Communication engineering cost estimation method and device, electronic equipment and storage medium

Abstract

The invention provides a communication project cost estimation method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring original project information of a to-be-estimated communication project; processing the original project information to obtain a project feature vector; inputting the project feature vector into a pre-constructed vertical domain large model to obtain an initial cost estimation result of the communication engineering output by the vertical domain large model, the vertical domain large model being obtained by training based on a sample project feature vector of the sample communication engineering and an initial cost estimation result label of the sample communication engineering; based on the project feature vector, performing retrieval in a pre-constructed historical communication engineering case database, and determining a target historical communication engineering case similar to the communication engineering; and based on the target historical communication engineering case, correcting and explaining the initial cost estimation result, and generating a final cost estimation result. According to the communication engineering cost estimation method provided by the invention, the efficiency and accuracy of cost estimation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a communication engineering cost estimation method and device, an electronic device and a storage medium. BACKGROUND

[0002] Communication engineering is the core link of information infrastructure construction, and the accuracy of its cost estimation is crucial to project cost control and decision-making. With the increasing complexity of 5G, fiber network and other projects, data forms are diversified. At present, the existing communication engineering cost estimation method mainly relies on cost engineers to manually check the quota manual, manually analyze design drawings and construction schemes to complete cost estimation, which is low in efficiency and poor in accuracy. SUMMARY

[0003] The present application provides a communication engineering cost estimation method, device, electronic device and storage medium to solve the defects of low efficiency and poor accuracy of the existing communication engineering cost estimation method.

[0004] The present application provides a communication engineering cost estimation method, comprising: obtaining the original project information of the communication engineering to be estimated; processing the original project information to obtain a project feature vector; inputting the project feature vector into a pre-constructed vertical domain large model to obtain an initial cost estimation result of the communication engineering output by the vertical domain large model, the vertical domain large model being trained based on a sample project feature vector of a sample communication engineering and an initial cost estimation result label of the sample communication engineering; based on the project feature vector, searching in a pre-constructed historical communication engineering case database to determine a target historical communication engineering case similar to the communication engineering; based on the target historical communication engineering case, correcting and explaining the initial cost estimation result to generate a final cost estimation result.

[0005] In some embodiments, the project feature vector includes a structured feature vector and a text feature vector; the vertical domain large model includes an embedding layer, a text encoder, a fusion encoder and a prediction layer; the embedding layer is used to convert the structured feature vector into a first dense vector sequence; the text encoder is used to convert the text feature vector into a second dense vector sequence; the fusion encoder is used to deep fuse the first dense vector sequence and the second dense vector sequence based on an attention mechanism to obtain a fusion feature vector sequence; and the prediction layer is used to predict the cost of the communication engineering based on the fusion feature vector sequence to obtain the initial cost estimation result of the communication engineering.

[0006] In some embodiments, the vertical domain large model further comprises a Transformer decoding layer, which is configured to model the sequence of fusion feature vectors as a time sequence.

[0007] In some embodiments, the original project information is processed to obtain a project feature vector, including: The original project information is preprocessed to obtain project information, including structured data and unstructured data; The structured data is parsed to obtain a structured feature vector, and the unstructured data is processed by natural language processing to obtain a text feature vector; Real-time external data related to the communication engineering is obtained, and the external data is processed to obtain an external feature vector; The structured feature vector, the text feature vector and the external feature vector are integrated to obtain the project feature vector.

[0008] In some embodiments, the historical communication engineering case database comprises a plurality of historical communication engineering cases; based on the project feature vector, a target historical communication engineering case similar to the communication engineering is determined by searching in a pre-constructed historical communication engineering case database, including: A plurality of historical project feature vectors of the plurality of historical communication engineering cases are determined; The similarity between the project feature vector and the plurality of historical project feature vectors is calculated; Based on the similarity between the project feature vector and the plurality of historical project feature vectors, one or more target historical communication engineering cases similar to the communication engineering are determined from the plurality of historical communication engineering cases.

[0009] In some embodiments, the initial cost estimate result is corrected and explained based on the target historical communication engineering case to generate a final cost estimate result, including: Based on a preset prompt engineering template, the initial cost estimate result, the target historical communication engineering case and specific analysis instructions are integrated to generate an input prompt; The input prompt is input into a large language model to obtain a cost estimate result output by the large language model.

[0010] In some embodiments, the training process of the vertical domain large model comprises: Sample project information of a sample communication engineering is obtained, and the sample project information is processed to obtain a sample project feature vector; An initial cost estimate result label of the sample communication engineering is determined; Freeze the bottom layer parameters of the initial vertical domain large model; Input the sample project feature vector into the initial vertical domain large model to obtain an initial cost prediction result of the sample communication engineering output by the initial vertical domain large model; Based on the initial cost prediction result and the initial cost estimation result label, a loss function value is calculated, and based on the loss function value, the middle-high layer parameters of the initial vertical domain large model are iteratively optimized to obtain the vertical domain large model.

[0011] The application also provides a communication engineering cost estimation device, comprising: An acquisition unit is configured to acquire original project information of a communication engineering to be estimated; A processing unit is configured to process the original project information to obtain a project feature vector; An estimation unit is configured to input the project feature vector into a pre-constructed vertical domain large model to obtain an initial cost estimation result of the communication engineering output by the vertical domain large model, wherein the vertical domain large model is obtained by training based on a sample project feature vector of a sample communication engineering and an initial cost estimation result label of the sample communication engineering; A retrieval unit is configured to retrieve a target historical communication engineering case similar to the communication engineering based on the project feature vector in a pre-constructed historical communication engineering case database; A correction unit is configured to correct and explain the initial cost estimation result based on the target historical communication engineering case to generate a final cost estimation result.

[0012] The application also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the communication engineering cost estimation method according to any one of the above when executing the program.

[0013] The application also provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the communication engineering cost estimation method according to any one of the above.

[0014] The communication engineering cost estimation method, device, electronic equipment and storage medium provided by the application, by obtaining the original project information of the communication engineering to be estimated; processing the original project information to obtain a project feature vector; inputting the project feature vector into a pre-constructed vertical large model to obtain an initial cost estimation result of the communication engineering output by the vertical large model; based on the project feature vector, searching in a pre-constructed historical communication engineering case database to determine a target historical communication engineering case similar to the communication engineering; based on the target historical communication engineering case, correcting and explaining the initial cost estimation result to generate a final cost estimation result, improving the efficiency and accuracy of cost estimation. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0016] Figure 1 is a flowchart of the communication engineering cost estimation method provided by the embodiment of the application.

[0017] Figure 2 is a flowchart of the training process of the vertical large model provided by the embodiment of the application.

[0018] Figure 3 is a structural diagram of the communication engineering cost estimation device provided by the embodiment of the application.

[0019] Figure 4 is a structural diagram of the electronic equipment provided by the embodiment of the application. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of the application more clear, the technical solutions in the application will be described clearly and completely in combination with the drawings in the application. Obviously, the described embodiments are some embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor belong to the protection scope of the application.

[0021] The terms "first," "second," etc., used in this invention are used to distinguish similar objects, not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, the first object can be one or more. Furthermore, in this invention, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0022] Currently, existing methods for estimating the cost of telecommunications engineering projects rely on manual methods, which have the following drawbacks: First, manual processing is inefficient and inaccurate, with long estimation cycles and susceptibility to subjective factors. Second, the lack of multimodal data processing capabilities makes it difficult to automatically associate legends with quantities in drawings and to understand special process requirements from text, resulting in information silos and estimation errors. In addition, the rules are poorly adaptable and cannot respond in a timely manner to dynamic changes in factors such as material prices and local policies, which seriously affects the accuracy of cost estimation.

[0023] To address this, embodiments of the present invention provide a method, apparatus, electronic device, and storage medium for estimating the cost of communication engineering projects. The method involves: acquiring the original project information of the communication engineering project to be estimated; processing the original project information to obtain a project feature vector; inputting the project feature vector into a pre-constructed vertical domain large-scale model to obtain the initial cost estimate result of the communication engineering project output by the vertical domain large-scale model; based on the project feature vector, searching a pre-constructed historical communication engineering case database to identify target historical communication engineering cases similar to the current communication engineering project; and based on the target historical communication engineering cases, correcting and interpreting the initial cost estimate result to generate the final cost estimate result. Embodiments of the present invention can improve the efficiency and accuracy of cost estimation.

[0024] Figure 1 This is a flowchart illustrating the communication engineering cost estimation method provided in an embodiment of the present invention. Figure 1 As shown, a method for estimating the cost of communication engineering projects is provided, including the following steps: steps 110 to 150. This method's steps are merely one possible implementation of the present invention.

[0025] Step 110: Obtain the original project information of the communication engineering project to be estimated; Among them, telecommunications engineering specifically refers to engineering projects related to the construction of information infrastructure. Telecommunications engineering includes, but is not limited to, 5G network deployment, fiber-to-the-home, data center interconnection, base station construction, and pipeline construction. Its characteristics include numerous project sites, long distances, and wide coverage.

[0026] In this context, "original project information" refers to the initial, unsystematic, multi-source, and multi-format data required to initiate a cost estimation task. Original project information includes, but is not limited to: design drawings, construction plans, bills of quantities, material lists, and equipment model lists.

[0027] Optionally, external data related to communication engineering can be acquired in real time. External data includes, but is not limited to, environmental data, policy information, material prices, equipment prices, and labor cost indices.

[0028] Step 120: Process the original project information to obtain the project feature vector; Optionally, the original project information is preprocessed to obtain project information; features are extracted from the project information to obtain an initial project feature vector; external data is preprocessed to obtain preprocessed external data; features are extracted from the preprocessed external data to obtain an external feature vector; and the initial project feature vector and the external feature vector are fused to obtain a project feature vector.

[0029] Optionally, the project feature vector includes time features.

[0030] In some embodiments, step 120 processes the original project information to obtain a project feature vector, including: Step 121: Preprocess the original project information to obtain project information, which includes structured data and unstructured data; Structured data refers to highly organized data that follows predefined formats and fixed patterns, such as bills of quantities, bills of materials, and budget quota libraries; unstructured data refers to data without a predefined data model or organizational structure, such as design drawings, construction plans, and technical specification documents.

[0031] Optionally, the original project information can be preprocessed by data cleaning, noise reduction, standardization, normalization, etc.

[0032] Step 122: Parse the structured data to obtain structured feature vectors, and perform natural language processing on the unstructured data to obtain text feature vectors; Structured feature vectors are numerical vectors parsed from structured data. Each dimension of a structured feature vector represents a specific, quantifiable engineering attribute, such as the length of a particular type of optical cable, the number of base stations required, or the diameter of a duct. Textual feature vectors include features such as special processes, technical requirements, and construction environment.

[0033] Optionally, optical character recognition (OCR) technology can be applied to recognize unstructured data and extract text, such as extracting legends and quantity annotations from design drawings.

[0034] Optionally, Natural Language Processing (NLP) technology is used to process the text extracted by OCR. Named entity recognition is used to identify key terms such as specific processes and designated material brands from the text. Relationship extraction is then used to understand the connections between entities, such as process A causing a 10% increase in the amount of material B used. Finally, this semantic information is encoded into a text feature vector.

[0035] Step 123: Acquire external data related to communication engineering in real time, process the external data, and obtain external feature vectors; External data refers to real-time or near-real-time information that can dynamically affect project costs, in addition to project information, such as material market prices, labor cost indices, and local policy and tax documents.

[0036] Optionally, by calling the application programming interface (API) of an external database or web crawler, the latest external data such as material market prices, relevant policy documents, and macroeconomic indices can be obtained periodically or at the start of each forecasting task.

[0037] Optionally, the external feature vector includes time features.

[0038] Step 124: Integrate the structured feature vector, text feature vector, and external feature vector to obtain the project feature vector.

[0039] Optionally, the structured feature vector, text feature vector, and external feature vector can be concatenated to obtain the project feature vector.

[0040] Understandably, by dynamically accessing and characterizing external data sources, external feature vectors are obtained. By integrating structured feature vectors, text feature vectors, and external feature vectors, project feature vectors are obtained. This provides a comprehensive feature representation for the vertical domain large model, endowing it with the ability to perceive the real-time market environment. This lays the data foundation for the vertical domain large model to perform complex nonlinear reasoning, accurately capture cost-driving factors, and ultimately achieve high-precision cost estimation.

[0041] Step 130: Input the project feature vector into the pre-built vertical domain large model to obtain the initial cost estimate of the communication project output by the vertical domain large model. The vertical domain large model is trained based on the sample project feature vector of the sample communication project and the initial cost estimate label of the sample communication project. Among them, the vertical domain large model refers to a large language model that has been specially trained and is deeply specialized in the vertical field of communication transmission engineering cost.

[0042] Optionally, the initial cost estimate includes a forecast of the total project cost, as well as a detailed breakdown of the costs for each key item, such as material costs and labor costs.

[0043] In some embodiments, the project feature vector includes structured feature vectors and text feature vectors; the vertical domain large model includes an embedding layer, a text encoder, a fusion encoder, and a prediction layer; the embedding layer is used to transform the structured feature vectors into a first dense vector sequence; the text encoder is used to transform the text feature vectors into a second dense vector sequence; the fusion encoder is used to perform deep fusion of the first dense vector sequence and the second dense vector sequence based on an attention mechanism to obtain a fused feature vector sequence; the prediction layer is used to predict the cost of the communication project based on the fused feature vector sequence to obtain an initial cost estimate for the communication project.

[0044] The embedding layer maps discrete engineering numerical values ​​without direct semantic relationships into a continuous, high-dimensional vector space, forming a first dense vector sequence. This process allows the model to learn the intrinsic relationships between numerical values ​​and transform them into a format compatible with text vectors. The text encoder is a standard Transformer-based module that processes text feature vectors. It captures the internal contextual relationships of the text through a self-attention mechanism, transforming the input text sequence into a second dense vector sequence rich in semantic information.

[0045] The fusion encoder dynamically calculates the correlation weights between different elements in two dense vector sequences, generating a novel fusion feature vector sequence. The prediction layer, also known as the task head, is located at the top of the model. This layer receives the fusion feature vector sequence and, through one or more fully connected layers, maps the highly condensed information it contains onto the final prediction target, outputting a structured initial cost estimate.

[0046] In some embodiments, the vertical large model also includes a Transformer decoding layer, which is used to model the fused feature vector sequence as a time series.

[0047] Step 140: Based on the project feature vector, search the pre-built historical communication engineering case database to identify target historical communication engineering cases similar to the communication engineering projects. The historical communication engineering case database is a vector database. It is constructed by transforming the feature vectors of historical projects into high-dimensional vectors using an embedding model and storing them in the database.

[0048] In some embodiments, the historical communication engineering case database contains multiple historical communication engineering cases; step 140, based on project feature vectors, searches the pre-built historical communication engineering case database to determine target historical communication engineering cases similar to the communication engineering projects, including: Step 141: Determine the feature vectors of multiple historical projects in multiple historical communication engineering cases; Step 142: Calculate the similarity between the project feature vector and the feature vectors of multiple historical projects; Step 143: Based on the similarity between the project feature vector and the feature vectors of multiple historical projects, identify one or more target historical communication engineering cases that are similar to the communication engineering project from multiple historical communication engineering cases.

[0049] Optionally, the similarity between the project feature vector and the feature vectors of multiple historical projects is calculated based on the K-nearest neighbor algorithm. From multiple historical communication engineering cases, the K target historical communication engineering cases that are most similar to the communication engineering are determined, where K is a natural number greater than or equal to 1.

[0050] Step 150: Based on the target historical communication engineering case, revise and interpret the initial cost estimate results to generate the final cost estimate results.

[0051] The final cost estimate is not a single numerical value, but a comprehensive analytical report that can be used for decision-making.

[0052] Optionally, explanatory text is generated based on historical communication engineering cases; the initial cost estimate and the explanatory text are then integrated to generate the final cost estimate. The explanatory text includes at least data analysis information, risk warnings, and case comparison information.

[0053] In some embodiments, step 150, based on a target historical communication engineering case, modifies and interprets the initial cost estimate to generate a final cost estimate, including: Step 151: Based on the preset prompt engineering template, integrate the initial cost estimate results, target historical communication engineering cases, and specific analysis instructions to generate input prompts; Step 152: Input the input prompts into the large language model to obtain the cost estimation results output by the large language model.

[0054] The preset prompt engineering template is a carefully designed, structured text framework. It contains fixed instruction text and multiple placeholders to assemble information from different sources into a coherent input instruction that can be understood by a large language model.

[0055] Understandably, by combining historical facts to verify, correct, and interpret the prediction results of the large-scale vertical model, a credible, reliable, and decision-making final solution can be generated, thereby improving the accuracy of cost estimation.

[0056] In this embodiment of the invention, the original project information of the communication project to be estimated is obtained; the original project information is processed to obtain a project feature vector; the project feature vector is input into a pre-constructed vertical domain large model to obtain the initial cost estimate result of the communication project output by the vertical domain large model; based on the project feature vector, a search is performed in a pre-constructed historical communication project case database to determine target historical communication project cases similar to the communication project; based on the target historical communication project cases, the initial cost estimate result is corrected and interpreted to generate the final cost estimate result, thereby improving the efficiency and accuracy of cost estimation.

[0057] Figure 2 This is a flowchart illustrating the training process of a large vertical domain model provided in an embodiment of the present invention. Figure 2 As shown, in some embodiments, the training process of the large vertical model includes: Step 210: Obtain sample project information of the sample communication project, process the sample project information, and obtain the sample project feature vector; Step 220: Determine the initial cost estimate label for the sample communication project; Step 230: Freeze the underlying parameters of the initial vertical domain large model; Step 240: Input the feature vector of the sample project into the initial vertical domain large model to obtain the initial cost prediction result of the sample communication project output by the initial vertical domain large model; Step 250: Based on the initial cost prediction results and the initial cost estimate results labels, calculate the loss function value. Based on the loss function value, iteratively optimize the mid-to-high-rise parameters of the initial vertical domain large model to obtain the vertical domain large model.

[0058] Optionally, the sample item information is processed to obtain a sample item feature vector, including: The sample item information is processed to obtain the sample structured feature vector and the sample text feature vector; Obtain external data related to the sample communication project, process the external data, and obtain the external feature vector of the sample. By integrating the sample structured feature vector, sample text feature vector, and sample external feature vector, the sample item feature vector is obtained.

[0059] Optionally, the sample project feature vector includes sample structured feature vector and sample text feature vector; the initial vertical domain large model includes an initial embedding layer, an initial text encoder, an initial fusion encoder, and an initial prediction layer; the initial embedding layer is used to transform the sample structured feature vector into a first sample dense vector sequence; the initial text encoder is used to transform the sample text feature vector into a second sample dense vector sequence; the initial fusion encoder is used to perform deep fusion of the first sample dense vector sequence and the second sample dense vector sequence based on an attention mechanism to obtain a sample fusion feature vector sequence; the initial prediction layer is used to predict the cost of the sample communication project based on the sample fusion feature vector sequence to obtain the initial cost prediction result of the sample communication project.

[0060] The communication engineering cost estimation device provided in the embodiments of the present invention is described below. The communication engineering cost estimation device described below and the communication engineering cost estimation method described above can be referred to each other.

[0061] Figure 3 This is a schematic diagram of the communication engineering cost estimation device provided in an embodiment of the present invention, as shown below. Figure 3 As shown, the communication engineering cost estimation device 300 includes: Acquisition unit 310 is used to acquire the original project information of the communication project to be estimated; The processing unit 320 is used to process the original project information to obtain the project feature vector; The estimation unit 330 is used to input the project feature vector into the pre-built vertical domain large model to obtain the initial cost estimation result of the communication project output by the vertical domain large model. The vertical domain large model is trained based on the sample project feature vector of the sample communication project and the initial cost estimation result label of the sample communication project. The retrieval unit 340 is used to search a pre-built database of historical communication engineering cases based on project feature vectors to identify target historical communication engineering cases that are similar to communication engineering projects. The correction unit 350 is used to correct and interpret the initial cost estimate based on the target historical communication engineering case, and generate the final cost estimate.

[0062] Optionally, the project feature vector includes structured feature vectors and text feature vectors; the vertical domain large model includes an embedding layer, a text encoder, a fusion encoder, and a prediction layer; the embedding layer is used to transform the structured feature vectors into a first dense vector sequence; the text encoder is used to transform the text feature vectors into a second dense vector sequence; the fusion encoder is used to perform deep fusion of the first dense vector sequence and the second dense vector sequence based on an attention mechanism to obtain a fused feature vector sequence; the prediction layer is used to predict the cost of the communication project based on the fused feature vector sequence to obtain the initial cost estimate of the communication project.

[0063] Optionally, the vertical large model also includes a Transformer decoding layer, which is used to model the fused feature vector sequence as a time series.

[0064] Optionally, the original project information is processed to obtain a project feature vector, including: The original project information is preprocessed to obtain project information, which includes structured data and unstructured data. Structured data is parsed to obtain structured feature vectors, and unstructured data is processed by natural language to obtain text feature vectors. Real-time acquisition of external data related to communication engineering, processing of external data to obtain external feature vectors; By integrating structured feature vectors, text feature vectors, and external feature vectors, the project feature vector is obtained.

[0065] Optionally, the historical communication engineering case database contains multiple historical communication engineering cases; based on project feature vectors, a search is performed in the pre-built historical communication engineering case database to identify target historical communication engineering cases similar to the communication engineering projects, including: Determine the feature vectors of multiple historical projects from multiple historical communication engineering cases; Calculate the similarity between the project feature vector and the feature vectors of multiple historical projects; Based on the similarity between the project feature vector and the feature vectors of multiple historical projects, one or more target historical communication engineering cases similar to the communication engineering project are identified from multiple historical communication engineering cases.

[0066] Optionally, based on historical communication engineering cases, the initial cost estimate is revised and interpreted to generate a final cost estimate, including: Based on the preset prompt engineering template, the initial cost estimate, target historical communication engineering cases, and specific analysis instructions are integrated to generate input prompts; Input the prompts into the large language model to obtain the cost estimate output by the large language model.

[0067] Optionally, the training process for a large vertical model includes: Obtain sample project information from the sample communication project, process the sample project information, and obtain the sample project feature vector; Determine the label of the initial cost estimate result for the sample communication project; Freeze the underlying parameters of the initial vertical domain large model; The feature vectors of the sample projects are input into the initial vertical domain large model to obtain the initial cost prediction results of the sample communication engineering output by the initial vertical domain large model. Based on the initial cost prediction results and initial cost estimate results labels, the loss function value is calculated. Based on the loss function value, the mid-to-high-rise parameters of the initial vertical domain large model are iteratively optimized to obtain the vertical domain large model.

[0068] It should be noted that the communication engineering cost estimation device provided in this embodiment of the invention can implement all the method steps implemented in the above-mentioned communication engineering cost estimation method embodiment, and can achieve the same technical effect. Here, the parts that are the same as those in the method embodiment and the beneficial effects will not be described in detail.

[0069] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, such as... Figure 4 As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430, and a communication bus 440. The processor 410, communications interface 420, and memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a communication engineering cost estimation method. This method includes: acquiring the original project information of the communication engineering to be estimated; processing the original project information to obtain a project feature vector; inputting the project feature vector into a pre-built vertical domain large model to obtain the initial cost estimation result of the communication engineering output by the vertical domain large model, wherein the vertical domain large model is trained based on the sample project feature vectors of sample communication engineering projects and the labels of the initial cost estimation results of sample communication engineering projects; based on the project feature vectors, searching in a pre-built historical communication engineering case database to determine target historical communication engineering cases similar to the communication engineering project; and based on the target historical communication engineering cases, correcting and interpreting the initial cost estimation result to generate the final cost estimation result.

[0070] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, 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 a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0071] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0072] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating the cost of communication engineering projects, characterized in that, include: Obtain the original project information for the communication engineering project to be estimated; The original project information is processed to obtain the project feature vector; The project feature vector is input into a pre-built vertical domain large model to obtain the initial cost estimate of the communication project output by the vertical domain large model. The vertical domain large model is trained based on the sample project feature vector of the sample communication project and the initial cost estimate label of the sample communication project. Based on the project feature vector, a search is performed in a pre-built historical communication engineering case database to identify target historical communication engineering cases similar to the communication engineering project. Based on the target historical communication engineering case, the initial cost estimate is revised and interpreted to generate the final cost estimate.

2. The method for estimating the cost of communication engineering projects according to claim 1, characterized in that, The project feature vectors include structured feature vectors and text feature vectors; the vertical domain large model includes an embedding layer, a text encoder, a fusion encoder, and a prediction layer; the embedding layer is used to transform the structured feature vectors into a first dense vector sequence; the text encoder is used to transform the text feature vectors into a second dense vector sequence; the fusion encoder is used to perform deep fusion of the first dense vector sequence and the second dense vector sequence based on an attention mechanism to obtain a fused feature vector sequence; the prediction layer is used to predict the cost of the communication project based on the fused feature vector sequence to obtain an initial cost estimate of the communication project.

3. The method for estimating the cost of communication engineering projects according to claim 2, characterized in that, The vertical domain large model also includes a Transformer decoding layer, which is used to model the fused feature vector sequence as a time series.

4. The method for estimating the cost of communication engineering projects according to claim 1, characterized in that, The original project information is processed to obtain a project feature vector, including: The original project information is preprocessed to obtain project information, which includes structured data and unstructured data; The structured data is parsed to obtain structured feature vectors, and the unstructured data is processed by natural language to obtain text feature vectors. Real-time acquisition of external data related to the communication project; processing of the external data to obtain external feature vectors; The project feature vector is obtained by integrating the structured feature vector, the text feature vector, and the external feature vector.

5. The method for estimating the cost of communication engineering projects according to claim 1, characterized in that, The historical communication engineering case database contains multiple historical communication engineering cases; the step of searching the pre-built historical communication engineering case database based on the project feature vector to determine target historical communication engineering cases similar to the communication engineering case includes: Determine the feature vectors of multiple historical projects from the multiple historical communication engineering cases; Calculate the similarity between the feature vector of the project and the feature vectors of the multiple historical projects; Based on the similarity between the project feature vector and the feature vectors of the multiple historical projects, one or more target historical communication engineering cases similar to the communication engineering are identified from the multiple historical communication engineering cases.

6. The method for estimating the cost of communication engineering projects according to claim 1, characterized in that, The process of revising and interpreting the initial cost estimate based on the target historical communication engineering case to generate the final cost estimate includes: Based on a preset prompt engineering template, the initial cost estimate, the target historical communication engineering case, and specific analysis instructions are integrated to generate input prompts; The input prompts are fed into the large language model to obtain the cost estimate output by the large language model.

7. The method for estimating the cost of communication engineering projects according to claim 1, characterized in that, The training process of the large vertical model includes: Obtain sample project information from the sample communication project, process the sample project information, and obtain the sample project feature vector; Determine the initial cost estimate label for the sample communication project; Freeze the underlying parameters of the initial vertical domain large model; The feature vector of the sample project is input into the initial vertical domain large model to obtain the initial cost prediction result of the sample communication project output by the initial vertical domain large model; Based on the initial cost prediction results and the initial cost estimate results labels, a loss function value is calculated. Based on the loss function value, the mid-to-high-level parameters of the initial vertical domain large model are iteratively optimized to obtain the vertical domain large model.

8. A communication engineering cost estimation device, characterized in that, include: The acquisition unit is used to acquire the original project information of the communication engineering to be estimated. The processing unit is used to process the original project information to obtain the project feature vector; The estimation unit is used to input the feature vector of the project into a pre-built vertical domain large model to obtain the initial cost estimation result of the communication project output by the vertical domain large model. The vertical domain large model is trained based on the feature vector of the sample communication project and the label of the initial cost estimation result of the sample communication project. The retrieval unit is used to search a pre-built historical communication engineering case database based on the project feature vector to identify target historical communication engineering cases similar to the communication engineering project. The correction unit is used to correct and interpret the initial cost estimate based on the target historical communication engineering case, and generate the final cost estimate.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the communication engineering cost estimation method as described in any one of claims 1 to 7.

10. A non-transitory 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 communication engineering cost estimation method as described in any one of claims 1 to 7.

Images