Power distribution network project material consumption prediction method

Abstract

The invention relates to a power distribution network project material consumption prediction method, and the method is used for predicting the material consumption of a new project having high similarity with a historical project, and comprises the steps: obtaining a new project material receiving table, historical project basic information and historical material receiving information, and carrying out the matching of a historical similar project similar to the new project; determining material subclasses of newly added projects based on material picking conditions of historical similar projects, and carrying out material consumption prediction from a specific material coding dimension according to an expected consumption quantity of the material subclasses; based on the historical work order picking time and the new work order starting time, predicting the material picking time of each material subclass; and obtaining a final material consumption prediction result in combination with the expected consumption quantity of the predicted material coding dimension and the predicted material receiving time. Compared with the prior art, different strategies are adopted for different types of businesses to carry out sub-item prediction, and material consumption prediction with higher fineness and accuracy is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of material demand prediction, in particular to a power distribution network project material consumption prediction method. BACKGROUND

[0002] Material management is the core support for the efficient operation of enterprise supply chain, and the key is to achieve dynamic balance of procurement, inventory and supply through accurate demand prediction. In the power material procurement link, demand prediction deviation is easy to cause "high price purchase" or "emergency replenishment", leading to rising procurement cost; and the timeliness of supply is also deeply bound with prediction accuracy, and prediction deviation is easy to cause redundant inventory, occupy warehouse space and operating funds, and prediction deviation is small, which may cause shortage of supply, disrupt emergency repair or project planning.

[0003] The power main network material demand has the characteristics of strong planning, concentrated demand, high standardization, and the long-term planning of main network infrastructure projects is clear, and the material supply can be accurately delivered according to the engineering node, and the predictability is strong and the deviation is low. Compared with the main network, the power distribution network material demand has the characteristics of dispersed demand and high frequency, high suddenness and uncertainty, many types of materials, low standardization, and the short-term fluctuation of distribution network infrastructure projects is large, the long tail effect is obvious, the external factors are strongly dependent, and the prediction is more difficult. In recent years, some units in the industry have also made some explorations on the prediction of power distribution network material consumption, but there are still some one-sided or unscientific situations.

[0004] For example, the patent application with the application publication number CN106600032A proposes a method for predicting the demand of inventory materials, which is mainly based on initial investment plan data and historical material subcategory usage data. The business scope of the prediction is power construction projects, and there is no mention of non-project material demand. The prediction result is only to the level of material subcategory, and there is no prediction of specific materials. The material subcategory prediction based on the project has certain significance in the case of self-purchased materials, but if it is centralized procurement, the prediction result is difficult to be directly used by the centralized procurement organization, and there is still a lot of room for improvement.

[0005] For example, the patent application with the application publication number CN119599298A proposes a method for predicting the demand of power grid supply chain distribution network protocol inventory materials based on time series model. It mainly focuses on the prediction of distribution network protocol inventory material demand based on time series model, and combines material characteristic label analysis to predict different materials using different prediction models, and carries out model combination prediction according to different weights. However, the prediction does not explicitly combine the business characteristics of distribution network projects for targeted mining, and only relies on time series model for prediction, and is more focused on data processing. Due to the complexity of power distribution network material demand, even if the time series is optimized, the prediction result is also difficult to achieve ideal effect. SUMMARY

[0006] The purpose of the present application is to provide a power industry distribution network material consumption prediction method for optimizing power industry material procurement and supply plan management, supporting fine inventory management, and solving the problem of difficulty in predicting power industry distribution network material consumption, especially the method of predicting power distribution network infrastructure project material consumption.

[0007] The purpose of the present application can be achieved by the following technical solutions:

[0008] A power distribution network project material consumption prediction method, the method is used for predicting the material consumption of new projects with high similarity to historical projects, including distribution network infrastructure projects and technical transformation projects; the prediction steps include:

[0009] Obtain the new project material requisition form and the historical project basic information and historical material requisition information, match the historical similar projects similar to the new project;

[0010] Based on the material requisition information of the historical similar projects, determine the material subcategory of the new project, and based on the expected consumption quantity of the material subcategory, carry out material consumption prediction in the material code dimension;

[0011] Based on the historical work order material requisition time and the new work order start time, predict the material requisition time of each material subcategory;

[0012] Combine the predicted expected consumption quantity in the material code dimension and the predicted material requisition time to obtain the final material consumption prediction result.

[0013] As a preferred technical solution, the historical project basic information and historical material requisition information are obtained as follows:

[0014] Based on the classification of material consumption prediction, determine the historical project set according to the project code rules, and collect the basic information and historical material requisition information of each project in the historical project set;

[0015] Preprocess the historical project basic information to obtain a project-related basic information database, which includes the project name, project content, project type, project inclusion time, comprehensive plan issuance time, project start time, material receiving and shipping completion time, and material settlement completion time of each project;

[0016] Select the projects whose material use has been completed as effective projects, i.e., the projects whose material receiving and shipping completion time or material settlement completion time have values are the project-related basic information database that meets the prediction conditions;

[0017] The project history material taking information is preprocessed to obtain a project related material taking information library, which includes material number, material description, material group, material category, material subcategory, material taking date, and material taking quantity of each project material taking.

[0018] As a preferred technical solution, the matching and adding similar historical projects are as follows:

[0019] The historical project basic information and historical material taking information are integrated and input into a text large model to define the basic characteristics of each historical project.

[0020] The project information of the new project is input into the text large model to generate a semantic vector, and the cosine similarity of the semantic vector of the new project and each historical project is calculated to match the historical similar projects.

[0021] As a preferred technical solution, for the new capital construction project, the expected consumption quantity of the material subcategory is as follows:

[0022] The transformer capacity and line length information of the historical similar capital construction project is obtained, and the corresponding relationship between the transformer capacity and line length and the material subcategory consumption is input.

[0023] The material subcategory usage quantity of the historical similar capital construction project is obtained, and the expected consumption quantity of the specific material subcategory of the new project is calculated based on the transformer capacity and line length information of the new capital construction project and the corresponding relationship between the transformer capacity and line length and the material subcategory consumption, to obtain the expected consumption quantity of the material subcategory used by the project, and the calculation method is as follows:

[0024] For the material subcategory affected by the transformer capacity: the material subcategory usage quantity of the new capital construction project is equal to the transformer capacity of the new capital construction project divided by the transformer capacity of the historical similar project and multiplied by the material subcategory usage quantity of the historical similar capital construction project.

[0025] For the material subcategory affected by the line length: the material subcategory usage quantity of the new capital construction project is equal to the line length of the new capital construction project divided by the line length of the historical similar project and multiplied by the material subcategory usage quantity of the historical similar capital construction project.

[0026] As a preferred technical solution, for the new technical improvement project, the material subcategory usage quantity of the historical similar technical improvement project is obtained as the expected consumption quantity of the material subcategory used by the new project.

[0027] As a preferred technical solution, the expected consumption quantity of the material subcategory is used to carry out material consumption prediction in the material code dimension, and it is judged in turn whether the project start time, comprehensive plan issuance time, and project listing time exist to realize rolling prediction based on project progress, and the specific steps are as follows:

[0028] If there is a project start time, the material code dimension's material use distribution data under the material small class within a certain period of time after the project start time of the new project is taken as the benchmark;

[0029] If there is no project start time but there is a comprehensive plan delivery time, the material code dimension's material use distribution data under the material small class within a certain period of time after the comprehensive plan delivery time of the new project is taken as the benchmark;

[0030] If there is only a project listing time, the material code dimension's material use distribution data under the material small class within a certain period of time after the project listing time of the new project is taken as the benchmark;

[0031] The material use proportion of each material code dimension in the material small class is obtained from the material use distribution data, and the use records with a material use proportion less than a certain percentage are excluded according to the actual data analysis and business calculation results, and the material use proportion is recalculated to obtain the material use proportion that meets the recent material demand;

[0032] Based on the material use proportion that meets the recent material demand, the expected consumption quantity of the material small class is drilled down to the expected consumption quantity of each material code dimension, i.e., the expected consumption quantity of each material code dimension is equal to the expected consumption quantity of the material small class multiplied by the material use proportion of each material in the material use proportion that meets the recent material demand.

[0033] As a preferred technical solution, the material use time is based on the prediction of the material category, material code and material quantity of the new project, and further material use time prediction is carried out:

[0034] According to the business characteristics, the historical projects are classified, and the relationship between the project listing time, the comprehensive plan delivery time, the project start time and the material use posting time of each type of historical project is analyzed, i.e., the distribution rule of the material use time after the project in different stages is analyzed, and the material use distribution of different material small classes after the different project stages of different types of projects is obtained;

[0035] When the new project enters a certain project stage, such as project listing, comprehensive plan delivery and project start, according to the material use distribution of different material small classes after the different project stages of different types of projects, the material use time of the new project is distributed after the corresponding project stage time point, and the rolling change is carried out with the advancement of the project stage.

[0036] As a preferred technical solution, the material consumption prediction method adopts a time series model to predict material consumption for new projects that do not have similar rules as historical projects, including user engineering projects, daily operation and maintenance projects, and emergency repair projects.

[0037] As a preferred technical solution, the time series model is adopted to predict material consumption for the new projects, specifically as follows:

[0038] The material requisition information of the new projects that do not have similar rules as historical projects is preprocessed to obtain a data set of material requisition data preprocessing;

[0039] Abnormal data conditions are identified, and linear difference or mean insertion methods are selected to repair abnormal data in the preprocessed data set using same period data in previous years;

[0040] External factors including holidays are integrated into the data set after abnormality processing and repair through labeling to obtain a comprehensive data set containing external factors;

[0041] The comprehensive data set is input into the time series model to capture and predict seasonal patterns and trend components of material consumption, and the time series model prediction result is obtained;

[0042] The prediction result of the time series model is analyzed and arranged to obtain the prediction result of other types of material consumption prediction by monthly collection.

[0043] As a preferred technical solution, the time series model adopts a difference autoregressive moving average model.

[0044] Compared with the prior art, the present application has the following beneficial effects:

[0045] 1) The present application carries out power industry distribution network material consumption prediction based on industry and business characteristics, adopts different strategies for different types of business for sub-item prediction, collects historical data for material prediction using a semantic large model and cosine similarity method for new projects with high similarity to historical projects, and uses a time series model for prediction for projects with low similarity, which is more accurate and precise.

[0046] 2) The present application takes the project listing time of the new project as the basis, takes the material code dimension material consumption distribution data under the material small class within a certain period after the project listing time point, and based on the material consumption proportion that meets the recent material demand, the expected consumption quantity of the material small class is drilled down to the expected consumption quantity of each material code dimension. The expected consumption quantity of the material code dimension obtained by the present application is more accurate and precise than the material small class prediction achieved by the prior art.

[0047] 3) Based on the prediction of the material category, material code and the quantity of material taking of the new project, the time of material taking is further predicted. The distribution of the time of material taking in different stages of the project is analyzed, and it is concluded that the time of material taking of the new project is distributed after the corresponding stage of the project, and changes with the progress of the project. The predicted time of material taking is more in line with the long tail characteristics of the actual material taking time, and the prediction of material consumption in the time dimension is more accurate. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 A flowchart for predicting the consumption of materials of a new project by using the method for predicting the consumption of materials of a distribution network in the power industry according to the present application;

[0049] Figure 2 A flowchart for predicting the consumption of materials of a new project by using the method for predicting the consumption of materials of a distribution network in the power industry according to the present application;

[0050] Figure 3 A flowchart for predicting the consumption of materials of a new project by using the method for predicting the consumption of materials of a distribution network in the power industry according to the present application. DETAILED DESCRIPTION

[0051] The present application will be described in detail below with reference to the accompanying drawings and specific embodiments. The present embodiment is implemented on the basis of the technical solution of the present application, and detailed implementation and specific operation process are given, but the protection scope of the present application is not limited to the following embodiments.

[0052] Embodiment 1

[0053] As one of the specific embodiments of the present application, the present embodiment provides a method for predicting the consumption of materials of a power distribution network infrastructure project, as shown in Figure 1 , the specific process is as follows:

[0054] S101, based on the classification of material consumption prediction, the project set of infrastructure project class is determined according to the project code rule, and the historical infrastructure project basic information and historical material taking information of each project in the project set are collected.

[0055] The historical infrastructure project basic information is preprocessed to obtain the infrastructure project related basic information database, which contains the project name, project content, project type, project listing time, comprehensive plan issuing time, project starting time, material receiving and delivering completion time, material settlement completion time, transformer capacity, line length and other information of each project.

[0056] Select the project whose material use has been completed as the effective project, that is, the project whose "material receiving and delivery completion time" or "material settlement completion time" has a value is the project related basic information library that meets the prediction conditions.

[0057] Pretreat the historical material taking information of the capital construction project, including excluding offset data, outliers, and null values, to obtain a capital construction project related material taking information library, which includes material number, material description, material group, material category, material subcategory, material taking date, and material taking quantity information of each project.

[0058] S102, integrate the historical capital construction project basic information and the historical material taking information of the capital construction project and input the text large model to define the basic characteristics of each historical project.

[0059] Input the project information of the new project into the text large model to generate semantic vectors, and match the historical similar capital construction projects by calculating the cosine similarity between the semantic vectors of the new project and each historical project.

[0060] S103, obtain the material taking information of the historical similar capital construction project, take the historical use details of the material subcategory, and lock the material subcategory to determine the material subcategory of the new project.

[0061] S104, obtain the transformer capacity and line length information of the historical similar capital construction project, and input the corresponding relationship between the transformer capacity, line length, and the influence on the material subcategory consumption.

[0062] Obtain the material subcategory usage quantity of the historical similar capital construction project, and calculate the expected consumption quantity of the specific material subcategory of the new project based on the transformer capacity and line length information of the new capital construction project, and the corresponding relationship between the transformer capacity, line length, and the influence on the material subcategory consumption, to obtain the expected consumption quantity of the material subcategory used by the project. The calculation method is as follows:

[0063] ① Material subcategory affected by transformer capacity: material subcategory usage quantity of new capital construction project = (transformer capacity of new capital construction project / transformer capacity of historical similar project) * material subcategory usage quantity of historical similar capital construction project.

[0064] ② Material subcategory affected by line length: material subcategory usage quantity of new capital construction project = (line length of new capital construction project / line length of historical similar project) * material subcategory usage quantity of historical similar capital construction project.

[0065] S105, based on the expected consumption quantity of the material subcategory, carry out material consumption prediction in the material code dimension. Determine whether the project start time, comprehensive plan issuance time, and project inclusion time exist in sequence to realize rolling prediction based on project progress. The specific steps are as follows:

[0066] ①If there is a project start time, take the material code dimension's material use distribution data in a certain period after the project start time of the newly added capital construction project as the benchmark.

[0067] ②If there is no project start time but there is a comprehensive plan issuance time, take the material code dimension's material use distribution data in a certain period after the comprehensive plan issuance time of the newly added capital construction project as the benchmark.

[0068] ③If there is only a project listing time, take the material code dimension's material use distribution data in a certain period after the project listing time of the newly added capital construction project as the benchmark.

[0069] According to the actual data analysis and business estimation results, the use proportion of the material small class in a certain fixed time period shows more obvious continuity, and this fixed time period is taken as the specific time period of the aforementioned "a certain period after a certain time point".

[0070] From the material use distribution data, the material use proportion of each material code dimension in the material small class is obtained, and according to the actual data analysis and business estimation results, the use records with a material use proportion less than a certain percentage are excluded and the material use proportion is recalculated to reduce the impact of inventory historical material consumption or reuse on the material use proportion, and the material use proportion that meets the recent material demand is obtained.

[0071] From the material use proportion that meets the recent material demand, the expected consumption quantity of each material code dimension in the material small class is obtained. That is:

[0072] The expected consumption quantity of each material code dimension = the expected consumption quantity of the material small class * the material use proportion of each material in the material use proportion that meets the recent material demand, for example:

[0073] Table 1 Expected consumption quantity of each material code dimension in the material small class

[0074]

[0075]

[0076] S106, based on the above prediction of the material category, material code and material quantity related to the material use of the newly added capital construction project, further develop the material use time prediction.

[0077] According to the business characteristics, the historical infrastructure projects are classified, and the relationships between the project listing time, the comprehensive plan issuing time, the project starting time and the material receiving and accounting time of each type of historical infrastructure project are analyzed, that is, the distribution of the material receiving time after the infrastructure project in different stages is analyzed, and the distribution of the material receiving of different material subcategories in different types of projects in different project stages is obtained. The key materials such as power cable, high-voltage switch cabinet and insulated conductor are mostly completed within three months after the project is created. The overall distribution presents a long tail feature, that is, most of the material receiving is concentrated in the early stage of the project, and a small amount of material receiving is distributed in a longer time span. In this embodiment, the material receiving time is distributed within three months after the project is issued according to a certain algorithm.

[0078] When a new infrastructure project enters a certain project stage, such as project listing, comprehensive plan issuing and project starting, according to the distribution of the material receiving of different material subcategories in different types of projects in different project stages, the material receiving time of the new infrastructure project is distributed after the corresponding project stage time point, and the distribution is changed rolling with the progress of the infrastructure project stage.

[0079] S107、When there is a new distribution network infrastructure project, the future material receiving of the new project after reaching different project stages can be obtained through the above prediction process, and the monthly collection is performed.

[0080] Embodiment 2

[0081] As another embodiment of the present application, an implementation example of the material consumption prediction for the power distribution network technical improvement project is provided in this embodiment. The technical improvement project does not have the information of the transformer capacity and the line length, but has rich project content text description information, which is more conducive to text analysis, such as Figure 2 The specific process is as follows:

[0082] S201、Based on the classification of material consumption prediction, the project set of the technical improvement project class is determined according to the project coding rule, and the basic information and historical material information of each project in the project set are collected.

[0083] The historical technical improvement project basic information is preprocessed to obtain a technical improvement project related basic information database, which includes the project name, project content, project type, project listing time, comprehensive plan issuing time, project starting time, material receiving and shipping completion time, material settlement completion time and other information of each project.

[0084] The projects whose material receiving has been completed are selected as effective projects, that is, the technical improvement projects with the values of the aforementioned “material receiving and shipping completion time” or “material settlement completion time” are the technical improvement project related basic information database that meets the prediction conditions.

[0085] The historical material requisition information of the technical transformation project is preprocessed, including excluding offset data, outliers, null values, etc., to obtain a project-related material requisition information library, which includes material number, material description, material group, material category, material subcategory, material requisition date, and material requisition quantity.

[0086] S202, the historical technical transformation project basic information and the technical transformation project historical material requisition information are integrated and input into the AI large model to define the basic characteristics of each historical project. Since the text description of the technical transformation project content is very rich, the definition of the characteristics of the historical technical transformation project is also more rich.

[0087] The project information of the newly added technical transformation project is input into the AI large model for similarity calculation to match the historical similar technical transformation project.

[0088] S203, the material requisition situation of the historical similar technical transformation project is obtained, the historical use details of the material subcategory are taken, and the material subcategory is locked to determine the material subcategory of the newly added project.

[0089] S204, the material subcategory usage quantity of the historical similar technical transformation project is obtained as the expected consumption quantity of the material subcategory used by the newly added project.

[0090] From the material subcategory expected consumption quantity, the material consumption quantity prediction is carried out to the material code dimension. Whether the project start time, comprehensive plan issuance time, and project inclusion time exists is judged in turn to realize the rolling prediction based on the project progress. The specific steps are as follows:

[0091] ①If the project start time exists, the material code dimension material requisition distribution data of the material subcategory within a certain period of time after the project start time of the newly added technical transformation project is taken as the reference.

[0092] ②If the project start time does not exist but the comprehensive plan issuance time exists, the material code dimension material requisition distribution data of the material subcategory within a certain period of time after the comprehensive plan issuance time of the newly added technical transformation project is taken as the reference.

[0093] ③If only the project inclusion time exists, the material code dimension material requisition distribution data of the material subcategory within a certain period of time after the project inclusion time of the newly added technical transformation project is taken as the reference.

[0094] According to the actual data analysis and business calculation results, it is determined that the use proportion of the material subcategory requisition situation within a certain fixed time period presents more obvious continuity, and the fixed time period is taken as the specific time period within a certain period of time after the certain time point.

[0095] The material consumption proportion of each material code dimension in the material small class is obtained from the material consumption distribution data, and the material consumption proportion is recalculated by excluding the consumption records with a consumption proportion below a certain percentage according to the actual data analysis and business calculation results, so as to reduce the influence of the stock history material consumption or reuse on the material consumption proportion, and obtain the material consumption proportion in line with the recent material demand.

[0096] The expected consumption quantity of each material code dimension is obtained from the material consumption proportion in line with the recent material demand.

[0097] The expected consumption quantity of each material code dimension = the expected consumption quantity of the material small class * the material consumption proportion of each material in the material consumption proportion in line with the recent material demand

[0098] S205, based on the above prediction of the material consumption related material category, material code and material consumption quantity of the new technical transformation project, further carry out the material consumption time prediction.

[0099] According to the business characteristics, the historical technical transformation projects are classified, and the relationship between the project listing time, comprehensive plan issuing time, project starting time and material consumption posting time of each type of historical technical transformation project is analyzed, that is, the distribution rule of the material consumption time after the technical transformation project in different stages is analyzed, and the material consumption distribution of different material small classes in different project stages of different types of projects is obtained.

[0100] When the new technical transformation project enters a certain project stage, such as project listing, comprehensive plan issuing and project starting, according to the material consumption distribution of different material small classes in different project stages of different types of projects, the material consumption time of the new technical transformation project is distributed after the corresponding project stage time point, and the rolling change is carried out with the progress of the technical transformation project stage.

[0101] S206, when there is a new distribution network technical transformation project, the future material consumption of the new project after reaching different project stages can be obtained through the above prediction process, and the monthly collection is carried out.

[0102] Embodiment 3

[0103] As a third embodiment of the present application, in this embodiment, an implementation example of distribution network other type material consumption prediction is provided, and the material consumption of the user engineering project, daily operation and maintenance project and emergency repair project and other distribution network material consumption without obvious characteristics is classified as distribution network other type material consumption, and a time series prediction model is used for material consumption prediction, as shown in Figure 3 The specific steps are as follows:

[0104] S301, pre-process the other material information of the power distribution network, including excluding offset data, outliers, null values, etc., to form a data set of the pre-processed other material information of the power distribution network. The pre-processed data set includes material number, material description, material group, material category, material subcategory, material subcategory, material date, material quantity, etc.

[0105] S302, identify abnormal data conditions caused by special factors, select the same period data in previous years, and apply linear difference or mean insertion method to repair the abnormal data in the pre-processed data set to obtain a data set repaired by abnormal processing.

[0106] S303, integrate external factors into the data set, such as epidemic situation, holiday, etc., in order to better explain and predict material consumption. In this embodiment, important holidays such as Spring Festival, New Year's Day, May Day, National Day, etc. are added to the data source by tagging, and the external factors such as holidays are integrated into the data set repaired by abnormal processing to obtain a comprehensive data set containing external factors, i.e. the final power distribution network other material related information library meeting the time series prediction conditions. Integrate external factors into the data set to better explain and predict material consumption.

[0107] S304, input the comprehensive data set obtained by the foregoing processing, introduce the time series autoregressive moving average model (ARIMA) to capture and predict the seasonal pattern and trend component of material consumption, and obtain the model prediction result.

[0108] S305, analyze and arrange the prediction result of the ARIMA model to obtain the prediction result of the monthly collection of other material consumption prediction.

[0109] Integrate the prediction results of the foregoing each item of material consumption and collect them by month. The collected data is the power distribution network material consumption prediction result, which shows the predicted consumption of the power distribution network materials in each month in the predictable time range. The core information is as follows:

[0110] Table 2 Predicted consumption of power distribution network materials in each month in the predictable time range

[0111]

[0112] The above detailed the preferred embodiments of the present application. It should be understood that those skilled in the art can make many modifications and changes without creative labor according to the concept of the present application. Therefore, any technical solution obtained by logical analysis, reasoning or limited experiment based on the prior art according to the concept of the present application shall be within the protection scope determined by the claims.

Claims

1. A power distribution network project material consumption prediction method, characterized by, The method is used for material consumption of new projects with high similarity to historical projects, including distribution network infrastructure projects and technical transformation projects; the prediction step includes: Obtaining the material requisition form of the new project and the basic information and historical material information of the historical project, matching the similar historical similar projects similar to the new project; Based on the material requisition of the historical similar projects, the material subcategory of the new project is determined, and the material consumption prediction in the material code dimension is carried out based on the expected consumption quantity of the material subcategory; Based on the historical work order material requisition time and the new work order start time, the material requisition time of each material subcategory is predicted; The final material consumption prediction result is obtained by combining the predicted expected consumption quantity in the material code dimension and the predicted material requisition time.

2. The method of claim 1, wherein, The basic information and historical material information of the historical project are obtained as follows: Based on the classification of material consumption prediction, the historical project set is determined according to the project code rule, and the basic information and historical material information of each project in the historical project set are collected; The historical project basic information is preprocessed to obtain the project related basic information database, which includes the project name, project content, project type, project listing time, comprehensive plan issuance time, project start time, material receiving and shipping completion time, and material settlement completion time of each project; Select the project whose material requisition has been completed as the effective project, that is, the project related basic information database that meets the prediction condition has a value of material receiving and shipping completion time or material settlement completion time; The historical material information of the project is preprocessed to obtain the project related material information database, which includes the material number, material description, material group, material category, material subcategory, material requisition date, and material quantity of each project.

3. The method of claim 1, wherein, The matching of the similar historical similar projects similar to the new project is as follows: Integrate the historical project basic information and historical material information and input the text large model to define the basic characteristics of each historical project; Input the project information of the new project into the text large model to generate a semantic vector, and match the historical similar projects by calculating and comparing the cosine similarity of the semantic vectors of the new project and each historical project.

4. The method of claim 1, wherein, For new infrastructure projects, the expected consumption quantity of the material subcategory is as follows: Obtain the transformer capacity and line length information of the similar historical infrastructure projects, input the corresponding relationship between transformer capacity and line length and the influence of material subcategory consumption; Obtain the material subcategory usage quantity of the similar historical infrastructure projects, and calculate the expected consumption quantity of the specific material subcategory of the new project based on the transformer capacity and line length information of the new infrastructure project and the corresponding relationship between transformer capacity and line length and the influence of material subcategory consumption, to obtain the expected consumption quantity of the material subcategory used by the project, the calculation method is as follows: For material subcategories affected by transformer capacity: the material subcategory usage quantity of the new infrastructure project is equal to the transformer capacity of the new infrastructure project divided by the transformer capacity of the historical similar project and multiplied by the material subcategory usage quantity of the historical similar infrastructure project; For the material subcategory affected by the length of the line: the material subcategory usage quantity of the newly added infrastructure project is equal to the length of the newly added infrastructure project divided by the length of the historical similar project and multiplied by the material subcategory usage quantity of the historical similar infrastructure project.

5. The method of claim 1, wherein, For the newly added technical improvement project, the material subcategory usage quantity of the historical similar technical improvement project is obtained as the expected consumption quantity of the material subcategory used by the newly added project.

6. The method of claim 1, wherein, The expected consumption quantity of the material subcategory is used to carry out material consumption prediction in the material code dimension, and it is judged in turn whether the project start time, comprehensive plan issuance time, and project listing time exist, so as to realize rolling prediction based on project progress, and the specific steps are as follows: If the project start time exists, the material code dimension material consumption distribution data of the material subcategory within a certain period after the project start time of the newly added project is taken as the reference; If the comprehensive plan issuance time exists, the material code dimension material consumption distribution data of the material subcategory within a certain period after the comprehensive plan issuance time of the newly added project is taken as the reference; If only the project listing time exists, the material code dimension material consumption distribution data of the material subcategory within a certain period after the project listing time of the newly added project is taken as the reference; The material consumption proportion of each material code dimension in the material subcategory is obtained from the material consumption distribution data, and the material consumption records with a consumption proportion less than a certain percentage are excluded according to the actual data analysis and business calculation results, and the material consumption proportion is recalculated to obtain the material consumption proportion that meets the recent material demand; Based on the material consumption proportion that meets the recent material demand, the expected consumption quantity of the material subcategory is drilled down to the expected consumption quantity of each material code dimension, i.e., the expected consumption quantity of each material code dimension is equal to the expected consumption quantity of the material subcategory multiplied by the material consumption proportion of each material in the material consumption proportion that meets the recent material demand.

7. The method of claim 1, wherein, The material consumption time is based on the prediction of the material category, material code, and material quantity of the newly added project, and further material consumption time prediction is carried out: According to the business characteristics, the historical projects are classified, and the relationship between the project listing time, comprehensive plan issuance time, project start time, and material consumption posting time of each type of historical project is analyzed, i.e., the distribution rule of material consumption time after the project enters different stages is analyzed, and the material consumption distribution of different material subcategories in different project stages of different types of projects is obtained; When the newly added project enters a certain project stage, such as project listing, comprehensive plan issuance, and project start, the material consumption time of the newly added project is distributed after the corresponding project stage time point according to the material consumption distribution of different material subcategories in different project stages of different types of projects, and it changes rolling with the progress of the project stage. 8.The power distribution network project material consumption prediction method of claim 1, wherein, The material consumption prediction method adopts a time series model to predict material consumption for new projects that do not have similar rules as historical projects, including user engineering projects, daily operation and maintenance projects, and emergency repair projects.

9. The method of claim 8, wherein, The time series model is adopted to predict material consumption for the new projects, specifically as follows: The material requisition information of the new projects that do not have similar rules as historical projects is preprocessed to obtain a data set of preprocessed material requisition data; Abnormal data conditions are identified, and linear difference or mean insertion methods are selected to apply to the preprocessed data set to repair abnormal data; External factors including holidays are integrated into the data set after abnormality processing and repair by labeling to obtain a comprehensive data set containing external factors; The comprehensive data set is input into the time series model to capture and predict seasonal patterns and trend components of material consumption, and a time series model prediction result is obtained; The prediction result of the time series model is analyzed and arranged to obtain a prediction result of other types of material consumption prediction collected by month.

10. The method of claim 8, wherein, The time series model adopts a difference autoregressive moving average model.

Images