Material personalized recommendation method based on distribution network project characteristics
By building an integrated intelligent recommendation model and combining multiple recommendation algorithms, the complexity and diversified needs of material planning management in the power industry are solved, accurate matching and prediction are achieved, and the operational efficiency and cost-effectiveness of power grid projects are improved.
Patent Information
- Application Number
- CN202411002681.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-08
AI Technical Summary
In the power industry, material planning management is complex and relies on manual experience, resulting in low efficiency in material procurement and inventory management. The traditional demand planning method cannot effectively respond to the complexity and diversified needs of power grid projects, resulting in inaccurate and practical enough recommendation results.
Using the personalized material recommendation method based on the characteristics of distribution network projects, an integrated intelligent recommendation model is constructed by obtaining historical distribution network project data and material procurement data, combining factor analysis, fusion editing distance algorithm, Mini Batch K-Means clustering analysis and Sliding average method is used to build project recommendation models, material recommendation models, material technical specification recommendation models and intelligent bidding batch arrangement models to achieve accurate matching and prediction of material procurement prices.
It improves the accuracy and management efficiency of material demand identification, reduces supply and demand imbalances, optimizes procurement plans, significantly improves operational efficiency and cost-effectiveness, and provides strong support for the successful implementation of power grid projects.
Smart Images

Figure CN118917750A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of power industry and power grid project management, and mainly to a method for personalized material recommendation based on distribution network project characteristics. Background Art
[0002] As the basic business of the power industry, material planning management is an important link in the digital industrial chain. Material declaration is difficult and complicated, involving multiple systems such as Enterprise Resource Planning (ERP), ECP Enterprise Management Cloud Platform (ECP), State Grid Mall, etc.; more than ten approvals are required, and the operation process is long and complicated; it involves data attributes such as project investment, catalog, price, time, technical specifications, factory, procurement group, procurement method, delivery location, etc., which are related to multiple links of project management and supply chain. The demand for materials is large, and the traditional demand planning method relies more on experience and manpower. The limited efficiency restricts multiple links such as material procurement and inventory in the project. At present, with the advancement of science and technology, machine learning is gradually introduced into personalized recommendation methods. This method replaces the traditional manual experience-dependent demand planning through automation and intelligence, significantly improving the efficiency of material procurement and management, and reducing errors and costs.
[0003] For example, CN 115630227 A “Information recommendation method, device, electronic device and storage medium” discloses “an information recommendation method, device, electronic device and storage medium, relating to the field of communication technology. The information recommendation method provided by the embodiment of the present invention comprises: obtaining the user's historical chat information and current input information; training the initial recommendation model based on the historical chat information to obtain an intelligent recommendation model; the intelligent recommendation model generates multiple recommendation words according to the current input information; matching the multiple recommendation words with the current input information to obtain the matching degree of each recommendation word; when the matching degree of the recommendation word is higher than the matching threshold, the recommendation word is determined as the recommended text. On a terminal that requires a large amount of text editing, the recommended text that matches the current input content can be quickly and personalized recommended to help users complete a large amount of text editing, thereby eliminating a large amount of repeated input operations, significantly improving the efficiency and effectiveness of communication, and realizing personalized recommendation of information”, but the invention only trains the initial model based on the historical chat information and the current input information to obtain the intelligent recommendation model. The training data is not comprehensive enough, which may cause the recommendation results of the intelligent recommendation model to be inaccurate and not practical. Summary of the invention
[0004] In order to solve the above problems existing in the prior art, the present application provides a method for personalized material recommendation based on the characteristics of distribution network projects.
[0005] The technical solution of this application is as follows:
[0006] A personalized material recommendation method based on distribution network project characteristics, the method comprising:
[0007] Acquire historical distribution network project data and historical material procurement data, and construct a distribution network project library after data processing of the historical distribution network project data;
[0008] Constructing an integrated intelligent recommendation model, using historical distribution network project data and historical material procurement data as training sets to train the integrated intelligent recommendation model, and obtaining a trained integrated intelligent recommendation model;
[0009] Input user requirements, project-related information, and material procurement-related information into the trained integrated intelligent recommendation model to obtain corresponding output results, including project recommendation results, material recommendation results, material technical specification recommendation results, and bidding batch and procurement method recommendation results;
[0010] According to the winning bid price in the historical material procurement data, the historical price trend is analyzed by the sliding average method to obtain the predicted material procurement price;
[0011] The user completes the distribution network material planning management based on the output results and the corresponding predicted material procurement prices.
[0012] Preferably, the historical distribution network project data is processed to construct a distribution network project library, specifically:
[0013] Performing data cleaning on the relevant attribute information of the historical distribution network project, wherein the data cleaning includes processing missing values, abnormal values and unifying the data format;
[0014] Factor analysis is used to analyze and extract relevant attribute information, classify variables, obtain factor loading matrices corresponding to the attributes, obtain common factors based on the factor loading matrix, and use the common factors as new attributes of the distribution network project;
[0015] A project library of distribution network projects is constructed according to the new attributes.
[0016] Preferably, the integrated intelligent recommendation model includes a project recommendation model, a material recommendation model, a material technical specification recommendation model and an intelligent bidding batch arrangement model, specifically:
[0017] The project recommendation model uses a fusion edit distance algorithm, a cosine algorithm based on word frequency, and a serialization algorithm according to the project library of the distribution network project, and performs weighted processing to obtain the historical project recommendation results of the project;
[0018] The material recommendation model obtains a recommended material data set through data preprocessing, Mini Batch K-Means clustering analysis, association rule construction and statistical analysis;
[0019] The material technical specification book recommendation model is based on historical similar projects and material procurement data. By calculating the usage frequency and proportion of technical specifications of similar projects, the technical specifications with the highest usage frequency and the largest proportion are selected as the recommended material technical specifications.
[0020] The intelligent bidding batch arrangement model performs information matching based on the annual bidding batch arrangement to obtain recommended annual bidding batches.
[0021] Preferably, the project recommendation model uses a fusion edit distance algorithm, a cosine algorithm based on word frequency, and a serialization algorithm according to the project library of the distribution network project, and performs weighted processing to obtain the historical project recommendation results of the project, specifically:
[0022] The distribution network project library is used as a training set, and the edit distance algorithm, the cosine algorithm based on word frequency and the serialization algorithm are integrated for weighted processing. A text similarity function is established, and a project recommendation model is constructed and trained. Based on the trained project recommendation model, the historical project recommendation results of the project are obtained.
[0023] Preferably, the material recommendation model obtains the recommended material data set through data preprocessing, Mini Batch K-Means clustering analysis, association rule construction and statistical analysis, specifically:
[0024] The data preprocessing includes summarizing the quantity of material subcategories used in each project to obtain a data set of three fields: project definition, material subcategory code, and material subcategory quantity; performing matrix transposition on the data set, wherein the matrix has material subcategory codes as columns and material subcategory quantities as values; and performing data normalization processing on the transposed data set with the goal of having quantity data of different material subcategories at the same level;
[0025] Using the Mini Batch K-Means clustering algorithm, the items in the data set are clustered into different project groups based on material usage;
[0026] The association rule includes antecedents and contingents, wherein the antecedents are project groups obtained by merging by matching new attributes through project definitions, and the contingents are material sub-category codes. Association rule combinations are constructed through pre-set minimum support and minimum confidence.
[0027] Based on statistical methods, the material codes and their quantities under each association rule are analyzed; the recommended material set under each association rule is determined based on the number of items in the previous item of the association rule; the occurrence ratio and average quantity of each material code in the project are extracted to construct the final recommended material data set.
[0028] Preferably, the predicted material procurement price is obtained by analyzing the trend of historical winning bid prices by the sliding average method according to the winning bid prices in the historical material procurement data, specifically:
[0029] The size of the sliding window is determined according to the periodicity and volatility of the winning bid prices in the historical material procurement data; the winning bid price data of several periods are selected according to the selected sliding window size, and the average value of the winning bid price data is calculated; the average value is used as the predicted value of the price in the next period, that is, the predicted material procurement price.
[0030] A material personalized recommendation system based on the characteristics of a distribution network project, the system comprises a project data acquisition module, an intelligent recommendation module and a module for determining project material declaration data and revision, wherein:
[0031] The project data acquisition module is used to acquire distribution network project data and build a distribution network project library;
[0032] The intelligent recommendation module is internally provided with an integrated intelligent recommendation model for intelligently recommending items requiring personalized material recommendations;
[0033] The module for determining project material declaration data and revising is used to select intelligently recommended material information and edit and adjust the selected material information.
[0034] A personalized material recommendation system based on the characteristics of distribution network projects, wherein the integrated intelligent recommendation model includes a project recommendation model, a material recommendation model, a material technical specification recommendation model and an intelligent bidding batch arrangement model, wherein:
[0035] The project recommendation model uses a fusion edit distance algorithm, a cosine algorithm based on word frequency, and a serialization algorithm according to the project library of the distribution network project, and performs weighted processing to obtain the historical project recommendation results of the project;
[0036] The material recommendation model obtains a recommended material data set through data preprocessing, Mini Batch K-Means clustering analysis, association rule construction and statistical analysis;
[0037] The material technical specification recommendation model is based on historical similar projects and material procurement big data. By calculating the usage frequency and proportion of technical specifications of similar projects, the technical specifications with the highest usage frequency and the largest proportion are selected as the project material procurement demand submission and the recommended optimal technical specifications;
[0038] The intelligent bidding batch arrangement model uses the project information and material information to match the annual bidding batch data as input, and recommends and displays the latest annual bidding batch and procurement plan for the materials in combination with the specific needs of material procurement.
[0039] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a method for personalized material recommendation based on distribution network project characteristics as described in any one of the embodiments is implemented.
[0040] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a personalized material recommendation method based on distribution network project characteristics as described in any one of the embodiments.
[0041] Compared with the prior art, the present invention has the following beneficial effects:
[0042] 1) The present invention provides a personalized material recommendation method based on the characteristics of distribution network projects, which uses the dimensionality reduction idea of factor analysis to simplify the data structure, reduce information redundancy, improve the interpretability and management efficiency of data, and can accurately identify the material requirements related to the characteristics of specific projects;
[0043] 2) The present invention provides a personalized material recommendation method based on the characteristics of distribution network projects. The integrated intelligent recommendation model can effectively improve the accuracy and applicability of the recommendation system. The integrated intelligent recommendation model includes a project recommendation model, a material recommendation model, a material technical specification recommendation model and an intelligent bidding batch arrangement model. By combining multiple recommendation algorithms, it can cope with the complexity and diversified needs of power grid projects and achieve accurate matching;
[0044] 3) The present invention provides a personalized material recommendation method based on the characteristics of distribution network projects, which uses historical data and project characteristics to accurately predict material requirements, avoid supply and demand imbalance, and optimize procurement plans. This system can significantly improve operational efficiency and cost-effectiveness, and provide strong support for the successful implementation and sustainable operation of power grid projects. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 is a flow chart of a method in an embodiment of the present invention. DETAILED DESCRIPTION
[0046] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.
[0047] The present invention provides the following technical solution: a method for personalized material recommendation based on distribution network project characteristics.
[0048] Example 1
[0049] This embodiment provides a personalized material recommendation method based on distribution network project characteristics, and the specific steps include:
[0050] S1. Acquire historical distribution network project data and historical material procurement data, and construct a distribution network project library after data processing of the historical distribution network project data;
[0051] S11, the historical distribution network project data includes the total project investment plan, voltage level, project type, construction stage, project start time, project completion time, and project name keywords; the project library is associated with the historical project information through the project name and project definition; the historical material procurement data includes material code, material type, procurement amount, procurement quantity, and material description;
[0052] S12, build distribution network project library;
[0053] S121, performing data cleaning on the relevant attribute information of the historical distribution network project, wherein the data cleaning includes processing missing values, abnormal values and unifying the data format;
[0054] S122, screening and ranking the importance of project attributes through business data analysis, analyzing and extracting relevant attribute information through factor analysis, classifying variables, obtaining a factor loading matrix corresponding to the attributes, and obtaining common factors based on the factor loading matrix;
[0055] S1221. Factor analysis is based on the idea of dimensionality reduction. Under the condition of losing as little or no original data information as possible, the numerous complex variables are aggregated into a few independent common factors. These common factors can reflect the main information of the original numerous variables. While reducing the number of variables, they also reflect the internal connection between the variables, which is convenient for simplifying data analysis and interpretation.
[0056] In this embodiment, the factor analysis is expressed as:
[0057] x=Af+ε (1)
[0058] In the formula, x represents the observed variable data matrix p, each column is an observed variable, that is, p-dimensional distribution network project data; f is the extracted common factor vector, that is, the new attribute of the distribution network project; A is the factor loading matrix, which represents the linear relationship between the observed variables and the factors; ε is the specific factor vector, which represents the unique part of the observed variables that cannot be explained by the factors;
[0059] The factor loading matrix is obtained by the corresponding formula of the above factor analysis, the common factors are obtained based on the factor loading matrix, and the common factors are used as new attributes of the distribution network project;
[0060] S123, the new attribute includes field information, project keywords, voltage level, project cycle, investment scale, and project type;
[0061] S1231, project keywords;
[0062] Because there are many professional terms in distribution network projects, before obtaining project keywords by Chinese word segmentation of project names, it is necessary to build a distribution network project-specific dictionary, then perform Chinese word segmentation on the project names, extract keywords from the word segmentation results, and further, use the TF-IDF (term frequency-inverse document frequency) algorithm to extract keywords, calculate the words with the highest frequency, and use them as keywords for the project;
[0063] TF-IDF (Term Frequency-Inverse Document Frequency) is used to evaluate the importance of a word to a document set or a document in a corpus, and is expressed as:
[0064] TF-IDF=TF×IDF
[0065] Among them, TF (Term Frequency, TF) is the word frequency, which refers to the frequency of a given word appearing in the document, and is expressed as follows:
[0066]
[0067] Where TF t,d represents the frequency of word t appearing in document d, n t,d Indicates the number of times word t appears in document d, n *,d Represents the total number of words in document d;
[0068] IDF (Inverse Document Frequency, IDF) is the inverse document frequency, which is obtained by dividing the total number of documents by the number of documents containing the term, and then taking the logarithm of the quotient, which can be expressed as:
[0069]
[0070] In the formula, IDF t represents the inverse document frequency of word t, N represents the total number of documents in the collection, and N t Indicates the number of documents in the document collection that contain word t;
[0071] S2. Build an integrated intelligent recommendation model, use historical distribution network project data and historical material procurement data as training sets to train the integrated intelligent recommendation model, and obtain a trained integrated intelligent recommendation model;
[0072] The integrated intelligent recommendation model includes a project recommendation model, a material recommendation model, a material technical specification recommendation model and an intelligent bidding batch arrangement model;
[0073] S21, project recommendation model;
[0074] The project recommendation model uses a fusion edit distance algorithm, a cosine algorithm based on word frequency, and a serialization algorithm according to the project library of the distribution network project, and performs weighted processing to obtain the historical project recommendation results of the project;
[0075] Furthermore, the distribution network project library is used as a training set, and the edit distance algorithm, the cosine algorithm based on word frequency and the serialization algorithm are integrated for weighted processing, a text similarity function is established, and a project recommendation model is constructed and trained. The feature data of the project is used as input, and the historical project recommendation results of the project are obtained based on the trained project recommendation model, where:
[0076] S211, edit distance algorithm
[0077] The edit distance algorithm measures the minimum number of editing operations required to transform one string into another, including inserting, deleting, or replacing characters;
[0078] The edit distance algorithm is expressed as follows:
[0079]
[0080] In the formula, a and b represent strings, i and j represent lengths, and lev a,b (i-1,j)+1 means delete a i ,lev a,b (i,j-1)+1 means delete b i ,lev a,b (i-1,j-1)+1 means replace b i ;
[0081] S212, Cosine similarity algorithm based on word frequency
[0082] The cosine similarity algorithm evaluates the similarity between two vectors by calculating the cosine value of the angle between them, which can be expressed as:
[0083]
[0084] In the formula, the word frequency vector of the material is {keyword X i , word frequency Y i};
[0085] The specific steps are as follows: take a number of keywords from each distribution network project, merge them into a set, calculate the word frequency of each distribution network project for the words in this set; generate a word frequency vector for each distribution network project; calculate the cosine similarity of the distribution network projects to be compared, with a value range of [0, 1], where the larger the value, the more similar it is;
[0086] S213, serialization algorithm;
[0087] The idea of the serialization algorithm is to find the longest continuous matching subsequence that does not contain "junk" elements; the "junk" elements are elements that have no value in some sense, including blank lines or blank characters; it will be recursively applied to the left and right sequence fragments of the matching sequence to produce a match that appears to be "correct";
[0088] The project recommendation model described in this embodiment integrates the edit distance algorithm, the cosine algorithm based on word frequency and the serialization algorithm, performs weighted processing, and establishes a text similarity function. In this embodiment, the text similarity function is obtained by weighting the edit distance algorithm with a weight of 0.3, the cosine algorithm based on word frequency with a weight of 0.4, and the serialization algorithm with a weight of 0.3;
[0089] Construct a project recommendation model, use the distribution network project library as a training set, and obtain a trained project recommendation model; since the distribution project data is time series data and updates quickly, the weighted parameters can be adjusted to obtain a better model in the future by combining the model training results with the update data;
[0090] S22. Material recommendation model
[0091] The material recommendation model obtains the recommended material data set through data preprocessing, Mini Batch K-Means clustering analysis, association rule construction and statistical analysis;
[0092] S221, data preprocessing;
[0093] The data preprocessing includes summarizing the quantity of material subcategories used in each project, obtaining a data set of three fields: project definition, material subcategory code, and material subcategory quantity, and performing matrix transposition on the data set, wherein the matrix has the material subcategory code as a column and the material subcategory quantity as a value. If the project does not use the material, the material subcategory quantity is set to 0; secondly, the transposed data set is normalized with the goal of having the quantity data of different material subcategories at the same level;
[0094] S222. Use the Mini Batch K-Means clustering algorithm to cluster the items in the data set into different item groups based on material usage;
[0095] Mini Batch K-Means clustering algorithm is a variant of K-Means clustering algorithm. K-Means is a common clustering algorithm that distributes data points to different clusters by iteratively optimizing the center points of clusters. Mini Batch K-Means calculates cluster centers by randomly sampling small batches of data points, reducing the amount of calculation and memory requirements.
[0096] Specifically, in this embodiment, the Mini Batch K-Means clustering algorithm is specifically as follows: randomly select k data points as initial cluster centers; for each iteration step, randomly extract a small batch of data from the data set, calculate the distance from each data point to the current cluster center, and assign the data point to the nearest cluster; update the cluster center according to the newly assigned data point; repeat the above steps until the conditions for stopping iteration are met, such as reaching the maximum number of iterations or the cluster center no longer changes significantly;
[0097] S223, association rules;
[0098] The association rule includes an antecedent and a consequent, wherein the antecedent is a project group obtained by merging by matching new attributes through project definition, and the consequent is a material sub-category code. The association rule combination is constructed by pre-set minimum support and minimum confidence. In this embodiment, the minimum support is 0.5 and the minimum confidence is 0.5.
[0099] S224, constructing a recommended material data set;
[0100] Based on the statistical method, the material code and material quantity under each association rule are obtained, and the number of items is aggregated according to the antecedent and the continuation of the association rule. The material code that appears in the most items in each material subcategory under each association rule is obtained, and the proportion of the number of items in which the material code is located and the average number of the material code are obtained to form the recommended material data set;
[0101] S23, recommended model for material technical specifications;
[0102] The material technical specification book recommendation model is based on historical similar projects and material procurement data. By calculating the usage frequency and proportion of technical specifications of similar projects, the technical specifications with the highest usage frequency and the largest proportion are selected as the recommended material technical specifications.
[0103] S24, intelligent bidding batch arrangement model;
[0104] The intelligent bidding batch arrangement model performs information matching based on the annual bidding batch arrangement to obtain the recommended annual bidding batch;
[0105] Furthermore, the intelligent bidding batch arrangement model is based on the annual bidding batch arrangement. It intelligently arranges bidding batches and procurement methods according to the specific demand quantity, date and other data of material procurement. The annual bidding batch adopts the historical annual bidding batch data and the customized batch list, matches the annual bidding batch data according to the project information and material information, and recommends and displays the latest annual bidding batch of the material.
[0106] According to steps S21-S24, user requirements, project-related information, and material procurement-related information are input into the trained integrated intelligent recommendation model to obtain corresponding output results, including historical project recommendation results, recommended material data sets, recommended material technical specifications, and recommended annual bidding batches;
[0107] S3. Based on the winning bid prices in historical material procurement data, the historical price trends are analyzed by statistical methods to obtain the predicted material procurement prices;
[0108] The size of the sliding window is determined according to the periodicity and volatility of the winning bid prices in the historical material procurement data. In this embodiment, the statistical method is the sliding average method, which is a common technique used to smooth time series data to identify trends, reduce random fluctuations in data, and highlight trend changes in data. In this implementation, the analysis and prediction by the sliding average method is specifically to select winning bid price data of several periods according to the size of the selected sliding window, and calculate the average value of the winning bid price data; the average value is used as the predicted value of the price of the next period, that is, the predicted material procurement price;
[0109] S4. The user completes the distribution network material planning management based on the output results and the corresponding predicted material procurement prices.
[0110] Example 2
[0111] This embodiment provides a material personalized recommendation system based on the characteristics of a distribution network project. The system includes a project data acquisition module, an intelligent recommendation module, and a project material declaration data determination and revision module, wherein:
[0112] The project data acquisition module is used to acquire distribution network project data and build a distribution network database;
[0113] The intelligent recommendation module is internally provided with an integrated intelligent recommendation model for intelligently recommending items requiring personalized material recommendations;
[0114] The module for determining project material declaration data and revising is used to select intelligently recommended material information and edit and adjust the selected material information.
[0115] Preferably, the integrated intelligent recommendation model includes a project recommendation model, a material recommendation model, a material technical specification recommendation model and an intelligent bidding batch arrangement model, wherein:
[0116] The project recommendation model uses a fusion edit distance algorithm, a cosine algorithm based on word frequency, and a serialization algorithm according to the project library of the distribution network project, and performs weighted processing to obtain the historical project recommendation results of the project;
[0117] The material recommendation model obtains a recommended material data set through data preprocessing, Mini Batch K-Means clustering analysis, association rule construction and statistical analysis;
[0118] The material technical specification book recommendation model is based on historical similar projects and material procurement data. By calculating the usage frequency and proportion of technical specifications of similar projects, the technical specifications with the highest usage frequency and the largest proportion are selected as the recommended material technical specifications.
[0119] The intelligent bidding batch arrangement model uses the project information and material information to match the annual bidding batch data as input, and recommends and displays the latest annual bidding batch and procurement plan for the materials in combination with the specific needs of material procurement.
[0120] Example 3
[0121] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, a method for personalized material recommendation based on distribution network project characteristics as described in any one of Embodiment 1 is implemented.
[0122] Example 4
[0123] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, a personalized material recommendation method based on distribution network project characteristics as described in any one of Embodiment 1 is implemented.
[0124] It is worth noting that the system, electronic device and computer-readable storage medium described in the present invention are based on the same principles as the method described in Example 1, and will not be described in detail here.
[0125] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A personalized material recommendation method based on distribution network project characteristics, characterized in that: The method comprises: Acquire historical distribution network project data and historical material procurement data, and construct a distribution network project library after data processing of the historical distribution network project data; Constructing an integrated intelligent recommendation model, using historical distribution network project data and historical material procurement data as training sets to train the integrated intelligent recommendation model, and obtaining a trained integrated intelligent recommendation model; Input user requirements, project-related information, and material procurement-related information into the trained integrated intelligent recommendation model to obtain corresponding output results, including project recommendation results, material recommendation results, material technical specification recommendation results, and bidding batch and procurement method recommendation results; According to the winning bid price in the historical material procurement data, the historical price trend is analyzed by the sliding average method to obtain the predicted material procurement price; The user completes the distribution network material planning management based on the output results and the corresponding predicted material procurement prices.
2. According to claim 1, a personalized material recommendation method based on distribution network project characteristics is characterized in that: After processing the historical distribution network project data, a distribution network project library is constructed, specifically: Cleaning the relevant attribute information of the historical distribution network data, wherein the data cleaning includes processing missing values, abnormal values and unifying the data format; Factor analysis is used to analyze and extract relevant attribute information, classify variables, obtain factor loading matrices corresponding to the attributes, obtain common factors based on the factor loading matrix, and use the common factors as new attributes of the distribution network project; A project library of distribution network projects is constructed according to the new attributes.
3. The personalized material recommendation method based on distribution network project characteristics according to claim 1 is characterized in that: The integrated intelligent recommendation model includes a project recommendation model, a material recommendation model, a material technical specification recommendation model and an intelligent bidding batch arrangement model, wherein: The project recommendation model uses a fusion edit distance algorithm, a cosine algorithm based on word frequency, and a serialization algorithm according to the project library of the distribution network project, and performs weighted processing to obtain the historical project recommendation results of the project; The material recommendation model obtains a recommended material data set through data preprocessing, Mini Batch K-Means clustering analysis, association rule construction and statistical analysis; The material technical specification book recommendation model is based on historical similar projects and material procurement data. By calculating the usage frequency and proportion of technical specifications of similar projects, the technical specifications with the highest usage frequency and the largest proportion are selected as the recommended material technical specifications. The intelligent bidding batch arrangement model performs information matching based on the annual bidding batch arrangement to obtain recommended annual bidding batches.
4. The personalized material recommendation method based on distribution network project characteristics according to claim 3 is characterized in that: The project recommendation model uses a fusion edit distance algorithm, a cosine algorithm based on word frequency, and a serialization algorithm according to the project library of the distribution network project, and performs weighted processing to obtain the historical project recommendation results of the project, specifically: The distribution network project library is used as a training set, and the edit distance algorithm, the cosine algorithm based on word frequency and the serialization algorithm are integrated for weighted processing. A text similarity function is established, and a project recommendation model is constructed and trained. Based on the trained project recommendation model, the historical project recommendation results of the project are obtained.
5. The personalized material recommendation method based on distribution network project characteristics according to claim 3 is characterized in that: The material recommendation model obtains the recommended material data set through data preprocessing, Mini Batch K-Means clustering analysis, association rule construction and statistical analysis, specifically: The data preprocessing includes summarizing the quantity of material subcategories used in each project to obtain a data set of three fields: project definition, material subcategory code, and material subcategory quantity; performing matrix transposition on the data set, wherein the matrix has material subcategory codes as columns and material subcategory quantities as values; and performing data normalization processing on the transposed data set with the goal of having quantity data of different material subcategories at the same level; Using the Mini Batch K-Means clustering algorithm, the items in the data set are clustered into different project groups based on material usage; The association rule includes antecedents and contingents, wherein the antecedents are project groups obtained by merging by matching new attributes through project definitions, and the contingents are material sub-category codes. Association rule combinations are constructed through pre-set minimum support and minimum confidence. Based on statistical methods, analyze the material code and quantity under each association rule; According to the number of items in the previous item of the association rule, the recommended material set under each association rule is determined; the occurrence ratio and average quantity of each material code in the project are extracted to construct the final recommended material data set.
6. The personalized material recommendation method based on distribution network project characteristics according to claim 1 is characterized in that: According to the winning bid prices in the historical material procurement data, the trend of historical winning bid prices is analyzed by the sliding average method to obtain the predicted material procurement price, which is as follows: The size of the sliding window is determined according to the periodicity and volatility of the winning bid prices in the historical material procurement data; the winning bid price data of several periods are selected according to the selected sliding window size, and the average value of the winning bid price data is calculated; the average value is used as the predicted value of the price in the next period, that is, the predicted material procurement price.
7. A personalized material recommendation system based on distribution network project characteristics, characterized in that: The system includes a project data acquisition module, an intelligent recommendation module, and a project material declaration data determination and revision module, wherein: The project data acquisition module is used to acquire distribution network project data and build a distribution network database; The intelligent recommendation module is internally provided with an integrated intelligent recommendation model for intelligently recommending items requiring personalized material recommendations; The module for determining project material declaration data and revising is used to select intelligently recommended material information and edit and adjust the selected material information.
8. A material personalized recommendation system based on distribution network project characteristics according to claim 7, characterized in that: The integrated intelligent recommendation model includes a project recommendation model, a material recommendation model, a material technical specification recommendation model and an intelligent bidding batch arrangement model, wherein: The project recommendation model uses a fusion edit distance algorithm, a cosine algorithm based on word frequency, and a serialization algorithm according to the project library of the distribution network project, and performs weighted processing to obtain the historical project recommendation results of the project; The material recommendation model obtains a recommended material data set through data preprocessing, Mini Batch K-Means clustering analysis, association rule construction and statistical analysis; The material technical specification recommendation model is based on historical similar projects and material procurement big data. By calculating the usage frequency and proportion of technical specifications of similar projects, the technical specifications with the highest usage frequency and the largest proportion are selected as the project material procurement demand submission and the recommended optimal technical specifications; The intelligent bidding batch arrangement model uses the project information and material information to match the annual bidding batch data as input, and recommends and displays the latest annual bidding batch and procurement plan for the materials in combination with the specific needs of material procurement.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements a personalized material recommendation method based on distribution network project characteristics as described in any one of claims 1 to 6.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, a personalized material recommendation method based on distribution network project characteristics as described in any one of claims 1 to 6 is implemented.
