Traceable whole-process management method and system for industrial worker points

Abstract

The present invention is applicable to the field of manufacturing management technology, and provides a traceable full-process management method and system for industrial worker points, including the following steps: receiving a points update instruction, which includes a worker number, a points increase or decrease type, and a verification picture; determining basic points according to the points increase or decrease type, identifying environmental information for the verification picture, retrieving an environmental weight, and determining a job type weight and a historical performance weight based on the worker number; calculating a points increase or decrease value based on the basic points, the environmental weight, the job type weight, and the historical performance weight to obtain points update information; receiving a points redemption instruction, which includes redeeming goods, to obtain points redemption information; generating a points record ledger based on the points update information and the points redemption information, and analyzing the safety risk index of the worker based on the points record ledger. It can ensure the accuracy and traceability of points records, and through the introduction of multiple factors such as environmental weight and job type weight, the points management is made more comprehensive and objective.

Description

Technical Field

[0001] The present invention relates to the technical field of manufacturing management, and in particular, to a traceable whole-process management method and system for the points of industrial workers. Background Art

[0002] In modern industrial production, the management and incentive mechanisms for industrial workers are of great significance for improving production efficiency and ensuring production safety. Traditional worker management methods often rely on manual records and evaluations, which are not only inefficient but also easily affected by human factors, resulting in management results that are not objective and fair enough. With the continuous development of information technology, digital and intelligent management methods have gradually become the industry trend.

[0003] However, in the field of points management for industrial workers, there are still some problems. On the one hand, the points records of workers often lack unified standards and processes, resulting in inaccurate and opaque points calculations. Factors such as workers' work performance, job types, and working environments are not fully considered, making it difficult for points management to comprehensively reflect the actual contributions and risk levels of workers. On the other hand, there is also a lack of effective supervision and records in the points redemption link, and problems such as points abuse and unfair redemption are likely to occur, affecting the enthusiasm of workers and the management efficiency of enterprises.

[0004] Therefore, it is necessary to provide a traceable whole-process management method and system for the points of industrial workers to solve the above problems. Summary of the Invention

[0005] Aiming at the deficiencies of the existing technology, the purpose of the present invention is to provide a traceable whole-process management method and system for the points of industrial workers to solve the problems in the above background art.

[0006] The present invention is implemented as follows. A traceable whole-process management method for the points of industrial workers, the method includes the following steps:

[0007] Receive a points update instruction, where the points update instruction includes a worker number, a points increase or decrease type, and a verification picture;

[0008] Determine the basic points according to the points increase or decrease type, identify the environmental information from the verification picture, retrieve the environmental weight, and determine the job weight and historical performance weight based on the worker number;

[0009] Calculate the points increase or decrease value based on the basic points, environmental weight, job weight, and historical performance weight to obtain the points update information;

[0010] Receive a points redemption instruction, where the points redemption instruction includes the redeemed goods, to obtain the points redemption information;

[0011] Generate an integral record ledger based on integral update information and integral redemption information, and analyze the safety risk index of workers based on the integral record ledger.

[0012] Another object of the present invention is to provide a traceable integral whole-process management system for industrial workers, and the system includes:

[0013] A worker integral update module, configured to receive an integral update instruction, where the integral update instruction includes a worker number, an integral increase or decrease type, and a verification picture;

[0014] A relevant weight determination module, configured to determine a basic integral according to the integral increase or decrease type, identify environmental information from the verification picture, retrieve an environmental weight, and determine an occupation weight and a historical performance weight based on the worker number;

[0015] An integral update information module, configured to calculate an integral increase or decrease value based on the basic integral, environmental weight, occupation weight, and historical performance weight to obtain integral update information;

[0016] An integral redemption information module, configured to receive an integral redemption instruction, where the integral redemption instruction includes redeemed goods, to obtain integral redemption information;

[0017] A worker safety risk module, configured to generate an integral record ledger based on integral update information and integral redemption information, and analyze the safety risk index of workers based on the integral record ledger.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] By receiving integral update instructions and integral redemption instructions, the present invention realizes the standardization and automation of integral management. Information such as worker numbers, integral increase or decrease types, and verification pictures in the instructions ensures the accuracy and traceability of integral records. In the integral calculation process, not only the basic integral is considered, but also multiple factors such as environmental weight, occupation weight, and historical performance weight are introduced, making integral management more comprehensive and objective, and being able to more accurately reflect the actual contributions and risk levels of workers. By digitally recording and analyzing integral information to generate an integral record ledger, the management efficiency is greatly improved. At the same time, analyzing the safety risk index of workers based on the integral record ledger provides strong support for the enterprise's safety management. Through the integral redemption mechanism, workers can redeem goods with their integral, and this positive incentive method helps to improve the work enthusiasm and sense of belonging of workers. Description of the Drawings

[0020] Figure 1 It is a flowchart of a traceable integral whole-process management method for industrial workers.

[0021] Figure 2It is a flowchart for determining relevant weights in a traceable full-process management method for industrial worker points.

[0022] Figure 3 It is a flowchart for calculating the safety risk index in a traceable full-process management method for industrial worker points.

[0023] Figure 4 It is a flowchart for generating safety training information and work adjustment plans in a traceable full-process management method for industrial worker points.

[0024] Figure 5 It is a schematic structural diagram of a traceable full-process management system for industrial worker points. Specific implementation mode

[0025] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0026] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0027] As Figure 1 shown, an embodiment of the present invention provides a traceable full-process management method for industrial worker points, and the method includes the following steps:

[0028] S100, receiving a points update instruction, where the points update instruction includes a worker number, a points increase or decrease type, and a verification picture;

[0029] S200, determining basic points according to the points increase or decrease type, identifying environmental information from the verification picture, retrieving environmental weights, and determining job weights and historical performance weights based on the worker number;

[0030] S300, calculating the points increase or decrease value based on the basic points, environmental weights, job weights, and historical performance weights to obtain points update information;

[0031] S400, receiving a points redemption instruction, where the points redemption instruction includes redeemed goods, to obtain points redemption information;

[0032] S500, generating a points record ledger based on the points update information and the points redemption information, and analyzing the safety risk index of the worker based on the points record ledger.

[0033] It should be noted that in the field of integral management of industrial workers, the integral records of workers often lack unified standards and processes, resulting in inaccurate and opaque integral calculations. Factors such as workers' work performance, job types, and working environments are not fully considered, making it difficult for integral management to comprehensively reflect the actual contributions and risk levels of workers. In addition, the integral redemption link also lacks effective supervision and records, prone to problems such as integral abuse and unfair redemption, which affect the enthusiasm of workers and the management efficiency of enterprises. The embodiments of the present invention aim to solve the above problems.

[0034] In the embodiments of the present invention, an integral management platform will be constructed. By inputting an integral update instruction into the integral management platform, the integral update instruction includes the worker number, integral increase or decrease type, and verification picture. Each worker has a unique number, and the integral increase or decrease type is set. Each integral increase or decrease type corresponds to a basic integral. Behaviors such as participating in safety training, discovering safety hazards, and complying with safety operation procedures can increase the integral, while behaviors such as violating safety regulations and having safety accidents will deduct the integral. The verification picture refers to a picture proving the corresponding behavior, ensuring the accuracy and traceability of integral records.

[0035] After inputting the integral update instruction, the embodiments of the present invention will determine the basic integral according to the integral increase or decrease type, identify the environmental information from the verification picture, and retrieve the environmental weight. The environmental weight is a weight set according to the risk level of the working environment (the weight of a high-risk environment > the weight of a low-risk environment). Based on the worker number, the job type weight and historical performance weight are determined. The job type weight is a weight set according to the risk level of the job type (the weight of a high-risk job type > the weight of a low-risk job type), and the historical performance weight is a weight set according to the historical safety performance of the worker. This comprehensive consideration method makes integral management more comprehensive and objective, and can more accurately reflect the actual contributions and risk levels of workers. Then, the integral increase or decrease value can be calculated based on the basic integral, environmental weight, job type weight, and historical performance weight. The integral increase or decrease value = basic integral × job type weight × environmental weight × historical performance weight, and then the integral update information can be obtained.

[0036] When a worker needs to redeem an item, an integral redemption instruction is input. The integral redemption instruction includes the redeemed commodity, and the integral redemption information is obtained. Based on the integral update information and the integral redemption information, an integral record ledger is generated. This positive incentive method helps to improve the work enthusiasm and sense of belonging of workers. At the same time, the transparent integral record and redemption process also help to enhance the trust and satisfaction of workers. Finally, the safety risk index of workers will be analyzed according to the integral record ledger, providing strong support for the safety management of enterprises.

[0037] Such as Figure 2As shown, as a preferred embodiment of the present invention, the steps of identifying the verification picture to determine the environmental information, retrieving the environmental weight, and determining the job weight and historical performance weight based on the worker number specifically include:

[0038] S201, extract the background environment of the verification picture, and input the background environment into the environmental type library for feature matching;

[0039] S202, output the matched environmental information, and determine the environmental weight according to the environmental information;

[0040] S203, determine the job information and performance information of the most recently set time period according to the worker number, determine the job weight according to the job information, and determine the historical performance weight according to the performance information.

[0041] In the embodiment of the present invention, the background environment of the verification picture will be automatically extracted and input into the environmental type library for feature matching. The environmental type library is established in advance and contains several environmental types. Each environmental type corresponds to environmental features and an environmental weight. After feature matching, the matched environmental information will be output, and the corresponding environmental weight will be determined. In addition, the job information and performance information of the most recently set time period will be automatically determined according to the worker number, the job weight will be determined according to the job information, and the historical performance weight will be determined according to the performance information. For example, the most recently set time period is the most recent month.

[0042] As a preferred embodiment of the present invention, the steps of extracting the background environment of the verification picture and inputting the background environment into the environmental type library for feature matching specifically include:

[0043] Process the verification picture using a multi-scale multi-modal visual feature fusion method to obtain basic visual characteristics containing multi-scale information and multi-modal visual information;

[0044] Perform semantic segmentation on the verification picture based on deep learning semantic analysis to obtain a semantic segmentation map containing high-level semantic information;

[0045] Encode the semantic segmentation map into a semantic feature vector, and assign corresponding first fusion weights to the basic visual characteristics and the semantic feature vector based on the complexity of the current environment and the historical matching accuracy, and perform weighted fusion to obtain a multi-dimensional environmental feature set;

[0046] Establish an environmental type library, which contains information on multiple environmental types. Each environmental type corresponds to a set of feature descriptions and an environmental weight value;

[0047] Taking the environmental type library as a standard template, the feature sets of each environmental type in the environmental type library are matched with the multi-dimensional environmental feature set. During the matching process, matching weights are assigned to each type of feature according to the historical performance accuracy of each type of feature, and weighted calculations are performed to obtain the weighted similarity scores corresponding to each environmental type.

[0048] Compare the weighted similarity scores of all environmental types, and select the environmental type with the highest score as the best environmental matching result corresponding to the verification picture.

[0049] In the embodiments of the present invention, traditional image features (such as color, texture, edges) and deep learning-based semantic features (such as environmental semantic segmentation) are combined to make up for the limitations of a single feature mode. Traditional features ensure the capture of low-level visual information, while deep semantic features enhance the understanding of high-level semantics of the environment. The combination of the two makes the expression of environmental background information more comprehensive, significantly improving the accuracy and robustness of recognition.

[0050] Moreover, for different feature modalities, the weights are dynamically adjusted according to the historical matching accuracy and environmental complexity to ensure that the features with better performance are preferentially utilized in different application scenarios. This dynamic adjustment mechanism enables the system to adapt to various complex environments, reduces misjudgments caused by the failure of a single feature, and enhances the overall stability and reliability. The pre-constructed environmental type library contains various typical environmental categories and their feature descriptions, forming a standardized comparison template. Each environmental type in the library corresponds to a feature set and a risk weight, providing a unified basis for subsequent integral calculations. This library ensures the standardization and consistency of the matching process, facilitating data comparison and analysis across time and locations. Finally, the weighted similarity scoring comprehensively considers the matching situations of various feature modalities, taking into account the actual roles of different features in environmental recognition and avoiding the excessive influence of any single feature on the result. This flexible matching strategy improves the fault tolerance of environmental recognition, enabling accurate overall recognition even when some features are interfered with.

[0051] As a preferred embodiment of the present invention, the verification picture is processed using a multi-scale multi-modal visual feature fusion method to obtain the basic visual characteristics including multi-scale information and multi-modal visual information. The specific steps are as follows:

[0052] Convert the input verification picture into multiple color space channels, and take each color channel as an independent input image.

[0053] Using color histogram statistics, set the number of bins of the color histogram according to the color information complexity of the input image, and perform normalization to obtain the normalized color histogram vector corresponding to each color channel.

[0054] Set the number of scale levels, and successively perform Gaussian blur and smooth sampling on the verification images according to the number of scale levels to gradually reduce the image resolution, so as to form a multi-layer image pyramid and obtain an image set of multiple scale levels;

[0055] Apply a multi-directional texture filter to each scale-level image to obtain a filtered response map with multiple scales and multiple directions;

[0056] Apply a non-linear activation function to the filtered response map with multiple scales and multiple directions to obtain an activated filtered response map;

[0057] Divide the activated filtered response map into several local grid units, calculate the local statistical features within each local grid unit, and obtain the local statistical feature vectors corresponding to each scale and direction;

[0058] Successively splice the local statistical feature vectors corresponding to all scales and directions to form a complete set of texture feature vectors;

[0059] Apply a gradient operator to each scale-level image to calculate the gradient components in the horizontal and vertical directions, and obtain the gradient magnitude map and gradient direction map corresponding to each scale level;

[0060] Divide the gradient directions into a predetermined number of angular intervals, and statistically accumulate the gradient magnitudes in each angular interval region in the gradient magnitude map and gradient direction map of each scale-level image to obtain the direction gradient histogram feature vector corresponding to each scale level;

[0061] Calculate the sum of the overall gradient intensities of each scale level according to the gradient magnitude map corresponding to each scale level, and obtain the weight value of each scale level after normalization;

[0062] Weight the direction gradient histogram feature vector of each scale level with the weight value of each scale level, and splice the weighted results to obtain a comprehensive multi-scale edge feature;

[0063] Divide the gradient magnitude map and gradient direction map of the current scale level into several local grid units, and within each local grid unit, statistically accumulate the gradient magnitudes in different gradient directions to obtain the direction gradient histogram of all local units in the current scale level;

[0064] Weight the direction gradient histogram of all local units in the current scale level with the total edge intensity index weight of the current scale level to obtain the weighted edge feature vector of the current scale level;

[0065] Splice the weighted edge feature vectors of different scale levels to form an overall edge feature vector containing multi-scale information, and output a complete multi-scale weighted edge feature vector.

[0066] Calculate the variance of the pixel grayscale values within each window of a fixed size sliding window on the verification image to form a local variance matrix;

[0067] Statistically analyze the probability distribution of the grayscale levels of the verification image to obtain the overall entropy value;

[0068] Normalize the local variance matrix and fuse it with the overall entropy value to obtain a local importance weight matrix;

[0069] Assign initial second fusion weights to the normalized color histogram vectors corresponding to each color channel, the complete set of texture feature vectors, and the comprehensive multi-scale edge features, and perform weighted fusion to obtain the current fused multi-modal visual feature vector;

[0070] Adopt a self-supervised loss feedback mechanism to iteratively adjust the second fusion weights to obtain the final fusion weight vector and fusion feature vector.

[0071] In the embodiments of the present invention, three types of visual features, namely color, texture, and edge, are combined, and each type of feature is processed through multi-scale, multi-directional, and statistical encoding, enabling the detailed capture of rich information in environmental images. In particular, the multi-space conversion and adaptive binning strategy of the color channel effectively improve the expression ability of color details and adapt to diverse and complex environmental color changes. Through spatial pyramid hierarchical encoding and combination with local weight weighting, both global and local features can be taken into account, retaining the spatial structure information of the environmental image, thereby enhancing the discriminative power of the features and the accuracy of environmental matching. The fusion process adopts a self-supervised iterative mechanism to dynamically adjust the weights of each feature and outputs more robust comprehensive visual features. Finally, by combining the spatial pyramid structure and local weights, space-sensitive feature encoding is achieved, ensuring that the features contain both global information and retain key local details.

[0072] As a preferred embodiment of the present invention, the steps of performing semantic segmentation on the verification image based on deep learning semantic analysis to obtain a semantic segmentation map containing high-level semantic information specifically include:

[0073] Extract features from the images of each scale layer using a deep convolutional neural network to obtain the corresponding multi-dimensional feature representations of each scale layer;

[0074] Calculate the amount of information of the multi-dimensional feature representations corresponding to each scale layer, and assign corresponding information weights according to the amount of information;

[0075] Perform weighted summation on the multi-dimensional feature representations corresponding to all scale layers according to the corresponding information weights to obtain the fused multi-scale features;

[0076] Apply the spatial attention mechanism and the channel attention mechanism to the fused multi-scale features respectively to obtain the spatial attention matrix and the channel attention matrix;

[0077] Fuse the spatial attention matrix and the channel attention matrix by element-wise multiplication, and weight the fused multi-scale features with the fusion result to obtain the final weighted feature map;

[0078] Feed the final weighted feature map into the deep convolutional decoder network to convert the high-dimensional feature map into a semantic prediction map matching the size of the verification image, and obtain a preliminary semantic segmentation probability map;

[0079] Based on the preliminary semantic segmentation probability map, use self-supervised auxiliary optimization to learnable parameters. After the optimization is completed, output the optimized semantic segmentation probability map;

[0080] Calculate the gradient information of the optimized semantic segmentation probability map to obtain a gradient map;

[0081] Input the gradient map into the pre-trained boundary refinement network for boundary enhancement to obtain a boundary enhanced map;

[0082] Weight the boundary enhanced map according to the scale factor and superimpose it on the optimized semantic segmentation probability map to strengthen the class confidence of the boundary region, and obtain a semantic segmentation probability map after boundary refinement and fusion.

[0083] In the embodiment of the present invention, by performing multi-scale processing on the verification image and dynamically adjusting the weights of features at each scale, the solution can capture multi-level information from coarse to fine, from macro to micro. This multi-scale fusion effectively compensates for the deficiencies of single-scale information, enabling the model to more comprehensively understand the image content and improve the accuracy and robustness of semantic segmentation. And the dual attention mechanism weights and adjusts the features in the spatial position and channel dimensions, effectively highlighting the semantic key regions and important feature channels and suppressing irrelevant or interfering information. This greatly improves the model's perception ability of target semantics and enhances the ability to distinguish subtle semantic differences in complex environments.

[0084] Moreover, self-supervised auxiliary training and boundary refinement are introduced. The self-supervised mechanism uses data augmentation without relying on additional annotations, and guides the model to learn transformation-invariant semantic features through consistency constraints, enhancing the model's adaptability in unseen environments and complex scenarios. And by processing the gradient information of the semantic probability map and boundary enhancement, the clarity and accuracy of the segmentation boundary are significantly improved, reducing the situation of blurred and confused class boundaries. This refinement process ensures the refinement of the segmentation result, facilitating the accurate extraction and matching of subsequent environmental features.

[0085] As Figure 3 shown, as a preferred embodiment of the present invention, the step of analyzing the safety risk index of workers based on the integral record ledger specifically includes:

[0086] S501. Calculate the safety risk index RI of the worker, where RI = ;

[0087] S502. When the safety risk index RI is greater than the risk threshold, retrieve the corresponding type of point deduction;

[0088] S503. Analyze the type of point deduction to generate safety training information and a work adjustment plan.

[0089] In the embodiment of the present invention, the safety risk index can reflect whether a worker has a certain risk. The calculation formula of the safety risk index RI is: RI = , represents the deduction points for the i-th point deduction, represents the bonus points for the j-th point increase. TD is the time decay factor, TD = , λ is the decay coefficient, t is the time from the point deduction or bonus points to the present, and n and m represent the total number of point deductions and the total number of point increases within the recently set time period. Then, a determination will be made on RI. When the safety risk index RI is greater than the risk threshold, it indicates that there is a certain risk. The corresponding type of point deduction will be retrieved, and the type of point deduction will be automatically analyzed to generate safety training information and a work adjustment plan, so as to provide safety training for the worker in a timely manner and transfer the worker to a job with a low safety index.

[0090] As a preferred embodiment of the present invention, the step of analyzing the type of point deduction to generate safety training information and a work adjustment plan specifically includes:

[0091] Sort all types of point deductions according to the amount of deduction points, select the top N types of point deductions, and obtain several sets of point deduction types;

[0092] Given a safety database for the type of work, according to the work type information, retrieve a set of safety accident types related to the corresponding work type information from the safety database for the type of work to obtain a list of safety accident types corresponding to the work type;

[0093] Match and judge the set of point deduction types with the list of safety accident types corresponding to the work type to obtain the matching result of the point deduction type and the safety accident type;

[0094] Retrieve the safety training plan and key training content for the matching result from the safety training content library to obtain a set of safety training information for the worker;

[0095] Extract the unqualified safety items related to the list of safety accident types corresponding to the work type from the safety database for the type of work to obtain a list of unqualified safety items;

[0096] Screen out the set of job types that meet the standards from the job safety library, and obtain the set of qualified job types;

[0097] Based on the list of qualified job types, combined with the skills, work experience of the current workers and the employment needs of the enterprise, generate a specific work adjustment plan.

[0098] In the embodiment of the present invention, by selecting the top-ranked types of score reduction, focusing on the main potential safety hazards, avoiding data redundancy and noise interference, the pertinence and efficiency of the analysis are ensured. Matching with the safety accident types related to the job types makes the safety training and work adjustment more targeted and effective, and reduces the interference of irrelevant content. And according to the job type information and deduction types of different workers, customize the safety training content and work adjustment plan, avoiding the "one-size-fits-all" management. The personalized plan is more in line with the actual risk situation of the workers, helps to improve the training effect and job suitability, and thus effectively reduces the risk of safety accidents.

[0099] As Figure 4 shown, as a preferred embodiment of the present invention, the steps of analyzing the types of score reduction to generate safety training information and work adjustment plans specifically include:

[0100] S5031, determine the top N types of score reduction according to the deduction scores, and retrieve the safety accident types related to the job types of the corresponding workers;

[0101] S5032, match the types of score reduction and safety accident types to determine safety training information and unqualified safety items;

[0102] S5033, input the unqualified safety items into the job safety library, determine the qualified job types, and generate a work adjustment plan.

[0103] In the embodiment of the present invention, it will also determine the top N (for example, the top three) types of score reduction according to the deduction scores, retrieve the safety accident types related to the job types of the corresponding workers, and each job type has corresponding strongly related safety accident types. Match the types of score reduction and safety accident types. When a certain type is successfully matched, targeted training needs to be carried out for this type, and the unqualified safety items are determined according to the corresponding safety accident types. Finally, the unqualified safety items will be input into the job safety library, the qualified job types will be determined, and a work adjustment plan will be generated according to the qualified job types.

[0104] As a preferred embodiment of the present invention, the method further includes: determining the total points of each worker based on the point record ledger, determining the incentive level based on the total points and constructing a red and black list of workers, and displaying the red and black list of workers. The red list displays workers with higher points, and the black list displays workers with lower points or potential safety hazards, so as to encourage workers to actively participate in safety management. A safety trend chart of each worker will also be drawn based on the point record ledger, and the analysis results will be presented to the management in an intuitive way to help them better understand the safety status of workers and take corresponding measures. Warning information will be generated based on the safety trend chart. For example, when the safety risk index of a certain worker continues to rise, warning information will be generated.

[0105] As a preferred embodiment of the present invention, the method further includes: receiving the point redemption demand information input by the worker, where the point redemption demand information includes the item name and the redemption points; summarizing and counting all the point redemption demand information to generate worker demand information, where the worker demand information contains several item names, and each item name corresponds to a required point. In this way, workers can redeem more desirable items, and the humanization is better.

[0106] As Figure 5 shown, the embodiment of the present invention also provides a traceable full-process management system for industrial worker points. The system includes:

[0107] A worker point update module 100, configured to receive a point update instruction, where the point update instruction includes a worker number, a point increase or decrease type, and a verification picture;

[0108] A relevant weight determination module 200, configured to determine the basic points according to the point increase or decrease type, identify the environmental information from the verification picture, retrieve the environmental weight, and determine the job weight and historical performance weight based on the worker number;

[0109] A point update information module 300, configured to calculate the point increase or decrease value based on the basic points, environmental weight, job weight, and historical performance weight to obtain point update information;

[0110] A point redemption information module 400, configured to receive a point redemption instruction, where the point redemption instruction includes the redeemed commodity to obtain point redemption information;

[0111] A worker safety risk module 500, configured to generate a point record ledger based on the point update information and point redemption information, and analyze the safety risk index of the worker based on the point record ledger.

[0112] As a preferred embodiment of the present invention, the relevant weight determination module 200 includes:

[0113] An environmental feature matching unit is configured to extract the background environment of a verification picture and input the background environment into an environment type library for feature matching;

[0114] An environmental weight determining unit is configured to output the matched environmental information and determine an environmental weight according to the environmental information;

[0115] An occupational performance weight unit is configured to determine occupational information and performance information for a recently set time period according to a worker number, determine an occupational weight according to the occupational information, and determine a historical performance weight according to the performance information.

[0116] As a preferred embodiment of the present invention, the worker safety risk module 500 includes:

[0117] A safety risk index unit is configured to calculate a safety risk index RI of a worker, RI = , represents the deduction points for the i-th integral reduction, represents the bonus points for the j-th integral increase, TD is a time decay factor, TD = , λ is a decay coefficient, and t is the time from the deduction or bonus points to the present;

[0118] An integral type retrieval unit is configured to retrieve a corresponding integral reduction type when the safety risk index RI is greater than a risk threshold;

[0119] An integral type analysis unit is configured to analyze the integral reduction type and generate safety training information and a work adjustment plan.

[0120] As a preferred embodiment of the present invention, the integral type analysis unit includes:

[0121] A safety accident type subunit is configured to determine the top N integral reduction types according to the deduction points and retrieve safety accident types related to the corresponding worker's occupation;

[0122] A safety training information subunit is configured to match the integral reduction type and the safety accident type to determine safety training information and unqualified safety items;

[0123] A work adjustment plan subunit is configured to input the unqualified safety items into an occupational safety library, determine a qualified occupation, and generate a work adjustment plan.

[0124] The above only describes the preferred embodiments of the present invention in detail and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0125] It should be understood that although the steps in the flowcharts of the embodiments of the present invention are shown in sequence according to the indications of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in each embodiment may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0126] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0127] After considering the specification and the disclosure of the embodiments, those skilled in the art will readily think of other embodiments of the present disclosure. This application aims to cover any variations, uses or adaptations of the present disclosure. These variations, uses or adaptations follow the general principles of the present disclosure and include common general knowledge or conventional technical means in the technical field not disclosed in the present disclosure. The specification and the embodiments are only regarded as exemplary, and the true scope and spirit of the present disclosure are pointed out by the claims.

Claims

1. Traceable whole-process management method for industrial workers' points, characterized in that, The method includes the following steps: Receiving an integral update instruction, where the integral update instruction includes a worker number, an integral increase / decrease type, and a verification picture; Determining a basic integral according to the integral increase / decrease type, identifying environmental information for the verification picture, retrieving an environmental weight, and determining an occupation weight and a historical performance weight based on the worker number; Calculating an integral increase / decrease value based on the basic integral, the environmental weight, the occupation weight, and the historical performance weight to obtain integral update information; Receiving an integral exchange instruction, where the integral exchange instruction includes an exchanged commodity to obtain integral exchange information; Generating an integral record ledger based on the integral update information and the integral exchange information, and analyzing the safety risk index of the worker based on the integral record ledger; Among them, the step of identifying environmental information for the verification picture, retrieving an environmental weight, and determining an occupation weight and a historical performance weight based on the worker number specifically includes: Extracting the background environment of the verification picture and inputting the background environment into an environmental type library for feature matching; Outputting the matched environmental information and determining the environmental weight according to the environmental information; Determining the occupation information and the performance information in the most recent set time period according to the worker number, determining the occupation weight according to the occupation information, and determining the historical performance weight according to the performance information; The step of extracting the background environment of the verification picture and inputting the background environment into an environmental type library for feature matching specifically includes: Processing the verification picture using a multi-scale multi-modal visual feature fusion method to obtain basic visual characteristics including multi-scale information and multi-modal visual information; Performing semantic segmentation on the verification picture based on deep learning semantic analysis to obtain a semantic segmentation map including high-level semantic information; Encoding the semantic segmentation map into a semantic feature vector, and assigning corresponding first fusion weights to the basic visual characteristics and the semantic feature vector based on the complexity of the current environment and the historical matching accuracy, and performing weighted fusion to obtain a multi-dimensional environmental feature set; Establishing an environmental type library, where the environmental type library contains information on multiple environmental types, and each environmental type corresponds to a set of feature descriptions and an environmental weight value; Using the environmental type library as a standard template, matching the feature sets of each environmental type in the environmental type library with the multi-dimensional environmental feature set. During the matching process, assigning a matching weight to each type of feature based on the historical performance accuracy of each type of feature and performing weighted calculation to obtain a weighted similarity score corresponding to each environmental type; Comparing the weighted similarity scores of all environmental types and selecting the environmental type with the highest score as the best environmental matching result corresponding to the verification picture.

2. The traceable full-process management method for industrial worker points according to claim 1, wherein The step of processing the verification picture using a multi-scale multi-modal visual feature fusion method to obtain basic visual characteristics including multi-scale information and multi-modal visual information specifically includes: Converting the input verification picture into multiple color space channels, and taking each color channel as an independent input image; Using color histogram statistics, setting the number of bins of the color histogram according to the color information complexity of the input image and performing normalization to obtain a normalized color histogram vector corresponding to each color channel; Set the number of scale levels, and successively perform Gaussian blur and smooth sampling on the verification image according to the number of scale levels, gradually reducing the image resolution to form a multi-layer image pyramid, and obtaining an image set of multiple scale levels; Extract the edge features and texture features of each scale in the image set of multiple scale levels to obtain a texture feature vector and multi-scale edge features; Taking a sliding window of a fixed size as a unit on the verification image, calculate the variance of the pixel gray values in each window to form a local variance matrix; Statistically analyze the probability distribution of the gray levels of the verification image to obtain an overall entropy value; Normalize the local variance matrix and fuse it with the overall entropy value to obtain a local importance weight matrix; Assign initial second fusion weights to the normalized color histogram vectors, texture feature vectors, and multi-scale edge features corresponding to each color channel, and perform weighted fusion to obtain the currently fused multi-modal visual feature vector.

3. The traceable full-process management method for industrial worker points according to claim 2, wherein Extracting the edge features and texture features of each scale in the image set of multiple scale levels to obtain a texture feature vector and multi-scale edge features specifically includes the following steps: Apply a multi-directional texture filter to each scale-level image to obtain a multi-scale and multi-directional filtered response map; Apply a non-linear activation function to the multi-scale and multi-directional filtered response map of each scale level to obtain an activated filtered response map; Divide the activated filtered response map into several local grid units, and calculate local statistical features within each local grid unit to obtain local statistical feature vectors corresponding to each scale and direction; Successively splice the local statistical feature vectors corresponding to all scales and directions to form a complete texture feature vector set; Apply a gradient operator to each scale-level image to calculate the gradient components in the horizontal and vertical directions to obtain a gradient magnitude map and a gradient direction map corresponding to each scale level; Divide the gradient direction into a predetermined number of angular intervals, and statistically analyze the cumulative gradient magnitude in each angular interval region in the gradient magnitude map and gradient direction map of each scale-level image to obtain a directional gradient histogram feature vector of each scale level; Calculate the sum of the overall gradient intensities of each scale level according to the gradient magnitude map corresponding to each scale level, and normalize it to obtain the weight value of each scale level; Weight the directional gradient histogram feature vector of each scale level with the weight value of each scale level, and splice the weighted results to obtain comprehensive multi-scale edge features; Divide the gradient magnitude map and gradient direction map of the current scale level into several local grid units, and within each local grid unit, statistically analyze the sum of the gradient magnitudes in different gradient directions to obtain the directional gradient histogram of all local units of the current scale level; Weight the directional gradient histogram of all local units of the current scale level with the total edge intensity index weight of the current scale level to obtain a weighted edge feature vector of the current scale level; Splice the weighted edge feature vectors of different scale levels to form an overall edge feature vector containing multi-scale information, and output a complete multi-scale weighted edge feature vector.

4. The traceable full-process management method for industrial worker points according to claim 3, characterized in that, The steps of performing semantic segmentation on the verification image based on deep learning semantic analysis to obtain a semantic segmentation map containing high-level semantic information specifically include: Extract features from the images of each scale layer using a deep convolutional neural network to obtain the corresponding multi-dimensional feature representations for each scale layer; Calculate the information content of the multi-dimensional feature representations corresponding to each scale layer, and allocate corresponding information weights according to the information content; Perform weighted summation on the multi-dimensional feature representations corresponding to all scale layers according to the corresponding information weights to obtain the fused multi-scale features; Apply the spatial attention mechanism and the channel attention mechanism to the fused multi-scale features respectively to obtain the spatial attention matrix and the channel attention matrix; Fuse the spatial attention matrix and the channel attention matrix in an element-wise multiplication manner, and weight the fused multi-scale features with the fusion result to obtain the final weighted feature map; Feed the final weighted feature map into the deep convolutional decoder network to convert the high-dimensional feature map into a semantic prediction map matching the size of the verification image, and obtain the preliminary semantic segmentation probability map; Based on the preliminary semantic segmentation probability map, use self-supervised assistance to optimize the learnable parameters. After optimization, output the optimized semantic segmentation probability map; Calculate the gradient information of the optimized semantic segmentation probability map to obtain the gradient map; Input the gradient map into the pre-trained boundary refinement network for boundary enhancement to obtain the boundary enhancement map; Weight the boundary enhancement map according to the proportionality coefficient and superimpose it on the optimized semantic segmentation probability map to strengthen the class confidence of the boundary region, and obtain the semantic segmentation probability map after boundary refinement fusion.

5. The traceable full-process management method for industrial worker points according to claim 4, characterized in that The steps of analyzing the safety risk index of workers based on the integral record ledger specifically include: Calculate the safety risk index RI of workers, RI = , represents the deduction points for the i-th integral reduction, represents the bonus points for the j-th integral increase, TD is the time decay factor, TD = , λ is the decay coefficient, and t is the time elapsed since the deduction or bonus points were given; When the safety risk index RI is greater than the risk threshold, retrieve the corresponding integral reduction type; Analyze the integral reduction type to generate safety training information and a work adjustment plan.

6. The traceable full-process management method for industrial worker points according to claim 5, wherein The steps of analyzing the safety risk index of workers based on the integral record ledger specifically include: The steps of analyzing the integral reduction type to generate safety training information and a work adjustment plan specifically include: Sort all integral reduction types according to the amount of deduction points, select the top N integral reduction types, and obtain several integral reduction type sets; Given a job safety database, retrieve the set of safety accident types related to the corresponding job information from the job safety database according to the job information to obtain the list of safety accident types corresponding to the job; Match and judge the integral reduction type set with the list of safety accident types corresponding to the job to obtain the matching result of the integral reduction type and the safety accident type; Retrieve the safety training plan and key training content for the matching result from the safety training content library to obtain the set of safety training information for this worker; Extract the unqualified safety items related to the list of safety accident types corresponding to the job from the job safety library to obtain the list of unqualified safety items; Screen out the set of jobs that meet the standards on the list of unqualified safety items from the job safety library to obtain the set of jobs that meet the standards; Based on the list of jobs that meet the standards, combine the skills, working years and enterprise employment needs of the current worker to generate a specific work adjustment plan.

7. The traceable full-process management method for industrial worker points according to claim 6, characterized in that, Steps for analyzing the integral reduction type and generating safety training information and work adjustment plan specifically include: Determine the top N integral reduction types according to the deduction scores, and retrieve the types of safety accidents related to the corresponding worker's job type; Match the integral reduction types and the types of safety accidents to determine safety training information and unqualified safety items; Input the unqualified safety items into the job safety database, determine the qualified job types, and generate a work adjustment plan.

8. The traceable full-process management method for industrial worker points according to claim 7, characterized in that, The method further includes: Determine the total integral of each worker based on the integral record ledger, determine the incentive level based on the total integral, construct a worker red-black list, and display the worker red-black list; Draw a safety trend chart for each worker based on the integral record ledger, and generate warning information based on the safety trend chart; Receive the integral exchange demand information input by the worker, where the integral exchange demand information includes the item name and the exchange integral; Summarize and statistically analyze all the integral exchange demand information to generate worker demand information, where the worker demand information contains several item names, and each item name corresponds to a required integral.

9. Traceable industrial worker integral whole-process management system, the system is applied to the traceable industrial worker integral whole-process management method described in any one of claims 1 to 8, characterized in that, The system includes: A worker integral update module for receiving an integral update instruction, where the integral update instruction includes a worker number, an integral increase or decrease type, and a verification picture; A relevant weight determination module for determining the basic integral according to the integral increase or decrease type, identifying the environmental information from the verification picture, retrieving the environmental weight, and determining the job type weight and historical performance weight based on the worker number; An integral update information module for calculating the integral increase or decrease value based on the basic integral, environmental weight, job type weight, and historical performance weight to obtain integral update information; An integral exchange information module for receiving an integral exchange instruction, where the integral exchange instruction includes the exchanged commodity to obtain integral exchange information; A worker safety risk module for generating an integral record ledger based on the integral update information and integral exchange information, and analyzing the safety risk index of the worker based on the integral record ledger.

Images