Method and system for drawing power material relation graph and demand prediction based on entropy value

Abstract

The application relates to an entropy value-based power material relationship graph drawing and demand prediction method and system, which comprises the following steps: collecting power material data in a project set, performing graph structuring, and obtaining an initial material relationship graph; quantifying the initial material relationship graph based on the initial material relationship graph by adopting a matrix expression method, and further adopting an associated material grouping algorithm to perform associated material grouping; calculating the entropy value of the quantized initial material relationship graph based on the associated material grouping result, and removing noise points based on the maximum entropy reduction noise reduction principle to obtain a final material relationship graph; and predicting a material demand quantity interval based on the material relationship graph. Compared with the prior art, the application has the advantages of improving the relationship graph drawing speed and quality and the like.

Description

Technical Field

[0001] This invention relates to the field of power materials technology, and in particular to a method and system for drawing power material relationship maps and forecasting demand based on entropy values. Background Technology

[0002] Material relationship mapping is a new technological tool adopted by power companies for lean material management, providing technical support for scenarios such as material demand forecasting. Currently, research and patents related to mapping technology in the domestic power industry all revolve around optimizing the analysis algorithms for material relationships. However, there is a lack of targeted research and patents on how to improve the accuracy of sample data and ensure that the speed and quality of mapping are not affected by data noise.

[0003] Patent application CN118193749A discloses a method and apparatus for association processing of power data elements. This method involves collecting power data elements to form a dataset to be associated and processed; through frequent processing of the dataset, it identifies frequently associated itemsets and their corresponding relationships. While applicable to the power industry, including the analysis of material relationships, this method only covers the collection, clustering, and linking of sample data related to these relationships, without addressing data noise removal. Patent application CN118193749A also discloses a method and system for automatic relationship identification based on deep learning. This method involves collecting power system knowledge graph data, including entities and their relationships, converting the knowledge graph data into a graph data format, and training a multi-layered graph neural network model based on historical power system knowledge graph data to predict the types of relationships between entities in the knowledge graph. This method primarily involves the decomposition, combination, and reconstruction of different relationship graphs, without addressing specific methods for judging the accuracy of sample data or cleaning it.

[0004] In summary, there is a need for an automated data cleaning method to address the issue of sample data accuracy, thereby improving the accuracy of sample data and ensuring the speed and quality of graph plotting. Summary of the Invention

[0005] The purpose of this invention is to provide an automated noise reduction method and system for drawing power material relationship maps and forecasting demand based on entropy values.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for drawing power resource relationship maps and forecasting demand based on entropy values ​​includes the following steps:

[0008] Collect power material data from the project set, perform graph structuring, and obtain an initial material relationship map;

[0009] Quantify the initial material relationship graph using the matrix expression method, and further group the associated materials using the associated material grouping algorithm;

[0010] Calculate the entropy value of the quantified initial material relationship graph based on the grouping results of the associated materials, and eliminate noise points based on the principle of maximum entropy reduction and noise reduction to obtain the final material relationship graph;

[0011] Predict the material demand quantity interval based on the material relationship graph.

[0012] Furthermore, the steps of graph structuring include:

[0013] Convert the power material data into vertex functions and edge functions to form an initial material relationship graph, where one material in the power material data corresponds to one graph vertex, and the relative demand between two materials corresponds to the graph edge.

[0014] Furthermore, the expression of the vertex function is:

[0015]

[0016] In the formula, V(x) is the vertex function of material x, which is a mapping of the demand frequency of material x and reflects the absolute demand of the project for material x, 0 < V(x) ≤ 1. When V(x) = 1, material x is an essential material for the entire project set P; N(P ,

[0018] , n , ,

[0017] , n , , y|x , , x , i , , i , , i , , y|x , y|x ,

[0021] , x ,

[0020] ,

[0019] ) is the number of items of project P that uses material x x and N(P) is the number of items in the project set;

[0017] The expression of the edge function is:

[0018]

[0019] In the formula, E(x, y) is the edge function between material x and material y, which is a mapping of the demand correlation between material x and material y; N(P y|x ) is the number of items of P y|x and P y|x is the project that uses material y among the projects that use material x; N(P) is the number of items in the project set.

[0020] Furthermore, the steps of quantification using the matrix expression method include:

[0021] Construct the mapping from the vertex function V(x) to the demand vector where D n =(d1, d2,..., d i ,..., d <0000​​​​Demand frequency V(x) i ), where n is the number of material types covered by item set P;

[0022] Construct a mapping from the edge function E(x, y) to the correlation matrix. Complete the quantization process, where R n×n For the association matrix, the element r in the i-th row and j-th column is... i，j =E(x) i x j ), 1≤i≤n, 1≤j≤n, and r i，i =E(x) i x i ) = 1.

[0023] Furthermore, the step of grouping related materials includes:

[0024] Construction material x i Prerequisite material mapping And based on the aforementioned prerequisite material mapping Traverse the association matrix R n×n In the i-th row, filter the column indices where the element is 1 to find material x. i Prerequisite material x j The aforementioned prerequisite material mapping For all materials x i A collection of materials that are directly related;

[0025] Based on the aforementioned material mapping Construct mapping Traverse the association matrix R n×n Find the column index j of all elements in the i-th row and j-th column that satisfy R(i,j) = R(j,i) = 1, and obtain the material x. i Supporting materials x j ,Right now:

[0026]

[0027] Group the materials that are compatible with each other into one group, and then repeat the above steps to complete the grouping of all materials.

[0028] Furthermore, the expression for calculating the entropy value is:

[0029]

[0030] In the formula, H is the entropy value, reflecting the degree of concentration of related material groups, and m is the number of related material groups. Group the i-th associated material. x1, x2, ..., x NLet P be the itemset that covers the materials, and N be the number of materials.

[0031] Furthermore, the step of obtaining the final material relationship map includes:

[0032] 1) Determine if the entropy value is zero. If it is, it indicates that noise reduction is not needed, and the quantized initial material relationship map is used as the final material relationship map. If not, it indicates that noise reduction is needed, and proceed to the next step:

[0033] 2) For the item set P in the quantified initial material relationship map, delete the j-th item to form a subset P. -j Calculate subset P -j The entropy value H(P) -j ), and repeat this step to obtain the entropy values ​​of all subsets;

[0034] 3) Delete the item with the highest entropy value among all subsets;

[0035] 4) Calculate whether the entropy reduction of item set P exceeds the threshold. If yes, end the noise reduction operation and obtain the final material relationship map. If not, repeat steps 1)-4) based on the entropy-reduced item set P until the end, and obtain the final material relationship map. The expression for calculating the entropy reduction of item set P is:

[0036] Entropy reduction = H(P) - max{H(P)} -j )}

[0037] In the formula, H(P) is the entropy value of the item set P.

[0038] Furthermore, the step of predicting the range of material demand includes:

[0039] The types of quantitative relationships between material requirements are determined based on the aforementioned material relationship map;

[0040] The demand range for each material is calculated based on the aforementioned quantity relationship type to obtain the material demand range.

[0041] Furthermore, the types of quantitative relationships include interval-type quantitative relationships, formula-type quantitative relationships, and proportional-type quantitative relationships.

[0042] This invention also provides a power resource relationship mapping and demand forecasting system based on entropy, comprising:

[0043] Graph structuring module: Used to collect power material data, perform graph structuring, and obtain an initial material relationship graph;

[0044] Quantification and grouping module: used to quantify based on the initial material relationship map using matrix representation, and further use the associated material grouping algorithm to group associated materials;

[0045] Entropy Analysis Module: Used to calculate the entropy value of each associated material group and remove noise points based on the principle of maximum entropy reduction to obtain the final material relationship map;

[0046] Prediction module: Used to predict the demand for materials based on the material relationship map and the determined quantitative relationship type.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] (1) Automated sample data noise reduction: The identification and cleaning of data noise is fully automated, reducing the drawbacks of traditional data cleaning, which requires too much manual effort and cannot eliminate subjective experience bias.

[0049] (2) Applicable to any type of power material: As a dimensionless measure, entropy measures the level of regularity of the sample. As long as the regularity of the noise data is obvious, this method can be applied without being limited by specific projects or material types.

[0050] (3) Entropy reduction as a project management indicator: The entropy reduction reflects the accuracy of the sample data, so it can be used as a reverse indicator to evaluate the standardization of sample project management.

[0051] (4) Material demand range: Based on the quantitative relationship type between materials determined by the material relationship map, predict the material demand range to improve the accuracy of prediction. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0053] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0054] Example 1

[0055] This embodiment provides a method for drawing power resource relationship maps and forecasting demand based on entropy values. It aims to automate the cleaning and optimization of power resource data through entropy analysis technology, thereby improving the accuracy of resource demand forecasting. The basic principle of entropy analysis is to distinguish normal data from noisy data by utilizing the contrast of data regularities. Entropy, as a dimensionless metric, measures the degree of regularity inherent in a sample and is not limited by the specific project or the scale of resources. The technical path for generating the resource relationship map consists of five steps: graph data acquisition, graph structure definition, graph matrix mapping, graph sample denoising, and graph quantitative analysis. Specifically, as follows... Figure 1 As shown in the figure, the method includes the following steps:

[0056] (1) Graph data collection

[0057] The purpose of graph data collection is to obtain the input data required for analysis. As the basis for the entire analysis and research, the selection of the scope of the input data is very important. In this project, typical distribution network projects in the past three years are used as the sample domain, and only projects that have completed final accounts are collected. The procurement orders include all-caliber batch tenders, agreement inventory allocations, and e-commerce procurement. After the historical data is collected, preliminary cleaning work is carried out to剔除 blank values and outliers.

[0058] (2) Graph structure definition

[0059] The structure of the material relationship graph is used to specify the underlying logic of the internal laws of materials. The basic content is to construct graph vertices and connecting edges. One graph vertex corresponds to one material, and any two vertices are connected to form a set of graph edges. When there are n materials in the whole domain, the relationship graph corresponds to n vertices and the number of permutations One-way edges. For the project set P, and two materials x and y, let the number of items in the project set be N(P), the projects using material x be P x , the projects using material y be P y , define the vertex function and edge function as follows:

[0060] Vertex function V(x): The vertex function is a mapping of the demand frequency of a single material, reflecting the absolute demand of the project for the material. The expression of the vertex function Obviously, 0 < V(x) ≤ 1, and when V(x) = 1, material x is an essential material for the entire project set P.

[0061] Edge function E(x, y): The edge function is a mapping of the demand correlation between two materials, reflecting the relative demand between the two materials. Denote the projects that use material x and also use material y as P y|x , the expression of the edge function Obviously, 0 < E(x, y) ≤ 1.

[0062] (3) Graph matrix mapping

[0063] Using the definition rules of the graph structure, convert the massive samples into vertex functions and edge functions, completing the first step from unstructured data to structured data. However, it still cannot be directly processed quantitatively by a computer. Therefore, a quantitative expression form of the relationship graph needs to be designed. Here, the matrix expression method is adopted.

[0064] For the n materials x1, x1... x covered by the project set P n , define the demand relationship vector D of the materials n =(d1, d2,..., dn ), d i It is the i-th material x i Demand frequency V(x) i This constructs the mapping from vertex functions to demand vectors. Define the material relationship matrix R n×n The element in row i and column j, r i，j =E(x) i x j ), 1≤i≤n, 1≤j≤n, and it is stipulated that r i，i =E(x) i x i = 1. This constructs a mapping from the edge function to the correlation matrix.

[0065] Demand Relationship Vector D n and the association matrix R n×n The introduction of [the algorithm / mechanism] solves the problem of quantitative representation of relationship graphs. The quantitative analysis of association relationships is transformed from an abstract graph structure into logical operations that can be accepted by computers. Based on the definition of association relationships, an algorithm for grouping associated materials is designed. For the prerequisite material, let material x be [the algorithm / mechanism]. i Prerequisite material mapping Only need to traverse the association matrix R n×n In the i-th row, filter the column indices where the element is 1. This is the premise of material mapping. For all materials x i A collection of materials that are directly related;

[0066] To solve for material x i The supporting material set, design mapping It also uses a traversal algorithm, and mandates... Based on the transitivity of the supporting relationships, for Obviously there is That is, the matching materials of mutually complementary materials must be grouped in the same way.

[0067] Finally, repeat the above steps to complete the grouping of all materials.

[0068] (4) Image sample noise reduction

[0069] First, construct the entropy value of the association graph. The entropy value reflects the degree of disorder in a system, and in the material relationship graph, it is used to reflect the dispersion of material grouping. Assume that the item set P covers materials x1, x2, ..., x... N Related materials are grouped as follows: And there are Define entropy in yes The number of materials in each group, where N is the total number of materials. For There must be Therefore, H ≥ 0. The more concentrated the grouping of related materials, the smaller the entropy; the more dispersed the grouping, the larger the entropy. The entropy drops to zero if and only if all materials are grouped into one group. The maximum entropy reduction noise reduction principle applies to two item sets P1 and P2: if... Then H(P1) ≤ H(P2). Removing noisy items will inevitably cause entropy reduction, but different noisy items will have different entropy reduction rates. Therefore, the noisy items with the largest entropy reduction should be removed first. The specific algorithm is as follows:

[0070] Step 1: Subset Entropy Comparison: For the current itemset P, delete the j-th item to form a subset P. -j Horizontal comparison of H(P) -j );

[0071] Step 2: Finding the path to maximum entropy reduction: For the largest H(P)... -i ), P -i =max{H(P -j )}, delete the i-th item and repeat step 1;

[0072] Step 3 Threshold Trigger: Entropy Reduction H(P) - max{H(P)} -j If the threshold is exceeded, the process ends.

[0073] (5) Graph Quantitative Analysis

[0074] First, determine the type of quantitative relationship between material requirements based on the material relationship diagram; then, calculate the demand range for each material based on the type of quantitative relationship to obtain the material demand range. The types of quantitative relationships include interval-type, formula-type, and proportional-type quantitative relationships.

[0075] Interval-type quantitative relationships: When it is difficult to find suitable engineering characteristic factors for quantitative relationships, the confidence interval estimation method is adopted. With a confidence level of α, the confidence interval estimate of the mean demand is: in It is the t-quantile with n-1 degrees of freedom.

[0076] Formulaic quantitative relationships: Quantitative relationships can incorporate engineering characteristic factors, denoted as Y = (y1, y2, ..., y...). n ) T Engineering characteristic factors Establish a linear equation Y = βX + ε, with undetermined coefficients β = (β0, β1, ..., β... m ) T Solve the linear equation for the undetermined coefficients β = (X TX) -1 X T Y. The characteristic factor of the newly added project is X. n+1 The demand forecast range is

[0077] The implementation method of this invention is computer program development, deployment and application. The programming language is Python, and the library packages are NumPy multidimensional array library package, Matplotlib data visualization, Pandas data analysis library, and Scikit-learn machine learning. The necessary data structures and technologies include: (1) ndarray class, which includes various matrix operation functions, supports matrix dot product and star product, adaptive broadcast mechanism, easily realizes the solution of correlation matrix, and reduces the workload of development; (2) JIT just-in-time compilation technology, which makes the NumPy library package developed based on C language have a calculation speed close to that of compiled language; (3) Network library, which supports multiple types of graphs, has built-in shortest path and connectivity detection algorithms, supports the calculation of clustering coefficients, and analyzes the local clustering structure of graphs.

[0078] Example 2

[0079] This embodiment provides a power resource relationship mapping and demand forecasting system based on entropy, including:

[0080] Graph structuring module: Used to collect power material data, perform graph structuring, and obtain an initial material relationship graph;

[0081] Quantification and grouping module: used to quantify based on the initial material relationship map using matrix representation, and further use the associated material grouping algorithm to group associated materials;

[0082] Entropy Analysis Module: Used to calculate the entropy value of each associated material group and remove noise points based on the principle of maximum entropy reduction to obtain the final material relationship map;

[0083] Prediction module: Used to predict the demand for materials based on the material relationship map and the determined quantitative relationship type.

[0084] The rest are as in Example 1.

[0085] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0086] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0087] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0088] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0089] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0090] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0091] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for drawing power resource relationship maps and forecasting demand based on entropy values, characterized in that, Includes the following steps: Collect power material data from the project set, perform graph structuring, and obtain an initial material relationship map; The initial material relationship map is quantified using matrix representation, and then associated material grouping is performed using an associated material grouping algorithm. The entropy value of the quantified initial material relationship map is calculated based on the grouping results of associated materials, and noise points are removed based on the maximum entropy reduction noise reduction principle to obtain the final material relationship map. Based on the aforementioned material relationship map, the range of material demand is predicted; The steps for grouping related materials include: Construction material x i Prerequisite material mapping And based on the aforementioned prerequisite material mapping Traverse the association matrix R n×n In the i-th row, filter the column indices where the element is 1 to find material x. i Prerequisite material x j The aforementioned prerequisite material mapping For all materials x i A set of materials with direct relationships; material mapping based on the aforementioned premise. Construct mapping Traverse the association matrix R n×n Given the i-th row and j-th column, find all column indices j that satisfy R(i,j) = R(j,i) = 1. Group the materials that match each other into one group, and finally repeat the above steps to complete the grouping of all materials; The steps for obtaining the final material relationship map include: 1) Determine if the entropy value is zero. If it is, it indicates that noise reduction is not needed, and the quantized initial material relationship map is used as the final material relationship map. If not, it indicates that noise reduction is needed, and proceed to the next step: 2) For the item set P in the quantified initial material relationship map, delete the j-th item to form a subset P. -j Calculate subset P -j The entropy value H(P) -j ), and repeat this step to obtain the entropy values ​​of all subsets; 3) Delete the item with the highest entropy value among all subsets; 4) Calculate whether the entropy reduction of item set P exceeds the threshold. If yes, end the noise reduction operation and obtain the final material relationship map. If not, repeat steps 1)-4) based on the entropy-reduced item set P until the end, and obtain the final material relationship map. The expression for calculating the entropy reduction of item set P is: Entropy reduction = H(P) - max{H(P)} -j )} In the formula, H(P) is the entropy value of the item set P.

2. The method for drawing power resource relationship maps and forecasting demand based on entropy values ​​according to claim 1, characterized in that, The steps for graph structuring include: The power material data is converted into vertex functions and edge functions to form an initial material relationship graph, wherein one type of material in the power material data corresponds to one graph vertex, and the relative demand of two types of materials corresponds to graph edges.

3. The method for drawing power material relationship maps and forecasting demand based on entropy values ​​according to claim 2, characterized in that... In the formula, V(x) is the vertex function of material x, which is a mapping of the demand frequency of material x and reflects the absolute demand of the project for material x. 0 < V(x) ≤ 1. When V(x) = 1, material x is an essential material for the entire project set P; N(P x ) is the number of items of project P that use material x x , and N(P) is the number of items in the project set; In the formula, E(x, y) is the edge function between material x and material y, which is a mapping of the demand correlation between material x and material y; N(P y|x ) is P y|x The number of terms, P y|x For an item set that uses material x, there is also an item set that uses material y; N(P) is the number of items in the item set.

4. The method for drawing power resource relationship maps and forecasting demand based on entropy values ​​according to claim 2, characterized in that, The steps for quantization using matrix representation include: Construct the mapping from the vertex function V(x) to the demand vector. Where D n = (d1, d2, ..., d i , ..., d n ) represents the demand relationship vector, d i It is the i-th material x i Demand frequency V(x) i ), where n is the number of material types covered by item set P; Construct a mapping from the edge function E(x,y) to the correlation matrix. Complete the quantization process, where R n×n For the association matrix, the element r in the i-th row and j-th column is... i，j =E(x) i x j ), 1≤i≤n, 1≤j≤n, and r i，i =E(x) i x i ) = 1.

5. The method for drawing power resource relationship maps and forecasting demand based on entropy values ​​according to claim 1, characterized in that, The expression for calculating the entropy value is: In the formula, H is the entropy value, reflecting the degree of concentration of related material groups, and m is the number of related material groups. Group the i-th associated material. x1, x2, ..., x N Let P be the itemset that covers the materials, and N be the number of materials.

6. The method for drawing power resource relationship maps and forecasting demand based on entropy values ​​according to claim 1, characterized in that, The steps for predicting the range of material demand include: The types of quantitative relationships between material requirements are determined based on the aforementioned material relationship map; The demand range for each material is calculated based on the aforementioned quantity relationship type to obtain the material demand range.

7. The method for drawing power resource relationship maps and forecasting demand based on entropy values ​​according to claim 6, characterized in that, The types of quantitative relationships include interval-type quantitative relationships, formula-type quantitative relationships, and proportional-type quantitative relationships.

8. A forecasting system for drawing power resource relationship maps and demand forecasting methods based on entropy values ​​according to any one of claims 1-7, characterized in that, include: Graph structuring module: Used to collect power material data, perform graph structuring, and obtain an initial material relationship graph; Quantification and grouping module: used to quantify based on the initial material relationship map using matrix representation, and further use the associated material grouping algorithm to group associated materials; Entropy Analysis Module: Used to calculate the entropy value of each associated material group and remove noise points based on the principle of maximum entropy reduction to obtain the final material relationship map; Prediction module: Used to predict the demand for materials based on the material relationship map and the determined quantitative relationship type.

Images