Distribution network planning scheme generation method, device, terminal and medium based on large model

Abstract

This application discloses a method, device, terminal, and medium for generating distribution network planning schemes based on a large model, relating to the field of distribution network technology. The scheme provided in this application combines constraint embedding prompts with distribution network planning scenarios. First, by constructing constraint embedding prompts, natural language and mathematical constraints are unified, thereby achieving cross-modal unified modeling of human-understandable descriptions and machine-calculateable rules. In actual distribution network planning scenarios, based on the constructed constraint embedding prompt set, the corresponding mathematical constraints can be automatically matched and invoked for optimization calculations of the large model by inputting natural language descriptions, thereby generating planning schemes that meet actual engineering needs. This avoids the tedious process of manually converting planning requirements in natural language form into mathematical formulas, thus improving the efficiency of distribution network planning work.

Description

Technical Field

[0001] This application relates to the field of power distribution network technology, and in particular to a method, device, terminal and medium for generating power distribution network planning schemes based on large models. Background Technology

[0002] With the rapid development of large-scale grid connection of renewable energy and diversified loads (such as electric vehicles and smart buildings), distribution network planning needs to address the challenges of high-dimensional uncertainty and dynamic trade-offs among multiple objectives. Currently, conventional planning methods mainly include traditional methods based on algorithms such as genetic algorithms and mixed-integer linear programming, as well as AI optimization methods using artificial intelligence technologies (such as reinforcement learning and deep learning). However, these methods cannot directly address the need to manually translate planning requirements from natural language into mathematical constraints. This process is easily affected by subjective factors such as the technical level of the writers, leading to issues such as ambiguous constraint expressions and conflicting optimization objectives, delaying work progress and resulting in the current technical problem of low efficiency in distribution network planning. Summary of the Invention

[0003] This application provides a method, device, terminal, and medium for generating distribution network planning schemes based on large models, which is used to solve the technical problem of low efficiency in existing distribution network planning work.

[0004] To address the aforementioned technical problems, the first aspect of this application provides a method for generating distribution network planning schemes based on large models, comprising:

[0005] Obtain sample data for power distribution network planning;

[0006] Based on the power distribution network planning sample data, and combined with the natural language constraints and mathematical expression constraints associated with the power distribution network planning constraints, a set of constraint embedding prompt words is constructed.

[0007] Obtain distribution network planning demand information, extract information features through a preset large model, and obtain planning task target information corresponding to the distribution network planning demand information;

[0008] Based on the planning task objective information, the corresponding objective constraint embedding prompts and the planning scheme optimization objective function are determined. Based on the objective constraint embedding prompts and the planning scheme optimization objective function, the power distribution network planning scheme is generated through the reasoning operation of the large model.

[0009] Preferably, the expression for the constraint embedding prompt word set is specifically:

[0010]

[0011] In the formula, The identifier for the constraint embedding prompt word set, This serves as a sample for power distribution network planning. It is a set of constraints. It is a natural language description constraint. It is a constraint expressed by mathematical formulas. These are the weighting coefficients.

[0012] Preferably, before generating the distribution network planning scheme based on the target constraint embedded prompts and the optimized objective function of the planning scheme, combined with the inference calculations of the large model, the following steps are also included:

[0013] Based on the planning scheme, the objective function is optimized, and combined with the preset loss function, the weight parameters of the large model are fine-tuned through LoRA optimization.

[0014] Preferably, the fine-tuning of the weight parameters of the large model specifically includes:

[0015] The weight parameter matrix of the large model is decomposed into two low-rank matrices, where the first low-rank matrix is ​​a Gaussian distribution matrix and the second low-rank matrix is ​​a zero matrix.

[0016] One low-rank matrix is ​​fixed in sequence, and the other low-rank matrix is ​​iteratively adjusted by solving the least squares problem.

[0017] When the change in loss reaches the preset convergence condition or the number of iterations reaches the preset maximum number of iterations threshold, the updated weight parameter matrix is ​​obtained by multiplying the two low-rank matrices and summing the weight parameter matrix.

[0018] Preferably, the loss function is:

[0019]

[0020]

[0021] In the formula, This is the total loss coefficient. Represents the task loss coefficient. This serves as a sample for power distribution network planning. This refers to the predicted output value corresponding to the sample of the power distribution network planning scheme. Let A and B be the actual values ​​corresponding to the sample of the power distribution network planning scheme, and let A and B be two low-rank matrices.

[0022] Preferably, the step of optimizing the objective function based on the embedded prompts of the target constraints and the planning scheme, and generating the distribution network planning scheme through the inference calculation of the large model, includes:

[0023] The objective function is optimized based on the embedded prompts and the planning scheme according to the target constraints, and several planning scheme solutions are obtained through the reasoning operation of the large model;

[0024] By using Pareto dominance logic, the solution of the planning scheme is optimized to obtain the Pareto optimal solution, and the distribution network planning scheme corresponding to the Pareto optimal solution is determined.

[0025] Preferably, after obtaining the distribution network planning sample data, the process further includes:

[0026] The power distribution network planning sample data is preprocessed, wherein the preprocessing includes: data cleaning, normalization and noise reduction.

[0027] The second aspect of this application provides a distribution network planning scheme generation device based on a large model, comprising:

[0028] The data acquisition unit is used to acquire sample data for power distribution network planning.

[0029] The constraint prompt word set construction unit is used to construct a constraint embedding prompt word set based on the power distribution network planning sample data and the natural language constraints and mathematical expression constraints associated with the power distribution network planning constraints.

[0030] The planning target determination unit is used to obtain distribution network planning demand information, and extract information features through a preset large model to obtain planning task target information corresponding to the distribution network planning demand information;

[0031] The planning scheme generation unit is used to determine the corresponding target constraint embedding prompts and the planning scheme optimization objective function based on the planning task objective information, and to generate a distribution network planning scheme through the reasoning operation of the large model based on the target constraint embedding prompts and the planning scheme optimization objective function.

[0032] The third aspect of this application provides a distribution network planning scheme generation terminal based on a large model, including: a memory and a processor;

[0033] The memory is used to store program code, which is used to implement a method for generating a power distribution network planning scheme based on a large model, as provided in the first aspect of this application.

[0034] The processor is used to read and execute the program code.

[0035] The fourth aspect of this application provides a computer-readable storage medium storing program code, which is read and executed by a processor to implement a method for generating a distribution network planning scheme based on a large model, as provided in the first aspect of this application.

[0036] As can be seen from the above technical solutions, this application has the following advantages:

[0037] The solution provided in this application combines constraint embedding prompts with distribution network planning scenarios. First, by constructing constraint embedding prompts, it unifies natural language and mathematical constraints, thereby achieving cross-modal unified modeling of human-understandable descriptions and machine-calculateable rules. In actual distribution network planning scenarios, based on the constructed constraint embedding prompt set, the system can automatically match and call the corresponding mathematical constraints for large-scale model optimization calculations by inputting natural language descriptions, thus generating planning schemes that meet actual engineering needs. This avoids the tedious process of manually converting planning requirements in natural language form into mathematical formulas, thereby improving the efficiency of distribution network planning. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a flowchart illustrating an embodiment of a distribution network planning scheme generation method based on a large model provided in this application.

[0040] Figure 2 This is an overall logic block diagram of an embodiment of a distribution network planning scheme generation method based on a large model provided in this application.

[0041] Figure 3 This is a schematic diagram of an embodiment of a power distribution network planning scheme generation device based on a large model, provided in this application.

[0042] Figure 4 This is a schematic diagram of the structure of a terminal embodiment for generating a distribution network planning scheme based on a large model, as provided in this application. Detailed Implementation

[0043] This application provides a method, device, terminal, and medium for generating distribution network planning schemes based on large models, which is used to solve the technical problem of low efficiency in existing distribution network planning work.

[0044] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the embodiments described below are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0045] First, a detailed description of an embodiment of a distribution network planning scheme generation method based on a large model provided in this application is as follows:

[0046] Please see Figure 1 and Figure 2 This application provides an embodiment of a method for generating distribution network planning schemes based on large models, including:

[0047] Step 101: Obtain sample data for power distribution network planning;

[0048] It should be noted that, firstly, data related to the power distribution network is collected from various fields, mainly including: power system operation data, historical planning data, geospatial data (GIS), and environmental data. Data sources can include power companies, sensors, satellite remote sensing, and weather stations.

[0049] Furthermore, after acquiring data in step 101 and before executing step 102, the following steps can be added:

[0050] 1011. Preprocess the sample data of the power distribution network planning;

[0051] It should be noted that the collected original power distribution network planning sample data... Preprocessing is performed, including data cleaning, normalization, and noise reduction. The normalization process uses the Z-Score method to standardize each data point to ensure data consistency and comparability.

[0052]

[0053] In the formula, Indicates from the The first in the field Types of variable data, Indicates from the The first in the field The average value of the variable data, Indicates from the The first in the field The variance of the variable data This represents the standardized variable data.

[0054] Step 102: Based on the distribution network planning sample data, and combined with the natural language constraints and mathematical expression constraints associated with the distribution network planning constraints, construct a set of constraint embedding prompt words;

[0055] It should be noted that, based on the collected data, a constraint embedding prompt set project is constructed. The constraint embedding prompts mentioned in this embodiment are a technical tool that structurally encodes multi-dimensional constraints in distribution network planning, including physical, economic, and policy constraints, into a combination of natural language and mathematical formulas. This unified encoding of these multi-dimensional constraints into structured input includes human-understandable natural language descriptions and machine-calculateable mathematical formulas. Its core purpose is to achieve low data dependency through human-understandable descriptions and machine-calculateable rules. The expression for the constraint embedding prompts is as follows:

[0056]

[0057] In the formula, To constrain the identifier of the embedded prompt word set, This serves as a sample for power distribution network planning. It is a set of constraints. It is a natural language description constraint. It is a constraint expressed by mathematical formulas. These are weighting coefficients, and the weighting percentages in the formula can be dynamically adjusted as needed.

[0058] Step 103: Obtain distribution network planning demand information, extract information features through a preset large model, and obtain the planning task target information corresponding to the distribution network planning demand information;

[0059] It should be noted that by obtaining actual power distribution network planning demand information, information features are extracted through a pre-set large model to extract key planning information, namely planning task target information.

[0060] When using LoRA (Low-Rank Adaptation) for fine-tuning large models, the loss function can be set as follows:

[0061]

[0062]

[0063] in, Let Frobenius norm be the matrix. The first term... Represents mission loss. The task of inputting into a large model, This represents the predicted value output by the large model for task X. The first term represents the actual value of task X. The second term represents the regularization term, which prevents overfitting. This loss function is minimized. This allows us to obtain the optimal fine-tuning strategy.

[0064] The process before step 104 also includes:

[0065] Step 1031: Optimize the objective function according to the planning scheme, and fine-tune the weight parameters of the large model by combining the preset loss function and using LoRA optimization.

[0066] More specifically, fine-tuning the weight parameters of a large model includes:

[0067] The weight parameter matrix of the large model is decomposed into two low-rank matrices, where the first low-rank matrix is ​​a Gaussian distribution matrix and the second low-rank matrix is ​​a zero matrix.

[0068] One low-rank matrix is ​​fixed in sequence, and the other low-rank matrix is ​​iteratively adjusted by solving the least squares problem.

[0069] When the change in loss reaches the preset convergence condition or the number of iterations reaches the preset maximum number of iterations threshold, the updated weight parameter matrix is ​​obtained by multiplying the two low-rank matrices and summing the weight parameter matrix.

[0070] The specific steps are as follows:

[0071] a. Parameter initialization: Assume the weight matrix of the pre-trained model is... ,in It is the input dimension. This is the output dimension. Define two low-rank matrices. ,in It is a low-rank dimension.

[0072] Random initialization: matrix Initialize using a Gaussian distribution (mean 0, standard deviation ). ,matrix Initialize to a zero matrix.

[0073]

[0074] Domain knowledge initialization: If the target task is related to the pre-training task, it can be initialized using partial information from the pre-training weights. and .

[0075] b. Iterative optimization: Repeat the following steps until convergence: Fixed ,renew According to the current The estimated value is updated by solving a least-squares problem to minimize the objective function.

[0076]

[0077] fixed ,renew :

[0078]

[0079] in It's the learning rate, usually set to... ~ , , for right , The gradient.

[0080] c. Convergence Criterion: Determine whether the objective function has reached the convergence condition: continuous loss change. or relative loss change rate Or it may reach the maximum number of iterations.

[0081] d. Output: The final updated weight matrix is ​​obtained. :

[0082]

[0083] Retain most of the parameters of the pre-trained model, only replace or add them. and Used for target task inference. Supports dynamic switching between different tasks. and This enables multi-task adaptation, and therefore, it can also determine the corresponding task types based on pre-planned task samples. and This allows for direct reuse when facing the same distribution network planning tasks in practice, further improving the efficiency of planning work.

[0084] LoRA fine-tuning can significantly reduce computing resource consumption in power grid planning. It only requires adjusting key parameters to adapt to different scenarios, ensuring that the planning scheme is both flexible and innovative and in line with actual engineering constraints, avoiding decision-making biases caused by over-reliance on data or experience.

[0085] Step 104: Based on the planning task objective information, determine the corresponding target constraint embedding prompts and the planning scheme optimization objective function. Then, based on the target constraint embedding prompts and the planning scheme optimization objective function, generate the distribution network planning scheme through the reasoning calculation of the large model.

[0086] It should be noted that, based on the planning task objective information obtained in the previous steps, the corresponding planning scheme optimization objective function is determined. Furthermore, by combining the constraint embedding prompt word set project constructed in the previous steps, target constraint embedding prompt words associated with the planning task objective information are matched. For example, the planning requirements proposed in the obtained distribution network planning document include three optimization objectives: total investment cost, average power supply availability, and carbon emissions. Based on these three optimization objectives, the overall optimization objective function can be preliminarily determined. An example of the optimization objective function is as follows:

[0087]

[0088] In the formula, Indicates the planning scheme, These represent different sub-objective functions. The total investment cost, The unit capacity purchase cost of the equipment. The unit annual operation and maintenance cost of the equipment; For the equipment's uptime, For average power availability, Indicates carbon emissions. Carbon emission intensity per unit of electricity generated Carbon emissions per unit capacity Indicates the amount of electricity generated. Indicates the construction capacity. Indicates the first The weight of each sub-objective.

[0089] Then, by combining the pre-built constraint embedding prompt set project, the constraint embedding prompts corresponding to the three optimization objectives are matched, so that the mathematical constraints in the constraint embedding prompts can be used as the constraint conditions associated with the optimization objective function, thereby forming a complete optimization model equation containing the objective function and constraint conditions. Based on the optimization model equation, the power distribution network planning scheme is generated by using the operation mechanism of the large model.

[0090] Furthermore, when the objective function constructed in step 104 is a multi-objective optimization function as provided in the example above, multi-objective optimization can be performed using the Pareto dominance relation. The Pareto dominance relation is a core concept in multi-objective optimization, used to compare the merits of two solutions. It is defined as: [Solution...] Dominant Solution (recorded as) If and only if No inferior to in all objectives (Right now , ), and is strictly superior to at least one objective. (Right now , This relationship, through dual conditions, ensures the overall superiority of the solution and forms the basis for selecting Pareto optimal solutions. By introducing the Pareto dominance relation, the planning model can screen out the non-dominated solution set, i.e., the Pareto front, thereby providing decision-makers with a series of optimal trade-offs and helping to improve the execution efficiency of distribution network planning.

[0091] Furthermore, after generating the distribution network planning scheme, the generated planning scheme can be evaluated by power flow calculation and simulation to determine whether it meets the actual engineering feasibility. For schemes that do not meet the conditions, they can be returned to the large model for re-optimization, so that the whole process forms a generation-feedback closed-loop optimization.

[0092] This application proposes a constraint-embedded mathematical-linguistic hybrid expression method. Through constraint embedding prompt word engineering, it achieves seamless compatibility between physical rules and large-scale language models (LLMs). It unifies human-understandable natural language descriptions and machine-computable mathematical rules across modalities. Based on a pre-constructed constraint embedding prompt word set, a pre-trained large model can be guided by natural language prompts to generate planning schemes that meet actual engineering needs through natural language constraints. This avoids the tedious process of manually converting natural language planning requirements into mathematical formulas, ensuring the rationality and accuracy of the model's generated results under physical rule constraints. Simultaneously, a low-rank adaptive multi-objective optimization strategy is employed, combined with LoRA fine-tuning technology and Pareto ranking, significantly reducing computational complexity while balancing the efficiency and accuracy of multi-objective optimization, thus improving the overall performance of the model.

[0093] The above is a detailed description of an embodiment of a distribution network planning scheme generation method based on a large model provided in this application. The following is a detailed description of an embodiment of a distribution network planning scheme generation device based on a large model provided in this application.

[0094] Please see Figure 3 This application provides an embodiment of a distribution network planning scheme generation device based on a large model, comprising:

[0095] Data acquisition unit 201 is used to acquire distribution network planning sample data;

[0096] The constraint prompt word set construction unit 202 is used to construct a constraint embedding prompt word set based on the distribution network planning sample data and the natural language constraints and mathematical expression constraints associated with the distribution network planning constraints.

[0097] The planning target determination unit 203 is used to obtain distribution network planning demand information, extract information features through a preset large model, and obtain planning task target information corresponding to the distribution network planning demand information.

[0098] The planning scheme generation unit 204 is used to determine the corresponding target constraint embedding prompts and planning scheme optimization objective functions based on the planning task objective information. It generates a distribution network planning scheme through reasoning and calculation of the large model based on the target constraint embedding prompts and planning scheme optimization objective functions.

[0099] like Figure 4 As shown, this application provides an embodiment of a distribution network planning scheme generation terminal based on a large model. The terminal implementation types include, but are not limited to, personal computers, industrial computers, servers, and embedded intelligent devices. The main components of the terminal include: a memory 33 and a processor 31, wherein the memory 33 and the processor 31 can be connected through a communication bus 34.

[0100] The memory 33 is used to store program code, which is used to implement a method for generating a power distribution network planning scheme based on a large model as provided in the above embodiments;

[0101] Processor 31 is used to read and execute program code.

[0102] This application also provides an embodiment of a computer-readable storage medium, in which program code is stored, which is read and executed by a processor to implement a method for generating a distribution network planning scheme based on a large model as provided in the above embodiment.

[0103] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the terminals, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0104] In the several embodiments provided in this application, it should be understood that the disclosed terminals, devices, and methods can be implemented in other ways. For example, the device 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 an indirect coupling or communication connection between devices or units through some interfaces, and may be electrical, mechanical, or other forms.

[0105] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0106] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0107] 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.

[0108] 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.

[0109] 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 described in 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.

[0110] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 this application.

Claims

1. A method for generating distribution network planning schemes based on a large model, characterized in that, include: Obtain sample data for power distribution network planning; Based on the power distribution network planning sample data, and combined with the natural language constraints and mathematical expression constraints associated with the power distribution network planning constraints, a set of constraint embedding prompt words is constructed. Obtain distribution network planning demand information, extract information features through a preset large model, and obtain planning task target information corresponding to the distribution network planning demand information; Based on the planning task objective information, the corresponding objective constraint embedding prompts and the planning scheme optimization objective function are determined. Based on the objective constraint embedding prompts and the planning scheme optimization objective function, the power distribution network planning scheme is generated through the reasoning operation of the large model. The expression for embedding the constraint prompt word set is specifically as follows: ； In the formula, The identifier for the constraint embedding prompt word set, This serves as a sample for power distribution network planning. It is a set of constraints. It is a natural language description constraint. It is a constraint expressed by mathematical formulas. These are the weighting coefficients.

2. The method for generating distribution network planning schemes based on a large model according to claim 1, characterized in that, Before generating the distribution network planning scheme by optimizing the objective function based on the embedded prompts of the target constraints and the planning scheme, and through the inference calculations of the large model, the following steps are also included: Based on the planning scheme, the objective function is optimized, and combined with the preset loss function, the weight parameters of the large model are fine-tuned through LoRA optimization.

3. The method for generating a distribution network planning scheme based on a large model according to claim 2, characterized in that, The fine-tuning of the weight parameters of the large model specifically includes: The weight parameter matrix of the large model is decomposed into two low-rank matrices, where the first low-rank matrix is ​​a Gaussian distribution matrix and the second low-rank matrix is ​​a zero matrix. One low-rank matrix is ​​fixed in sequence, and the other low-rank matrix is ​​iteratively adjusted by solving the least squares problem. When the change in loss reaches the preset convergence condition or the number of iterations reaches the preset maximum number of iterations threshold, the updated weight parameter matrix is ​​obtained by multiplying the two low-rank matrices and summing the weight parameter matrix.

4. The method for generating a distribution network planning scheme based on a large model according to claim 2, characterized in that, The loss function is: ； ； In the formula, This is the total loss coefficient. Represents the task loss coefficient. This serves as a sample for power distribution network planning. This refers to the predicted output value corresponding to the sample of the power distribution network planning scheme. Let A and B be the actual values ​​corresponding to the sample of the power distribution network planning scheme, and let A and B be two low-rank matrices.

5. The method for generating distribution network planning schemes based on a large model according to claim 1, characterized in that, The step of optimizing the objective function based on the embedded prompts and the planning scheme according to the target constraints, and generating the distribution network planning scheme through the inference calculation of the large model, includes: The objective function is optimized based on the embedded prompts and the planning scheme according to the target constraints, and several planning scheme solutions are obtained through the reasoning operation of the large model; By using Pareto dominance logic, the solution of the planning scheme is optimized to obtain the Pareto optimal solution, and the distribution network planning scheme corresponding to the Pareto optimal solution is determined.

6. The method for generating a distribution network planning scheme based on a large model according to claim 1, characterized in that, After obtaining the distribution network planning sample data, the following is also included: The power distribution network planning sample data is preprocessed, wherein the preprocessing includes: data cleaning, normalization and noise reduction.

7. A device for generating distribution network planning schemes based on a large model, characterized in that, include: The data acquisition unit is used to acquire sample data for power distribution network planning. The constraint prompt word set construction unit is used to construct a constraint embedding prompt word set based on the power distribution network planning sample data and the natural language constraints and mathematical expression constraints associated with the power distribution network planning constraints. The planning target determination unit is used to obtain distribution network planning demand information, and extract information features through a preset large model to obtain planning task target information corresponding to the distribution network planning demand information; The planning scheme generation unit is used to determine the corresponding target constraint embedding prompts and the planning scheme optimization objective function based on the planning task objective information, and generate a distribution network planning scheme through the reasoning operation of the large model based on the target constraint embedding prompts and the planning scheme optimization objective function. The expression for embedding the constraint prompt word set is specifically as follows: ； In the formula, The identifier for the constraint embedding prompt word set, This serves as a sample for power distribution network planning. It is a set of constraints. It is a natural language description constraint. It is a constraint expressed by mathematical formulas. These are the weighting coefficients.

8. A distribution network planning scheme generation terminal based on a large model, characterized in that, include: memory and processor; The memory is used to store program code, which is used to implement the method for generating a distribution network planning scheme based on a large model as described in any one of claims 1 to 6; The processor is used to read and execute the program code.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that is read and executed by a processor to implement a method for generating a distribution network planning scheme based on a large model as described in any one of claims 1 to 6.

Images