro integrated power grid based on a cloud-edge collaborative architecture as described in claim 6, is characterized in that... The mathematical expression for the objective function of the main network in the optimization model is as follows: in, For power supply During the period Those who have made contributions , Time periods The power purchased and the power sold. For time period Network power loss For power supply The unit power generation cost coefficient, and Time periods Electricity purchase and sales price coefficients, This is the network loss reduction factor; The mathematical expression for the objective function of the distribution network in the optimization model is as follows: in, For time period The system operation risk value, For distribution network losses, For node voltage, This is the voltage reference value. , , These are the weighting coefficients; The mathematical expression for the objective function of the microgrid in the optimization model is as follows: in, , Microgrid time periods The power purchased and the power sold. For energy storage power, In order to reduce the load, This represents the offset of the transferable load. and These are the actual and preferred load curves, respectively. This is a comfort weighting coefficient; and This is the energy storage usage cost coefficient. In order to reduce the load penalty factor, This is the penalty factor for transferable load offset. This represents the absolute value operator.

9. The multi-time-scale coordinated optimization scheduling method for a master-distribution-micro integrated power grid based on a cloud-edge collaborative architecture as described in claim 1, characterized in that, The day-ahead scheduling model, intraday rolling optimization model, and real-time scheduling model all incorporate equipment availability to weighted adjust the output of energy storage and adjustable loads. Specifically, the upper limit of energy storage power, the upper limit of adjustable load response, and the upper limit of the feasible domain of the real-time control variable in the real-time scheduling model are multiplied by availability, thereby achieving dynamic adjustment of the upper limit of equipment output. The mathematical expression for availability is as follows: in, For the first Equipment availability For equipment failure rate, This refers to the equipment repair rate.

10. The multi-time-scale coordinated optimization scheduling method for a master-distribution-microgrid integrated power grid based on a cloud-edge collaborative architecture as described in claim 1, characterized in that, The mathematical expression for the system's operational risk value is as follows: in, As a risk weighting factor, This is a comprehensive system risk index; This is an indicator of voltage over-limit risk. This serves as an indicator of line overload risk. As a power deviation risk indicator; The mathematical expressions for each energy storage risk indicator are as follows: in, For the first Branch power; This is the upper limit of line power; Number of branch roads; This is the upper limit of line power; For the first Node voltage; Reference voltage; For permissible voltage deviation limits; The number of nodes participating in the evaluation; For the first Measured power of the line; For the first Thermal stability or operating limits of each line; and They are time points The predicted value and the actual power; As a normalized benchmark; To avoid extremely small positive numbers with a denominator of zero; Number of time points; The mean; For the first Each energy storage unit at time The state of charge; For the first The upper and lower limits of operation for each energy storage unit; This represents the number of energy storage units.

Images