Power grid material demand information prediction method and device based on multi-source data fusion

Abstract

The invention relates to a power grid material demand information prediction method and device based on multi-source data fusion. The method comprises the following steps: acquiring power grid material description information corresponding to each power grid material project object; determining clustering index information corresponding to the power grid material project object according to the power grid material description information corresponding to the power grid material project object; according to a preset clustering rule and the clustering index information corresponding to each power grid material project object, performing clustering processing on each power grid material project object to obtain a plurality of clustering clusters; determining material recommendation list information corresponding to the clustering cluster according to the power grid material description information of each power grid material project object in the clustering cluster; and inputting the power grid material description information of each power grid material project object in the clustering cluster and the material recommendation list information corresponding to the clustering cluster into a material demand prediction model to obtain a material demand prediction result of the clustering cluster. The method can improve the accuracy of power grid material demand prediction.

Description

Technical Field

[0001] The present application relates to the field of big data processing technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for predicting power grid material demand information based on multi-source data fusion. Background Art

[0002] Grid material demand forecasting is a critical prerequisite for grid material management, ensuring the supply of materials for grid construction and operation and improving resource allocation efficiency. With the development of intelligent grids, grid material demand forecasting is gradually being integrated with artificial intelligence algorithms.

[0003] Traditionally, power grid material demand forecasting has primarily relied on historical consumption data and static single-item attributes. For example, these methods employ traditional time series analysis or basic machine learning models. These methods rely on structured data such as historical shipment records and item type labels, mining temporal trends or using simple classification rules to predict future demand for power grid materials over a short or long period of time. However, these methods struggle to cope with complex power grid scenarios, resulting in low accuracy in power grid material demand forecasting. Summary of the Invention

[0004] Based on this, it is necessary to provide a power grid material demand information prediction method, device, computer equipment, computer-readable storage medium and computer program product based on multi-source data fusion to improve the accuracy of power grid material demand prediction in response to the above technical problems.

[0005] In a first aspect, the present application provides a method for predicting power grid material demand information based on multi-source data fusion, comprising:

[0006] Obtain power grid material description information corresponding to each power grid material project object; the power grid material description information at least includes project basic feature information, material attribute feature information, supply chain feature information and material recommendation feature information; determine clustering index information corresponding to the power grid material project object based on the power grid material description information corresponding to the power grid material project object; the clustering index information at least includes demand urgency index information, material complexity index information and material recommendation index information; cluster each power grid material project object according to preset clustering rules and the clustering index information corresponding to each power grid material project object to obtain multiple cluster clusters; determine material recommendation list information corresponding to the cluster cluster based on the power grid material description information of each power grid material project object in the cluster cluster; input the power grid material description information of each power grid material project object in the cluster cluster and the material recommendation list information corresponding to the cluster cluster into a material demand forecasting model to obtain a material demand forecasting result for the cluster cluster.

[0007] In one embodiment, the determining of the material recommendation list information corresponding to the cluster cluster based on the power grid material description information of the power grid material project objects included in the cluster cluster includes: selecting a material recommendation model matching the cluster cluster from a material recommendation model library based on the cluster characteristics of the cluster cluster; the cluster characteristics include the indicator distribution interval of the cluster indicator information corresponding to each of the power grid material project objects in the cluster cluster; and inputting the power grid material description information of each of the power grid material project objects in the cluster cluster into the material recommendation model matching the cluster cluster to obtain the material recommendation list information corresponding to the cluster cluster.

[0008] In one embodiment, the power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster are input into a material demand forecasting model to obtain the material demand forecasting result of the cluster, including: determining a material demand forecasting model that matches the cluster according to the cluster characteristics of the cluster; inputting the power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster into the material demand forecasting model that matches the cluster to obtain the material demand forecasting result of the cluster.

[0009] In one embodiment, the determining of clustering index information corresponding to the power grid material project object based on the power grid material description information corresponding to the power grid material project object includes: inputting the power grid material description information corresponding to the power grid material project object into a demand urgency calculation model to obtain the demand urgency index information; the demand urgency calculation model is used to determine the demand urgency index information based on the project importance information, seasonal risk information, policy-driven information and supply chain resilience information in the power grid material description information; inputting the power grid material description information corresponding to the power grid material project object into a logistics complexity calculation model to obtain the material complexity index information; the logistics complexity calculation model is used to determine the material complexity index information based on the material technical parameter information, material compatibility risk information, supplier concentration information, alternative material information and full life cycle cost information in the power grid material description information; inputting the power grid material description information corresponding to the power grid material project object into a material recommendation calculation model to obtain the material recommendation index information; the material recommendation calculation model is used to determine the material recommendation index information based on material matching information, project similarity information and material cost-effectiveness information.

[0010] In one embodiment, clustering is performed on each of the power grid material project objects according to the clustering index information corresponding to each of the power grid material project objects according to preset clustering rules, including: determining a membership model that matches the clustering index information according to quantitative demand information of the clustering index information; inputting the clustering index information into the matching membership model to obtain the membership corresponding to the clustering index information; the membership model is used to convert the clustering index information into the corresponding membership through a membership function that matches the clustering index information; and clustering is performed on each of the power grid material project objects according to the preset clustering rules and the membership corresponding to each of the power grid material project objects.

[0011] In one embodiment, the obtaining of the power grid material description information corresponding to each power grid material project object includes: the obtaining of the power grid material description information corresponding to each power grid material project object includes: obtaining multi-source data of power grid materials; the multi-source data of power grid materials include at least two types of power grid material data of project text description data, historical material consumption time series data, supply chain logistics status data and model data of a material recommendation model library; standardizing each of the power grid material data to obtain each standardized data; according to the project identification of the power grid material project object, semantically fusing each of the standardized data belonging to the same power grid material project object to obtain a structured feature matrix corresponding to each of the power grid material project objects; each row of the structured feature matrix corresponds to each of the power grid material project objects, and each column of the structured feature matrix corresponds to each of the standardized data; the structured feature matrix corresponding to each of the power grid material project objects is used as the power grid material description information corresponding to each of the power grid material project objects.

[0012] In a second aspect, the present application further provides a power grid material demand information prediction device based on multi-source data fusion, comprising: an acquisition module for acquiring power grid material description information corresponding to each power grid material project object; the power grid material description information includes at least project basic feature information, material attribute feature information, supply chain feature information and material recommendation feature information;

[0013] An indicator determination module is used to determine clustering indicator information corresponding to the power grid material project object based on the power grid material description information corresponding to the power grid material project object; the clustering indicator information at least includes demand urgency indicator information, material complexity indicator information and material recommendation index information;

[0014] A clustering module, configured to perform clustering processing on each of the power grid material project objects according to a preset clustering rule and the clustering indicator information corresponding to each of the power grid material project objects to obtain a plurality of cluster clusters;

[0015] A list determination module is used to determine the recommended material list information corresponding to the cluster according to the power grid material description information of each power grid material project object in the cluster;

[0016] The prediction module is used to input the power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster into the material demand prediction model to obtain the material demand prediction result of the cluster.

[0017] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0018] In a fourth aspect, the present application also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor.

[0019] In a fifth aspect, the present application also provides a computer program product, comprising a computer program, which implements the steps of the above method when executed by a processor.

[0020] The above-mentioned power grid material demand information prediction method, device, computer equipment, computer-readable storage medium and computer program product based on multi-source data fusion obtains power grid material description information corresponding to each power grid material project object; the power grid material description information at least includes project basic feature information, material attribute feature information, supply chain feature information and material recommendation feature information; based on the power grid material description information corresponding to the power grid material project object, the clustering index information corresponding to the power grid material project object is determined; the clustering index information at least includes demand urgency index information, material complexity index information and material recommendation index information; according to the preset clustering rules and the clustering index information corresponding to each power grid material project object, each power grid material project object is clustered to obtain multiple cluster clusters; based on the power grid material description information of each power grid material project object in the cluster cluster, the material recommendation list information corresponding to the cluster cluster is determined; the power grid material description information of each power grid material project object in the cluster cluster and the material recommendation list information corresponding to the cluster cluster are input into the material demand prediction model to obtain the material demand prediction result of the cluster cluster. In this way, by obtaining multi-source power grid material description information covering project basic characteristics, material attribute characteristics, supply chain characteristics and material recommendation characteristics, comprehensive coverage of power grid material demand related information is achieved; by constructing multi-dimensional indicators of demand urgency, material complexity and material recommendation degree, the ability to characterize complex power grid scenarios is significantly improved, providing highly discriminative input features for subsequent cluster analysis; corresponding material recommendation lists are generated for clusters, and the power grid material description information and material recommendation list information of each cluster are jointly input into the material demand forecasting model to obtain comprehensive and accurate material demand forecasting results, which can cope with demand fluctuations in complex power grid scenarios, thereby improving the accuracy of power grid material demand forecasting. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 This is a diagram of an application environment of a method for predicting power grid material demand information based on multi-source data fusion in one embodiment;

[0023] Figure 2 1 is a flow chart of a method for predicting power grid material demand information based on multi-source data fusion in one embodiment;

[0024] Figure 3Schematic diagram of a flow chart of a method for predicting power grid material demand information based on multi-source data fusion in another embodiment;

[0025] Figure 4 This is a structural block diagram of a power grid material demand information prediction device based on multi-source data fusion in one embodiment;

[0026] Figure 5 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0028] Current methods for forecasting power grid material demand are mainly based on historical consumption data and static single-item attributes, such as traditional time series analysis or basic machine learning models. These methods mainly rely on structured data such as historical delivery records and item type labels, and predict the demand for power grid materials in the future or over a long period of time by mining time trends or simple classification rules. They do not consider the real-time status of the supply chain, environmental factors, and standardized material inventory rules. As a result, the forecasting model is prone to insufficient dynamic adaptability and is unable to cope with demand fluctuations in complex supply environments.

[0029] Furthermore, while current power grid material demand forecasting methods can group projects using clustering algorithms, these clustering methods are limited to one-dimensional time series or simple classification features, failing to fully integrate heterogeneous data from multiple sources. Furthermore, the classification of project types is not sufficiently refined, leading to the forced merging of subcategories with significantly different demand within the same project category. Sometimes, due to insufficient feature dimensionality, these subcategories cannot be effectively distinguished, resulting in forecast bias. These shortcomings collectively lead to insufficiently accurate forecast results, making it difficult to support refined material scheduling and cost optimization.

[0030] The power grid material demand information prediction method based on multi-source data fusion provided in the embodiment of the present application can be applied to Figure 1In the application environment shown, the terminal 102 communicates with the server 104 via a network. The data storage system can store data that the server 104 needs to process. The data storage system can be integrated on the server 104 or placed on the cloud or other network servers. Terminal 102 obtains grid material description information corresponding to each grid material project object; the grid material description information includes at least basic project feature information, material attribute feature information, supply chain feature information, and material recommendation feature information; terminal 102 determines clustering index information corresponding to the grid material project object based on the grid material description information corresponding to the grid material project object; the clustering index information includes at least demand urgency index information, material complexity index information, and material recommendation index information; terminal 102 clusters each grid material project object according to a preset clustering rule and the clustering index information corresponding to each grid material project object to obtain a plurality of clusters; terminal 102 determines material recommendation list information corresponding to the cluster cluster based on the grid material description information of each grid material project object in the cluster; terminal 102 inputs the grid material description information of each grid material project object in the cluster cluster and the material recommendation list information corresponding to the cluster cluster into a material demand forecasting model to obtain a material demand forecast result for the cluster cluster. Terminal 102 may be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, and Internet of Things devices. The server 104 may be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0031] In an exemplary embodiment, Figure 2 As shown in the figure, a method for predicting power grid material demand information based on multi-source data fusion is provided. Figure 1 The terminal 102 in FIG. 1 is used as an example for explanation, including:

[0032] Step S202: Obtain the power grid material description information corresponding to each power grid material project object.

[0033] Among them, the power grid material project objects are specific material demand projects involved in the process of power grid construction, operation and maintenance, and inspection, and are the basic units for material demand forecasting and management.

[0034] It should be noted that the materials in this application may include standard materials. Standard materials are produced or purchased in accordance with unified technical standards, specifications, models, performance parameters, and other requirements in the power grid field. They have standardized properties and can be used in power grid construction, operation and maintenance, and repair. They can ensure the versatility, interchangeability, and quality consistency of the materials, facilitating material management, demand forecasting, and efficient supply.

[0035] The power grid material description information is a collection of various information used to describe the power grid material project object. The power grid material description information at least includes project basic feature information, material attribute feature information, supply chain feature information, and material recommendation feature information.

[0036] Among them, the basic characteristic information of the project is information that describes the basic attributes of the power grid material project itself, including project type number, geographical location code, project budget amount, power grid link (such as transmission, distribution), implementation time, etc.

[0037] Among them, material attribute characteristic information is information reflecting the characteristics of the materials involved in the project, including historical consumption frequency, material complexity, material type, specifications, models, technical parameters, etc.

[0038] Among them, supply chain characteristic information refers to various types of information related to material supply, including logistics delay risk score, supplier information, inventory status, logistics transportation conditions, procurement cycle and other characteristics of the supply chain links.

[0039] Among them, material recommendation feature information is feature information used to assist in material recommendation decisions, including indicators related to material recommendation, such as the idle material matching degree and the framework agreement discount rate. The idle material matching degree refers to the degree of compatibility between the materials required for the power grid material project and the existing idle materials in terms of type, specification, and performance. It is a quantitative indicator that measures whether the idle materials can meet project needs. A higher value indicates a closer match. The framework agreement discount rate is the price discount ratio given by the supplier for specific materials under the framework agreement procurement model. It reflects the cost benefit that can be obtained by purchasing the materials through the framework agreement. A higher discount rate means a relatively lower procurement cost.

[0040] Step S204: determining clustering index information corresponding to the power grid material project object based on the power grid material description information corresponding to the power grid material project object.

[0041] Among them, the clustering index information is a set of quantitative indicators extracted from the power grid material description information and used to measure the similarity of the characteristics of power grid material project objects, providing a judgment basis for the clustering processing of project objects.

[0042] The clustering index information includes at least demand urgency index information, material complexity index information, and material recommendation index information.

[0043] The demand urgency index information is quantitative information reflecting the urgency of the demand for materials of the power grid material project.

[0044] Among them, the material complexity index information is information that quantifies the complexity of materials based on their technical parameters, compatibility, supply sources and other attributes.

[0045] Among them, the material recommendation index information is information that quantifies the degree to which the material is suitable for the project by comprehensively considering factors such as the matching degree between the material and the project and cost-effectiveness, and is used to evaluate the adaptability of the material recommendation.

[0046] In the specific implementation, a three-dimensional clustering index is constructed based on the project type of different power grid material project objects, including demand urgency index information, material complexity index information and material recommendation index information; each power grid material project object is mapped to the three-dimensional clustering space through the membership function to generate a three-dimensional membership vector, and then each power grid material project object is clustered according to the preset clustering rules and the membership vector corresponding to each power grid material project object to obtain multiple cluster clusters.

[0047] Step S206 , clustering the power grid material project objects according to the preset clustering rules and the clustering indicator information corresponding to the power grid material project objects to obtain a plurality of clusters.

[0048] Among them, the preset clustering rules are pre-set criteria or algorithm logic for determining the clustering affiliation of power grid material project objects, covering the weight distribution of clustering indicators, similarity measurement methods (such as distance calculation methods), cluster division thresholds, etc., providing a clear operational basis for the clustering processing of project objects.

[0049] In a specific implementation, clustering processing of each power grid material project object may include a process of grouping and classifying multiple power grid material project objects according to the similarity of their clustering index information based on preset clustering rules, thereby aggregating project objects with similar characteristics into the same group.

[0050] Among them, the cluster is a collection of power grid material project objects with similar clustering indicator information formed after clustering processing.

[0051] Step S208 : determining the recommended material list information corresponding to the cluster according to the grid material description information of each grid material project object in the cluster.

[0052] In a specific implementation, each cluster is associated with at least one material recommendation model in the material recommendation model library, and material recommendation list information adapted to the characteristics of the cluster is automatically generated through the material recommendation model.

[0053] Among them, the recommended material list information is a list of information generated based on the common needs of the power grid material project objects within the cluster, combined with their power grid material description information (such as project characteristics, material attributes, recommended characteristics, etc.), which includes the types, specifications, quantity recommendations, etc. of materials suitable for the cluster, and is used to clarify the scope of recommended materials corresponding to the cluster.

[0054] In a specific implementation, the recommended material list information may include the material model, inventory location, supplier agreement terms and priority score.

[0055] Step S210 : inputting the grid material description information of each grid material project object in the cluster and the material recommendation list information corresponding to the cluster into the material demand forecasting model to obtain the material demand forecasting result of the cluster.

[0056] Among them, the material demand forecasting model is an artificial intelligence model used to predict the demand for power grid materials. It can receive the power grid material description information and the corresponding material recommendation list information in the cluster. According to the material recommendation list information, the recommended material range that is compatible with the cluster can be screened out, so that the material demand forecast can be carried out from the limited recommended material range combined with the power grid material description information, thus eliminating the need to traverse all materials, thereby improving the efficiency and accuracy of material forecasting.

[0057] In specific implementation, the training process of the material demand prediction model may include: collecting grid material description information (including project basic characteristics, material attribute characteristics, etc.) of historical grid material projects and corresponding material recommendation list information, and using the actual material demand results of these projects as label data; then, combining the historical grid material description information and the material recommendation list information as input features, and the actual demand results as output labels, and dividing the training set and the validation set; then, using the training set to train the initial model, and continuously adjusting the model parameters through algorithms such as back propagation to gradually reduce the error between the predicted value output by the model and the actual demand results; at the same time, using the validation set to monitor the training effect of the model to avoid problems such as overfitting. After multiple iterative optimizations, the model performance reaches the preset standard and the training is completed.

[0058] Among them, the material demand prediction result of the cluster is the prediction output of the material demand prediction model on the quantity, time and other aspects of the materials required by the cluster, which may include material demand, time distribution and priority sorting.

[0059] In the above-mentioned power grid material demand information prediction method based on multi-source data fusion, comprehensive coverage of power grid material demand related information is achieved by obtaining multi-source power grid material description information covering project basic characteristics, material attribute characteristics, supply chain characteristics and material recommendation characteristics; by constructing multi-dimensional indicators of demand urgency, material complexity and material recommendation degree, the ability to characterize complex power grid scenarios is significantly improved, and high-discrimination input features are provided for subsequent clustering analysis; corresponding material recommendation lists are generated for clusters, and the power grid material description information and material recommendation list information of each cluster are jointly input into the material demand prediction model to obtain comprehensive and accurate material demand prediction results, which can cope with demand fluctuations in complex power grid scenarios, thereby improving the accuracy of power grid material demand prediction.

[0060] In another embodiment, obtaining power grid material description information corresponding to each power grid material project object includes: obtaining multi-source data of power grid materials; the multi-source data of power grid materials includes at least two types of power grid material data among project text description data, historical material consumption time series data, supply chain logistics status data and model data of a material recommendation model library; standardizing each power grid material data to obtain each standardized data; semantically fusion of each standardized data belonging to the same power grid material project object according to the project identification of the power grid material project object to obtain a structured feature matrix corresponding to each power grid material project object; each row of the structured feature matrix corresponds to each power grid material project object, and each column of the structured feature matrix corresponds to each standardized data; and using the structured feature matrix corresponding to each power grid material project object as the power grid material description information corresponding to each power grid material project object.

[0061] In the specific implementation, multi-source heterogeneous data can be obtained in real time from the power grid material project demand pool as power grid material multi-source data, including a variety of power grid material data, such as: project text description, historical material consumption time series data, supply chain logistics status data, and model data of the material recommendation model library; then the power grid material multi-source data is standardized and semantically fused to generate a structured feature matrix.

[0062] Project description data includes project name, project type, project location, technical requirements, and a summary of approval documents. For example, the description of a "Rural Power Grid Reconstruction" project includes: "Project Number: GD-2025-015; Type: Infrastructure; Location: XX County, Guangdong Province; Requirements: Add 50 pole-mounted circuit breakers and replace old aluminum-core cables with copper-core insulated cables."

[0063] Among them, the historical material consumption time series data includes material ID, collection time, collection quantity, material unit price, using department, related project ID, etc.

[0064] Supply chain logistics status data includes supplier name, order number, material transportation progress, estimated arrival time, actual arrival time, and reasons for logistics delays. For example, the logistics status for order number "SUP-250401" is: supplier "XX Cable Factory", estimated arrival time 2025-04-10, actual arrival time 2025-04-15, and the reason for the delay is "typhoon causing highway closure."

[0065] In addition, multi-source data on power grid materials may also include equipment failure rate data, which includes equipment model, installation time, cumulative operating time, number of failures, failure type, maintenance records, etc. For example, equipment failure rate data may include: equipment model "SWITCH-35kV-002" has accumulated 3 failures in 2025, and the failure type is "poor contact due to contact oxidation."

[0066] Among them, the material recommendation model library may include multiple material recommendation models: idle material standard list recommendation model, reverse material standard list recommendation model, long-term unfulfilled contract standard material list recommendation model, centralized procurement material standard list recommendation model, reserve material standard list recommendation model, effective framework agreement material standard list recommendation model, etc.

[0067] The model data of the material recommendation model library may include historical values ​​of model input and output, data range, data type, data format, etc.

[0068] In specific implementations, standardization of multi-source data on power grid materials can be performed separately for text data, numerical data, supply chain data, and model data in the multi-source data. For example, for standardization of text data, the BERT model can be used to extract keywords (such as "transformer" and "insulated cable") from project text description data and map them to unified material codes; for unstructured text, it can be converted into structured tags; for standardization of numerical data, the "collected quantity" and "unit price" in historical material consumption time series data can be standardized using standard scores (Z-scores) to eliminate dimensional differences. For example, the "accumulated operating time" in equipment failure rate data can be segmented and normalized; for standardization of supply chain data, the causes of logistics delays in the supply chain logistics status data can be encoded as discrete variables; for standardization of model data, the output results of each material recommendation model in the material recommendation model library can be quantified into continuous values ​​from 0 to 1.

[0069] In a specific implementation, each power grid material data is standardized to obtain standardized data. Each standardized data includes: standardized project text description data, standardized historical material consumption time series data, standardized supply chain logistics status data, and standardized material recommendation model library model data. The standardized project text description data can be used as basic project feature information, the standardized historical material consumption time series data can be used as material attribute feature information, the standardized supply chain logistics status data can be used as supply chain feature information, and the standardized material recommendation model library model data can be used as material recommendation feature information.

[0070] After obtaining each standardized data set, the standardized data belonging to the same power grid material project object is semantically fused based on the project identifier of the power grid material project object to obtain a structured feature matrix corresponding to each power grid material project object. Specifically, the standardized project text description data, the standardized historical material consumption time series data, and the standardized supply chain logistics status data can be associated through the project ID to construct a project-material-supply chain relationship map. For example, the project ID "GD-2025-015" is associated with "copper core insulated cable demand" in the project text description data, "TR-10kV-001 collection record" in the historical material consumption time series data, and "cable arrival delayed by 5 days" in the supply chain logistics status data.

[0071] Next, a structured feature matrix is ​​constructed. Each row in the matrix represents a power grid material project object, and each column represents a standardized data point. Each standardized data point can include power grid material descriptive information, such as basic project characteristics, material attribute characteristics, supply chain characteristics, and material recommendation characteristics. Therefore, each column of the structured feature matrix can include: basic project characteristics (project type code, geographic location code, and budget amount, etc.), material attribute characteristics (historical consumption frequency, material complexity, etc.), supply chain characteristics (logistics delay risk score, etc.), and material recommendation characteristics (idle material matching degree, framework agreement discount rate, etc.).

[0072] In another embodiment, clustering index information corresponding to the power grid material project object is determined based on the power grid material description information corresponding to the power grid material project object, including: inputting the power grid material description information corresponding to the power grid material project object into a demand urgency calculation model to obtain demand urgency index information; the demand urgency calculation model is used to determine the demand urgency index information based on the project importance information, seasonal risk information, policy-driven information and supply chain resilience information in the power grid material description information; inputting the power grid material description information corresponding to the power grid material project object into a logistics complexity calculation model to obtain material complexity index information; the logistics complexity calculation model is used to determine the material complexity index information based on the material technical parameter information, material compatibility risk information, supplier concentration information, alternative material information and full life cycle cost information in the power grid material description information; inputting the power grid material description information corresponding to the power grid material project object into a material recommendation calculation model to obtain material recommendation index information; the material recommendation calculation model is used to determine the material recommendation index information based on material matching information, project similarity information and material cost-effectiveness information.

[0073] Exemplarily, the demand urgency calculation model is expressed as:

[0074] ;

[0075] in, It is the demand urgency indicator information; The project importance information has a value range of [0,1]. The project importance level can be determined according to the project strategic level. For example, the national key project (such as UHV project) is 1.0, the provincial key project is 0.8, and the conventional project is 0.5. Seasonal risk information, including seasonal risk coefficient, with a value range of [0,1], calculated based on historical meteorological data; Government-driven information, including policy-driven factors, with a value range of [1.0, 1.2], such as 1.2 for government subsidy projects, 1.1 for carbon neutrality-related projects, and 1.0 for ordinary projects; It is the supply chain resilience information, including the supply chain resilience score, with a value range of [0,1].

[0076] ;

[0077] in, Score the supplier's lead time (lead time ≤ 7 days is 1.0, and decreases by 0.2 for every additional 7 days); Supplier diversity, calculated as the Shannon entropy of the number of suppliers to the total market share: ; is the logistics redundancy, based on the proportion of alternative transportation routes.

[0078] in, 、 、 、 They are dynamic weights, which can be preset values ​​determined based on user experience or calculated using the entropy weight method: Construct a historical data matrix ( is the number of items); standardize the matrix and calculate the Entropy value of the indicator : ; Calculate weights: .

[0079] For example, the logistics complexity calculation model is expressed as:

[0080] ;

[0081] in, It is the material complexity index information;

[0082] is the material technical parameter information, including the heterogeneity of technical parameters, with a value range of [0,1], which can be calculated by the Gini-Simpson index: ;

[0083] Material compatibility risk information, including compatibility conflict risk, with a value range of [0,1];

[0084] The supplier concentration information includes supplier concentration, which ranges from [0 to 1] and is determined based on the Herfindahl index: ;

[0085] Information about alternative materials, including the availability of alternative materials, with a value range of [0,1];

[0086] is the life cycle cost information, including the life cycle cost fluctuation, with a value range of [0,1], which can be expressed as: .

[0087] in, 、 、 、 、 The corresponding weights are pre-set according to user experience values, or determined by random forest feature importance analysis.

[0088] For example, the material recommendation calculation model can be expressed as:

[0089] ;

[0090] in, Material recommendation index information; Material matching degree information, including rule engine matching degree, with a value range of [0,1], determined based on the number of matching standard material types and the total number of required material types; is the item similarity information, including collaborative filtering recommendation score, with a value range of [0,1], , For the project Cosine similarity with the current item, For the project The selection rate of a certain material.

[0091] in, is the cost-effectiveness information of materials, including the cost-effectiveness score, which ranges from [0,1] and can be expressed as: , 、 and are weight coefficients, Score the technical performance. is the maintenance cost discount rate.

[0092] in, 、 、 They are the corresponding weights, which can be pre-set through user experience values ​​or optimized through reinforcement learning, where the state space of reinforcement learning is project type, supply chain status, and inventory level; the action space is to adjust the weight combination; and the reward function predicts the accuracy.

[0093] The technical solution of the above embodiment significantly improves the ability to depict complex power grid scenarios and improves the efficiency and accuracy of power grid material demand forecasting by constructing a three-dimensional indicator system of demand urgency index, material complexity index and standard material recommendation degree.

[0094] In another embodiment, clustering processing is performed on each power grid material project object according to the preset clustering rules and the clustering index information corresponding to each power grid material project object, including: determining a membership model that matches the clustering index information according to the quantitative demand information of the clustering index information; inputting the clustering index information into the matching membership model to obtain the membership corresponding to the clustering index information; the membership model is used to convert the clustering index information into the corresponding membership through the membership function that matches the clustering index information; clustering processing is performed on each power grid material project object according to the preset clustering rules and the membership corresponding to each power grid material project object.

[0095] In the specific implementation, a three-dimensional clustering index is constructed based on different project types, including demand urgency index information, material complexity index information and material recommendation index information; each power grid material project object is mapped to the three-dimensional clustering space through the membership function to generate a three-dimensional membership vector.

[0096] Among them, the quantitative demand information is the specific requirements and standards for the numerical processing of clustering indicator information, including the numerical change trend, value range, precision, data type, etc. of the indicator, which is used to standardize the quantification process of clustering indicator information and ensure that it can be effectively input into the membership model for processing.

[0097] In the specific implementation, the demand urgency index information, material complexity index information and material recommendation index information are combined The original value of is mapped to the interval [0,1] to construct a fuzzy three-dimensional clustering space; among them, the demand urgency index information The quantitative demand information is "urgent changes near the threshold need to be responded to quickly", so the S-type membership function can be used to calculate the membership of the demand urgency index information:

[0098] ;

[0099] in is the slope parameter, which controls the steepness of the function. The center point parameter controls the function offset position and sets the center point through historical data analysis. =0.6, slope =10, so When >0.7, the membership is close to 1. When <0.4, the membership is close to 0.

[0100] Therefore, the demand urgency indicator information The matching membership model can use the following S-type membership function to convert the demand urgency index information Convert to the corresponding membership:

[0101] .

[0102] Due to the material complexity index information Information on material recommendation indicators The quantitative demand information is "no need to quickly respond to emergency changes near the threshold", so the material complexity index information can be calculated by the following trapezoidal membership function Information on material recommendation indicators Membership degree:

[0103] ;

[0104] in, 、 、 、 are the threshold parameters and define the membership change interval.

[0105] Determine material complexity index information based on expert experience and project classification The threshold interval is =0.3, =0.5, =0.7, =0.9, then the membership model matching the material complexity index information can be transformed into the material complexity index information through the following trapezoidal membership function. Convert to the corresponding membership:

[0106] ;

[0107] Determine material recommendation index information based on expert experience and project classification The threshold interval =0.2, =0.4, =0.6, =0.8, ensuring high recommendation ( >0.6) The membership degree is 1, then the material recommendation index information The matching membership model can be used to convert the material recommendation index information into Convert to the corresponding membership:

[0108] .

[0109] Thus, each power grid material project object can be The membership function is mapped to the corresponding membership, and the membership of the demand urgency, material complexity and material recommendation degree are calculated respectively to generate a three-dimensional membership vector Through the S-type and trapezoidal membership functions, the demand urgency, material complexity and standard recommendation indicators are mapped to the three-dimensional cluster space, generating a membership vector with a clear semantic interpretation. Through the S-type and trapezoidal membership functions, the demand urgency, material complexity and standard recommendation indicators are mapped to the three-dimensional space, generating a membership vector with a clear semantic interpretation, providing highly discriminative input features for subsequent cluster analysis.

[0110] In the specific implementation, the three-dimensional clustering space is divided into grid units according to the preset resolution. Combined with the fuzzy C-means clustering algorithm, the grid density threshold and boundary conditions are dynamically adjusted to generate multimodal clusters. Each cluster is associated with at least one material recommendation model, and the material recommendation list information adapted to the cluster characteristics is automatically generated. By combining dynamic grid division with fuzzy C-means clustering, the recognition accuracy of multimodal demand patterns is improved. Among them, the preset resolution is the grid unit length of each dimension (such as = 0.1), the three-dimensional clustering space is divided into 10 10 10 = 1000 cubic grid cells, Mathematical expression for the range of grid cells: .

[0111] Count the number of items in each grid cell , the density is defined as .

[0112] The dynamic adjustment rule of density threshold is: , then merge adjacent sparse grids to form a larger unit, that is, perform sparse grid merging; if , then the grid is pressed Further subdivision is performed by performing dense grid splitting; for cross-boundary projects, the grid boundaries are adjusted according to the membership gradient to ensure data continuity, that is, boundary condition optimization is performed.

[0113] The input data of the fuzzy C-means clustering algorithm is the coordinates of the center point of the grid cell and its density weight ; According to the grid density weight, optimize the objective function:

[0114] ;

[0115] in: is the number of clusters, is the number of items, For the project Cluster The membership degree of is the fuzzy factor, are the cluster center coordinates.

[0116] Automatically select the optimal number of clusters based on the entropy of the density distribution:

[0117] ;

[0118] Among them, each cluster after clustering is a continuous subspace, representing a type of material demand pattern. For example: Cluster A: High urgency ( >0.8), high complexity ( >0.7), recommended framework agreement materials ( >0.6); Cluster B: medium urgency (0.5< ≤0.8), low complexity ( ≤0.4)、Recommended idle materials( >0.8). Multiple clusters can exist in the same three-dimensional clustering space, reflecting different demand scenarios (such as emergency repair, routine upgrade, and strategic reserve).

[0119] In another embodiment, based on the power grid material description information of the power grid material project objects included in the cluster cluster, the material recommendation list information corresponding to the cluster cluster is determined, including: based on the cluster characteristics of the cluster cluster, selecting a material recommendation model that matches the cluster cluster from a material recommendation model library; the cluster characteristics include the indicator distribution interval of the cluster indicator information corresponding to each power grid material project object in the cluster cluster; the power grid material description information of each power grid material project object in the cluster cluster is input into the material recommendation model that matches the cluster cluster to obtain the material recommendation list information corresponding to the cluster cluster.

[0120] In specific implementations, each cluster is matched with an appropriate material recommendation model from the material recommendation model library, such as the standard list recommendation model for idle materials, the reverse standard list recommendation model for materials, the standard list recommendation model for materials under long-term unfulfilled contracts, the standard list recommendation model for centralized procurement materials, the standard list recommendation model for reserve materials, and the standard list recommendation model for materials under effective framework agreements. The standard list recommendation model for idle materials identifies reusable idle materials (e.g., unused circuit breakers of the same model in inventory) based on inventory turnover, material expiration dates, and technology obsolescence risk. The reverse standard list recommendation model generates a list of reusable materials based on returned equipment inspection reports (e.g., the recyclability of copper cores in decommissioned transformers). The standard list recommendation model for long-term unfulfilled contracts analyzes contract overdue duration and supplier credit ratings to recommend priority backlog materials (e.g., cable procurement contracts overdue by 90 days). The standard list recommendation model for centralized procurement materials matches current project requirements based on procurement framework agreement catalogs (e.g., insulator models purchased centrally). The recommended model for a standard inventory of reserve materials is used to dynamically adjust inventory thresholds based on emergency response plans (such as standards for the storage of materials to combat ice disasters). The recommended model for a standard inventory of materials under an effective framework agreement is used to recommend the most cost-effective materials within the agreement (such as discounted batches of copper cables during the agreement period) based on the agreement's validity period and price fluctuations.

[0121] The cluster characteristics include the index distribution interval of the cluster index information corresponding to each power grid material project object in the cluster. In other words, the cluster characteristics can be the index information of each demand urgency in the cluster. Index distribution range, each material complexity index information The index distribution range and the recommended index information of each material For example, the cluster characteristics of a cluster may include: >0.7, >0.6, which means that the are all greater than 0.7, the cluster are all greater than 0.6. Therefore, according to the cluster characteristics ( interval), you can select the material recommendation model with the highest priority.

[0122] For example, the matching relationship between cluster features and material recommendation models can be shown in Table 1.

[0123] Table 1

[0124]

[0125] For example, weighted recommendations can be made for material selection records of historically similar clusters. The recommendation score of the material recommendation model can be expressed as:

[0126] ;

[0127] in, , The rule matching degree includes the matching degree between cluster characteristics and the model. The coordinated generation of standard material lists based on rules and data significantly improves the comprehensiveness of power grid material demand forecasting.

[0128] The recommended material list information may include material model, recommendation reason, priority score, and supplier information. For example, the recommended material list information may be as shown in Table 2.

[0129] Table 2

[0130]

[0131] In another embodiment, the power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster are input into the material demand forecasting model to obtain the material demand forecasting result of the cluster, including: determining the material demand forecasting model that matches the cluster according to the cluster characteristics of the cluster; inputting the power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster into the material demand forecasting model that matches the cluster to obtain the material demand forecasting result of the cluster.

[0132] In the specific implementation, the material demand forecasting model can be adaptively selected based on the indicator distribution (i.e., cluster characteristics) of the cluster demand urgency index, material complexity index, and standard material recommendation index to balance the prediction accuracy and computational efficiency, and the model parameters can be dynamically updated through incremental learning methods based on real-time updated multi-source data streams.

[0133] In the specific implementation, the logic of dynamically selecting the most suitable material demand forecasting model based on the three-dimensional indicator distribution of the cluster is as follows:

[0134] (1) For high urgency and high complexity clusters: If the urgency of the project requirements in the cluster is high (e.g. >0.7) and the technical complexity is prominent (e.g. >0.6), the LSTM (Long Short-Term Memory) model is preferred as the material demand forecasting model. LSTM can capture nonlinear dependencies in time series, such as the sudden fluctuations and multi-stage dependencies in material demand during emergency repair projects.

[0135] (2) For clusters with medium urgency and strong rule dependency: For clusters with moderate urgency (e.g. 0.4< ≤0.7) and the standard recommendation is high (e.g. >0.6), the XGBoost (Extreme Gradient Boosting) model was selected as the material demand forecasting model. XGBoost uses feature cross-pollination and tree ensembles to effectively analyze material combination patterns under centralized procurement agreements, such as the cost-time balance when coordinating supply from multiple suppliers.

[0136] (3) For low urgency and periodic demand clusters: If the demand within the cluster is obviously periodic (such as annual maintenance plan) and the urgency is low ( <0.4), using the Prophet time series model as a material demand forecasting model. Prophet accurately predicts the consumption rhythm of routine maintenance materials, such as the regular replacement needs of insulators and cables, by decomposing trend, seasonality, and holiday effects.

[0137] (4) For small sample and high recommendation cluster: For clusters with small data volume (e.g., number of samples <100) but high standard recommendation matching degree (e.g. 0.8) cluster, a grey neural network was selected as the material demand forecasting model. This model combines grey system theory with neural networks and uses a small amount of historical data to infer long-term demand trends for framework agreement materials, such as multi-year procurement plans for strategic reserve materials.

[0138] Therefore, the material demand forecasting model is adaptively selected according to cluster characteristics, so that the material demand forecasting model can adapt to the demand forecasting of complex scenarios, effectively avoid prediction bias, and improve the accuracy of power grid material demand forecasting. It can provide an accurate data foundation and strong technical support for subsequent power grid material management and other related work.

[0139] In practice, model selection can be further optimized by calculating the entropy (a measure of data dispersion) of the distribution of indicators (cluster characteristics) within a cluster. Clusters with high entropy (dispersed distribution) are selected with more complex models (such as LSTM), while clusters with low entropy (concentrated distribution) are selected with lightweight models (such as Prophet), thus achieving an adaptive balance between accuracy and efficiency.

[0140] It should be noted that to adapt to real-time multi-source data streams (such as new project entry and supply chain status changes), incremental learning methods can be used to continuously optimize model parameters to avoid the computational burden and delay caused by repeated training of the entire data. Among them, incremental learning implementation methods include:

[0141] (1) Online Gradient Descent: Applicable to parameterized models, whenever new data arrives, the model calculates the gradient of the loss function based on the current parameters and fine-tunes the weights in the opposite direction of the gradient. For example, the LSTM model dynamically adjusts the connection strength of hidden layer neurons by receiving real-time logistics delay data to quickly respond to the impact of supply chain disruptions on demand.

[0142] (2) XGBoost incremental training: Based on the existing tree structure, additional subtrees generated by new data are trained. The new subtrees are optimized only for recent data features, and regularization terms are used to control model complexity and prevent overfitting. For example, in response to sudden reverse material demand, the newly added subtrees can quickly learn the reuse patterns of returned equipment.

[0143] (3) Prophet model expansion: By expanding the historical time series window and refitting seasonal and trend terms, it adapts to changes in long-term demand patterns. For example, after adding monthly maintenance project data, the model automatically adjusts the seasonal component to reflect more detailed cyclical fluctuations.

[0144] In the specific implementation, the grid material description information of each grid material project object within a cluster and the corresponding recommended material list information are input into the material demand forecasting model to generate a cluster-specific material demand forecast. This forecast includes the forecast results for each cluster and the associated recommended material list information. By comparing the deviation between actual supply chain consumption data and the predicted material demand in real time, a feedback loop is established to optimize core parameters. The specific implementation process is as follows: The forecast results for each cluster include material demand quantity, time distribution, and priority ranking. Based on the recommended material list information for each cluster, the recommended material model, inventory location, supplier agreement terms, and priority score are listed. The recommended material list information is updated in real time via a dynamic rule engine. For example, when the inventory of a certain agreed material falls below a safety threshold, it is automatically replaced with a suboptimal recommendation. Actual supply chain consumption data (such as material outbound records and logistics arrival times) is collected in real time through IoT devices and the ERP system and compared item by item with the forecast results. The absolute deviation, relative deviation rate, and time deviation are calculated. A loss function is constructed based on the deviation rate, and the weights of the three indicators are dynamically adjusted using the gradient descent method. After each parameter update, the optimization effect is verified through historical data backtesting and online simulation to ensure that the prediction accuracy is improved and there are no negative fluctuations.

[0145] This application significantly improves the ability to depict complex power grid scenarios by constructing a three-dimensional indicator system of demand urgency index, material complexity index and standard material recommendation, while maintaining computational efficiency and being embedded in a real-time prediction system. Through S-type and trapezoidal membership functions, the demand urgency, material complexity and standard recommendation indicators are mapped to three-dimensional space, generating a membership vector with a clear semantic interpretation, providing highly discriminative input features for subsequent clustering analysis. By combining dynamic grid division with fuzzy C-means clustering, the accuracy of multimodal demand pattern recognition is improved, and a standard material list is generated based on rules and data, significantly improving the comprehensiveness of power grid material demand forecasting. Finally, the prediction model is adaptively selected based on the feature vector, so that the prediction model can adapt to demand forecasting in complex scenarios, effectively avoiding prediction bias, and improving the accuracy of power grid material demand forecasting. It can provide an accurate data foundation and strong technical support for subsequent power grid material management and other related work.

[0146] In another embodiment, Figure 3 As shown in the figure, a method for predicting power grid material demand information based on multi-source data fusion is provided. Figure 1 Taking the terminal 102 in FIG. 1 as an example, the method includes the following steps:

[0147] Step S302: acquiring multi-source data of power grid materials, and performing standardization processing on each power grid material data to obtain each standardized data.

[0148] In step S304, based on the project identifier of the power grid material project object, the standardized data belonging to the same power grid material project object are semantically fused to obtain the structured feature matrix corresponding to each power grid material project object, and the structured feature matrix corresponding to each power grid material project object is used as the power grid material description information corresponding to each power grid material project object.

[0149] Step S306: input the power grid material description information corresponding to the power grid material project object into the demand urgency calculation model to obtain demand urgency index information, and input the power grid material description information corresponding to the power grid material project object into the logistics complexity calculation model to obtain material complexity index information, and input the power grid material description information corresponding to the power grid material project object into the material recommendation calculation model to obtain material recommendation index information.

[0150] Step S308 : determining a membership model that matches the clustering index information according to the quantitative requirement information of the clustering index information, and inputting the clustering index information into the matching membership model to obtain the membership corresponding to the clustering index information.

[0151] Step S310 : performing clustering processing on each power grid material item object according to a preset clustering rule and a membership degree corresponding to each power grid material item object to obtain a plurality of clusters.

[0152] Step S312: selecting a material recommendation model that matches the cluster from a material recommendation model library based on the cluster characteristics of the clustered cluster.

[0153] Step S314: input the power grid material description information of each power grid material project object in the cluster into a material recommendation model that matches the cluster, and obtain material recommendation list information corresponding to the cluster.

[0154] Step S316: Determine a material demand forecasting model that matches the cluster according to the cluster characteristics of the cluster.

[0155] Step S318: input the power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster into the material demand forecasting model matching the cluster to obtain the material demand forecasting result of the cluster.

[0156] It should be noted that the specific limitations of the above steps can be found in the specific limitations of a method for predicting power grid material demand information based on multi-source data fusion mentioned above.

[0157] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0158] Based on the same inventive concept, the embodiments of the present application also provide a device for predicting power grid material demand information based on multi-source data fusion, which is used to implement the aforementioned method for predicting power grid material demand information based on multi-source data fusion. The implementation solution provided by the device is similar to the implementation solution described in the aforementioned method. Therefore, the specific limitations of one or more embodiments of the device for predicting power grid material demand information based on multi-source data fusion provided below can be found in the above-mentioned limitations of the method for predicting power grid material demand information based on multi-source data fusion, and will not be repeated here.

[0159] In an exemplary embodiment, Figure 4 As shown, a power grid material demand information prediction device based on multi-source data fusion is provided, comprising:

[0160] Acquisition module 410 is used to obtain grid material description information corresponding to each grid material project object; the grid material description information includes at least basic project feature information, material attribute feature information, supply chain feature information, and material recommendation feature information. Index determination module 420 is used to determine clustering index information corresponding to the grid material project object based on the grid material description information corresponding to the grid material project object; the clustering index information includes at least demand urgency index information, material complexity index information, and material recommendation index information. Clustering module 430 is used to cluster each grid material project object according to preset clustering rules and the clustering index information corresponding to each grid material project object, thereby obtaining multiple clusters. List determination module 440 is used to determine the recommended material list information corresponding to the cluster based on the grid material description information of each grid material project object in the cluster. Prediction module 450 is used to input the grid material description information of each grid material project object in the cluster and the recommended material list information corresponding to the cluster into a material demand prediction model to obtain a material demand prediction result for the cluster.

[0161] In one embodiment, the list determination module 440 is specifically used to select a material recommendation model that matches the cluster cluster from the material recommendation model library based on the cluster characteristics of the cluster cluster; the cluster characteristics include the indicator distribution range of the cluster indicator information corresponding to each power grid material project object in the cluster cluster; the power grid material description information of each power grid material project object in the cluster cluster is input into the material recommendation model that matches the cluster cluster to obtain the material recommendation list information corresponding to the cluster cluster.

[0162] In one embodiment, the prediction module 450 is specifically used to determine a material demand prediction model that matches the cluster cluster based on the cluster characteristics of the cluster cluster; input the power grid material description information of each power grid material project object in the cluster cluster and the material recommendation list information corresponding to the cluster cluster into the material demand prediction model that matches the cluster cluster to obtain the material demand prediction result of the cluster cluster.

[0163] In one embodiment, the index determination module 420 is specifically used to input the power grid material description information corresponding to the power grid material project object into the demand urgency calculation model to obtain demand urgency index information; the demand urgency calculation model is used to determine the demand urgency index information based on the project importance information, seasonal risk information, policy-driven information and supply chain resilience information in the power grid material description information; the power grid material description information corresponding to the power grid material project object is input into the logistics complexity calculation model to obtain material complexity index information; the logistics complexity calculation model is used to determine the material complexity index information based on the material technical parameter information, material compatibility risk information, supplier concentration information, alternative material information and full life cycle cost information in the power grid material description information; the power grid material description information corresponding to the power grid material project object is input into the material recommendation calculation model to obtain material recommendation index information; the material recommendation calculation model is used to determine the material recommendation index information based on the material matching information, project similarity information and material cost-effectiveness information.

[0164] In one embodiment, the clustering module 430 is specifically used to determine a membership model that matches the clustering index information based on the quantitative demand information of the clustering index information; input the clustering index information into the matching membership model to obtain the membership corresponding to the clustering index information; the membership model is used to convert the clustering index information into the corresponding membership through a membership function that matches the clustering index information; and cluster each power grid material project object according to the preset clustering rules and the membership corresponding to each power grid material project object.

[0165] In one embodiment, the acquisition module 410 is specifically used to acquire multi-source data of power grid materials; the multi-source data of power grid materials include at least two types of power grid material data among project text description data, historical material consumption time series data, supply chain logistics status data and model data of the material recommendation model library; each power grid material data is standardized to obtain each standardized data; according to the project identification of the power grid material project object, each standardized data belonging to the same power grid material project object is semantically fused to obtain a structured feature matrix corresponding to each power grid material project object; each row of the structured feature matrix corresponds to each power grid material project object, and each column of the structured feature matrix corresponds to each standardized data; the structured feature matrix corresponding to each power grid material project object is used as the power grid material description information corresponding to each power grid material project object.

[0166] Each module in the aforementioned multi-source data fusion-based power grid material demand information forecasting device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a computer device memory in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0167] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless means. The wireless means can be implemented via Wi-Fi, mobile cellular networks, near-field communication (NFC), or other technologies. When executed by the processor, the computer program implements a method for predicting power grid material demand information based on multi-source data fusion. The display unit of the computer device is used to produce a visual image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse. Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0168] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0169] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0170] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0171] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0172] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0173] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for predicting power grid material demand information based on multi-source data fusion, characterized in that: The method comprises: Obtaining grid material description information corresponding to each grid material project object; the grid material description information includes at least project basic feature information, material attribute feature information, supply chain feature information and material recommendation feature information; Determine clustering index information corresponding to the power grid material project object according to the power grid material description information corresponding to the power grid material project object; the clustering index information includes at least demand urgency index information, material complexity index information and material recommendation index information; performing clustering processing on each of the power grid material project objects according to a preset clustering rule and the clustering indicator information corresponding to each of the power grid material project objects to obtain a plurality of cluster clusters; Determining the recommended material list information corresponding to the cluster according to the grid material description information of each grid material project object in the cluster; The power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster are input into a material demand forecasting model to obtain a material demand forecasting result of the cluster.

2. The method according to claim 1, characterized in that The determining, based on the grid material description information of the grid material project object included in the cluster, the material recommendation list information corresponding to the cluster includes: Selecting a material recommendation model that matches the cluster from a material recommendation model library according to the cluster characteristics of the cluster; the cluster characteristics include an indicator distribution interval of the cluster indicator information corresponding to each of the power grid material project objects in the cluster; The power grid material description information of each of the power grid material project objects in the cluster is input into a material recommendation model that matches the cluster to obtain material recommendation list information corresponding to the cluster.

3. The method according to claim 2, characterized in that The step of inputting the grid material description information of each grid material project object in the cluster and the material recommendation list information corresponding to the cluster into the material demand forecasting model to obtain the material demand forecasting result of the cluster includes: Determining a material demand forecasting model that matches the cluster according to the cluster characteristics of the cluster; The power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster are input into a material demand forecasting model matching the cluster to obtain a material demand forecasting result of the cluster.

4. The method according to claim 1, wherein The determining, based on the power grid material description information corresponding to the power grid material project object, clustering index information corresponding to the power grid material project object includes: Inputting the power grid material description information corresponding to the power grid material project object into a demand urgency calculation model to obtain the demand urgency index information; the demand urgency calculation model is used to determine the demand urgency index information based on the project importance information, seasonal risk information, policy driving information and supply chain resilience information in the power grid material description information; Inputting the power grid material description information corresponding to the power grid material project object into the logistics complexity calculation model to obtain the material complexity index information; the logistics complexity calculation model is used to determine the material complexity index information based on the material technical parameter information, material compatibility risk information, supplier concentration information, alternative material information and full life cycle cost information in the power grid material description information; The power grid material description information corresponding to the power grid material project object is input into a material recommendation degree calculation model to obtain the material recommendation degree index information; the material recommendation degree calculation model is used to determine the material recommendation degree index information based on material matching information, project similarity information and material cost-effectiveness information.

5. The method according to claim 1, wherein The clustering process of each of the power grid material project objects according to the clustering indicator information corresponding to each of the power grid material project objects according to the preset clustering rule includes: Determining a membership model that matches the clustering indicator information according to the quantitative requirement information of the clustering indicator information; Inputting the clustering index information into the matching membership model to obtain the membership corresponding to the clustering index information; the membership model is used to convert the clustering index information into the corresponding membership through a membership function matching the clustering index information; Clustering is performed on each of the power grid material project objects according to the preset clustering rule and the membership degree corresponding to each of the power grid material project objects.

6. The method according to claim 1, characterized in that The obtaining of the grid material description information corresponding to each grid material project object includes: Acquire multi-source data of power grid materials; the multi-source data of power grid materials includes at least two types of power grid material data selected from project text description data, historical material consumption time series data, supply chain logistics status data, and model data of a material recommendation model library; Performing standardization processing on each of the power grid material data to obtain each standardized data; According to the project identifier of the power grid material project object, semantic fusion is performed on each of the standardized data belonging to the same power grid material project object to obtain a structured feature matrix corresponding to each of the power grid material project objects; each row of the structured feature matrix corresponds to each of the power grid material project objects, and each column of the structured feature matrix corresponds to each of the standardized data; The structured feature matrix corresponding to each of the power grid material project objects is used as the power grid material description information corresponding to each of the power grid material project objects.

7. A device for predicting power grid material demand information based on multi-source data fusion, characterized in that: The device comprises: An acquisition module is used to obtain the grid material description information corresponding to each grid material project object; the grid material description information at least includes project basic feature information, material attribute feature information, supply chain feature information and material recommendation feature information; An indicator determination module is used to determine clustering indicator information corresponding to the power grid material project object based on the power grid material description information corresponding to the power grid material project object; the clustering indicator information at least includes demand urgency indicator information, material complexity indicator information and material recommendation index information; A clustering module, configured to perform clustering processing on each of the power grid material project objects according to a preset clustering rule and the clustering indicator information corresponding to each of the power grid material project objects to obtain a plurality of cluster clusters; A list determination module is used to determine the recommended material list information corresponding to the cluster according to the power grid material description information of each power grid material project object in the cluster; The prediction module is used to input the power grid material description information of each power grid material project object in the cluster and the material recommendation list information corresponding to the cluster into the material demand prediction model to obtain the material demand prediction result of the cluster.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images