Microgrid operation optimization method and system

Abstract

The invention discloses an operation optimization method and system for a micro-grid, and relates to the technical field of micro-grid optimization, and the method comprises the steps: taking the minimum operation cost of the micro-grid as an optimization target, constructing a micro-grid operation model, obtaining a plurality of training operation constraint data of the micro-grid, carrying out the regional feature marking of each training operation constraint data, and obtaining a micro-grid operation model; the method comprises the steps of obtaining a corresponding region feature set, training a preset initial region recognition model by adopting the region feature set to generate a target region recognition model, and solving a microgrid operation model by adopting the target region recognition model based on a large M relaxation method to obtain corresponding optimal scheduling parameters. The technical problems that a traditional microgrid operation optimization method mainly depends on construction of a physical model for optimization, but along with increase of the problem dimension, the calculated amount is sharply expanded, operation optimization on the microgrid is difficult to perform efficiently, and the reliability of microgrid operation is reduced are solved.

Description

Technical Field

[0001] This invention relates to the field of microgrid optimization technology, and in particular to a method and system for optimizing the operation of a microgrid. Background Technology

[0002] As the global energy structure transitions towards cleaner and lower-carbon energy, the penetration rate of distributed renewable energy sources, such as wind and solar power, in the power system is continuously increasing. Microgrids, as a localized autonomous energy system integrating distributed generation, energy storage, controllable loads, and energy management systems, can be connected to the grid for energy efficiency optimization or operated in islanded mode to ensure power supply to critical loads, significantly improving power supply reliability, energy utilization efficiency, and system flexibility. However, the intermittency and uncertainty of high-proportion renewable energy sources, the complexity of coordinating diverse and heterogeneous resources, and the conflict between multiple objectives—operational safety and economics—pose significant challenges to the efficient operation of microgrids.

[0003] Currently, traditional microgrid operation optimization methods mainly rely on building physical models for optimization (such as using optimization tools like Mixed Integer Linear Programming (MILP) and Second-Order Cone Programming (SOCP) to build accurate mathematical models including unit combination, power balance, and network topology constraints). However, as the problem dimensions increase, the computational load will expand dramatically, making it difficult to efficiently optimize the operation of microgrids and reducing the reliability of microgrid operation. Summary of the Invention

[0004] This invention provides a method and system for optimizing the operation of microgrids, which solves the technical problem that traditional microgrid operation optimization methods mainly rely on building physical models for optimization. However, as the problem dimension increases, the amount of computation will increase dramatically, making it difficult to efficiently optimize the operation of microgrids and reducing the reliability of microgrid operation.

[0005] The first aspect of this invention provides a method for optimizing the operation of a microgrid, comprising:

[0006] A microgrid operation model is constructed with the goal of minimizing the operating cost of the microgrid.

[0007] Multiple training operation constraint data of the microgrid are obtained, and regional feature labels are applied to each training operation constraint data to obtain the corresponding regional feature set;

[0008] The preset initial region recognition model is trained using the region feature set to generate a target region recognition model;

[0009] Based on the Big M relaxation method, the target area identification model is used to solve the microgrid operation model to obtain the corresponding optimized scheduling parameters.

[0010] Optionally, the training runtime constraint data includes training reference aggregate power and training injection power, and the step of performing regional feature labeling on each of the training runtime constraint data to obtain the corresponding regional feature set includes:

[0011] A preset flexible region identification function is used to flexibly label the regions of each of the training reference aggregate powers to obtain multiple target reference aggregate powers;

[0012] A preset safe region identification function is used to mark the safe regions of each of the training injection powers to obtain multiple target injection powers;

[0013] A region feature set is generated using each of the target reference aggregate power and each of the target injection power.

[0014] Optionally, the step of using a preset flexible region identification function to flexibly label the regions of each of the training reference aggregated powers to obtain multiple target reference aggregated powers includes:

[0015] Each of the training reference aggregated power is input into a preset flexible region recognition function to obtain multiple target function values;

[0016] When the objective function value corresponding to the training reference aggregate power is a preset flexible region threshold, the training reference aggregate power is labeled with a preset first flexible label to obtain the corresponding target reference aggregate power.

[0017] When the objective function value corresponding to the training reference aggregate power is not the flexible region threshold, the training reference aggregate power is marked with a preset second flexible label to obtain the corresponding target reference aggregate power.

[0018] Optionally, the step of marking safe regions for each of the training injection powers using a preset safe region identification function to obtain multiple target injection powers includes:

[0019] Each of the training injection powers is input into a preset safe region identification function to obtain multiple safe indication values;

[0020] When the safety indication value corresponding to the training injection power is less than or equal to the preset safety threshold, the training injection power is marked with a preset first safety label to obtain the corresponding target injection power.

[0021] When the safety indication value corresponding to the training injection power is greater than the safety threshold, the training injection power is marked with a preset second safety label to obtain the corresponding target injection power.

[0022] Optionally, the step of training a preset initial region recognition model using the region feature set to generate a target region recognition model includes:

[0023] The preset initial region recognition model is trained using the region feature set as input, and the training region data is output.

[0024] Based on the preset cross-entropy loss function, the loss function value of the region feature set is calculated using the training region data;

[0025] When the loss function value is greater than or equal to the preset standard loss function value, the network parameters of the initial region recognition model are adjusted using stochastic gradient descent until the loss function value is less than the standard loss function value.

[0026] When the loss function value is less than the standard loss function value, a target region recognition model is generated.

[0027] Optionally, the step of solving the microgrid operation model using the target region identification model based on the Big M relaxation method to obtain the corresponding optimized scheduling parameters includes:

[0028] The large M relaxation method is used to perform constraint transformation on the microgrid operation model to obtain the corresponding flexible region constraint conditions and safe region constraint conditions.

[0029] By combining the flexible region constraints, the safe region constraints, and the microgrid operation model, the corresponding target microgrid operation model is obtained.

[0030] An optimization solver is used to solve the target microgrid operation model to obtain the corresponding optimized scheduling parameters.

[0031] A second aspect of the present invention provides a microgrid operation optimization system, comprising:

[0032] The module is used to construct a microgrid operation model with the goal of minimizing the operating cost of the microgrid.

[0033] The acquisition module is used to acquire multiple training operation constraint data of the microgrid, and to perform regional feature labeling on each of the training operation constraint data to obtain the corresponding regional feature set;

[0034] The training module is used to train the preset initial region recognition model using the region feature set to generate the target region recognition model;

[0035] The optimization module is used to solve the microgrid operation model based on the Big M relaxation method and the target area identification model to obtain the corresponding optimized scheduling parameters.

[0036] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the microgrid operation optimization method as described in any of the preceding claims.

[0037] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the microgrid operation optimization method as described in any of the preceding claims.

[0038] The fifth aspect of the present invention provides a computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein when the program instructions are executed by a computer, the computer performs the microgrid operation optimization method as described in any of the preceding claims.

[0039] As can be seen from the above technical solutions, the present invention has the following advantages:

[0040] This invention constructs a microgrid operation model with the goal of minimizing microgrid operating costs. It acquires multiple training operation constraint data for the microgrid, labels each training constraint data with regional features to obtain a corresponding regional feature set, and uses this regional feature set to train a pre-set initial regional identification model to generate a target regional identification model. Based on the Big M relaxation method, the target regional identification model is used to solve the microgrid operation model to obtain the corresponding optimized scheduling parameters. This overcomes the technical problem that traditional microgrid operation optimization methods mainly rely on constructing physical models for optimization. However, as the problem dimension increases, the computational load expands dramatically, making efficient microgrid operation optimization difficult and reducing the reliability of microgrid operation. Compared with traditional microgrid operation optimization methods, this invention uses the Big M relaxation method combined with a target regional identification model to solve the microgrid operation model and obtain optimized scheduling parameters, reducing the computational load of microgrid operation optimization and improving the reliability of microgrid operation. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 The flowchart illustrates the steps of a microgrid operation optimization method provided in Embodiment 1 of the present invention.

[0043] Figure 2 The flowchart illustrates the steps of a microgrid operation optimization method provided in Embodiment 2 of the present invention.

[0044] Figure 3 This is a structural block diagram of a microgrid operation optimization system provided in Embodiment 3 of the present invention;

[0045] Figure 4 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention. Detailed Implementation

[0046] This invention provides a method and system for optimizing the operation of microgrids, which addresses the technical problem that traditional microgrid operation optimization methods mainly rely on building physical models for optimization. However, as the problem dimension increases, the computational load expands dramatically, making it difficult to efficiently optimize the operation of microgrids and reducing the reliability of microgrid operation.

[0047] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0048] Please see Figure 1 , Figure 1 This is a flowchart illustrating the steps of a microgrid operation optimization method provided in Embodiment 1 of the present invention.

[0049] The present invention provides a method for optimizing the operation of a microgrid, comprising:

[0050] Step 101: Construct a microgrid operation model with the goal of minimizing the operating cost of the microgrid.

[0051] In this embodiment of the invention, with the goal of minimizing the operating cost of the microgrid, a microgrid operation model is constructed by setting bus active power balance constraints, bus reactive power balance constraints, branch voltage relationship constraints, and safety limit constraints.

[0052] It should be noted that the power grid operation model is as follows:

[0053]

[0054] in, Let t be the electricity price of the upstream main grid. Let t be the power purchased from the upstream main network. The power generation cost for the flexibility resources of the i-node. Let be the power generation capacity of the flexible resources of node i at time t. Let be the active power flow of the line between node j and node k at time t. Let be the active power flow of the line between node j and node i at time t. Let be the per-unit resistance value of the line between node j and node i. Let be the squared value of the current in the line between node j and node i at time t. Let t be the power generation capacity of the flexible resources at node j. Let t be the active load of node j. Let be the reactive power flow of the line between node j and node k at time t. Let be the per-unit reactance value of the line between node j and node i. Let be the reactive power of the flexibility resources at node j at time t. Let the reactive load of node j at time t be... Let be the voltage at node j at time t. Let be the voltage at node i at time t. For branch road collection, Let be the minimum voltage at node i at time t. Let be the maximum voltage at node i at time t. For a set of nodes, Let be the minimum current between node i and node j at time t. Let be the maximum current between node i and node j at time t, where i is the first index of the node, j is the second index of the node, k is the third index of the node, T represents all scheduling periods, and t is the index of time. A collection of flexible resources.

[0055] Step 102: Obtain multiple training operation constraint data of the microgrid, and perform regional feature labeling on each training operation constraint data to obtain the corresponding regional feature set.

[0056] Training operation constraint data refers to the data collected during the training of the initial region identification model, which reflects the boundary conditions of microgrid operation, including training reference aggregated power and training injected power.

[0057] The regional feature set refers to the dataset formed by associating and integrating all target reference aggregate power and target injection power according to data dimensions (time, node).

[0058] In this embodiment of the invention, a preset flexible region identification function is used to flexibly label the regions of each training reference aggregated power to obtain multiple target reference aggregated powers. A preset safe region identification function is used to safely label the regions of each training injection power to obtain multiple target injection powers. A region feature set is generated using each target reference aggregated power and each target injection power.

[0059] Step 103: Train the preset initial region recognition model using the region feature set to generate the target region recognition model.

[0060] In this embodiment of the invention, a preset initial region recognition model is trained using a region feature set as input, and training region data is output. Based on a preset cross-entropy loss function, the loss function value of the region feature set is calculated using the training region data. When the loss function value is greater than or equal to a preset standard loss function value, the network parameters of the initial region recognition model are adjusted using stochastic gradient descent until the loss function value is less than the standard loss function value. When the loss function value is less than the standard loss function value, a target region recognition model is generated.

[0061] Step 104: Based on the Big M relaxation method, the target area identification model is used to solve the microgrid operation model to obtain the corresponding optimized scheduling parameters.

[0062] Optimized scheduling parameters refer to the key parameters obtained by solving the target microgrid operation model, which are used to guide the operation and scheduling of the microgrid. These parameters include, but are not limited to, the power purchased and sold between the microgrid and the upper-level main grid, and the adjustment amount of controllable loads.

[0063] In this embodiment of the invention, the Big M relaxation method is used to perform constraint transformation on the microgrid operation model, resulting in corresponding flexible region constraints and safe region constraints. The flexible region constraints, safe region constraints, and the microgrid operation model are then combined to obtain the corresponding target microgrid operation model. An optimization solver is used to solve the target microgrid operation model to obtain the corresponding optimized scheduling parameters.

[0064] In this embodiment of the invention, a microgrid operation model is constructed with the goal of minimizing the operating cost of the microgrid. Multiple training operation constraint data for the microgrid are obtained, and each training operation constraint data is labeled with regional features to obtain a corresponding regional feature set. This regional feature set is used to train a preset initial regional identification model to generate a target regional identification model. Based on the Big M relaxation method, the target regional identification model is used to solve the microgrid operation model to obtain the corresponding optimized scheduling parameters. This overcomes the technical problem that traditional microgrid operation optimization methods mainly rely on constructing physical models for optimization. However, as the problem dimension increases, the computational load expands dramatically, making it difficult to efficiently optimize microgrid operation and reducing the reliability of microgrid operation. Compared with traditional microgrid operation optimization methods, this invention uses the Big M relaxation method combined with a target regional identification model to solve the microgrid operation model and obtain optimized scheduling parameters, reducing the computational load of microgrid operation optimization and improving the reliability of microgrid operation.

[0065] Please see Figure 2 , Figure 2 This is a flowchart illustrating the steps of a microgrid operation optimization method provided in Embodiment 2 of the present invention.

[0066] The present invention provides a method for optimizing the operation of a microgrid, comprising:

[0067] Step 201: Construct a microgrid operation model with the goal of minimizing the operating cost of the microgrid.

[0068] In this embodiment of the invention, with the goal of minimizing the operating cost of the microgrid, a microgrid operation model is constructed by setting bus active power balance constraints, bus reactive power balance constraints, branch voltage relationship constraints, and safety limit constraints.

[0069] Step 202: Obtain multiple training operation constraint data of the microgrid, and perform regional feature labeling on each training operation constraint data to obtain the corresponding regional feature set.

[0070] Furthermore, the training runtime constraint data includes the training reference aggregate power and the training injected power, and step 202 includes the following sub-steps:

[0071] S11. Use a preset flexible region recognition function to perform flexible region labeling on each training reference aggregate power to obtain multiple target reference aggregate powers.

[0072] Furthermore, S11 includes the following sub-steps:

[0073] S111. Input the aggregated power of each training reference into the preset flexible region recognition function to obtain multiple target function values.

[0074] Training reference aggregated power refers to the total power reference value of a load aggregator (such as an electric vehicle cluster or an energy storage-controllable load joint cluster) in a microgrid during a specific time period, which is collected for training the initial area identification model. It is used to reflect the overall power demand of the aggregator and is the basic data for judging the flexibility of the area. For example, the total charging power reference value of all electric vehicles in a certain area in 1 hour.

[0075] In this embodiment of the invention, the aggregated power of each training reference is input into a preset flexible region recognition function one by one to obtain multiple target function values.

[0076] It should be noted that the flexible region identification function is as follows:

[0077]

[0078] in, The objective function value, To train the reference aggregate power, Let d be the power requirement of the flexible resource d, where d is the index of the flexible resource, and LA is the distributed resource set. To provide operational constraints for flexible resource d. A collection of flexible resources.

[0079] S112. When the objective function value corresponding to the training reference aggregate power is the preset flexible region threshold, the training reference aggregate power is labeled with the preset first flexible label to obtain the corresponding target reference aggregate power.

[0080] The flexible region threshold is a critical value used to determine whether the training reference aggregate power is in the flexible region, and its value is 0.

[0081] The first flexible label refers to the identifier used to mark the attribute of the training reference aggregate power region, and the label content is specifically "flexible region feasible".

[0082] In this embodiment of the invention, when the objective function value corresponding to the training reference aggregate power is 0, the training reference aggregate power is marked by a preset first flexible label to obtain the corresponding target reference aggregate power.

[0083] S113. When the objective function value corresponding to the training reference aggregate power is not the flexible region threshold, the training reference aggregate power is labeled with a preset second flexible label to obtain the corresponding target reference aggregate power.

[0084] The second flexible label refers to the identifier used to mark the region attribute of the training reference aggregation power; specifically, the label content is "flexible region not feasible".

[0085] In this embodiment of the invention, when the objective function value corresponding to the training reference aggregate power is not 0, a preset second flexible label is called to mark the training reference aggregate power to obtain the corresponding target reference aggregate power.

[0086] S12. Use a preset safe region identification function to mark the safe region for each training injection power to obtain multiple target injection powers.

[0087] Furthermore, S12 includes the following sub-steps:

[0088] S121. Input each training injection power into the preset safe area identification function to obtain multiple safe indication values.

[0089] Training injected power refers to the microgrid bus power data collected during the training of the initial region identification model, including active injected power and reactive injected power.

[0090] The safety indication value refers to the degree of safety risk in microgrid operation corresponding to the training injected power.

[0091] In this embodiment of the invention, each training injection power is input into a preset safe area identification function one by one to obtain multiple safe indication values.

[0092] It should be noted that the specific function for identifying the secure area is as follows:

[0093]

[0094] in, These are the operating status values ​​of the microgrid. This is the third constraint value. This is the second constraint value. This is the fourth constraint value. This is the first constraint value. To train the injected active power, The reactive power injected for training power injection.

[0095] S122. When the safety indication value corresponding to the training injection power is less than or equal to the preset safety threshold, the training injection power is marked with the preset first safety label to obtain the corresponding target injection power.

[0096] The safety threshold refers to the critical value at which the training injection power is within the safe range.

[0097] The first safety label refers to the identifier that marks the safety attributes of the training injection power region, and the label content is "safe operating area".

[0098] In this embodiment of the invention, when the safety indication value corresponding to the training injection power is less than or equal to a preset safety threshold, the preset first safety label is called to mark the training injection power to obtain the corresponding target injection power.

[0099] S123. When the safety indication value corresponding to the training injection power is greater than the safety threshold, the training injection power is marked by a preset second safety label to obtain the corresponding target injection power.

[0100] The second safety label refers to the identifier that marks the safety attributes of the training injection power region, and the label content is "non-safe operating area".

[0101] In this embodiment of the invention, when the safety indication value corresponding to the training injection power is greater than the safety threshold, a preset second safety label is called to mark the training injection power to obtain the corresponding target injection power.

[0102] S13. Generate a region feature set using the reference aggregation power and injection power of each target.

[0103] In this embodiment of the invention, a region feature set is generated using each target reference aggregation power and each target injection power, wherein the region feature set includes multiple target reference aggregation powers and multiple target injection powers to generate the region feature set.

[0104] Step 203: Train the preset initial region recognition model using the region feature set to generate the target region recognition model.

[0105] Furthermore, step 203 includes the following sub-steps:

[0106] S21. Use the regional feature set as input to train the preset initial region recognition model and output the training region data.

[0107] Training region data refers to the data output by the initial region recognition model after training with the region feature set.

[0108] In this embodiment of the invention, the regional feature set is input into a preset initial regional recognition model for training to obtain training regional data.

[0109] S22. Based on the preset cross-entropy loss function, calculate the loss function value of the region feature set using the training region data.

[0110] In this embodiment of the invention, the training region data and the region feature set are input into a preset cross-entropy loss function to obtain the corresponding loss function value.

[0111] It should be noted that the cross-entropy loss function is as follows:

[0112]

[0113] in, The value of the loss function. For the first A true value, For the first One predicted value, The total number of samples, This is the index of the sample.

[0114] S23. When the loss function value is greater than or equal to the preset standard loss function value, the network parameters of the initial region recognition model are adjusted using the stochastic gradient descent method until the loss function value is less than the standard loss function value.

[0115] The standard loss function value refers to the loss threshold set based on the microgrid area identification accuracy requirements (such as the error tolerance for judging area attributes in actual operation) and historical training data. It is the core standard for determining whether the initial area identification model training has converged.

[0116] It should be noted that the optimization algorithm for adjusting the initial region identification model network parameters using stochastic gradient descent works by calculating the gradient of the loss function with respect to each network parameter, and then updating the parameters along the negative gradient direction at a preset learning rate to gradually reduce the loss function value. This can effectively improve the model's prediction accuracy for microgrid region attributes.

[0117] In this embodiment of the invention, when the loss function value is greater than or equal to the preset standard loss function value, the network parameters of the initial region recognition model are adjusted using the stochastic gradient descent method, and the process jumps to execute S21-S23 until the loss function value is less than the standard loss function value.

[0118] S24. When the loss function value is less than the standard loss function value, a target region recognition model is generated.

[0119] In this embodiment of the invention, if the loss function value is less than the standard loss function value, a target region recognition model is generated.

[0120] It should be noted that the target region identification model introduces a multi-layer neural network regression model for the identification of flexible and secure regions. The specific expression of the multi-layer neural network regression model is as follows:

[0121]

[0122] in, For the i-th in the l-th layer s One hidden neuron, Let i be the linear transformation matrix of the i0th layer. For the i-th s Layer and js Linear transformation matrix between layers, For layer set, This is the set of neurons in the l-th layer. For the j-th element in the l-th layer s One hidden neuron, As the first index of the layer, For the output of the model, This is a vector composed of all neurons in the Lth layer. For all bias matrices in the Lth layer, For the j-th layer of layer L s The transformation matrix corresponding to each neuron For the j-th layer of the L-1 level s The transformation matrix corresponding to each neuron As input to the model, For, i s j is the first index of the neuron. s i is the second index of the neuron, and i0 is the second index of the layer.

[0123] Simultaneously, a sigmoid activation function (i.e., a sigmoid activation function) is used to replace the output layer of the multi-layer neural network regression model to construct a network-parameterized target region identification model. This enables the target region identification model to identify flexible and safe operating regions of the microgrid. The specific expression of the target region identification model is as follows:

[0124]

[0125] in, The threshold for the classifier. For This is an indicator function for the classifier threshold.

[0126] Step 204: Use the Big M relaxation method to perform constraint transformation on the microgrid operation model to obtain the corresponding flexible region constraint conditions and safe region constraint conditions.

[0127] In this embodiment of the invention, the large M relaxation method is used to transform the nonlinear part of the target microgrid operation model into large M relaxation conditions, thereby obtaining the corresponding flexible region constraints and safe region constraints.

[0128] It is worth mentioning that the Big M relaxation method was used to transform the ReLU activation function of the target microgrid operation model by introducing binary variables. This transformation of the ReLU activation function was applied to all activation functions in the transformation formula to achieve a complete transformation of the neural network. The transformation formula is as follows:

[0129]

[0130] in, Let m be the output value of the m-th neuron in the neural network. For the i-th layer of layer l s ReLU relaxed binary variables for each neuron For the i-th layer of layer l s The minimum value of a number of neurons in the large M-relaxation method. For the i-th layer of layer l s The maximum value of the number of neurons in the M-relaxation method, where m is the fourth index of the neuron.

[0131] The sigmoid activation function of the target microgrid operation model cannot be linearized; therefore, when performing the transformation... The output of the target microgrid operation model when the predetermined probability lower bound is met.

[0132]

[0133] It can be binary converted to:

[0134]

[0135] For example, When it is 0.5, when it satisfies

[0136]

[0137] Output The value is 1.

[0138] Finally, the target microgrid model is converted into data-driven flexible region constraints and safe region constraints.

[0139] It should be noted that the specific constraints for the flexible region are as follows:

[0140]

[0141] in, Let t be the flexibility power of node j aggregation at time t. Let be the feasible range of flexibility power for node j.

[0142] It should be noted that the specific constraints of the safe area are as follows:

[0143]

[0144] in, The active power injected into node j at time t. The reactive power injected into node j at time t. This is a safe zone obtained based on learning.

[0145] Step 205: Combine the flexible region constraints, the safe region constraints, and the microgrid operation model to obtain the corresponding target microgrid operation model.

[0146] In this embodiment of the invention, the flexible region constraint condition, the safe region constraint condition, and the microgrid operation model are coupled to obtain the corresponding target microgrid operation model.

[0147] Step 206: Use an optimization solver to solve the target microgrid operation model to obtain the corresponding optimized scheduling parameters.

[0148] Optimization solvers refer to commercial solvers that support solving mixed-integer linear programming problems, including but not limited to Cplex (mathematical programming solver) and Gurobi (mathematical optimization engine).

[0149] In this embodiment of the invention, Cple or Gurobi is invoked to solve the target microgrid operation model to obtain the corresponding optimized scheduling parameters.

[0150] In this embodiment of the invention, a microgrid operation model is constructed with the goal of minimizing the operating cost of the microgrid. Multiple training operation constraint data for the microgrid are obtained, and each training operation constraint data is labeled with regional features to obtain a corresponding regional feature set. This regional feature set is used to train a preset initial regional identification model to generate a target regional identification model. Based on the Big M relaxation method, the target regional identification model is used to solve the microgrid operation model to obtain the corresponding optimized scheduling parameters. This overcomes the technical problem that traditional microgrid operation optimization methods mainly rely on constructing physical models for optimization. However, as the problem dimension increases, the computational load expands dramatically, making it difficult to efficiently optimize microgrid operation and reducing the reliability of microgrid operation. Compared with traditional microgrid operation optimization methods, this invention uses the Big M relaxation method combined with a target regional identification model to solve the microgrid operation model and obtain optimized scheduling parameters, reducing the computational load of microgrid operation optimization and improving the reliability of microgrid operation.

[0151] Please see Figure 3 , Figure 3 This is a structural block diagram of a microgrid operation optimization system provided in Embodiment 3 of the present invention.

[0152] This invention provides a microgrid operation optimization system, comprising:

[0153] Module 301 is used to construct a microgrid operation model with the goal of minimizing the operating cost of the microgrid.

[0154] The acquisition module 302 is used to acquire multiple training operation constraint data of the microgrid, and to perform regional feature labeling on each training operation constraint data to obtain the corresponding regional feature set;

[0155] Training module 303 is used to train a preset initial region recognition model using a region feature set to generate a target region recognition model;

[0156] The optimization module 304 is used to solve the microgrid operation model based on the big M relaxation method and the target area identification model to obtain the corresponding optimized scheduling parameters.

[0157] Furthermore, the training runtime constraint data includes training reference aggregate power and training injection power, and the acquisition module 302 includes:

[0158] The first labeling submodule is used to perform flexible region labeling on each training reference aggregate power using a preset flexible region recognition function to obtain multiple target reference aggregate powers;

[0159] The second labeling submodule is used to label the safe regions of each training injection power using a preset safe region identification function, so as to obtain multiple target injection powers.

[0160] A submodule is constructed to generate a region feature set using the aggregated power of each target reference and the injected power of each target.

[0161] Furthermore, the first tagging submodule includes:

[0162] The flexible region identification unit is used to input the aggregated power of each training reference into a preset flexible region identification function to obtain multiple target function values;

[0163] The flexible region labeling unit is used to label the training reference aggregate power with a preset first flexible label when the objective function value corresponding to the training reference aggregate power is a preset flexible region threshold, so as to obtain the corresponding target reference aggregate power.

[0164] When the objective function value corresponding to the training reference aggregate power is not the flexible region threshold, the training reference aggregate power is labeled with a preset second flexible label to obtain the corresponding target reference aggregate power.

[0165] Furthermore, the second tagging submodule includes:

[0166] The safe zone identification unit is used to input each training injection power into a preset safe zone identification function to obtain multiple safe indication values.

[0167] The safe area marking unit is used to mark the training injection power with a preset first safety label when the safety indication value corresponding to the training injection power is less than or equal to a preset safety threshold, so as to obtain the corresponding target injection power.

[0168] When the safety indication value corresponding to the training injection power is greater than the safety threshold, the training injection power is marked with a preset second safety label to obtain the corresponding target injection power.

[0169] Furthermore, training module 303 includes:

[0170] The training submodule is used to train a preset initial region recognition model by taking the region feature set as input and output training region data.

[0171] The loss submodule is used to calculate the loss function value of the region feature set based on the preset cross-entropy loss function and the training region data.

[0172] The iterative optimization submodule is used to adjust the network parameters of the initial region recognition model using stochastic gradient descent when the loss function value is greater than or equal to the preset standard loss function value, until the loss function value is less than the standard loss function value.

[0173] When the loss function value is less than the standard loss function value, a target region recognition model is generated.

[0174] Furthermore, module 304 is optimized, including:

[0175] The conversion submodule is used to perform constraint conversion processing on the microgrid operation model using the Big M relaxation method to obtain the corresponding flexible region constraint conditions and safe region constraint conditions.

[0176] The combined submodule is used to combine the flexible area constraints, the safe area constraints, and the microgrid operation model to obtain the corresponding target microgrid operation model.

[0177] The optimization solution submodule is used to solve the target microgrid operation model using an optimization solver to obtain the corresponding optimized scheduling parameters.

[0178] Please see Figure 4 , Figure 4 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention.

[0179] An electronic device according to an embodiment of the present invention includes: a memory 401 and a processor 402. The memory 401 stores a computer program. When the computer program is executed by the processor 402, the processor 402 executes the microgrid operation optimization method as described in any of the above embodiments.

[0180] Memory 401 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 401 has storage space 403 for program code 413 for performing any of the method steps described above. For example, storage space 403 for program code may include individual program codes 413 for implementing the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When run by a computing processing device, this code causes the computing processing device to perform the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When this code is run by a computing device, it causes the computing device to perform the various steps in the microgrid operation optimization method described above.

[0181] Embodiment 5 of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the microgrid operation optimization method as described in any of the above embodiments.

[0182] Embodiment 6 of the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer performs the microgrid operation optimization method as described in any of the above embodiments.

[0183] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0184] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0185] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0186] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0187] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0188] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the operation of a microgrid, characterized in that, include: A microgrid operation model is constructed with the goal of minimizing the operating cost of the microgrid. Multiple training operation constraint data of the microgrid are obtained, and regional feature labels are applied to each training operation constraint data to obtain the corresponding regional feature set; The preset initial region recognition model is trained using the region feature set to generate a target region recognition model; Based on the Big M relaxation method, the target area identification model is used to solve the microgrid operation model to obtain the corresponding optimized scheduling parameters.

2. The microgrid operation optimization method according to claim 1, characterized in that, The training operation constraint data includes training reference aggregate power and training injection power. The step of performing regional feature labeling on each of the training operation constraint data to obtain the corresponding regional feature set includes: A preset flexible region identification function is used to flexibly label the regions of each of the training reference aggregate powers to obtain multiple target reference aggregate powers; A preset safe region identification function is used to mark the safe regions of each of the training injection powers to obtain multiple target injection powers; A region feature set is generated using each of the target reference aggregate power and each of the target injection power.

3. The microgrid operation optimization method according to claim 2, characterized in that, The step of using a preset flexible region identification function to flexibly label the regions of each of the training reference aggregated powers to obtain multiple target reference aggregated powers includes: Each of the training reference aggregated power is input into a preset flexible region recognition function to obtain multiple target function values; When the objective function value corresponding to the training reference aggregate power is a preset flexible region threshold, the training reference aggregate power is labeled with a preset first flexible label to obtain the corresponding target reference aggregate power. When the objective function value corresponding to the training reference aggregate power is not the flexible region threshold, the training reference aggregate power is marked with a preset second flexible label to obtain the corresponding target reference aggregate power.

4. The microgrid operation optimization method according to claim 2, characterized in that, The step of marking the training injection power with a preset safe region identification function to obtain multiple target injection powers includes: Each of the training injection powers is input into a preset safe region identification function to obtain multiple safe indication values; When the safety indication value corresponding to the training injection power is less than or equal to the preset safety threshold, the training injection power is marked with a preset first safety label to obtain the corresponding target injection power. When the safety indication value corresponding to the training injection power is greater than the safety threshold, the training injection power is marked with a preset second safety label to obtain the corresponding target injection power.

5. The microgrid operation optimization method according to claim 1, characterized in that, The step of training a preset initial region recognition model using the region feature set to generate a target region recognition model includes: The preset initial region recognition model is trained using the region feature set as input, and the training region data is output. Based on the preset cross-entropy loss function, the loss function value of the region feature set is calculated using the training region data; When the loss function value is greater than or equal to the preset standard loss function value, the network parameters of the initial region recognition model are adjusted using stochastic gradient descent until the loss function value is less than the standard loss function value. When the loss function value is less than the standard loss function value, a target region recognition model is generated.

6. The microgrid operation optimization method according to claim 1, characterized in that, The step of solving the microgrid operation model using the target region identification model based on the Big M relaxation method to obtain the corresponding optimized scheduling parameters includes: The large M relaxation method is used to perform constraint transformation on the microgrid operation model to obtain the corresponding flexible region constraint conditions and safe region constraint conditions. By combining the flexible region constraints, the safe region constraints, and the microgrid operation model, the corresponding target microgrid operation model is obtained. An optimization solver is used to solve the target microgrid operation model to obtain the corresponding optimized scheduling parameters.

7. A microgrid operation optimization system, characterized in that, include: The module is used to construct a microgrid operation model with the goal of minimizing the operating cost of the microgrid. The acquisition module is used to acquire multiple training operation constraint data of the microgrid, and to perform regional feature labeling on each of the training operation constraint data to obtain the corresponding regional feature set; The training module is used to train the preset initial region recognition model using the region feature set to generate the target region recognition model; The optimization module is used to solve the microgrid operation model based on the Big M relaxation method and the target area identification model to obtain the corresponding optimized scheduling parameters.

8. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the steps of the microgrid operation optimization method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the microgrid operation optimization method as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, wherein when the program instructions are executed by a computer, the computer performs the microgrid operation optimization method as described in any one of claims 1-6.

Images