Micro-energy network collaborative optimization method and device based on event triggering and packet loss compensation

Abstract

This invention discloses a method and apparatus for collaborative optimization of microgrids based on event triggering and packet loss compensation. The method includes: constructing a collaborative operation model of the microgrid system using the alternating direction multiplier method; constructing an augmented Lagrangian function based on the collaborative operation model; iteratively solving the augmented Lagrangian function to obtain the scheduling scheme of each microgrid; each iteration includes: each microgrid solving its local coupling variables and comparing them with the coupling variables solved in the previous round; if the deviation exceeds a preset threshold, noise is added to the coupling variables before sending them to the scheduling center; otherwise, multiplexing symbols are sent; the scheduling center receives the data sent by each microgrid; if data packets are lost, the coupling variables are linearly extrapolated and predicted based on historical momentum; if the data is a multiplexing symbol, the coupling variables from the previous round are reused; the consistency variables and dual variables are updated based on the coupling variables of each microgrid using the Nesterov accelerated gradient method, and fed back to the microgrids that have not lost packets.

Description

Technical Field

[0001] This invention relates to the field of microgrid optimization, and in particular to a method and apparatus for collaborative optimization of microgrids based on event triggering and packet loss compensation. Background Technology

[0002] Microgrids integrate multiple energy forms such as electricity, gas, and heat. Through coordinated operation, they can effectively reduce the peak-valley difference in the system and improve the renewable energy absorption rate. However, microgrids are characterized by dispersed source and load entities and weak communication infrastructure. Overcoming interference from weak communication environments and ensuring privacy and security while achieving efficient coordinated operation are the core technological bottlenecks to ensure that their advantages are fully realized.

[0003] Existing optimization methods for microgrid collaborative operation are mainly divided into centralized and distributed approaches. While centralized optimization methods can theoretically yield the global optimum, they face drawbacks in rural scenarios, including computational complexity, stringent communication requirements, and a high risk of privacy data leakage for each participant. Distributed optimization, particularly strategies based on the alternating direction multiplier method, offers a feasible path to address these shortcomings by decomposing the large problem into subproblems that each participant can solve independently. Nevertheless, existing distributed collaborative algorithms heavily rely on ideal synchronous communication networks and fail to consider the "weak communication" characteristics of remote areas, such as limited bandwidth and susceptibility to packet loss. Data loss can easily cause the algorithm to oscillate. Furthermore, existing privacy protection methods typically employ differential privacy protection by adding fixed noise, which introduces unavoidable steady-state errors, severely sacrificing the system's economic optimality.

[0004] Therefore, there is an urgent need for a new technical solution to address the technical problem of how to improve the efficiency and stability of microgrid collaborative operation in weak communication environments. Summary of the Invention

[0005] This invention provides a microgrid collaborative optimization method and apparatus based on event triggering and packet loss compensation, which addresses the technical problem of how to improve the efficiency and stability of microgrid collaborative operation in weak communication environments.

[0006] To achieve the above objectives, this invention provides a microgrid collaborative optimization method based on event triggering and packet loss compensation, comprising: A cooperative operation model for the microgrid system is constructed using the alternating direction multiplier method; an augmented Lagrangian function is constructed based on the cooperative operation model; the augmented Lagrangian function is iteratively solved to obtain the scheduling scheme for each microgrid; each iteration includes: Each microgrid solves the local coupling variables and compares them with the coupling variables solved in the previous round. If the deviation exceeds the preset threshold, noise is added to the coupling variables and they are sent to the scheduling center; otherwise, multiplexing symbols are sent. The dispatch center receives data sent by each microgrid. If data packets are lost, it performs linear extrapolation prediction of the coupling variables based on historical momentum. If the data is a reused variable, it reuses the coupling variables from the previous round. Based on the coupling variables of each microgrid and combined with the Nesterov accelerated gradient method, it updates the consistency variables and dual variables and feeds them back to the microgrids that have not lost packets.

[0007] Preferably, the cooperative operation model of a microgrid system constructed using the alternating direction multiplier method includes: Construct a microgrid model and operational constraints; based on the microgrid model and operational constraints, construct a local optimization model for each microgrid with the goal of minimizing operating costs; decouple the collaborative operation model of the microgrid system into a global objective function and a local optimization model, use the power exchanged between interconnecting lines of microgrids as a consistency variable, and employ the alternating direction multiplier method to construct a collaborative operation model in the form of a distributed optimization problem.

[0008] Preferably, constructing the augmented Lagrangian function based on the cooperative operation model includes: The cooperative running model in the form of a distributed optimization problem is represented as: ; in, For microgrid decision variables; For the first Local decision variables of a microgrid; For consistency variables; The number of microgrids; This is a local optimization model for microgrids; and A coefficient matrix describing the energy coupling relationship between microgrids; For the first A compact convex set consisting of all physical constraints within a microgrid; For the coupling variables of the microgrid; Constructing an augmented Lagrangian function based on a collaborative operation model in the form of a distributed optimization problem. : ; in, For Micro Energy Network The dual variable; The penalty parameter is T; T represents the transpose of the matrix. It represents the square of the 2-norm.

[0009] Preferably, the coupling variables for solving the local problem in each microgrid include: Each microgrid solves its own cost minimization problem in parallel: utilizing the consistency variables from the previous iteration. and dual variables Update local decision variables With coupling variables , The update is represented as: .

[0010] Preferably, the comparison with the coupling variables from the previous solution includes: definition For the first The first microgrid in the The coupling variables calculated in the next iteration The coupling variable is the one that was successfully transmitted and confirmed by the dispatch center in the last time the microgrid was used; and Compare and determine the trigger flag. : ; ; in, The preset threshold; This is the initial threshold constant; As the attenuation factor, For the first The decay factor for the next iteration; Denotes the 2-norm; when At this time, it indicates that the microgrid is sending the latest data. ;when When, it indicates that the microgrid is sending a multiplexer.

[0011] Preferably, adding noise to the coupling variables includes: On MicroNet Upload coupling variables Forward, towards Injection scale dynamically decays with the number of iterations of Laplace noise Generate encrypted variables , will variables It is sent to the scheduling center as a new coupling variable. Represented as: ; ; in, express Obeying position parameter 0 and scale parameter 0 The Laplace distribution of; This is the initial noise base. This is the decay index.

[0012] Preferably, the dispatch center receives data sent by each microgrid. If data packets are lost, the coupling variables are predicted linearly based on historical momentum. If the data is a reused variable, the coupling variables from the previous round are reused, including: The dispatch center determines the link status based on the data sent by each microgrid. And based on the link status For coupling variables Perform differentiated updates: If the data arrives Then, regarding historical momentum Perform a correction update to adjust the prediction direction: ; in, For the first The historical momentum of the next iteration represents the changing trend and numerical inertia of the coupled variables in the previous continuous iteration process; For the first -1 iterations of historical momentum; The momentum retention coefficient; For momentum update coefficients; If data packets are lost Then, linear extrapolation prediction is performed on the coupled variables based on historical momentum: ; If the data is a reusable character Then the coupling variables from the previous round are reused, and the momentum is reset: .

[0013] Preferably, updating the consistency variables and dual variables based on the coupling variables of each microgrid using the Nesterov accelerated gradient method, and feeding this update back to the microgrids that have not experienced packet loss, includes: The dispatch center solves the global coordination problem based on the coupling variables of each microgrid to update the consistency variables. : ; Define the overall objective function of the cooperative operation model as follows: After each global update, if at the 1st Round iteration and the first The trend of the overall objective function during -1 iterations is as follows: ; in, For the microgrid decision variables obtained from the k-th iteration, These are the decision variables for the microgrid obtained from the (k-1)th iteration. If the overall objective function value is not monotonically decreasing and the current direction of inertia deviates from the direction of gradient descent, then a restart operation is performed to clear the accumulated momentum. ; in, The dynamic inertia coefficient; For the first Accelerated time steps in the next iteration; The Nesterov accelerated gradient method is used to update the consistency variables. : ; in, To accelerate the time step; These are the consistency variables from the previous iteration; For accelerated consistency variables; The control center updates the dual variables of each microgrid uniformly based on the coupling variables of each microgrid and the degree of boundary coupling constraint exceedance. : ; The dual variable is updated using the Nesterov accelerated gradient method. : ; The consistency variables and dual variables are fed back to the micro-energy network where no packet loss has occurred, i.e., the link state. Microgrids consisting of 1s and 0s.

[0014] Preferably, when iteratively solving for the augmented Lagrange function, a convergence criterion is also included; When the original residual and dual residual The iteration converges when all values ​​are below the set convergence threshold. ; in, This is the original convergence threshold; This is the dual convergence threshold.

[0015] The present invention also provides a microgrid collaborative optimization device based on event triggering and packet loss compensation, which is used to implement the method of the present invention.

[0016] The present invention has the following beneficial effects: This invention presents a microgrid collaborative optimization method based on event triggering and packet loss compensation. It introduces an event-triggered communication mechanism based on a dynamic threshold, triggering transmission only when data deviation exceeds a threshold. This effectively avoids frequent communication caused by small numerical fluctuations, reducing bandwidth pressure in weak communication environments. Simultaneously, the attenuation threshold ensures convergence accuracy in the later stages of iteration. A differential privacy protection mechanism based on dynamic noise attenuation overcomes the drawbacks of traditional methods that introduce steady-state errors by adding fixed noise. Larger noise is injected in the early stages of iteration to mask sensitive information and prevent reverse reconstruction, while the noise automatically attenuates to near zero in the later stages, achieving both privacy protection and precise scheduling. Finally, an anti-packet loss compensation mechanism based on hybrid momentum extrapolation is introduced to address data arrival and packet loss issues. The invention employs differentiated updates for momentum correction, linear extrapolation prediction, and zero-order preservation in three states (not triggered), utilizing historical momentum to fill data gaps and correct prediction directions. This effectively overcomes the algorithm oscillation and divergence problem caused by packet loss in weak communication, improving the system's robustness in packet loss environments. Furthermore, the method integrates a Nesterov acceleration strategy with adaptive restart, using historical gradient information to accelerate convergence and promptly clearing accumulated momentum when the objective function is not monotonically decreasing. Only the updated consistency variables and dual variables are fed back to the microgrid without packet loss, avoiding algorithm divergence caused by asymmetric information updates. This achieves efficient, stable, and privacy-preserving collaborative optimization scheduling of the microgrid system in weak communication environments.

[0017] The microgrid collaborative optimization device based on event triggering and packet loss compensation of the present invention, used in the method of the present invention, has the same beneficial effects as the method of the present invention.

[0018] In addition to the objectives, features, and advantages described above, the present invention has other objectives, features, and advantages. The invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0019] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the iterative solution process of a preferred embodiment of the present invention.

[0020] Figure 2 This is a diagram showing the power output of each device in a preferred embodiment of the microgrid A of the present invention.

[0021] Figure 3 This is a diagram of the interactive power of inter-microgrid interconnection lines according to a preferred embodiment of the present invention; (a) is the interactive power of the power interconnection line, (b) is the interactive power of the gas interconnection line, and (c) is the interactive power of the heat interconnection line.

[0022] Figure 4These are comparison diagrams of algorithm residual convergence trajectories under different communication and privacy scenarios in the preferred embodiment of the present invention; (a) is a comparison of the original residual convergence trajectory, and (b) is a comparison of the dual residual convergence trajectory.

[0023] Figure 5 This is a schematic diagram of the communication state of coupled variables according to a preferred embodiment of the present invention.

[0024] Figure 6 These are comparison diagrams of the interactive power iteration trajectories of tie lines under different scenarios in the preferred embodiment of the present invention; (a) is the power iteration trajectory of the AC power tie line of the microgrid, and (b) is the power iteration trajectory of the AC thermal tie line of the microgrid. Detailed Implementation

[0025] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0026] In a preferred embodiment of the present invention, a microgrid collaborative optimization method based on event triggering and packet loss compensation is provided, comprising: A collaborative operation model for the microgrid system is constructed using the alternating direction multiplier method. An augmented Lagrangian function is then constructed based on the collaborative operation model. The augmented Lagrangian function is iteratively solved to obtain the scheduling scheme for each microgrid.

[0027] In a preferred embodiment of the present invention, the cooperative operation model of the microgrid system constructed using the alternating direction multiplier method includes: Construct a microgrid model and operational constraints; based on the microgrid model and operational constraints, construct a local optimization model for each microgrid with the goal of minimizing operating costs; decouple the collaborative operation model of the microgrid system into a global objective function and a local optimization model, use the power exchanged between interconnecting lines of microgrids as a consistency variable, and employ the alternating direction multiplier method to construct a collaborative operation model in the form of a distributed optimization problem.

[0028] In a preferred embodiment of the present invention, the construction of the microgrid model and operational constraints include: First, parameters and data on energy supply, energy equipment, and energy demand in the microgrid are collected. Energy supply includes distributed photovoltaic, wind power, hydropower, distribution network, natural gas network, and heating network. Energy equipment includes two types of energy conversion equipment: natural gas cogeneration units and electric heating equipment, as well as three types of energy storage equipment: energy storage equipment, thermal storage devices, and gas storage devices. Energy demand includes electrical load, thermal load, and natural gas load.

[0029] Based on the collected parameters and data, a microgrid model and operational constraints are constructed, including: Each microgrid integrates renewable energy and thermal-electric multi-energy coupling equipment; the microgrids themselves and between microgrids achieve multi-energy flow mutual support through distribution networks, heating networks and gas transmission networks.

[0030] (1) Natural Gas Cogeneration Unit (CHP) Model: This model consumes natural gas while simultaneously generating electricity and heat. ; ; in, , , They are respectively At any given moment, CHP's electrical power, thermal power, and natural gas consumption power; , These are electrical efficiency and thermal efficiency, respectively.

[0031] The equipment constraints for natural gas combined heat and power units include: ; in, for A 0-1 variable indicating whether CHP is on or off at any given time. , These are the upper and lower limits of power generation. , These are the upper and lower limits of thermal power. This represents the CHP ramp rate.

[0032] (2) Electric heating model: ; in, , They are respectively The heating power and electrical power consumption of the electric heating equipment at all times. The thermal efficiency of the electric heating device.

[0033] The constraints on electric heating equipment include: ; in, for The 0-1 variable indicating whether the electric heating device is in the on or off state at any given time. , These are the upper and lower limits of the electric heating power.

[0034] (3) Constraints on photovoltaic, wind power, and hydropower equipment include: ; in, , , for The output power of photovoltaic, wind power, and hydropower equipment at all times. , , This refers to the upper limit of the power generation capacity of photovoltaic, wind power, and hydropower equipment.

[0035] (4) Energy storage device model: ; in, for The amount of electricity stored in the energy storage device at any given time. , for The charging and discharging power at any given time , These are the charging and discharging efficiencies, respectively.

[0036] The constraints of energy storage devices are: ; in, This represents the maximum power for energy storage charging and discharging. , Represent The charging and discharging status indicators for energy storage at all times. , These are the maximum and minimum values ​​of the energy storage rate. For energy storage capacity.

[0037] (5) Thermal storage device: The operating characteristics of thermal energy storage devices are basically the same as those of energy storage devices; they smooth out fluctuations by charging or releasing thermal energy. ; in, for The heat storage device stores heat at all times. and They are respectively The power of heat charging and discharging at any given time. and These represent the efficiency of heat charging and heat dissipation, respectively.

[0038] The equipment constraints of thermal storage devices are similar to those of energy storage devices, so they can be constructed by referring to the constraints of energy storage devices and existing related technologies.

[0039] (6) Gas storage device: The model characteristics of gas storage devices are roughly similar to those of energy storage and thermal storage devices. The model is as follows: ; in, for The amount of gas stored in the gas storage device at any given time. and These are the charging and discharging capacities of the gas storage device, respectively. and These are the charging and discharging power coefficients of the gas storage device, respectively.

[0040] The constraints of gas storage devices are similar to those of energy storage devices, so they can be constructed by referring to the constraints of energy storage devices and existing related technologies.

[0041] After constructing the device model and constraints, the multi-energy transmission network within the system is modeled: (1) Power network model: A linearized DistFlow model is used to describe the power flow distribution within and between microgrids. This model is applied to any branch of the interconnections within and between microgrids. Satisfying both current flow constraints and security constraints: ; In the formula: , They are respectively Flow through branch roads during the period The active and reactive power; , They are the downstream branches respectively The active and reactive power; , They are nodes The active load and active power injection; For nodes The set of child nodes The child node number; , They are nodes The reactive load and reactive power injection; For nodes Voltage amplitude; , For branch resistance and reactance; The reference voltage; , For the safe operating range of node voltage; This represents the line's apparent power transmission limit.

[0042] (2) Heating network model: Microgrid heating systems primarily transfer heat energy through hot water pipes. The model focuses on heat power balance and heat loss and temperature delay characteristics during transmission. For the heating network nodes... Its constraints are described as follows: ; in, , They are nodes The heat source injection power and heat load; For pipelines The transmitted heat power; For the heating network and nodes The set of connected adjacent nodes. The serial number of the adjacent heating network node; This is the specific heat capacity of water; For pipelines mass flow rate; , These are the supply water temperature and the return water temperature, respectively. , These are the inlet and outlet temperatures of the pipeline, respectively. The heat transfer coefficient; This refers to the length of the pipe. The ambient temperature.

[0043] (3) Natural gas network model: The steady-state power flow of the natural gas pipeline network is described by the Weymouth equation. For the nodes of the gas network... and pipelines It is necessary to meet the requirements of node gas volume balance and pipeline pressure constraints: ; In the formula: , They are nodes The amount of gas injected and the amount of gas consumed; For flow through the pipeline Gas flow rate; For the gas network and nodes The set of connected adjacent nodes. The sequence number of the adjacent gas network node; , For nodes , The pressure; This is the pipeline constant; , These are the upper and lower limits of the allowable pressure for the node.

[0044] In a preferred embodiment of the present invention, the local optimization model for each microgrid, with the goal of minimizing operating costs, includes: Based on the aforementioned equipment model and constraints, and the multi-energy flow transmission network model, a local optimization model is constructed for each microgrid with the objective of minimizing operating costs. This local optimization model aims to minimize the overall operating cost within the scheduling cycle T1. The operating cost consists of the following components: grid interaction costs, equipment operation and maintenance, voltage deviation penalty costs, and renewable energy curtailment penalty costs.

[0045] The local optimization model is represented as: ; ; in, For grid interaction costs, , They are respectively Time-of-use electricity pricing , For purchasing and selling electricity; For equipment operation and maintenance costs, A collection of controllable devices. For equipment serial number, For unit power operation and maintenance cost, Provide power to the equipment; Penalty cost for voltage deviation For the system node set, For node sequence number, This is the voltage deviation penalty coefficient. This represents the actual voltage amplitude at the node. This is the reference voltage. To avoid the penalties of abandoning new energy sources, A collection of renewable energy units, For unit serial number, The unit is the penalty coefficient for wind and solar power curtailment; and These represent the theoretical maximum available power output and the actual power output of renewable energy sources, respectively.

[0046] In a preferred embodiment of the present invention, constructing the augmented Lagrangian function based on the cooperative operation model includes: The cooperative running model in the form of a distributed optimization problem is represented as: ; in, For microgrid decision variables; For the first Local decision variables of a microgrid (including the output of each device, energy storage status, etc.); For consistency variables; The number of microgrids; This is a local optimization model for microgrids; and A coefficient matrix describing the energy coupling relationship between microgrids; For the first A compact convex set consisting of all physical constraints within a microgrid; For the coupling variables of the microgrid; Constructing an augmented Lagrangian function based on a collaborative operation model in the form of a distributed optimization problem. : ; in, For Micro Energy Network The dual variable; The penalty parameter is T; T represents the transpose of the matrix. It represents the square of the 2-norm.

[0047] In a preferred embodiment of the present invention, based on the augmented Lagrangian function, the large system optimization problem is decomposed into a parallel computational subproblem and a global coordination problem.

[0048] See Figure 1 In a preferred embodiment of the present invention, when iteratively solving the augmented Lagrange function, each iteration includes steps A1 to A3: A1. Each microgrid solves the local coupling variables and compares them with the coupling variables solved in the previous round. If the deviation exceeds the preset threshold, noise is added to the coupling variables and then sent to the scheduling center; otherwise, multiplexing symbols are sent.

[0049] In the preferred embodiment A1 of the present invention, the coupling variables for solving the local problem of each microgrid include: Each microgrid solves its own cost minimization problem in parallel: utilizing the consistency variables from the previous iteration. and dual variables Update local decision variables With coupling variables , The update is represented as: ; In the preferred embodiment of the present invention, A1, the comparison with the coupling variables from the previous solution includes: definition For the first The first microgrid in the The coupling variables calculated in the next iteration The coupling variable is the one that was successfully transmitted and confirmed by the dispatch center in the last time the microgrid was used; and Compare and determine the trigger flag. : ; ; in, To ensure convergence accuracy in the later stages of iteration, the preset threshold is designed to decay exponentially with the number of iterations. This is the initial threshold constant; As the attenuation factor, For the first The decay factor for the next iteration; Denotes the 2-norm; when At this time, it indicates that the microgrid is sending the latest data. ;when When, it indicates that the microgrid is sending a multiplexer.

[0050] In the preferred embodiment A1 of the present invention, adding noise to the coupling variable includes: Microgrids involve a large amount of energy consumption data related to production and daily life. Directly uploading coupled variables could lead to the leakage of sensitive information such as user load characteristics and production plans. To balance privacy and scheduling accuracy, this invention introduces differential privacy, employing a differential privacy protection mechanism based on dynamic noise attenuation. For the data characteristics of the iterative process of the alternating direction multiplier method, a Laplace mechanism is used to generate noise masking.

[0051] On MicroNet Upload coupling variables Forward, towards Injection scale dynamically decays with the number of iterations of Laplace noise Generate encrypted variables , will variables It is sent to the scheduling center as a new coupling variable. Represented as: ; ; in, express Obeying the position parameter is Scale parameters are Laplace distribution, scale parameter Designed to be power-law decay form; This is the initial noise baseline, which is related to the system capacity baseline and the initial privacy budget. This is the decay index.

[0052] In the early stages of iteration A larger gradient is used to mask sensitive features in the gradient and prevent reverse reconstruction; as iterations proceed, As the decay approaches zero, the transmission variables tend to balance, and the individual user characteristics become blurred. Noise tends to disappear, ensuring that the algorithm can accurately converge to the optimal solution.

[0053] A2. The dispatch center receives data sent by each microgrid. If data packets are lost, the coupling variables are predicted by linear extrapolation based on historical momentum. If the data is a reused variable, the coupling variables from the previous round are reused.

[0054] To overcome the problem of distributed algorithm oscillation and divergence caused by data packet loss when uploading coupled variables to the control center in a weak communication environment, this invention introduces a prediction compensation mechanism based on hybrid momentum extrapolation to achieve robust iteration against packet loss.

[0055] A2 specifically includes: The dispatch center determines the link status based on the data sent by each microgrid. And based on the link status For coupling variables Perform differentiated updates: If the data arrives Then, regarding historical momentum Perform a correction update to adjust the prediction direction: ; in, For the first The historical momentum of the next iteration characterizes the changing trend and numerical inertia of the coupled variables in the previous continuous iteration process, aiming to provide a reasonable evolution direction for data extrapolation under the condition of communication absence; For the first -1 iterations of historical momentum; The momentum retention factor is set to -0.81. The momentum update coefficient is set to 0.15. If data packets are lost Then, linear extrapolation predictions are performed on the coupled variables based on historical momentum to fill data gaps: ; If the data is a reusable character This indicates that the transmitter is in a steady state, so the coupling variables from the previous round are reused, and the momentum is reset: ; A3. The scheduling center updates the consistency variables and dual variables based on the coupling variables of each microgrid and the Nesterov accelerated gradient method, and feeds them back to the microgrids that have not lost packets.

[0056] After completing communication compensation, to improve convergence efficiency under weak communication, the control center adopts a Nesterov acceleration strategy with adaptive restart to update the consistency and dual variables. The Nesterov accelerated gradient method improves convergence speed by introducing an inertia term and using historical gradient information to correct the current update step size.

[0057] A3 specifically includes: The dispatch center solves the global coordination problem based on the coupling variables of each microgrid to update the consistency variables. : ; Define the overall objective function of the cooperative operation model as follows: To prevent numerical overshoot caused by momentum accumulation during iteration, the overall objective function is... Monitor accordingly. After each global update, if at the [number]th [time]... Round iteration and the first The trend of the overall objective function during -1 iterations is as follows: ; in, For the microgrid decision variables obtained from the k-th iteration, These are the decision variables for the microgrid obtained from the (k-1)th iteration. If the overall objective function value is not monotonically decreasing and the current direction of inertia deviates from the direction of gradient descent, then a restart operation is performed to clear the accumulated momentum. ; in, The dynamic inertia coefficient; For the first Accelerated time steps in the next iteration; This adaptive restart mechanism can detect cost trends and reset the dynamic inertia coefficient in a timely manner, causing the algorithm to degenerate into standard gradient descent, thereby finding the correct descent direction again.

[0058] Subsequently, based on the current and historical iteration values, the Nesterov accelerated gradient method is used to update the consistency variables. : ; in, To accelerate the time step; These are the consistency variables from the previous iteration; For accelerated consistency variables; The control center updates the dual variables of each microgrid uniformly based on the coupling variables of each microgrid and the degree of boundary coupling constraint exceedance. : ; The dual variable is updated using the Nesterov accelerated gradient method. : ; To avoid algorithm divergence caused by asymmetric information updates, consistency variables and dual variables are fed back to the microgrid where no packet loss has occurred, i.e., the link state. Microgrids consisting of 1s and 0s.

[0059] In a preferred embodiment of the present invention, the convergence criterion for iteratively solving the augmented Lagrangian function includes: When the original residual and dual residual The iteration converges when all values ​​are below the set convergence threshold. ; in, This is the original convergence threshold; This is the dual convergence threshold.

[0060] This invention presents a microgrid collaborative optimization method based on event triggering and packet loss compensation. It introduces an event-triggered communication mechanism based on a dynamic threshold, triggering transmission only when data deviation exceeds a threshold. This effectively avoids frequent communication caused by small numerical fluctuations, reducing bandwidth pressure in weak communication environments. Simultaneously, the attenuation threshold ensures convergence accuracy in the later stages of iteration. A differential privacy protection mechanism based on dynamic noise attenuation overcomes the drawbacks of traditional methods that introduce steady-state errors by adding fixed noise. Larger noise is injected in the early stages of iteration to mask sensitive information and prevent reverse reconstruction, while the noise automatically attenuates to near zero in the later stages, achieving both privacy protection and precise scheduling. Finally, an anti-packet loss compensation mechanism based on hybrid momentum extrapolation is introduced to address data arrival and packet loss issues. The invention employs differentiated updates for momentum correction, linear extrapolation prediction, and zero-order preservation in three states (not triggered), utilizing historical momentum to fill data gaps and correct prediction directions. This effectively overcomes the algorithm oscillation and divergence problem caused by packet loss in weak communication, improving the system's robustness in packet loss environments. Furthermore, the method integrates a Nesterov acceleration strategy with adaptive restart, using historical gradient information to accelerate convergence and promptly clearing accumulated momentum when the objective function is not monotonically decreasing. Only the updated consistency variables and dual variables are fed back to the microgrid without packet loss, avoiding algorithm divergence caused by asymmetric information updates. This achieves efficient, stable, and privacy-preserving collaborative optimization scheduling of the microgrid system in weak communication environments.

[0061] In a preferred embodiment of the present invention, a microgrid collaborative optimization device based on event triggering and packet loss compensation is also provided, the device being used to implement the method of the present invention.

[0062] The microgrid collaborative optimization device based on event triggering and packet loss compensation of the present invention, used in the method of the present invention, has the same beneficial effects as the method of the present invention.

[0063] Verification section: To verify the feasibility and superiority of the method of this invention, a microgrid system was used as the research object, and a simulation was conducted on the MATLAB platform. The simulation system consists of three electrically interconnected and thermally coupled microgrids: microgrid A, microgrid B, and microgrid C. The microgrids are interconnected and energy exchanged through power lines, hot water supply trunk lines, and medium-pressure gas pipelines.

[0064] Figure 2 This is a diagram showing the power output of each device in Microgrid A. Figure 2 As can be seen from the curve trend, throughout the entire 24-hour dispatch cycle, the sum of positive power supply and the sum of negative power consumption in each period perfectly envelop and fit the black load curve, strictly meeting the real-time power balance constraints. Renewable energy sources such as photovoltaics and wind power have higher output during the daytime, and energy storage devices charge. During nighttime peak loads or when new energy output is insufficient, energy storage devices discharge to suppress power fluctuations. CHP power generation operates flexibly according to demand, and the tie-line and interactive power are dynamically adjusted according to changes in energy supply and demand. This demonstrates that the method of this invention can coordinate the operation of multiple energy forms, achieve efficient and low-carbon energy dispatch, and improve the system's economy and reliability.

[0065] Figure 3 This is a power interaction diagram of inter-microgrid tie lines, comprising three sub-diagrams: (a) power interaction of electricity tie lines, (b) power interaction of gas tie lines, and (c) power interaction of heat tie lines. Each sub-diagram shows the dynamic changes in power flow across different tie lines at different times, reflecting the mutual support between microgrids through the interaction of multiple energy flows including electricity, heat, and gas. For example, in... Figure 3 In (a), the power interconnection line between microgrids A and C shows a negative value at night and a positive value during the daytime peak period. This indicates that each microgrid, based on the peak and valley characteristics of its own new energy output, uses the interconnection line to carry out cross-regional mutual assistance of electricity, gas and heat, which effectively improves the comprehensive utilization efficiency of energy and the economic benefits of system operation.

[0066] Figure 4The diagram presents a comparison of the algorithm residual convergence trajectories under different communication and privacy scenarios, comprising two sub-plots: (a) comparison of the original residual convergence trajectory and (b) comparison of the dual residual convergence trajectory. Simulations were conducted under three scenarios with varying communication and privacy conditions to investigate the effectiveness of the proposed method. Scenario 1 uses the traditional ADMM algorithm in an ideal communication environment; Scenario 2 uses the traditional ADMM algorithm with a 2% packet loss rate; and Scenario 3 uses the proposed method with a 2% packet loss rate. From the convergence trajectories of (a) the original residual and (b) the dual residual, it is evident that Scenario 3 exhibits faster and more stable residual convergence without significant oscillations; Scenario 2 shows significant residual oscillations and slow convergence; while Scenario 1 demonstrates stable convergence, it is difficult to achieve in reality. This comparison demonstrates that the proposed method exhibits strong robustness to packet loss in weak communication environments. Combined with the privacy protection mechanism of event triggering and dynamic noise attenuation, it achieves both privacy protection and precise scheduling in weak communication environments while ensuring the algorithm's convergence stability.

[0067] Figure 5 This is a diagram showing the communication state distribution of coupled variables. The first letter (A, B, C) of the coupled variables represents the data sender (i.e., the microgrid region); the middle letters (P, G, H) represent three different energy flow coupling types: electricity, gas, and heat, respectively; the suffixes (_AB, _AC, _BA, etc.) represent the orientation of the cross-regional boundary of this energy interaction (for example, "A-P_AB" indicates that the electricity coupled variable flowing from region A to region B represents the electrical power of region A on the tie line AB). As can be seen from the diagram, in the early stages of iteration ( The gradient of the variable changes drastically, and the system maintains high-frequency effective data transmission to establish the convergence direction; as the iteration tends to a steady state ( When the update frequency of a large number of variables falls below the dynamic threshold, they enter a quiescent state, reducing the interaction frequency. To address packet loss during communication, momentum prediction successfully fills these data gaps, ensuring the algorithm's continuous operation even with missing data. This demonstrates that the proposed method, through dynamic threshold determination, significantly reduces unnecessary communication frequency, effectively lowers bandwidth consumption in weak communication environments, while retaining necessary communication to ensure algorithm convergence accuracy.

[0068] Figure 6Comparison of the power iteration trajectories of interconnected lines under different scenarios is presented, including two sub-plots: (a) the power iteration trajectory of the AC power interconnection line in a microgrid, and (b) the power iteration trajectory of the AC thermal interconnection line in a microgrid. Comparing the trajectories of scenarios 1 and 3 reveals that the standard algorithm's iteration path is smooth and has a clear directionality, making it vulnerable to gradient attacks that could reconstruct the load characteristics within the region. Scenario 3, however, exhibits significant random oscillations in the early stages of iteration. The injected Laplace noise effectively masks the true power interaction values ​​and gradient directions, thus achieving privacy protection for the data within each microgrid. As the number of iterations increases, thanks to the dynamic decay design of the noise scale, the oscillation amplitude of the scenario 3 curve gradually decreases and eventually coincides with the endpoint of scenario 1. This demonstrates that the method of this invention can achieve both privacy protection and precise scheduling.

[0069] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A microgrid collaborative optimization method based on event triggering and packet loss compensation, characterized in that, include: A collaborative operation model for a microgrid system is constructed using the alternating direction multiplier method. An augmented Lagrange function is constructed based on the aforementioned collaborative operation model; The augmented Lagrangian function is solved iteratively to obtain the scheduling scheme for each microgrid; Each iteration includes: Each microgrid solves the local coupling variables and compares them with the coupling variables solved in the previous round. If the deviation exceeds a preset threshold, noise is added to the coupling variables and they are sent to the scheduling center; otherwise, a multiplexed symbol is sent. The dispatch center receives data sent by each microgrid. If data packets are lost, it performs linear extrapolation prediction of the coupling variables based on historical momentum. If the data is a reused variable, it reuses the coupling variables from the previous round. Based on the coupling variables of each microgrid, it updates the consistency variables and dual variables using the Nesterov accelerated gradient method and feeds them back to the microgrids that have not lost packets. Constructing the augmented Lagrange function based on the aforementioned cooperative operation model includes: The cooperative running model in the form of a distributed optimization problem is represented as: ； in, For microgrid decision variables; For the first Local decision variables of a microgrid; For consistency variables; The number of microgrids; This is a local optimization model for microgrids; and A coefficient matrix describing the energy coupling relationship between microgrids; For the first A compact convex set consisting of all physical constraints within a microgrid; For the coupling variables of the microgrid; An augmented Lagrange function is constructed based on the cooperative operation model of the distributed optimization problem form. : ； in, For Micro Energy Network The dual variable; The penalty parameter is T; T represents the transpose of the matrix. Represents the square of the 2-norm; The local coupling variables for each microgrid solution include: Each microgrid solves its own cost minimization problem in parallel: utilizing the consistency variables from the previous iteration. and dual variables Update local decision variables With coupling variables , The update is represented as: ； The comparison with the coupling variables from the previous solution includes: definition For the first The first microgrid in the The coupling variables calculated in the next iteration The coupling variable is the one that was successfully transmitted and confirmed by the dispatch center in the last time the microgrid was used; and Compare and determine the trigger flag. : ； ； in, The preset threshold; This is the initial threshold constant; As the attenuation factor, For the first The decay factor for the next iteration; Denotes the 2-norm; when At this time, it indicates that the microgrid is sending the latest data. ;when When, it indicates that the microgrid is sending a multiplexer; Adding noise to the coupling variables includes: On MicroNet Upload coupling variables Forward, towards Injection scale dynamically decays with the number of iterations of Laplace noise Generate encrypted variables , will variables It is sent to the scheduling center as a new coupling variable. Represented as: ； ； in, express Obeying position parameter 0 and scale parameter 0 The Laplace distribution of; This is the initial noise base. This is the decay index.

2. The microgrid collaborative optimization method based on event triggering and packet loss compensation according to claim 1, characterized in that, The cooperative operation model of a microgrid system constructed using the alternating direction multiplier method includes: Construct a microgrid model and operational constraints; based on the microgrid model and operational constraints, construct a local optimization model for each microgrid with the goal of minimizing operating costs; decouple the collaborative operation model of the microgrid system into a global objective function and the local optimization model, and use the power exchanged between interconnecting lines of microgrids as a consistency variable, employing the alternating direction multiplier method to construct a collaborative operation model in the form of a distributed optimization problem.

3. The microgrid collaborative optimization method based on event triggering and packet loss compensation according to claim 2, characterized in that, The dispatch center receives data sent by each microgrid. If data packets are lost, it performs linear extrapolation prediction of the coupling variables based on historical momentum. If the data is a reused variable, it reuses the coupling variables from the previous round, including: The dispatch center determines the link status based on the data sent by each microgrid. And based on the link status For coupling variables Perform differentiated updates: If the data arrives Then, regarding historical momentum Perform a correction update to adjust the prediction direction: ； in, For the first The historical momentum of the next iteration represents the changing trend and numerical inertia of the coupled variables in the previous continuous iteration process; For the first -1 iterations of historical momentum; The momentum retention coefficient; For momentum update coefficients; If data packets are lost Then, linear extrapolation prediction is performed on the coupled variables based on historical momentum: ； If the data is a reusable character Then the coupling variables from the previous round are reused, and the momentum is reset: 。 4. The microgrid collaborative optimization method based on event triggering and packet loss compensation according to claim 3, characterized in that, The consistency and dual variables are updated based on the coupling variables of each microgrid using the Nesterov accelerated gradient method, and then fed back to the microgrids that have not experienced packet loss, including: The dispatch center solves the global coordination problem based on the coupling variables of each microgrid to update the consistency variables. : ； Define the overall objective function of the cooperative operation model as follows: After each global update, if at the 1st Round iteration and the first The trend of the overall objective function during -1 iterations is as follows: ； in, For the microgrid decision variables obtained from the k-th iteration, These are the decision variables for the microgrid obtained from the (k-1)th iteration. If the overall objective function value is not monotonically decreasing and the current direction of inertia deviates from the direction of gradient descent, then a restart operation is performed to clear the accumulated momentum. ； in, The dynamic inertia coefficient; For the first Accelerated time steps in the next iteration; The Nesterov accelerated gradient method is used to update the consistency variables. : ； in, To accelerate the time step; These are the consistency variables from the previous iteration; For accelerated consistency variables; The control center updates the dual variables of each microgrid uniformly based on the coupling variables of each microgrid and the degree of boundary coupling constraint exceedance. : ； The dual variable is updated using the Nesterov accelerated gradient method. : ； The consistency variables and dual variables are fed back to the micro-energy network where no packet loss has occurred, i.e., the link state. Microgrids consisting of 1s and 0s.

5. The microgrid collaborative optimization method based on event triggering and packet loss compensation according to claim 4, characterized in that, The iterative solution of the augmented Lagrangian function also includes a convergence criterion. When the original residual and dual residual The iteration converges when all values ​​are below the set convergence threshold. ； in, This is the original convergence threshold; This is the dual convergence threshold.

6. A microgrid collaborative optimization device based on event triggering and packet loss compensation, characterized in that, The apparatus is used to implement the method according to any one of claims 1 to 5.

Images