A method for evaluating load transfer strategies in distribution systems based on expert preference alignment

Abstract

This invention provides a method, apparatus, equipment, and medium for evaluating load transfer strategies in power distribution systems based on expert preference alignment. It relates to the field of power distribution system fault technology. The method involves constructing a training set for an evaluation model and, through a pre-defined large language model, jointly constructing a candidate set of evaluation criteria with experts. Based on this candidate set, an assertion set for the evaluation model is obtained using the large language model. An expert assertion set is then acquired after experts perform manual binarization evaluation of the training set. Based on the consistency measure between the expert assertion set and the evaluation model assertion set, the optimal evaluation criterion is selected from the candidate set. Finally, the load transfer strategy to be evaluated is evaluated using the optimal evaluation criterion. This invention overcomes the subjectivity and inefficiency of traditional evaluation methods, automates and intelligentizes the evaluation process, improves the accuracy, objectivity, and reliability of strategy evaluation, and provides strong support for the safe and stable operation of power distribution systems.

Description

Technical Field

[0001] This invention relates to the field of power distribution system fault technology, and specifically to a method, apparatus, equipment and medium for evaluating power distribution system load transfer strategies based on expert preference alignment. Background Technology

[0002] With economic growth and improved living standards, the stable operation of power systems has become particularly crucial. Research on distribution network fault recovery and power supply reliability improvement not only helps to quickly restore power and reduce outage losses, but also prevents faults by optimizing maintenance plans. However, current methods rely too heavily on the experience of operators and lack economic efficiency and reliability guarantees. The combination of power grid technology and large language modeling technology has become a major trend in future power grid development. However, how to effectively and accurately evaluate the output quality of these models has become a new challenge. The core of evaluating large language modeling technology lies in the subjectivity of standard setting, the complexity of evaluation, and the interpretability of results. We aim to construct an efficient and reliable evaluation system by combining automated and manual verification methods, while paying attention to and effectively managing the "standard drift" phenomenon in the evaluation process to ensure the long-term adaptability and accuracy of the evaluation tool. Therefore, there is an urgent need to design an evaluation tool to help experts generate and evaluate the output of large language modeling technology.

[0003] The application of machine learning algorithms in evaluating load transfer strategies in power distribution systems is primarily based on data-driven methods. It utilizes historical data to predict and assess the effectiveness of different load transfer strategies. The basic idea is as follows: First, relevant data is collected from the power distribution system, including load data, equipment status, and weather conditions. The collected data is then processed and transformed to extract representative features. These features may include historical load curves, peak electricity consumption times, and seasonal characteristics, in order to better capture the impact of load transfer strategies. However, traditional machine learning algorithms struggle to utilize power sector dispatching regulations and are prone to getting trapped in local optima. Summary of the Invention

[0004] The purpose of this invention is to provide a method, apparatus, equipment, and medium for evaluating load transfer strategies in power distribution systems that aligns with expert preferences. This invention addresses the problems of difficulty in quantifying expert experience, low evaluation efficiency, and strong subjectivity in the existing load transfer strategy evaluation process. It enables automated and intelligent evaluation of load transfer strategies, accurately and efficiently evaluating load transfer strategies and achieving a high degree of consistency with expert preferences.

[0005] To achieve the above objectives, this invention provides a method for evaluating load transfer strategies in distribution systems based on expert preference alignment, comprising:

[0006] The load transfer strategy to be evaluated interacts with the simulation environment to generate a fault recovery result report, and together with the corresponding load transfer strategy, constructs an evaluation model training set.

[0007] The text of the regulatory procedure is semantically understood by a pre-set large language model, generating multiple text-based evaluation criteria, which together with the text-based evaluation criteria provided by experts constitute a candidate set of evaluation criteria.

[0008] Based on the candidate set of evaluation criteria, the training set of the evaluation model is binarized and evaluated using a large language model to obtain the assertion set of the evaluation model.

[0009] Obtain the set of expert assertions generated after experts manually binarize the training set of the evaluation model;

[0010] Based on the consistency measure between the expert assertion set and the evaluation model assertion set, the optimal evaluation standard is selected from the candidate set of evaluation standards.

[0011] The load transfer strategy to be evaluated is evaluated using the optimal evaluation criteria.

[0012] According to the present invention, a load transfer strategy evaluation method for distribution systems based on expert preference alignment interacts with a simulation environment to generate a fault recovery result report, and constructs an evaluation model training set together with the corresponding load transfer strategy, including:

[0013] In the simulation environment, a fault scenario is constructed, and the load transfer strategy to be evaluated is loaded and executed. The operation data, fault recovery process and performance evaluation data of the power distribution system are obtained to form a fault recovery result report. The fault recovery result report and the corresponding load transfer strategy are used together to construct the evaluation model training set.

[0014] According to the present invention, a load transfer strategy evaluation method for power distribution systems based on expert preference alignment is provided. The control procedure text includes the operation procedure, fault handling process, safety specifications, and technical standards of the power distribution system.

[0015] According to the present invention, a load transfer strategy evaluation method for distribution systems oriented towards expert preference alignment is provided. Based on a candidate set of evaluation criteria, the evaluation model training set is binarized and evaluated using a large language model to obtain an evaluation model assertion set, including:

[0016] For each evaluation criterion in the candidate set of evaluation criteria, input each load transfer strategy and the corresponding fault recovery result report from the evaluation model training set into the large language model;

[0017] The large language model is used to determine whether the load transfer strategy meets the evaluation criteria, and the binary result of whether it meets or does not is output.

[0018] The binarized results of all load transfer strategies under all evaluation criteria are summarized to form the assertion set of the evaluation model.

[0019] According to the present invention, a load transfer strategy evaluation method for distribution systems based on expert preference alignment is used to select the optimal evaluation criterion from a candidate set of evaluation criteria based on a consistency metric between the expert assertion set and the evaluation model assertion set. The method includes:

[0020] Statistical methods were used to calculate the coverage and false rejection rate of the evaluation model assertion set and the expert assertion set;

[0021] The consistency index between the model assertion set and the expert assertion set is obtained based on the coverage and false rejection rate.

[0022] The candidate set of evaluation criteria is sorted according to the consistency index, and the evaluation criterion with the highest consistency index is selected as the optimal evaluation criterion.

[0023] According to the present invention, a method for evaluating load transfer strategies in a distribution system based on expert preference alignment is provided, wherein the expression for the consistency index is:

[0024]

[0025]

[0026] In the formula, Represents a set of assertions; for F coverage, for F False rejection rate for F Consistency indicators; For indicator functions; if targeting load transfer strategies If the expert believes that it does not meet the evaluation criteria provided by the expert, and the large language model also believes that it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe that it does not meet the evaluation criteria provided by the experts, then =1; if targeting load transfer strategy If the experts believe it meets one of the evaluation criteria provided by the experts, and the large language model believes it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe it meets the evaluation criteria provided by the experts, then =1.

[0027] Secondly, the present invention provides an evaluation device for load transfer strategies in a power distribution system oriented towards expert preference alignment, comprising:

[0028] The simulation interaction unit is used to interact with the load transfer strategy to be evaluated and the simulation environment, generate a fault recovery result report, and jointly construct an evaluation model training set with the corresponding load transfer strategy.

[0029] The evaluation criteria generation unit is used to perform semantic understanding of the regulatory procedure text through a pre-set large language model, generate multiple text-based evaluation criteria, and form a candidate set of evaluation criteria together with the text-based evaluation criteria provided by experts.

[0030] The first acquisition unit is used to perform binarization evaluation of the evaluation model training set based on the evaluation criterion candidate set and through the large language model to obtain the evaluation model assertion set.

[0031] The second acquisition unit is used to acquire the expert assertion set formed after the experts perform manual binarization evaluation of the training set of the evaluation model.

[0032] The screening unit is used to select the optimal evaluation criteria from the candidate set of evaluation criteria based on the consistency measure between the expert assertion set and the evaluation model assertion set.

[0033] The evaluation unit is used to evaluate the load transfer strategy to be evaluated using the optimal evaluation criteria.

[0034] According to the present invention, a load transfer strategy evaluation device for a power distribution system oriented towards expert preference alignment is provided, wherein the simulation interaction unit is specifically used for:

[0035] In the simulation environment, a fault scenario is constructed, and the load transfer strategy to be evaluated is loaded and executed. The operation data, fault recovery process and performance evaluation data of the power distribution system are obtained to form a fault recovery result report. The fault recovery result report and the corresponding load transfer strategy are used together to construct the evaluation model training set.

[0036] According to the present invention, a load transfer strategy evaluation device for a power distribution system oriented towards expert preference alignment is provided. The control procedure text includes the operation procedure, fault handling process, safety specifications, and technical standards of the power distribution system.

[0037] According to the present invention, a load transfer strategy evaluation device for a power distribution system oriented towards expert preference alignment is provided, wherein the first acquisition unit is specifically used for:

[0038] For each evaluation criterion in the candidate set of evaluation criteria, input each load transfer strategy and the corresponding fault recovery result report from the evaluation model training set into the large language model;

[0039] The large language model is used to determine whether the load transfer strategy meets the evaluation criteria, and the binary result of whether it meets or does not is output.

[0040] The binarized results of all load transfer strategies under all evaluation criteria are summarized to form the assertion set of the evaluation model.

[0041] According to the present invention, a load transfer strategy evaluation device for a power distribution system is provided, wherein the screening unit is specifically used for:

[0042] Statistical instruments were used to calculate the coverage and false rejection rate of the evaluation model assertion set and the expert assertion set.

[0043] The consistency index between the model assertion set and the expert assertion set is obtained based on the coverage and false rejection rate.

[0044] The candidate set of evaluation criteria is sorted according to the consistency index, and the evaluation criterion with the highest consistency index is selected as the optimal evaluation criterion.

[0045] According to the present invention, a load transfer strategy evaluation device for a distribution system oriented towards expert preference alignment is provided, wherein the expression for the consistency index is:

[0046]

[0047]

[0048] In the formula, Represents a set of assertions; for F coverage, for F False rejection rate for F Consistency indicators; For indicator functions; if targeting load transfer strategies If the expert believes that it does not meet the evaluation criteria provided by the expert, and the large language model also believes that it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe that it does not meet the evaluation criteria provided by the experts, then =1; if targeting load transfer strategy If the experts believe it meets one of the evaluation criteria provided by the experts, and the large language model believes it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe it meets the evaluation criteria provided by the experts, then =1.

[0049] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the first aspect of the method for evaluating load transfer strategies of a power distribution system oriented towards expert preference alignment.

[0050] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the expert preference-aligned power distribution system load transfer strategy evaluation method of the first aspect.

[0051] This invention provides a method, apparatus, equipment, and medium for evaluating load transfer strategies in power distribution systems based on expert preference alignment. It realizes an evaluation method for load transfer strategies in power distribution systems based on expert preference alignment, effectively integrating expert experience and knowledge into the strategy evaluation process. This overcomes the subjectivity and inefficiency problems of traditional evaluation methods, automates and intelligentizes the evaluation process, and improves the accuracy, objectivity, and reliability of strategy evaluation, providing strong support for the safe and stable operation of power distribution systems.

[0052] Compared with the prior art, the present invention has the following advantages:

[0053] 1. Improve evaluation efficiency: By combining the powerful semantic understanding capabilities of large language models with expert knowledge, the evaluation of load transfer strategies is automated, reducing human intervention and improving evaluation efficiency.

[0054] 2. Enhance the accuracy of evaluation: By selecting evaluation criteria that are highly consistent with expert assertions, the evaluation results are ensured to be consistent with expert preferences, thereby improving the accuracy and reliability of the evaluation.

[0055] 3. Quantify expert knowledge: Transform experts’ experience and knowledge into quantifiable evaluation standards, solving the problem that expert experience is difficult to quantify in traditional methods, and facilitating the inheritance and sharing of knowledge.

[0056] 4. Improve model generalization ability: By continuously optimizing the evaluation model, collecting more fault recovery cases, and expanding the training set of the evaluation model, the applicability and generalization ability of the model under different fault scenarios and load transfer strategies have been improved.

[0057] 5. Promote human-machine collaboration: This invention realizes a strategy evaluation method for human-machine collaboration, which not only gives full play to the efficiency of artificial intelligence, but also retains the professional judgment of experts, thus ensuring the scientificity and rationality of the evaluation results.

[0058] 6. Strong applicability: This invention is not only applicable to the evaluation of load transfer strategies in power distribution systems, but can also be extended to fields such as power distribution network planning scheme evaluation, distributed power source access strategy evaluation, load management and demand response strategy evaluation. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in this 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 some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0060] In the attached diagram:

[0061] Figure 1 This is a flowchart of the power distribution system load transfer strategy evaluation method oriented towards expert preference alignment according to the present invention;

[0062] Figure 2 This is a structural block diagram of the power distribution system load transfer strategy evaluation device oriented towards expert preference alignment according to the present invention;

[0063] Figure 3 This is a schematic diagram of the structure of the electronic device of the present invention. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0065] The following detailed description of some embodiments of the present invention will be provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other. The order of steps described in the embodiments is merely illustrative and should not be considered as a limitation. Those skilled in the art can adjust the order of steps without compromising logic.

[0066] Example 1

[0067] Please see Figure 1 This embodiment provides a method for evaluating load transfer strategies in distribution systems that is aligned with expert preferences. The method includes the following steps:

[0068] S101. Interact with the load transfer strategy to be evaluated with the simulation environment to generate a fault recovery result report, and jointly construct an evaluation model training set with the corresponding load transfer strategy.

[0069] Specifically, load transfer strategies are used for power distribution system fault recovery. However, before use, they need to be evaluated to predict and assess their effectiveness. S101 specifically includes: constructing various possible fault scenarios in a simulation environment to simulate the operation of the power distribution system under different fault scenarios; loading and executing the load transfer strategy to be evaluated to simulate the entire power distribution system fault recovery process; acquiring the power distribution system's operating data, fault recovery process, and performance evaluation data to form a fault recovery result report; and constructing an evaluation model training set by combining the fault recovery result report and the corresponding load transfer strategy. Each sample in the evaluation model training set contains the load transfer strategy itself and the fault recovery result report, providing complete data support for subsequent evaluation processes.

[0070] The operational data includes key parameters such as voltage, current, power, and frequency, as well as status information such as switch actions and protection actions. Performance evaluation data includes key indicators such as fault isolation time, total power restoration time, restored load capacity, unrestored load capacity, distribution system loss changes, node voltage deviation, and equipment load rate.

[0071] Total power restoration time: The time from the occurrence of a fault to the restoration of power to all recoverable loads.

[0072] Restored load capacity: The total load capacity for which power supply was successfully restored.

[0073] Unrestored load capacity: A list of loads and their capacity for which power cannot be restored.

[0074] Changes in power distribution system losses: A comparison of power distribution system losses before and after the implementation of load shifting strategies.

[0075] Node voltage deviation: Whether the load node voltage is within the allowable range.

[0076] Equipment load rate: Whether the transfer of power will cause the line or equipment to be overloaded.

[0077] S102. Semantically understand the regulatory procedure text through a pre-set large language model, generate multiple text-based evaluation criteria, and combine them with the text-based evaluation criteria provided by experts to form a candidate set of evaluation criteria.

[0078] Specifically, the preset large language model can be a pre-trained large language model. The control procedure text includes the operation procedures, fault handling procedures, safety specifications, and technical standards of the power distribution system. Before S102, the control procedure text can also be pre-processed to remove noise and redundant information, such as irrelevant comments and formatting information, and extract the effective text content.

[0079] Specifically, in S102, using large language model semantic analysis technology, key judgment criteria related to the evaluation of load transfer strategies for fault recovery are extracted, generating multiple text-based judgment criteria for fault recovery load transfer strategies (hereinafter referred to as evaluation criteria). These evaluation criteria cover aspects such as security, reliability, and economy.

[0080] In this embodiment, the evaluation criteria for large language model generation include:

[0081] Safety standards: whether the operating parameters such as current and voltage are within safe ranges, and whether equipment overload or overheating is avoided.

[0082] Reliability standards: whether power restoration time is minimized and whether power is restored to critical loads first.

[0083] Economic criteria: whether it reduces network losses in the power distribution system and whether it reduces operating costs.

[0084] Specifically, in S102, experts, based on their practical experience and professional knowledge, summarized and refined a series of strategy evaluation criteria, such as constraints on certain key indicators and standardized requirements for operational procedures, providing evaluation standards in text form. The evaluation criteria generated by the large language model are then merged with the evaluation criteria provided by the experts to form a candidate set of evaluation criteria. This candidate set contains multiple possible evaluation criteria, providing a basis for subsequent screening.

[0085] S103. Based on the candidate set of evaluation criteria, the training set of the evaluation model is binarized and evaluated using a large language model to obtain the assertion set of the evaluation model.

[0086] Specifically, S103 includes:

[0087] S1031. For each evaluation criterion in the candidate set of evaluation criteria, input each load transfer strategy and corresponding fault recovery result report from the evaluation model training set into the large language model.

[0088] S1032. Use a large language model to determine whether the load transfer strategy meets the evaluation criteria, and output a binary result indicating whether it meets or does not meet the criteria.

[0089] S1033. Summarize the binarized results of all load transfer strategies under all evaluation criteria to form an evaluation model assertion set. The evaluation model assertion set records the evaluation results of each load transfer strategy under each evaluation criterion.

[0090] S104. Obtain the expert assertion set formed after experts perform manual binarization evaluation on the training set of the evaluation model.

[0091] Specifically, experts, based on their professional knowledge and experience, comprehensively assess the rationality, safety, and effectiveness of load transfer strategies. Experts review each load transfer strategy and its corresponding fault recovery report in the evaluation model training set, assigning a "compliant" or "non-compliant" rating to each strategy, thus forming an expert assertion set. To ensure the objectivity and authority of the evaluation results, multiple experts can evaluate the same load transfer strategy, and the majority opinion or a weighted average of their evaluations can be used.

[0092] S105. Based on the consistency measure between the expert assertion set and the evaluation model assertion set, select the optimal evaluation standard from the candidate set of evaluation standards.

[0093] Specifically, S105 includes:

[0094] S1051. Using statistical methods, calculate the coverage and false rejection rate of the evaluation model assertion set and the expert assertion set.

[0095] S1052. Based on the coverage and false rejection rate, obtain the consistency index between the evaluation model assertion set and the expert assertion set.

[0096] The statistical methods used to calculate the relevant expressions for Coverage, False Rejection Rate (FFR), and Alignment metrics are as follows:

[0097]

[0098]

[0099] In the formula, This represents an assertion. Let... y Let be a binary vector, where {0,1} indicates that experts believe the load transfer strategy... Does it meet the evaluation criteria provided by experts (0 for non-compliance, 1 for compliance)? Assumptions It is a set of assertions (including) j (One assertion). For indicator functions; if targeting load transfer strategies If the expert believes that it does not meet the evaluation criteria provided by the expert, and the large language model also believes that it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe that it does not meet the evaluation criteria provided by the experts, then =1; if targeting load transfer strategy If the experts believe it meets one of the evaluation criteria provided by the experts, and the large language model believes it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe it meets the evaluation criteria provided by the experts, then =1.

[0100] S1053. Sort the candidate set of evaluation criteria according to the consistency index, and select the evaluation criteria with the highest consistency index as the optimal evaluation criteria. That is, select the evaluation criteria that are most consistent with the expert evaluation results and determine them as the optimal evaluation criteria.

[0101] In this embodiment, a threshold for the consistency index can be set, and only evaluation criteria with consistency index values ​​higher than the threshold can be selected. When multiple evaluation criteria meet the consistency index requirements, the interpretability, operability, and practical application needs of the evaluation criteria can be comprehensively considered to ultimately determine the optimal evaluation criterion.

[0102] S106. Evaluate the load transfer strategy to be evaluated using the optimal evaluation criteria.

[0103] Specifically, optimal evaluation criteria are used to automatically evaluate load transfer strategies. The load transfer strategies to be evaluated and fault recovery results reports are input into a large language model. Based on the optimal evaluation criteria, it is determined whether the load transfer strategy conforms to expert preferences, and the evaluation results are output. The evaluation results can also be fed back to decision-makers, providing a reference for fault recovery decisions in the power distribution system and improving the efficiency and accuracy of load transfer strategy evaluation.

[0104] Specifically, as the operating environment of the power distribution system changes and control regulations are updated, new evaluation standards may emerge, necessitating timely updates to the candidate set of evaluation standards. Furthermore, by incorporating the latest expert assertions, consistency indices should be recalculated to ensure the evaluation results remain aligned with expert preferences. Additionally, collecting more power distribution system fault recovery cases expands the training set of the evaluation model, improving the generalization ability of the evaluation method and enabling it to adapt to a wider range of fault scenarios and load transfer strategies.

[0105] Example 2

[0106] like Figure 2 As shown, this embodiment provides a load transfer strategy evaluation device for power distribution systems oriented towards expert preference alignment, including:

[0107] The simulation interaction unit 201 is used to interact with the load transfer strategy to be evaluated with the simulation environment, generate a fault recovery result report, and jointly construct an evaluation model training set with the corresponding load transfer strategy.

[0108] The evaluation criteria generation unit 202 is used to perform semantic understanding of the regulatory procedure text through a preset large language model, generate multiple text-based evaluation criteria, and together with the text-based evaluation criteria provided by experts, form a candidate set of evaluation criteria.

[0109] The first acquisition unit 203 is used to perform binarization evaluation of the evaluation model training set based on the evaluation criterion candidate set and through a large language model to obtain the evaluation model assertion set.

[0110] The second acquisition unit 204 is used to acquire the expert assertion set formed after the experts perform manual binarization evaluation of the training set of the evaluation model.

[0111] The screening unit 205 is used to select the optimal evaluation standard from the candidate set of evaluation standards based on the consistency measure between the expert assertion set and the evaluation model assertion set.

[0112] Evaluation unit 206 is used to evaluate the load transfer strategy to be evaluated using the optimal evaluation criteria.

[0113] Specifically, the simulation interaction unit 201 is used for:

[0114] In the simulation environment, a fault scenario is constructed, and the load transfer strategy to be evaluated is loaded and executed. The operation data, fault recovery process and performance evaluation data of the power distribution system are obtained to form a fault recovery result report. The fault recovery result report and the corresponding load transfer strategy are used together to construct the evaluation model training set.

[0115] The control and regulation documents include the power distribution system's operating procedures, fault handling procedures, safety specifications, and technical standards.

[0116] The first acquisition unit 203 is specifically used for:

[0117] For each evaluation criterion in the candidate set of evaluation criteria, input each load transfer strategy and the corresponding fault recovery result report from the evaluation model training set into the large language model;

[0118] The large language model is used to determine whether the load transfer strategy meets the evaluation criteria, and the binary result of whether it meets or does not is output.

[0119] The binarized results of all load transfer strategies under all evaluation criteria are summarized to form the assertion set of the evaluation model.

[0120] The filtering unit 205 is specifically used for:

[0121] Statistical instruments were used to calculate the coverage and false rejection rate of the evaluation model assertion set and the expert assertion set.

[0122] The consistency index between the model assertion set and the expert assertion set is obtained based on the coverage and false rejection rate.

[0123] The candidate set of evaluation criteria is sorted according to the consistency index, and the evaluation criterion with the highest consistency index is selected as the optimal evaluation criterion.

[0124] The expression for the consistency index is:

[0125]

[0126]

[0127] In the formula, Represents a set of assertions; for F coverage, for F False rejection rate for F Consistency indicators; For indicator functions; if targeting load transfer strategies If the expert believes that it does not meet the evaluation criteria provided by the expert, and the large language model also believes that it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe that it does not meet the evaluation criteria provided by the experts, then =1; if targeting load transfer strategy If the experts believe it meets one of the evaluation criteria provided by the experts, and the large language model believes it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe it meets the evaluation criteria provided by the experts, then =1.

[0128] Example 3

[0129] Figure 3 An example is a schematic diagram of the structure of an electronic device, such as... Figure 3 As shown, the electronic device may include a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other via the communication bus 840. The memory 830 stores a computer program that can run on the processor 810. When the processor 810 executes the computer program, it implements the expert preference-aligned power distribution system load transfer strategy evaluation method of Embodiment 1.

[0130] Example 4

[0131] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the expert preference-aligned power distribution system load transfer strategy evaluation method of Embodiment 1.

[0132] It should be noted that the computer-readable storage medium in this embodiment can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0133] In this embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this embodiment, the computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable program. This propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof.

[0134] The computer-readable storage medium described above can be used to write computer programs for executing this embodiment in one or more programming languages ​​or combinations thereof. These programming languages ​​include object-oriented programming languages—such as Java, Python, and C++—and conventional procedural programming languages—such as C or similar programming languages. The program can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0135] In summary, the method, apparatus, equipment, and medium for evaluating load transfer strategies in distribution systems based on expert preference alignment provided by this invention employ a large-model evaluation criterion screening design, effectively offering a new solution for evaluating large language models and overcoming the limitations of past fully automated methods in understanding complex expert needs and evaluation criteria. Ultimately, manual expert evaluation continuously improves the evaluation criteria and enhances the accuracy of the evaluation. With expert participation, fewer and more accurate assertion sets are generated, demonstrating higher evaluation consistency. This, to a certain extent, ensures expert preference calibration in load transfer strategy evaluation, significantly improving the relevance and practicality of the strategy evaluation and providing strong support for the optimized scheduling of distribution systems.

[0136] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the embodiments disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. It should be understood that the invention is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for evaluating load transfer strategies in power distribution systems based on expert preference alignment, characterized in that, include: The load transfer strategy to be evaluated interacts with the simulation environment to generate a fault recovery result report, and together with the corresponding load transfer strategy, constructs an evaluation model training set. The text of the regulatory procedure is semantically understood by a pre-set large language model, generating multiple text-based evaluation criteria, which together with the text-based evaluation criteria provided by experts constitute a candidate set of evaluation criteria. Based on the candidate set of evaluation criteria, the training set of the evaluation model is binarized and evaluated using the large language model to obtain the assertion set of the evaluation model. Obtain the expert assertion set formed after experts perform manual binarization evaluation on the training set of the evaluation model; Based on the consistency measure between the expert assertion set and the evaluation model assertion set, the optimal evaluation standard is selected from the candidate set of evaluation standards. The load transfer strategy to be evaluated is evaluated using the aforementioned optimal evaluation criteria.

2. The method for evaluating load transfer strategies in distribution systems based on expert preference alignment as described in claim 1, characterized in that, The process involves interacting with the load transfer strategy to be evaluated in the simulation environment to generate a fault recovery result report, and jointly constructing an evaluation model training set with the corresponding load transfer strategy, including: A fault scenario is constructed in the simulation environment, and the load transfer strategy to be evaluated is loaded and executed; the operation data, fault recovery process and performance evaluation data of the power distribution system are obtained to form the fault recovery result report; the fault recovery result report and the corresponding load transfer strategy are used together to construct the training set of the evaluation model.

3. The method for evaluating load transfer strategies in distribution systems based on expert preference alignment as described in claim 1, characterized in that, The control and regulation text includes the power distribution system's operation procedures, fault handling procedures, safety specifications, and technical standards.

4. The method for evaluating load transfer strategies in distribution systems based on expert preference alignment as described in claim 1, characterized in that, Based on the candidate set of evaluation criteria, the evaluation model training set is binarized and evaluated using the large language model to obtain the evaluation model assertion set, including: For each evaluation criterion in the candidate set of evaluation criteria, input each load transfer strategy and corresponding fault recovery result report from the training set of the evaluation model into the large language model; The large language model is used to determine whether the load transfer strategy meets the evaluation criteria, and a binary result of whether it meets or does not is output. The binarized results of all load transfer strategies under all evaluation criteria are summarized to form the assertion set of the evaluation model.

5. The method for evaluating load transfer strategies in distribution systems based on expert preference alignment as described in claim 1, characterized in that, The step of selecting the optimal evaluation criterion from the candidate evaluation criterion set based on the consistency measure between the expert assertion set and the evaluation model assertion set includes: Statistical methods were used to calculate the coverage and false rejection rate of the assertion set of the evaluation model and the expert assertion set. The consistency index between the evaluation model assertion set and the expert assertion set is obtained based on the coverage and false rejection rate. The candidate set of evaluation criteria is sorted according to the consistency index, and the evaluation criterion with the highest consistency index is selected as the optimal evaluation criterion.

6. The method for evaluating load transfer strategies in distribution systems based on expert preference alignment according to claim 5, characterized in that, The expression for the consistency index is: In the formula, Represents a set of assertions; for F coverage, for F False rejection rate for F Consistency indicators; For indicator functions; if targeting load transfer strategies If the expert believes that it does not meet the evaluation criteria provided by the expert, and the large language model also believes that it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe that it does not meet the evaluation criteria provided by the experts, then =1; if targeting load transfer strategy If the experts believe it meets one of the evaluation criteria provided by the experts, and the large language model believes it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe it meets the evaluation criteria provided by the experts, then =1.

7. A load transfer strategy evaluation device for power distribution systems oriented towards expert preference alignment, characterized in that, include: The simulation interaction unit is used to interact with the load transfer strategy to be evaluated and the simulation environment, generate a fault recovery result report, and jointly construct an evaluation model training set with the corresponding load transfer strategy. The evaluation criteria generation unit is used to perform semantic understanding of the regulatory procedure text through a pre-set large language model, generate multiple text-based evaluation criteria, and form a candidate set of evaluation criteria together with the text-based evaluation criteria provided by experts. The first acquisition unit is used to perform binarization evaluation of the evaluation model training set based on the evaluation criterion candidate set and through the large language model to obtain the evaluation model assertion set. The second acquisition unit is used to acquire the expert assertion set formed after the experts perform manual binarization evaluation on the training set of the evaluation model; A screening unit is used to select the optimal evaluation criterion from the candidate set of evaluation criteria based on the consistency measure between the expert assertion set and the evaluation model assertion set. The evaluation unit is used to evaluate the load transfer strategy to be evaluated using the optimal evaluation criteria.

8. The power distribution system load transfer strategy evaluation device oriented towards expert preference alignment according to claim 7, characterized in that, The simulation interaction unit is specifically used for: A fault scenario is constructed in a simulation environment, the load transfer strategy to be evaluated is loaded and executed, and the operation data, fault recovery process and performance evaluation data of the power distribution system are obtained to form the fault recovery result report. The fault recovery result report and the corresponding load transfer strategy are used together to construct the training set of the evaluation model.

9. The power distribution system load transfer strategy evaluation device oriented towards expert preference alignment according to claim 7, characterized in that, The control and regulation text includes the power distribution system's operation procedures, fault handling procedures, safety specifications, and technical standards.

10. The power distribution system load transfer strategy evaluation device oriented towards expert preference alignment according to claim 7, characterized in that, The first acquisition unit is specifically used for: For each evaluation criterion in the candidate set of evaluation criteria, input each load transfer strategy and corresponding fault recovery result report from the training set of the evaluation model into the large language model; The large language model is used to determine whether the load transfer strategy meets the evaluation criteria, and a binary result of whether it meets or does not is output. The binarized results of all load transfer strategies under all evaluation criteria are summarized to form the assertion set of the evaluation model.

11. The power distribution system load transfer strategy evaluation device oriented towards expert preference alignment according to claim 7, characterized in that, The filtering unit is specifically used for: Using statistical instruments, the coverage and false rejection rate of the assertion set of the evaluation model and the expert assertion set are calculated. The consistency index between the evaluation model assertion set and the expert assertion set is obtained based on the coverage and false rejection rate. The candidate set of evaluation criteria is sorted according to the consistency index, and the evaluation criterion with the highest consistency index is selected as the optimal evaluation criterion.

12. The power distribution system load transfer strategy evaluation device oriented towards expert preference alignment according to claim 11, characterized in that, The expression for the consistency index is: In the formula, Represents a set of assertions; for F coverage, for F False rejection rate for F Consistency indicators; For indicator functions; if targeting load transfer strategies If the expert believes that it does not meet the evaluation criteria provided by the expert, and the large language model also believes that it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe that it does not meet the evaluation criteria provided by the experts, then =1; if targeting load transfer strategy If the experts believe it meets one of the evaluation criteria provided by the experts, and the large language model believes it does not meet one of the evaluation criteria provided by the large language model, then... =1; if targeting load transfer strategy If the experts believe it meets the evaluation criteria provided by the experts, then =1.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the power distribution system load transfer strategy evaluation method oriented towards expert preference alignment as described in any one of claims 1 to 6.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the expert preference-aligned power distribution system load transfer strategy evaluation method as described in any one of claims 1 to 6.

Images