A tobacco project management method and system based on the Internet of Things
Through the Internet of Things-based tobacco project management method, using image acquisition and standard image set scoring, electronic archives and directed graphs with power are constructed, which solves the problems of subjective judgment and lack of norms in traditional project management, and achieves efficient and accurate projects Management and risk identification.
Patent Information
- Application Number
- CN202410630232.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-21
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-05-21
AI Technical Summary
Traditional project management methods rely on subjective judgment and lack standardized evaluation mechanisms and visualization tools, which leads to poor objectivity, accuracy and consistency of evaluation results, and lack of standardization of project processing processes, which can easily lead to confusion and delays.
Adopt the tobacco project management method based on the Internet of Things, obtain project information through IoT devices, match image collection with standard image sets for scoring, build electronic archives and powered directed graphs, realize the processing and scoring of project flow nodes, and combine visualization tools. Demonstrate project processing flow and time.
It improves the objectivity and accuracy of the evaluation results, forms a standardized project management process, enhances the visualization ability of project management, helps managers to identify and solve potential risks in a timely manner, and ensures that the project is completed on time and efficiently.
Smart Images

Figure CN118607900B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of project management, and in particular to a tobacco project management method and system based on the Internet of Things. Background Art
[0002] The Internet of Things (IoT) refers to a network system that uses Internet technology and devices such as sensors to connect various physical objects, achieve interconnection between them, and realize intelligent and automated data exchange and collaborative processing.
[0003] In traditional methods, project managers often rely on subjective judgment and experience to evaluate projects, which is easily affected by personal bias and errors, resulting in low objectivity and accuracy of evaluation results. In addition, in traditional methods, project managers may not be able to fully record relevant information of the project, including pictures, data, evaluation results, etc., so they cannot provide complete and true project information, which causes trouble for managers' decision-making and reference. In traditional methods, there is a lack of standardized evaluation mechanisms and indicator systems, resulting in poor comparability and consistency of evaluation results. Different managers may adopt different evaluation standards, making it difficult to conduct comprehensive evaluation and comparative analysis. In traditional methods, the project processing process usually lacks clear processes and specifications, which can easily lead to confusion, delays and errors. Project managers may not be able to clearly understand the order and flow of project processing, and it is difficult to identify and solve potential problems and risks in a timely manner. In traditional methods, due to the lack of visualization tools and analysis methods, it is difficult for project managers to intuitively understand the overall situation and key links of the project, and it is difficult to monitor and manage the progress of the project in a timely manner, which increases the difficulty and risk of project management. Summary of the invention
[0004] The purpose of the present invention is to propose a tobacco project management method and system based on the Internet of Things in view of the problems existing in the background technology.
[0005] The technical solution of the present invention is a tobacco project management method based on the Internet of Things, comprising:
[0006] Acquire target project information, and collect images of the target site corresponding to the target project information based on the Internet of Things device, so as to obtain target project image data;
[0007] Matching the target project image data with a preset standard image set library to obtain a standard image set corresponding to the target project image data, and scoring the target project image data according to the project score corresponding to the standard image set to obtain a first target project score set;
[0008] Building a target project electronic file based on the target project information, the target project image data, a standard image set corresponding to the target project image data, and the first target project scoring set;
[0009] Based on the project processing flow, a plurality of project transfer nodes are constructed, the project transfer nodes correspond to the project processing positions, the target project electronic files are transmitted to the plurality of project transfer nodes in a processing order, the project transfer nodes process and score the target project electronic files, thereby obtaining a second target project score set;
[0010] A directed graph is constructed based on the project processing positions, the directed graph includes a plurality of nodes and a plurality of directed edges, the directed edge is used to connect two adjacent nodes, and the direction of the directed edge corresponds to the project flow direction, the node corresponds to the project flow node, and the weight of the directed edge is assigned according to the processing time of the electronic archive of the target project, so as to obtain a weighted directed graph;
[0011] A comprehensive score of the target project is calculated based on the first target project score set and the second target project score set, and a shortest processing time of the target project is calculated based on the weighted directed graph, and multiple project flow nodes corresponding to the shortest processing time of the target project are recorded, and the weighted directed graph is displayed through a visualization tool.
[0012] Preferably, matching the target project image data with a preset standard image set library to obtain a standard image set corresponding to the target project image data includes:
[0013] Extracting the image features of the target project image data and the standard image features of the standard image set library through a trained feature extraction model;
[0014] The original distance between the target project image data and the standard image set library is calculated based on the cosine distance, and the cosine distance calculation formula is as follows:
[0015]
[0016] Among them, d(q,g i ) represents the original distance between the target project image data and the i-th standard image in the standard image set library, q represents the target project image data, g i represents the i-th standard image in the standard image set library, x q represents the image features of the target item image data, x gi represents the standard image feature of the i-th standard image in the standard image set library;
[0017] Based on the original distance, an initial list in ascending order is obtained. Based on the initial list, k standard images that are most similar to the target project image data are obtained from the standard image set library, thereby obtaining a similar image set. The initial list is as follows:
[0018] L(q,G)={d(q,g0),d(q,g1),...,d(q,g N )};
[0019] Wherein, L(q,G) represents the initial list, and N represents the total number of standard images in the standard image set library;
[0020] The similar image set is divided into a hard adjacent image set and a soft adjacent image set, and the most similar standard image is selected from the similar image set, and k / 2 standard images that are most similar to the most similar standard image in the standard image set library are obtained, so as to obtain a standard image similarity set. The division expression is as follows:
[0021]
[0022] Among them, N h (q, k) represents the hard neighboring image set, N s (q, k) represents the soft neighboring image set, N(q, k) represents the similar image set, N(g i , k) represents the k standard images in the standard image set library that are most similar to the i-th standard image in the standard image set library;
[0023] Determine whether the intersection of the hard neighboring image set and the standard image similarity set is greater than a preset threshold. If so, add the standard image similarity set to the hard neighboring image set to obtain a final hard neighboring image set, and merge the final hard neighboring image set with the soft neighboring image set to obtain a standard image set corresponding to the target project image data.
[0024] Preferably, the feature extraction model includes a channel reorganization module, a global relationship modeling branch module, a local feature extraction branch module and a feature fusion module. The channel reorganization module performs the input feature F i Use point-by-point convolution and integrate and reorganize along the channel dimension, and transform the input feature F i Cut into two parts of equal size, denoted as F l and F g , F g and F lare respectively transmitted to the global relationship modeling branch module and the local feature extraction branch module, the global relationship modeling branch module and the local feature extraction branch module run in parallel, and the global relationship modeling branch module uses the Transformer encoder to g The global relationship of is globally modeled to obtain the first output feature X g The local feature extraction branch module uses ResNet to extract F l Perform local modeling to obtain the second output feature X l , and the local feature extraction branch module is composed of a residual network, and the feature fusion module is used to transform the first output feature X g and the second output feature X l The image features are obtained by fusion.
[0025] Preferably, the project transfer node includes a cloud user node, a cloud interaction node, a block generation node and a scoring endorsement node. The cloud user node is used to receive a service request from a person in charge of a project processing position. The cloud interaction node is used for the alliance chain to interact with the cloud storage end. The block generation node is used to package the processing and scoring results of the target project electronic archive into blocks and transmit them to the blockchain network. The scoring endorsement node is used to perform evidence verification and dynamic operation authentication on the cloud storage end. The person in charge of the project processing position initiates an endorsement request to the scoring endorsement node through the cloud user node. The scoring endorsement node performs corresponding evidence verification and authentication operations according to the request type, and feeds back the result to the cloud user node. The cloud user node determines the validity of the processing and scoring of the target project electronic archive based on the feedback result. If it is determined to be valid, a packaging request is initiated to the block generation node, thereby storing the processing and scoring results of the target project electronic archive in the corresponding blockchain.
[0026] Preferably, a directed graph is constructed based on the project processing positions, the directed graph includes a plurality of nodes and a plurality of directed edges, the directed edge is used to connect two adjacent nodes, and the direction of the directed edge corresponds to the project flow direction, the node corresponds to the project flow node, and the weight of the directed edge is assigned according to the processing time of the target project electronic file, so as to obtain a weighted directed graph, including:
[0027] According to the target project processing position, determine the node corresponding to each project processing position;
[0028] Connecting two adjacent nodes with a directed edge, wherein the direction of the directed edge corresponds to the direction of project flow;
[0029] Assigning weights to the directed edges according to the processing time recorded in the target project electronic archive;
[0030] Repeat the above steps until all the target project processing positions are processed, thereby obtaining a weighted directed graph.
[0031] Preferably, the calculating the comprehensive score of the target project based on the first target project score set and the second target project score set includes:
[0032] Establishing a joint scoring matrix based on the second target project scoring set, and performing matrix decomposition on the joint scoring matrix using a matrix decomposition technique, thereby decomposing the joint scoring matrix into the product of a potential factor matrix of a responsible person and a potential factor matrix of a project scoring point;
[0033] Using the k-means clustering method to cluster the potential factor matrix of the person in charge and the potential factor matrix of the project scoring points, thereby obtaining cluster persons in charge and cluster project scoring points;
[0034] The professionalism of the person in charge is calculated based on the historical scoring data, thereby obtaining the scoring professionalism of the person in charge. The professionalism calculation formula is as follows:
[0035]
[0036] in, Indicates the professionalism of the person in charge in the project scoring points of the target project, represents the score of the project scoring point i in the target project by the person in charge u, represents the highest score of item score point i in the target item, represents the average score of item score point i in the target project, H n represents the total number of ratings approved by the person in charge u, V n Indicates the total number of approvals and disapprovals of the ratings of the person in charge u;
[0037] The calculation formula for the score of the person in charge is as follows:
[0038]
[0039] Among them, Z f (f u ) represents the professional rating of the person in charge, N m represents the maximum number of scores among all the scores given to the project by the person in charge u, and N represents the number of scores given by the person in charge u.
[0040] Preferably, the calculating the comprehensive score of the target project based on the first target project score set and the second target project score set includes:
[0041] The weight of the cluster leader is calculated based on the score expertise of the leader. The weight calculation formula of the cluster leader is as follows:
[0042]
[0043] Among them, a u represents the weight of the person in charge of the cluster, and M represents the number of persons in the cluster;
[0044] The cluster leader's score for the cluster item score point is calculated based on the cluster leader's weight. The score calculation formula for the cluster item score point is as follows:
[0045]
[0046] Among them, r′ represents the score of the cluster item score point, r u ' i represents the score of the project scoring point in the target project by the cluster leader, represents the i-th item score point in the target item in the cluster j, represents the number of scoring points of the i-th item in the target item in the cluster j;
[0047] Calculate the first average score of the first target project score set, and obtain the second average score of the second target project score set based on the score of the cluster leader on the cluster project score point, and simply weight the first average score and the second average score to obtain the comprehensive score of the target project.
[0048] Preferably, the calculating the shortest processing time of the target project based on the weighted directed graph includes:
[0049] Create an empty shortest path set, set the shortest processing time of the starting node in the weighted directed graph to 0, and set the shortest processing time of other nodes to infinity;
[0050] Adding the starting node to the shortest path set, and updating the shortest processing time of all neighboring nodes of the starting node;
[0051] For each neighbor node, calculate the path length from the current node to the neighbor node and compare it with the current shortest processing time of the neighbor node. If the path length is less than the current shortest processing time, update the shortest processing time of the neighbor node.
[0052] A node with the minimum shortest processing time is selected from the nodes that have not been added to the shortest path set, and it is added to the shortest path set. The above operation is repeated until the shortest path set includes all nodes or no nodes are added, and the shortest processing time of the target item is obtained based on the shortest path set.
[0053] The technical solution of the present invention is a tobacco project management system based on the Internet of Things, which is applicable to the tobacco project management method based on the Internet of Things, including:
[0054] A target project information acquisition module 1, wherein the target project information acquisition module 1 is used to acquire relevant information of the target project;
[0055] Image acquisition module 2, the image acquisition module 2 is used to collect images of the target site through the Internet of Things device to obtain image data of the target project;
[0056] The standard image set matching module 3 is used to match the target project image data with a preset standard image set library to find a standard image set that matches the target project image;
[0057] Scoring module 4, the scoring module 4 is used to score the target project image data according to the project scores corresponding to the standard image set to obtain a first target project score set;
[0058] A target project electronic archive construction module 5, wherein the target project electronic archive construction module 5 is used to construct an electronic archive of the target project based on the target project information, the target project image data and the first target project score set;
[0059] The project transfer node and processing module 6 are used to construct a plurality of project transfer nodes corresponding to the project processing positions. The target project electronic files are transmitted to the project transfer nodes in a processing order. The project transfer nodes process and score the target project electronic files to obtain a second target project score set.
[0060] A directed graph construction module 7, which is used to construct a weighted directed graph based on the project processing position, to represent the process and order of project processing, wherein the nodes represent the project flow nodes, the directed edges represent the processing order between adjacent nodes, and the directed edges are assigned weights according to the processing time of the target project electronic archive;
[0061] A comprehensive score and shortest processing time calculation module 8, which is used to calculate the comprehensive score of the target project based on the first target project score set and the second target project score set, calculate the shortest processing time of the target project using a weighted directed graph, and record the corresponding multiple project flow nodes;
[0062] The visualization display module 9 is used to display the weighted directed graph through a visualization tool, and visualize the project processing flow and the shortest processing time.
[0063] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:
[0064] 1. The present invention collects images of the target site through the Internet of Things device, and can obtain information of the target project in real time, including tobacco growth conditions, site environment, etc., so that project managers can timely understand the project progress and environmental changes, which is conducive to timely adjustment of management strategies and taking countermeasures. By matching the target project image data with a preset standard image set library and scoring the target project image data according to the project score corresponding to the standard image set, a standardized evaluation mechanism is realized. This standardized evaluation method can avoid the interference of subjective factors and improve the objectivity and accuracy of the evaluation results. The electronic archive of the target project can comprehensively record the relevant information of the target project, including project information, image data, evaluation results and other aspects. This recording method can improve the integrity and authenticity of the information and provide a more comprehensive reference for project managers.
[0065] 2. The present invention can form a process-based management model for the project management process by constructing a project processing flow and multiple project flow nodes, making the project processing process more standardized and orderly. This process-based management method can improve the efficiency and accuracy of management. By processing and scoring the electronic files of the target project through multiple project flow nodes, the target project can be comprehensively evaluated from different angles. This comprehensive evaluation method can improve the reliability and accuracy of the evaluation results. By constructing a weighted directed graph, the relationship between the project processing flow and each processing position can be intuitively displayed. This visual process can help managers understand the order and flow of project processing more clearly, so as to better monitor and manage project progress. The directed graph can reveal possible delay nodes or bottlenecks in the project processing process, so as to help managers identify and resolve potential risks in a timely manner. Through the analysis of the directed graph, targeted measures can be taken to ensure that the project can be completed on time and efficiently.
[0066] 3. The present invention can obtain a comprehensive score for the target project by comprehensively calculating the first target project score set and the second target project score set. This comprehensive evaluation can comprehensively consider the influence of multiple factors, so as to more comprehensively evaluate the quality and value of the target project, and recording multiple project flow nodes corresponding to the shortest processing time can help managers identify key links and key positions in the project processing process. By focusing on and monitoring these nodes, it can ensure that the project can proceed smoothly and deal with possible problems and risks in a timely manner. Moreover, the use of visualization tools to display weighted directed graphs can intuitively present the project processing flow and the relationship between each processing link. This visualization can help managers better understand and grasp the overall situation of the project, and support decision-making and communication. BRIEF DESCRIPTION OF THE DRAWINGS
[0067] Figure 1 A schematic flow chart of an overall method in an embodiment of the present invention;
[0068] Figure 2 A schematic diagram of a process for matching target project image data in an embodiment of the present invention;
[0069] Figure 3 A schematic diagram of a process of constructing a directed graph in an embodiment of the present invention;
[0070] Figure 4 A schematic diagram of a flow chart of calculating a comprehensive score of a target project in an embodiment of the present invention;
[0071] Figure 5 A schematic diagram of a process for calculating the shortest processing time of a target item in an embodiment of the present invention;
[0072] Figure 6 A schematic diagram of the overall system flow in an embodiment of the present invention.
[0073] Figure numerals: 1. Target project information acquisition module; 2. Image acquisition module; 3. Standard image set matching module; 4. Scoring module; 5. Target project electronic archive construction module; 6. Project flow node and processing module; 7. Directed graph construction module; 8. Comprehensive scoring and shortest processing time calculation module; 9. Visual display module. DETAILED DESCRIPTION
[0074] Embodiment 1, as Figure 1 As shown, the present invention proposes a tobacco project management method based on the Internet of Things, comprising:
[0075] S1. Obtain target project information, and collect images of the target site corresponding to the target project information based on the Internet of Things device, thereby obtaining target project image data;
[0076] S2, matching the target project image data with a preset standard image set library to obtain a standard image set corresponding to the target project image data, and scoring the target project image data according to the project score corresponding to the standard image set to obtain a first target project score set;
[0077] S3, constructing an electronic file of the target project based on the target project information, the target project image data, the standard image set corresponding to the target project image data, and the first target project scoring set;
[0078] S4. Based on the project processing flow, multiple project transfer nodes are constructed. The project transfer nodes correspond to the project processing positions. The electronic files of the target projects are transmitted to the multiple project transfer nodes in the processing order. The project transfer nodes process and score the electronic files of the target projects, thereby obtaining a second target project score set.
[0079] S5. Construct a directed graph based on the project processing positions. The directed graph includes multiple nodes and multiple directed edges. The directed edge is used to connect two adjacent nodes. The direction of the directed edge corresponds to the direction of the project flow. The node corresponds to the project flow node. The weight of the directed edge is assigned according to the processing time of the target project electronic file, thereby obtaining a weighted directed graph.
[0080] S6. Calculate the comprehensive score of the target project based on the first target project score set and the second target project score set, calculate the shortest processing time of the target project based on the weighted directed graph, record multiple project flow nodes corresponding to the shortest processing time of the target project, and display the weighted directed graph through a visualization tool.
[0081] In the present invention, the standard image set library refers to a database containing a series of standard images, which represent reference samples of different qualities, characteristics or categories. The library is usually created and maintained by professionals or experts in related fields, and is used to compare, match, score or classify with the target image. In the target project, the standard image set library can be a predefined image collection, which contains a plurality of standard images that have been scored or classified. These standard images are usually representative and reflect the specific attributes, qualities or categories of interest to some extent. The target project electronic archive refers to a collection of target project related information, image data and scoring results. Structured data is an electronic file that is integrated, archived and stored according to certain formats and specifications. In the target project, an electronic archive of the target project can be constructed based on the target project information, target project image data, the standard image set corresponding to the target project image data and the first target project scoring set. The target project information includes basic information such as the project name, project description, project leader, project start time, and expected completion time; the target project image data includes the collected original image data and processed image data; the standard image set and the first target project scoring set are used to match and score the target project images, thereby supporting the assessment and monitoring of project quality.
[0082] Embodiment 2, as Figure 2-5 As shown, the tobacco project management method based on the Internet of Things proposed by the present invention, compared with the first embodiment, this embodiment also includes:
[0083] Match the target project image data with the preset standard image set library to obtain the standard image set corresponding to the target project image data, including:
[0084] A1. Extracting the image features of the target project image data and the standard image features of the standard image set library through the trained feature extraction model;
[0085] A2. Calculate the original distance between the target project image data and the standard image set library based on the cosine distance. The cosine distance calculation formula is as follows:
[0086]
[0087] Among them, d(q,g i ) represents the original distance between the target project image data and the i-th standard image in the standard image set library, q represents the target project image data, g i represents the i-th standard image in the standard image set library, x q Represents the image features of the target item image data, x gi Represents the standard image features of the i-th standard image in the standard image set library;
[0088] A3. Based on the original distance, an initial list in ascending order is obtained. Based on the initial list, k standard images that are most similar to the target project image data are obtained from the standard image set library, thereby obtaining a similar image set. The initial list is as follows:
[0089] L(q,G)={d(q,g0),d(q,g1),...,d(q,g N )};
[0090] Where L(q,G) represents the initial list, and N represents the total number of standard images in the standard image set library;
[0091] A4. Divide the similar image set into a hard adjacent image set and a soft adjacent image set, select the most similar standard image from the similar image set, and obtain the k / 2 standard images that are most similar to the most similar standard image in the standard image set library, thereby obtaining a standard image similarity set. The division expression is as follows:
[0092]
[0093] Among them, N h (q, k) represents the hard neighboring image set, N s (q,k) represents the soft neighboring image set, N(q,k) represents the similar image set, and N(g i ,k) represents the k standard images in the standard image set library that are most similar to the i-th standard image in the standard image set library;
[0094] A5. Determine whether the intersection of the hard neighboring image set and the standard image similarity set is greater than a preset threshold. If so, add the standard image similarity set to the hard neighboring image set to obtain a final hard neighboring image set, and merge the final hard neighboring image set with the soft neighboring image set to obtain a standard image set corresponding to the target project image data.
[0095] In an optional embodiment, the feature extraction model includes a channel reorganization module, a global relationship modeling branch module, a local feature extraction branch module and a feature fusion module. The channel reorganization module performs input feature F i Use point-by-point convolution and integrate and reorganize along the channel dimension, and transform the input feature F i Cut into two parts of equal size, denoted as F l and F g , F g and F l They are respectively transmitted to the global relationship modeling branch module and the local feature extraction branch module. The global relationship modeling branch module and the local feature extraction branch module run in parallel, and the global relationship modeling branch module uses the Transformer encoder to gThe global relationship of is globally modeled to obtain the first output feature X g The local feature extraction branch module uses ResNet to extract F l Perform local modeling to obtain the second output feature X l , and the local feature extraction branch module is composed of a residual network, and the feature fusion module is used to transform the first output feature X g and the second output feature X l The image features are obtained by fusion.
[0096] It should be noted that the Transformer encoder is a neural network structure based on the self-attention mechanism, which is composed of multiple Encoder layers. In the above optional embodiment, the global relationship modeling branch module uses the Transformer encoder to model the global relationship to extract the relationship information between different regions in the image; ResNet (Residual Network) is a deep residual network composed of multiple Residual Blocks. ResNet solves the gradient disappearance and gradient explosion problems in deep neural networks by introducing cross-layer connections (shortcut connection). In the above optional embodiment, the local feature extraction branch module is composed of multiple ResNets, which are used to model local features to extract local information in the image.
[0097] In an optional embodiment, the project transfer node includes a cloud user node, a cloud interaction node, a block generation node and a scoring endorsement node. The cloud user node is used to receive service requests from project processing position leaders. The cloud interaction node is used for the alliance chain to interact with the cloud storage end. The block generation node is used to package the processing and scoring results of the target project electronic archives into blocks and transmit them to the blockchain network. The scoring endorsement node is used to perform cloud storage end evidence verification and dynamic operation authentication. The project processing position leader initiates an endorsement request to the scoring endorsement node through the cloud user node. The scoring endorsement node performs corresponding evidence verification and authentication operations based on the request type, and feeds back the results to the cloud user node. The cloud user node determines the validity of the processing and scoring of the target project electronic archives based on the feedback results. If it is determined to be valid, a packaging request is initiated to the block generation node, thereby storing the processing and scoring results of the target project electronic archives in the corresponding blockchain.
[0098] It should be noted that blockchain is a decentralized distributed database technology consisting of multiple nodes. Each node has a complete copy of the data, which can verify and store all transaction records and data to ensure that the data cannot be tampered with. The consortium chain is different from the public blockchain. The consortium chain is a blockchain network maintained by several members. These members are usually organizations or individuals with a certain trust relationship, who jointly formulate the operating rules of the blockchain. In blockchain technology, endorsement refers to the process by which nodes in the blockchain network authenticate and confirm the validity of transactions or data. The endorsing nodes need to perform verification and authentication operations to ensure that the transactions or data comply with the rules and standards of the blockchain network. Dynamic operation authentication is an authentication method based on multi-dimensional information such as user behavior, device status, and environmental conditions. It can verify and authorize user identities in real time and dynamically, improving the security and credibility of authentication.
[0099] In an optional embodiment, a directed graph is constructed based on the project processing positions. The directed graph includes multiple nodes and multiple directed edges. The directed edge is used to connect two adjacent nodes. The direction of the directed edge corresponds to the project flow direction. The node corresponds to the project flow node. The weight of the directed edge is assigned according to the processing time of the target project electronic file, so as to obtain a weighted directed graph, including:
[0100] B1. According to the target project processing position, determine the node corresponding to each project processing position;
[0101] B2. Connect two adjacent nodes with directed edges, where the direction of the directed edge corresponds to the direction of project flow;
[0102] B3. Assign weights to directed edges based on the processing time recorded in the target project's electronic archive;
[0103] B4. Repeat the above steps until all target project processing positions are processed, thereby obtaining a weighted directed graph.
[0104] In an optional embodiment, calculating a comprehensive score of the target item based on the first target item score set and the second target item score set includes:
[0105] C1. Establish a joint scoring matrix based on the second target project scoring set, and use matrix decomposition technology to decompose the joint scoring matrix, so as to decompose the joint scoring matrix into the product of the potential factor matrix of the person in charge and the potential factor matrix of the project scoring point;
[0106] C2. Use the k-means clustering method to cluster the potential factor matrix of the person in charge and the potential factor matrix of the project scoring points, so as to obtain the cluster person in charge and the cluster project scoring points;
[0107] C3. Calculate the professionalism of the person in charge based on the historical scoring data to obtain the scoring professionalism of the person in charge. The professionalism calculation formula is as follows:
[0108]
[0109] in, Indicates the professionalism of the person in charge in the project scoring points of the target project, represents the score of project score point i in the target project by person in charge u, represents the highest score of item score point i in the target project, represents the average score of item i in the target project, H n V represents the total number of approvals for the person in charge u. n Indicates the total number of approvals and disapprovals for the rating of person u;
[0110] The formula for calculating the professionalism of the person in charge is as follows:
[0111]
[0112] Among them, Z f (f u ) represents the professionalism of the person in charge, N m It represents the maximum number of ratings given by all project managers u, and N represents the number of ratings given by manager u.
[0113] In an optional embodiment, calculating a comprehensive score of the target item based on the first target item score set and the second target item score set includes:
[0114] C4. Calculate the weight of the cluster leader based on the leader's scoring expertise. The weight calculation formula for the cluster leader is as follows:
[0115]
[0116] Among them, a u represents the weight of the cluster leader, and M represents the number of leaders in the cluster;
[0117] C5. Calculate the cluster leader's score for the cluster project score points based on the cluster leader's weight. The score calculation formula for the cluster project score points is as follows:
[0118]
[0119] Among them, r′ represents the score of the cluster item score point, r u ' i represents the cluster leader's score for the project scoring points in the target project, represents the score point of the i-th item in the target item in cluster j, represents the number of scoring points of the i-th item in the target item in cluster j;
[0120] C6. Calculate the first average score of the first target project score set, and obtain the second average score of the second target project score set based on the cluster leader's score on the cluster project score points. Perform simple weighting on the first average score and the second average score to obtain a comprehensive score for the target project.
[0121] In an optional embodiment, calculating the shortest processing time of the target project based on the weighted directed graph includes:
[0122] D1. Create an empty shortest path set, set the shortest processing time of the starting node in the weighted directed graph to 0, and set the shortest processing time of other nodes to infinity;
[0123] D2. Add the starting node to the shortest path set, and update the shortest processing time of all neighboring nodes of the starting node;
[0124] D3. For each neighbor node, calculate the path length from the current node to the neighbor node, and compare it with the current shortest processing time of the neighbor node. If the path length is less than the current shortest processing time, update the shortest processing time of the neighbor node.
[0125] D4. Select a node with the minimum shortest processing time from the nodes that are not added to the shortest path set, add it to the shortest path set, repeat the above operation until the shortest path set includes all nodes or no nodes are added, and obtain the shortest processing time of the target project based on the shortest path set.
[0126] Embodiment three, as Figure 6 As shown, a tobacco project management system based on the Internet of Things proposed by the present invention is applicable to the tobacco project management method based on the Internet of Things, including:
[0127] Target project information acquisition module 1, the target project information acquisition module 1 is used to obtain relevant information of the target project;
[0128] Image acquisition module 2, image acquisition module 2 is used to collect images of the target site through the Internet of Things device to obtain image data of the target project;
[0129] The standard image set matching module 3 is used to match the target project image data with a preset standard image set library to find a standard image set that matches the target project image;
[0130] Scoring module 4, scoring module 4 is used to score the target project image data according to the project scores corresponding to the standard image set to obtain a first target project score set;
[0131] A target project electronic archive construction module 5, the target project electronic archive construction module 5 is used to construct an electronic archive of the target project based on the target project information, the target project image data and the first target project score set;
[0132] The project transfer node and processing module 6 are used to construct a plurality of project transfer nodes corresponding to the project processing positions. The electronic files of the target projects are transmitted to the project transfer nodes in the processing order. The project transfer nodes process and score the electronic files of the target projects to obtain a second target project score set.
[0133] A directed graph construction module 7 is used to construct a weighted directed graph based on the project processing positions, which is used to represent the process and sequence of project processing. The nodes represent the project flow nodes, and the directed edges represent the processing sequence between adjacent nodes. The directed edges are weighted according to the processing time of the target project electronic archives.
[0134] A comprehensive score and shortest processing time calculation module 8 is used to calculate the comprehensive score of the target project based on the first target project score set and the second target project score set, calculate the shortest processing time of the target project using a weighted directed graph, and record the corresponding multiple project flow nodes;
[0135] Visualization display module 9 is used to display the weighted directed graph through visualization tools, and visualize the project processing flow and the shortest processing time.
[0136] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.
Claims
1. A tobacco project management method based on the Internet of Things, It is characterized in that include: Acquire target project information, and collect images of the target site corresponding to the target project information based on the Internet of Things device, so as to obtain target project image data; Matching the target project image data with a preset standard image set library to obtain a standard image set corresponding to the target project image data, and scoring the target project image data according to the project score corresponding to the standard image set to obtain a first target project score set; Building a target project electronic file based on the target project information, the target project image data, a standard image set corresponding to the target project image data, and the first target project scoring set; Based on the project processing flow, a plurality of project transfer nodes are constructed, the project transfer nodes correspond to the project processing positions, the target project electronic files are transmitted to the plurality of project transfer nodes in a processing order, the project transfer nodes process and score the target project electronic files, thereby obtaining a second target project score set; A directed graph is constructed based on the project processing position, the directed graph includes a plurality of nodes and a plurality of directed edges, the directed edge is used to connect two adjacent nodes, and the direction of the directed edge corresponds to the project flow direction, the node corresponds to the project flow node, and the weight of the directed edge is assigned according to the processing time of the target project electronic file, so as to obtain a weighted directed graph, including: According to the target project processing position, determine the node corresponding to each project processing position; Connecting two adjacent nodes with a directed edge, wherein the direction of the directed edge corresponds to the direction of project flow; Assigning weights to the directed edges according to the processing time recorded in the target project electronic archive; Repeat the above steps until all the target project processing positions are processed, thereby obtaining a weighted directed graph; Calculating a comprehensive score of the target project based on the first target project score set and the second target project score set, calculating a shortest processing time of the target project based on the weighted directed graph, recording a plurality of project flow nodes corresponding to the shortest processing time of the target project, and displaying the weighted directed graph through a visualization tool; The step of matching the target project image data with a preset standard image set library to obtain a standard image set corresponding to the target project image data includes: Extracting the image features of the target project image data and the standard image features of the standard image set library through a trained feature extraction model; The original distance between the target project image data and the standard image set library is calculated based on the cosine distance, and the cosine distance calculation formula is as follows: Among them, d(q,g i ) represents the original distance between the target project image data and the i-th standard image in the standard image set library, q represents the target project image data, g i represents the i-th standard image in the standard image set library, x q represents the image features of the target item image data, x gi represents the standard image feature of the i-th standard image in the standard image set library; Based on the original distance, an initial list in ascending order is obtained. Based on the initial list, k standard images that are most similar to the target project image data are obtained from the standard image set library, thereby obtaining a similar image set. The initial list is as follows: L(q,G)={d(q,g 0 ),d(q,g 1 ),...,d(q,g N )}; Wherein, L(q,G) represents the initial list, and N represents the total number of standard images in the standard image set library; The similar image set is divided into a hard adjacent image set and a soft adjacent image set, and the most similar standard image is selected from the similar image set, and k / 2 standard images that are most similar to the most similar standard image in the standard image set library are obtained, so as to obtain a standard image similarity set. The division expression is as follows: Among them, N h (q, k) represents the hard neighboring image set, N s (q, k) represents the soft neighboring image set, N(q, k) represents the similar image set, N(g i , k) represents the k standard images in the standard image set library that are most similar to the i-th standard image in the standard image set library; Determine whether the intersection of the hard adjacent image set and the standard image similarity set is greater than a preset threshold, and if so, add the standard image similarity set to the hard adjacent image set to obtain a final hard adjacent image set, and merge the final hard adjacent image set with the soft adjacent image set to obtain a standard image set corresponding to the target project image data; The project transfer node includes a cloud user node, a cloud interaction node, a block generation node and a scoring endorsement node. The cloud user node is used to receive a service request from a project processing position leader. The cloud interaction node is used for the alliance chain to interact with the cloud storage end. The block generation node is used to package the processing and scoring results of the target project electronic archive into blocks and transmit them to the blockchain network. The scoring endorsement node is used to perform evidence verification and dynamic operation authentication on the cloud storage end. The project processing position leader initiates an endorsement request to the scoring endorsement node through the cloud user node. The scoring endorsement node performs corresponding evidence verification and authentication operations according to the request type and feeds back the result to the cloud user node. The cloud user node determines the validity of the processing and scoring of the target project electronic archive based on the feedback result. If it is determined to be valid, a packaging request is initiated to the block generation node, thereby storing the processing and scoring results of the target project electronic archive in the corresponding blockchain; The calculating the comprehensive score of the target project based on the first target project score set and the second target project score set includes: Establishing a joint scoring matrix based on the second target project scoring set, and performing matrix decomposition on the joint scoring matrix using a matrix decomposition technique, thereby decomposing the joint scoring matrix into the product of a potential factor matrix of a responsible person and a potential factor matrix of a project scoring point; Using the k-means clustering method to cluster the potential factor matrix of the person in charge and the potential factor matrix of the project scoring points, thereby obtaining cluster persons in charge and cluster project scoring points; The professionalism of the person in charge is calculated based on the historical scoring data, thereby obtaining the scoring professionalism of the person in charge. The professionalism calculation formula is as follows: in, Indicates the professionalism of the person in charge in the project scoring points of the target project, represents the score of the project scoring point i in the target project by the person in charge u, represents the highest score of item score point i in the target item, represents the average score of item score point i in the target project, H n represents the total number of ratings approved by the person in charge u, V n Indicates the total number of approvals and disapprovals of the ratings of the person in charge u; The calculation formula for the score of the person in charge is as follows: Among them, Z f (f u ) represents the professional rating of the person in charge, N m represents the maximum number of scores given by the person in charge u to the project, and N represents the number of scores given by the person in charge u; The weight of the cluster leader is calculated based on the score expertise of the leader. The weight calculation formula of the cluster leader is as follows: Among them, a u represents the weight of the person in charge of the cluster, and M represents the number of persons in the cluster; The cluster leader's score for the cluster item score point is calculated based on the cluster leader's weight. The score calculation formula for the cluster item score point is as follows: Among them, r′ represents the score of the cluster item score point, r u ' i represents the score of the project scoring point in the target project by the cluster leader, represents the i-th item score point in the target item in the cluster j, represents the number of scoring points of the i-th item in the target item in the cluster j; Calculate a first average score of the first target project score set, and obtain a second average score of the second target project score set based on the score of the cluster project score point by the cluster leader, and simply weight the first average score and the second average score to obtain a comprehensive score of the target project; The calculating the shortest processing time of the target project based on the weighted directed graph includes: Create an empty shortest path set, set the shortest processing time of the starting node in the weighted directed graph to 0, and set the shortest processing time of other nodes to infinity; Adding the starting node to the shortest path set, and updating the shortest processing time of all neighboring nodes of the starting node; For each neighbor node, calculate the path length from the current node to the neighbor node and compare it with the current shortest processing time of the neighbor node. If the path length is less than the current shortest processing time, update the shortest processing time of the neighbor node. A node with the minimum shortest processing time is selected from the nodes that have not been added to the shortest path set, and it is added to the shortest path set. The above operation is repeated until the shortest path set includes all nodes or no nodes are added, and the shortest processing time of the target item is obtained based on the shortest path set.
2. According to the tobacco project management method based on the Internet of Things according to claim 1, It is characterized in that The feature extraction model includes a channel reorganization module, a global relationship modeling branch module, a local feature extraction branch module and a feature fusion module. The channel reorganization module performs input feature F i Use point-by-point convolution and integrate and reorganize along the channel dimension, and transform the input feature F i Cut into two parts of equal size, denoted as F l and F g , F g and F l are respectively transmitted to the global relationship modeling branch module and the local feature extraction branch module, the global relationship modeling branch module and the local feature extraction branch module run in parallel, and the global relationship modeling branch module uses the Transformer encoder to g The global relationship of is globally modeled to obtain the first output feature X g The local feature extraction branch module uses ResNet to extract F l Perform local modeling to obtain the second output feature X l , and the local feature extraction branch module is composed of a residual network, and the feature fusion module is used to transform the first output feature X g and the second output feature X l The image features are obtained by fusion.
3. A tobacco project management system based on the Internet of Things, which is applicable to a tobacco project management method based on the Internet of Things as described in any one of claims 1-2, It is characterized in that include: A target project information acquisition module (1), wherein the target project information acquisition module (1) is used to acquire relevant information of the target project; An image acquisition module (2), the image acquisition module (2) being used to acquire images of a target site through an Internet of Things device to obtain image data of a target project; A standard image set matching module (3), wherein the standard image set matching module (3) is used to match the target project image data with a preset standard image set library to find a standard image set that matches the target project image; A scoring module (4), the scoring module (4) being used to score the target project image data according to the project scores corresponding to the standard image set, to obtain a first target project score set; A target project electronic archive construction module (5), the target project electronic archive construction module (5) being used to construct an electronic archive of the target project based on the target project information, the target project image data and the first target project score set; A project transfer node and a processing module (6), wherein the project transfer node and the processing module (6) are used to construct a plurality of project transfer nodes corresponding to project processing positions, the target project electronic files are transmitted to the project transfer nodes in a processing order, and the project transfer nodes process and score the target project electronic files to obtain a second target project score set; A directed graph construction module (7), the directed graph construction module (7) is used to construct a weighted directed graph based on the project processing position, used to represent the process and sequence of project processing, the nodes represent the project flow nodes, the directed edges represent the processing sequence between adjacent nodes, and the directed edges are assigned weights according to the processing time of the target project electronic archive; A comprehensive score and shortest processing time calculation module (8), the comprehensive score and shortest processing time calculation module (8) is used to calculate the comprehensive score of the target project based on the first target project score set and the second target project score set, calculate the shortest processing time of the target project using a weighted directed graph, and record the corresponding multiple project flow nodes; A visualization display module (9), wherein the visualization display module (9) is used to display a weighted directed graph through a visualization tool, and to visualize the project processing flow and the shortest processing time.
