Method and system for generating feasibility research report

Abstract

The invention provides a feasibility research report generation method and system, and relates to the technical field of electric power report generation, and the method comprises the steps: obtaining the electric power requirement data of an electric power project; inputting the power requirement data into the trained report generation model to obtain a power feasibility research report; wherein the report generation model comprises an embedding generation unit and a report acquisition unit; the embedding generation unit is used for performing embedding processing on the power requirement data to obtain a plurality of requirement text embedding units; and the report acquisition unit is used for analyzing each required text embedding to obtain an electric power feasibility research report. The report generation model is subjected to fine tuning training through the electric power requirement data training data set marked with the electric power feasibility research report, so that the electric power feasibility research report can be directly obtained after the electric power requirement data is input into the report generation model, the efficiency of obtaining the electric power feasibility research report is improved, and the accuracy of the electric power feasibility research report is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of report generation, and in particular to a feasibility study report generation method and system. Background Art

[0002] With the continuous development of society, the power industry has become one of the pillar industries in modern society. With the continuous deepening of information construction in the power industry, higher requirements are put forward for the planning, construction and management of power information systems. Feasibility study reports (referred to as feasibility reports) are an important basis for decision-making in information projects. The quality of their preparation directly affects the successful implementation of the project. At present, the process of writing feasibility reports related to the power industry is usually completed by writers with professional knowledge of the power industry. However, even when writers with professional knowledge of the power industry write power feasibility reports, there is still a problem of low efficiency in writing power feasibility reports due to the complexity and tediousness of power industry knowledge. At the same time, since there may be omissions when personnel handle power industry knowledge, there is a problem of low quality of the written power feasibility reports. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a feasibility study report generation method and system, which can solve the technical problems in the prior art of low efficiency and low quality of power feasibility study reports when manually writing power feasibility study reports.

[0004] In order to achieve the above objectives, the technical solutions adopted in the embodiments of the present invention are as follows:

[0005] In a first aspect, an embodiment of the present invention provides a method for generating a feasibility study report, comprising:

[0006] Obtaining power requirement data for the power project; wherein the power requirement data includes: power load characteristics, voltage level requirements, power quality requirements and energy efficiency indicators;

[0007] The power requirement data is input into the trained report generation model to obtain a power feasibility report; wherein the report generation model includes an embedding generation unit and a report acquisition unit; the embedding generation unit is used to embed the power requirement data to obtain multiple requirement text embeddings; the report acquisition unit is used to analyze the embeddings of each requirement text to obtain the power feasibility report; the report generation model is obtained by fine-tuning and training a power requirement data training dataset labeled with a power feasibility report.

[0008] Furthermore, an embodiment of the present invention provides a first possible implementation of the first aspect, wherein the fine-tuning training step of the report generation model includes:

[0009] obtaining power industry data, and determining power requirement data for a plurality of power projects based on the power industry data;

[0010] Marking the power feasibility study reports corresponding to the power requirement data of each power project, and establishing a sample data set based on the marked power requirement data of each power project; wherein the sample data set includes a training data set;

[0011] The training data set is input into the large language model, and the large language model is fine-tuned based on the LoRA technology to obtain the report generation model after fine-tuning training.

[0012] Furthermore, an embodiment of the present invention provides a second possible implementation of the first aspect, wherein the step of fine-tuning the large language model based on LoRA technology to obtain the report generation model after fine-tuning includes:

[0013] Setting initial values ​​of training parameters of a large language model, and setting a low-rank decomposition matrix in the large language model;

[0014] The large language model is iteratively trained based on the training data set, and a loss function is calculated after each round of iteration. When the loss function is less than a preset function value, the training is stopped to obtain the report generation model.

[0015] Furthermore, the embodiment of the present invention provides a third possible implementation of the first aspect, wherein the sample data set also includes a test data set;

[0016] The feasibility study report generation method further includes:

[0017] Determining each of the actual power feasibility studies based on the power requirement data of each of the power projects corresponding to the test data set;

[0018] Inputting the test data set into the report generation model after fine-tuning and training to obtain a forecast power feasibility study report corresponding to each power requirement data;

[0019] Based on the predicted power feasibility report and the actual power feasibility report corresponding to each of the power requirement data, performance evaluation indicators are obtained to evaluate the performance of the report generation model after fine-tuning training; wherein, the performance evaluation indicators include: BELU indicator and ROUGE indicator.

[0020] In a second aspect, an embodiment of the present invention further provides a feasibility study report generation system, comprising:

[0021] Data acquisition module: used to obtain power requirement data of power projects; wherein the power requirement data includes: power load characteristics, voltage level requirements, power quality requirements and energy efficiency indicators;

[0022] Report generation module: used to input the power requirement data into the trained report generation model to obtain a power feasibility report; wherein, the report generation model includes an embedding generation unit and a report acquisition unit; the embedding generation unit is used to embed the power requirement data to obtain multiple requirement text embeddings; the report acquisition unit is used to analyze the embeddings of each requirement text to obtain the power feasibility report; the report generation model is obtained by fine-tuning and training the power requirement data training data set marked with the power feasibility report.

[0023] Furthermore, the embodiment of the present invention provides a first possible implementation of the second aspect, wherein the feasibility study report generation system further includes: a report management module;

[0024] The report management module is used to store the power feasibility study report and record historical versions of the power feasibility study report.

[0025] Furthermore, the embodiment of the present invention provides a second possible implementation of the second aspect, wherein the feasibility study report generation system further includes: a collaborative editing module;

[0026] The collaborative editing module is used to call the power feasibility study report in the report management module and perform multi-user collaborative editing on the power feasibility study report.

[0027] Furthermore, the embodiment of the present invention provides a third possible implementation of the second aspect, wherein the feasibility study report generation system further includes: a data analysis module;

[0028] The data analysis module is used to obtain power industry related data, analyze the power industry related data, and generate a power data analysis chart.

[0029] Furthermore, the embodiment of the present invention provides a fourth possible implementation of the second aspect, wherein the feasibility study report generation system further includes: a template storage module;

[0030] The template storage module includes a requirement analysis unit, a template storage unit and a template calling unit; wherein the template storage unit stores a plurality of power feasibility report templates; the requirement analysis unit is used to analyze the power requirement data and determine the optimal power feasibility report template from the template storage unit; the template calling unit is used to input the optimal power feasibility report template into the report generation module, so that the report generation module generates a power feasibility report in the format of the optimal power feasibility report template.

[0031] Furthermore, the embodiment of the present invention provides a fifth possible implementation of the second aspect, wherein the template storage module further includes: a template editing unit;

[0032] The template editing unit is used to enable a user to manually edit the power feasibility study report template stored in the template storage unit.

[0033] An embodiment of the present invention provides a feasibility study report generation method and system, the method comprising: obtaining power requirement data of a power project; wherein the power requirement data includes: power load characteristics, voltage level requirements, power quality requirements, and energy efficiency indicators; inputting the power requirement data into a trained report generation model to obtain a power feasibility study report; wherein the report generation model includes an embedding generation unit and a report acquisition unit; the embedding generation unit is used to embed the power requirement data to obtain multiple requirement text embeddings; the report acquisition unit is used to analyze each requirement text embedding to obtain a power feasibility study report; the report generation model is obtained by fine-tuning and training a power requirement data training dataset labeled with the power feasibility study report. The present invention fine-tunes and trains the report generation model using the power requirement data training dataset labeled with the power feasibility study report, so that the fine-tuned and trained report generation model has the ability to quickly and accurately output the power feasibility study report based on the power requirement data. By inputting the power requirement data of the power project into the trained report generation model to obtain the power feasibility study report, the efficiency of obtaining the power feasibility study report is improved and the quality of the obtained feasibility study report is ensured.

[0034] Other features and advantages of the embodiments of the present invention will be described in the following description, or some features and advantages can be inferred or determined without doubt from the description, or can be learned by implementing the above-mentioned technologies of the embodiments of the present invention.

[0035] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 A schematic diagram showing the steps of a method for generating a feasibility study report provided by an embodiment of the present invention is shown;

[0038] Figure 2A schematic diagram of units of a report generation model in a feasibility study report generation method provided by an embodiment of the present invention is shown;

[0039] Figure 3 A schematic diagram showing the main modules of a feasibility study report generation system provided by an embodiment of the present invention is shown;

[0040] Figure 4 A schematic diagram of all modules in a feasibility study report generation system provided by an embodiment of the present invention is shown;

[0041] Figure 5 A unit schematic diagram of a template storage module in a feasibility study report generation system provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0043] This embodiment provides a feasibility study report generation method and system. For details, see Figure 1 The diagram below shows a step-by-step process for generating a feasibility study report. The method mainly includes:

[0044] Step S101: Obtain power requirement data of a power project; wherein the power requirement data includes: power load characteristics, voltage level requirements, power quality requirements, and energy efficiency indicators;

[0045] The power requirement data of the power project that the user wants to build obtained above is helpful for adaptively generating a power feasibility report, wherein the power requirement data includes: power load characteristics, voltage level requirements, power quality requirements and energy efficiency indicators; specifically, power load characteristics are used to determine the main transformer capacity and reactive power compensation scheme in the power feasibility report; voltage level requirements are used to determine the grid access scheme and transformer ratio in the power feasibility report; power quality requirements are used to determine the filtering device and voltage stabilization system in the power feasibility report; energy efficiency indicators are used to determine the precedent optimization scheme in the power feasibility report; in addition to the above-mentioned power load characteristics, voltage level requirements, power quality requirements and energy efficiency indicators, the power requirement data also includes: technology iteration requirements, load growth requirements, special power requirements and environmental protection energy efficiency requirements.

[0046] Step S103: Input the power requirement data into the trained report generation model to obtain a power feasibility report; Figure 2A schematic diagram of the units of a report generation model in a feasibility study report generation method is shown, wherein the report generation model includes an embedding generation unit 21 and a report acquisition unit 22; the embedding generation unit 21 is used to embed power requirement data to obtain multiple requirement text embeddings; the report acquisition unit 22 is used to analyze each requirement text embedding to obtain a power feasibility study report; the report generation model is fine-tuned and trained using a power requirement data training dataset labeled with power feasibility study reports to obtain the requirement text embeddings;

[0047] The embedding generation unit 21 tokenizes the power requirement data to obtain multiple power requirement tags, and embeds each power requirement tag through the embedding layer to obtain multiple requirement text embeddings; the report acquisition unit 22 extracts features from each requirement text embedding through a neural network to obtain multiple feature representations, and normalizes each feature representation to obtain a power feasibility study report; the report generation model usually adopts a large language model, and uses the content completion algorithm of the large language model to complete and obtain a complete power feasibility study report through the acquired power requirement data.

[0048] The above-mentioned feasibility report generation method provided by an embodiment of the present invention fine-tunes the report generation model through a power requirement data training data set marked with a power feasibility report, so that the fine-tuned report generation model has the ability to quickly and accurately output a power feasibility report based on the power requirement data. By inputting the power requirement data of the power project into the trained report generation model, the power feasibility report is obtained, which improves the efficiency of obtaining the power feasibility report and ensures the quality of the obtained feasibility report.

[0049] In one embodiment, this embodiment provides a specific implementation method of fine-tuning the report generation model:

[0050] Obtain power industry data and determine power requirement data for multiple power projects based on the power industry data;

[0051] In the above steps, power industry data (including grid operation data, equipment data, policies and regulations, etc.) is obtained from power companies such as State Grid Corporation of China and China Southern Power Grid Corporation, and the power industry data is cleaned, converted and stored. The obtained power industry data is analyzed to determine the power requirement data of multiple power projects.

[0052] Marking the power feasibility study reports corresponding to the power requirement data of each power project, and establishing a sample data set based on the marked power requirement data of each power project; wherein the sample data set includes a training data set;

[0053] Input the training data set into the large language model, perform fine-tuning training on the large language model based on LoRA technology, and obtain a report generation model after fine-tuning training;

[0054] Low-Rank Adaptation of Large Language Models (LoRA) technology is a method of fine-tuning large language models with extremely low resources. It uses a Linear layer on one side of the original large language model to reduce the data from e dimension to f dimension, with the parameter being the matrix A initialized with random Gaussian. Then, a Linear layer is used to increase the data from f dimension to e dimension, with the parameter being the matrix B initialized with 0. d is the embedding dimension of the large language model, and r is a hyperparameter much smaller than d. The parameter matrix d*d on one side of the large language model is converted to the parameter matrix A*B on the other side. Since f is much smaller than e, the number of parameters in the training process of the large language model is greatly reduced. This can not only take advantage of the powerful ability of pre-training the large language model, but also avoid the massive resource consumption caused by fine-tuning all parameters. It can also control memory usage and reduce computing time and manpower investment.

[0055] In one embodiment, this embodiment provides a specific implementation method for fine-tuning a large language model based on LoRA technology to obtain a report generation model after fine-tuning training:

[0056] Set the initial values ​​of the training parameters of the large language model and set the low-rank decomposition matrix in the large language model;

[0057] Set the learning rate, training batch, training rounds, optimizer (usually AdamW optimizer), weight decay coefficient, etc. during the large language model training process, and add a low-rank decomposition matrix to the Transformer-specific layer in the large language model.

[0058] Iteratively train the large language model based on the training dataset, calculate the loss function after each iteration, and stop training when the loss function is less than the preset function value to obtain the report generation model;

[0059] The loss function is calculated based on the obtained power feasibility report after each round of iteration (using the comparative loss function).

[0060] In one embodiment, the sample data set provided in this embodiment also includes a test data set;

[0061] Feasibility study report generation methods also include:

[0062] Determine each actual power feasibility study report based on the power requirement data of each power project corresponding to the test data set;

[0063] The test data set is input into the fine-tuned report generation model to obtain the predicted power feasibility report corresponding to each power requirement data;

[0064] Based on the predicted and actual power feasibility reports corresponding to each power requirement data, performance evaluation indicators are obtained to evaluate the performance of the fine-tuned report generation model. The performance evaluation indicators include the BELU indicator and the ROUGE indicator.

[0065] The above-mentioned BELU indicator is mainly used to evaluate the degree of precise matching between the generated text (forecasted power feasibility report) and the reference text (actual power feasibility report), and to measure the fluency and accuracy of the generated results; the above-mentioned ROUGE indicator focuses on evaluating the coverage of key information of the generated text (forecasted power feasibility report) and the reference text (actual power feasibility report), and is widely used in tasks such as text summarization and question-answering generation. Specifically, the recall rate and precision rate of the large language model in the training process are determined by the actual power feasibility report and the forecasted power feasibility report, and the value of the BELU indicator is determined based on the precision rate. The value range of the obtained BELU indicator is 0-1 (including 0 and 1 at both ends). If the value of the obtained BELU indicator is closer to 1, the performance of the report generation model is better; the value of the ROUGE indicator is determined based on the recall rate. The value range of the obtained ROUGE indicator is 0-1 (including 0 and 1 at both ends). If the value of the obtained ROUGE indicator is closer to 1, the performance of the report generation model is better.

[0066] This embodiment also provides a feasibility study report generation system, for details, see Figure 3 The following is a schematic diagram of the main modules of a feasibility study report generation system, which mainly includes:

[0067] Data acquisition module 31: used to obtain power requirement data of power projects; wherein the power requirement data includes: power load characteristics, voltage level requirements, power quality requirements and energy efficiency indicators;

[0068] Report generation module 32: used to input the power requirement data into the trained report generation model to obtain a power feasibility report; wherein, the report generation model includes an embedding generation unit 21 and a report acquisition unit 22; the embedding generation unit 21 is used to embed the power requirement data to obtain multiple requirement text embeddings; the report acquisition unit 22 is used to analyze the embeddings of each requirement text to obtain a power feasibility report; the report generation model is obtained by fine-tuning and training the power requirement data training data set marked with the power feasibility report.

[0069] In one embodiment, see Figure 4 The schematic diagram of all modules in a feasibility study report generation system shown in FIG. The feasibility study report generation system provided in this embodiment further includes: a report management module 33;

[0070] The report management module 33 is used to store the power feasibility study report and record the historical versions of the power feasibility study report;

[0071] The above-mentioned report management module 33 supports the version history management of the power feasibility study report by adopting a distributed version control architecture. The report management module 33 maintains a complete version history tree for each power feasibility study report, recording the content, modifier and modification time of each modification of the power feasibility study report. At the same time, the report management module 33 adopts an incremental storage strategy, only saving the changed part of the power feasibility study report to reduce storage space occupancy. The report management module 33 supports version comparison, rollback and merge operations. Users can view the differences between any two versions of the power feasibility study report through the version history browser, and can choose to roll back the power feasibility study report to a specific version.

[0072] In one embodiment, the present embodiment provides a feasibility study report generation system, further comprising: a collaborative editing module 34;

[0073] The collaborative editing module 34 is used to call the power feasibility study report in the report management module and perform multi-user collaborative editing on the power feasibility study report;

[0074] The collaborative editing module 34 implements real-time collaborative editing of power feasibility reports based on the WebSocket protocol. It uses an operational transformation algorithm to handle concurrent editing operations, ensuring data consistency when multiple users are editing the report simultaneously. The collaborative editing module 34 supports up to 100 people editing the same power feasibility report online simultaneously, with editing delays of no more than 200 milliseconds. The collaborative editing module 34 assigns a different-colored cursor to each online user, displaying their editing position in real time. It also features a paragraph-level editing lock mechanism, which, when enabled, prevents users from editing the same paragraph simultaneously, preventing editing conflicts. Users can also discuss and coordinate through the built-in instant messaging function, improving team collaboration efficiency.

[0075] The collaborative editing module 34 is equipped with a semantic version difference comparison algorithm. First, it performs a structured analysis on the content of the power feasibility study report and generates a syntax tree representation. Then, it uses a tree difference algorithm to calculate the structural differences between the two versions of the power feasibility study report. Finally, it uses a word vector model to calculate the semantic similarity of the content. When it is detected that two users modify the same paragraph at the same time, the collaborative editing module 34 calculates the cosine similarity of the modified content. When the similarity is greater than or equal to the preset similarity, the collaborative editing module 34 determines it as a non-conflicting modification and automatically merges the modified content. When the similarity is less than the preset similarity, the collaborative editing module 34 marks it as a conflict and prompts the user to manually resolve it. This algorithm effectively reduces the false conflict rate and further improves the collaboration efficiency.

[0076] In one embodiment, the feasibility study report generation system provided in this embodiment further includes: a data analysis module 35;

[0077] The data analysis module 35 is used to obtain and analyze the power industry related data to generate a power data analysis chart;

[0078] The data analysis module 35 obtains power industry-related data and converts it into JSON format. Equipped with JSON structure parsing capabilities, it can handle nested objects and arrays. For standard-format power industry data, the data analysis module 35 automatically identifies the data type and unit, performing necessary unit conversions and normalizing values. For non-standard data, the system provides a data mapping configuration tool that allows users to define the mapping relationship between data fields and system variables. The data analysis results are stored in the data analysis module 35's cache database for use in generating and updating power feasibility studies.

[0079] The data analysis module 35 has built-in more than 20 chart templates, including line charts, bar charts, pie charts, heat maps, etc., which are suitable for different types of power data display. Based on the obtained data analysis results, the power data chart is automatically generated and added to the power feasibility report to enhance the professionalism of the report.

[0080] In one embodiment, see Figure 5 The schematic diagram of a unit of a template storage module in a feasibility study report generation system shown in FIG. The feasibility study report generation system provided in this embodiment further includes: a template storage module 36;

[0081] The template storage module 36 includes a requirement analysis unit 361, a template storage unit 362, and a template calling unit 363. The template storage unit 362 stores a variety of power feasibility study report templates. The requirement analysis unit 361 is used to analyze power requirement data and determine the optimal power feasibility study report template from the template storage unit 362. The template calling unit 363 is used to input the optimal power feasibility study report template into the report generation module 32, so that the report generation module 32 generates a power feasibility study report in the format of the optimal power feasibility study report template.

[0082] The template storage unit 362 of the template storage module 36 stores various feasibility study report templates for the power industry (such as the feasibility study report template for the construction of the power grid intelligent operation and maintenance system, the feasibility study report template for the construction of the distribution automation system, etc.). The requirement analysis unit 361 analyzes the power requirement data to determine the optimal power feasibility study report template. The template calling unit 363 is used to input the optimal power feasibility study report template into the report generation module 32, so that the report generation module 32 can generate a power feasibility study report in the format of the optimal power feasibility study report template, thereby ensuring the format standardization of the power feasibility study report.

[0083] In one embodiment, the template storage module 36 in the feasibility study report generation system provided in this embodiment further includes: a template editing unit 364;

[0084] The template editing unit 364 is used to enable the user to manually edit the power feasibility study report template stored in the template storage unit 362;

[0085] The template editing unit 364 allows the user to adaptively modify the power feasibility study report template, providing a flexible template modification mechanism.

[0086] The feasibility study report generation method and system provided in the embodiments of the invention are mainly aimed at power industry enterprises (such as power companies, energy suppliers, equipment manufacturers, etc.), power industry consulting companies (such as companies providing professional power consulting services, etc.), and institutions engaged in power industry information software development. The above-mentioned power industry enterprises usually need to prepare feasibility study reports for power information construction during their development process; power industry consulting companies need to efficiently generate feasibility study reports based on power issues raised by users, and institutions engaged in power industry information software development usually need to write feasibility study reports related to the power industry;

[0087] The feasibility report generation method and system can analyze the power requirement data based on the trained large language model to obtain the corresponding power feasibility report, thereby improving the efficiency of generating the power feasibility report and ensuring the quality of the obtained power feasibility report file; the power feasibility report is saved through the report management module 33, and the viewing, backtracking and version comparison of historical versions of the power feasibility report are realized; the collaborative editing module 34 is used to realize multi-user collaborative editing of the power feasibility report; the data analysis module 35 is used to provide visual charts for the power feasibility report, thereby improving the professionalism of the power feasibility report; and the template storage module 36 is used to make the structure and format of the generated power feasibility report more standardized.

[0088] In addition, in the description of the embodiments of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0089] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0090] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0091] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for generating a feasibility study report, characterized in that: include: Obtaining power requirement data for the power project; wherein the power requirement data includes: power load characteristics, voltage level requirements, power quality requirements and energy efficiency indicators; The power requirement data is input into the trained report generation model to obtain a power feasibility report; wherein the report generation model includes an embedding generation unit and a report acquisition unit; the embedding generation unit is used to embed the power requirement data to obtain multiple requirement text embeddings; the report acquisition unit is used to analyze the embeddings of each requirement text to obtain the power feasibility report; the report generation model is obtained by fine-tuning and training a power requirement data training dataset labeled with a power feasibility report.

2. The method for generating a feasibility study report according to claim 1, characterized in that: The fine-tuning training steps of the report generation model include: obtaining power industry data, and determining power requirement data for a plurality of power projects based on the power industry data; Marking the power feasibility study reports corresponding to the power requirement data of each power project, and establishing a sample data set based on the marked power requirement data of each power project; wherein the sample data set includes a training data set; The training data set is input into the large language model, and the large language model is fine-tuned based on the LoRA technology to obtain the report generation model after fine-tuning training.

3. The method for generating a feasibility study report according to claim 2, characterized in that: The step of performing model fine-tuning training on the large language model based on the LoRA technology to obtain the report generation model after fine-tuning training includes: Setting initial values ​​of training parameters of a large language model, and setting a low-rank decomposition matrix in the large language model; The large language model is iteratively trained based on the training data set, and a loss function is calculated after each round of iteration. When the loss function is less than a preset function value, the training is stopped to obtain the report generation model.

4. The method for generating a feasibility study report according to claim 2, characterized in that: The sample data set also includes a test data set; The feasibility study report generation method further includes: Determining each of the actual power feasibility studies based on the power requirement data of each of the power projects corresponding to the test data set; Inputting the test data set into the report generation model after fine-tuning and training to obtain a forecast power feasibility study report corresponding to each power requirement data; Based on the predicted power feasibility report and the actual power feasibility report corresponding to each of the power requirement data, performance evaluation indicators are obtained to evaluate the performance of the report generation model after fine-tuning training; wherein, the performance evaluation indicators include: BELU indicator and ROUGE indicator.

5. A feasibility study report generation system, characterized in that: include: Data acquisition module: used to obtain power requirement data of power projects; wherein the power requirement data includes: power load characteristics, voltage level requirements, power quality requirements and energy efficiency indicators; Report generation module: used to input the power requirement data into the trained report generation model to obtain a power feasibility report; wherein, the report generation model includes an embedding generation unit and a report acquisition unit; the embedding generation unit is used to embed the power requirement data to obtain multiple requirement text embeddings; the report acquisition unit is used to analyze the embeddings of each requirement text to obtain the power feasibility report; the report generation model is obtained by fine-tuning and training the power requirement data training data set marked with the power feasibility report.

6. The feasibility study report generation system according to claim 5, characterized in that: Also includes: Report management module; The report management module is used to store the power feasibility study report and record historical versions of the power feasibility study report.

7. The feasibility study report generation system according to claim 6, characterized in that: Also includes: Collaborative editing module; The collaborative editing module is used to call the power feasibility study report in the report management module and perform multi-user collaborative editing on the power feasibility study report.

8. The feasibility study report generation system according to claim 5, characterized in that: Also includes: Data analysis module; The data analysis module is used to obtain power industry related data, analyze the power industry related data, and generate a power data analysis chart.

9. The feasibility study report generation system according to claim 5, characterized in that: Also includes: Template storage module; The template storage module includes a requirement analysis unit, a template storage unit and a template calling unit; wherein the template storage unit stores a plurality of power feasibility report templates; the requirement analysis unit is used to analyze the power requirement data and determine the optimal power feasibility report template from the template storage unit; the template calling unit is used to input the optimal power feasibility report template into the report generation module, so that the report generation module generates a power feasibility report in the format of the optimal power feasibility report template.

10. The feasibility study report generation system according to claim 9, characterized in that: The template storage module further includes: a template editing unit; The template editing unit is used to enable a user to manually edit the power feasibility study report template stored in the template storage unit.

Images