Construction worker integral self-service exchange and online learning linkage method based on face recognition

Abstract

The invention provides a construction worker integral self-service exchange and online learning linkage method and system based on face recognition, and the method comprises the steps: deploying face recognition equipment at a construction site entrance, a key operation region and a self-service exchange terminal, and carrying out the face recognition of a construction worker through an online mode and / or an offline mode, generating a face scanning event signal after determining the identity of the current worker; the system linkage module receives the face scanning event signal and judges an event type according to a preset event-rule mapping table; for the behavior event meeting the integral obtaining condition, an integral obtaining module is linked, and integral calculation and distribution are carried out according to the current effective integral rule base; after the worker confirms the exchange, point deduction is completed, the terminal is driven to deliver goods or a receiving voucher is generated, and an incentive closed loop is completed. According to the method, a dynamic feedback closed-loop control system is constructed, and the technical defects that an existing system is rigid in rule and lagged in excitation, and weak links of building worker groups cannot be improved in a targeted mode are overcome.

Description

Technical Field

[0001] This invention relates to the field of construction engineering management technology, and more specifically, to a method and system for linking construction workers' points self-service redemption and online learning based on facial recognition. Background Technology

[0002] Currently, with the deepening of smart construction site construction, using information technology and intelligent methods to manage and incentivize construction workers has become an industry trend. Among them, point-based management, as a common behavioral incentive model, is widely used in worker safety behavior standards, skills training, and daily attendance. Traditional point systems typically use static rules, such as fixed point values ​​to record events like attendance and training completion. Workers can redeem accumulated points for goods or small amounts of cash.

[0003] However, existing technical solutions have significant drawbacks, making it difficult to construct an efficient, adaptive, and sustainable incentive loop. First, the rigidity of the point system is a core issue. Most systems use preset, fixed point distribution standards, failing to dynamically adjust based on the management priorities at different project stages (e.g., different risks during foundation construction versus high-altitude work), seasonal characteristics (e.g., anti-slip and anti-electric shock training during the rainy season), or unexpected safety rectification needs. This leads to a disconnect between incentive measures and management objectives, failing to accurately guide worker behavior.

[0004] Secondly, incentives are delayed and lack specificity. Traditional systems typically passively record points after a behavior occurs, resulting in a long incentive feedback cycle. More importantly, the system lacks the ability to deeply analyze group behavior data and cannot automatically identify common knowledge gaps or high-risk behavior patterns among workers (e.g., a persistently low pass rate for certain safety operating procedures exams). Consequently, it cannot proactively adjust incentive resources (such as temporarily increasing point rewards for relevant training) for targeted reinforcement.

[0005] Furthermore, existing solutions often focus on short-term behavioral triggers while neglecting the formation of long-term behavioral habits. Points redemption is often treated as a simple exchange of goods, failing to connect with workers' learning and growth curves and intrinsic motivation. From a behavioral science and neurobiological perspective, effective incentives should simulate a positive feedback loop of "reward-behavior reinforcement," and be personalized based on individual differences in the effectiveness of incentives (similar to the "sensitivity" of the nervous system) to avoid incentive fatigue or resource waste.

[0006] Furthermore, data silos often exist between identity verification and various subsystems (attendance, learning, and redemption), resulting in poor linkage. Although there have been attempts to use card swiping, fingerprint, or facial recognition for attendance, these identity verification events have not formed a real-time, intelligent closed-loop linkage with points calculation, dynamic rule adjustment, and personalized incentive push.

[0007] Therefore, the industry urgently needs a closed-loop linkage system and method that can deeply integrate identity authentication, behavior collection, intelligent analysis and dynamic incentives, and optimize rules in real time based on data feedback, so as to accurately improve the safety literacy and skill level of construction workers. Summary of the Invention

[0008] In view of this, in order to solve the problems mentioned in the background technology, a method and system for self-service redemption of construction workers' points and online learning based on facial recognition is proposed.

[0009] The objective of this invention can be achieved through the following technical solutions:

[0010] This invention provides a method for linking self-service points redemption and online learning for construction workers based on facial recognition, including:

[0011] Facial recognition devices are deployed at construction site entrances, key work areas, and self-service redemption terminals to perform facial recognition on construction workers in online and / or offline modes. After confirming the current worker's identity, a facial recognition event signal is generated. The facial recognition event signal includes at least attendance entry events and redemption verification events, and the event data is uploaded in real time.

[0012] The system linkage module receives the face recognition event signal and determines the event type according to the preset event-rule mapping table. For behavior events that meet the conditions for obtaining points, the linkage module calculates and distributes points according to the currently effective points rule library. At the same time, the learning behavior data of all workers on the online learning platform is obtained synchronously as the basis for distributing points. The learning behavior data includes the course completion status of all workers, the exam pass rate of all workers and their corresponding exam categories.

[0013] The system analyzes the behavioral and learning data of all workers in real time to identify group weaknesses; based on the identification results, it dynamically adjusts the intensity of point rewards for specific behaviors in the point rule base, and pushes the adjusted incentive strategy to the target worker group through the message push module associated with the facial recognition device.

[0014] When a redemption request is initiated at a self-service redemption terminal or on a mobile device, facial recognition is used for secondary identity verification. After successful verification, the system's linked modules query real-time points and display a list of redeemable goods. After the worker confirms the redemption, the points are deducted, and the terminal is driven to issue the goods or generate a collection voucher, thus completing the incentive loop.

[0015] Preferably, as one possible implementation, the dynamic adjustment of the point reward intensity for specific behaviors in the point rule base specifically includes:

[0016] Based on the learning data within a preset period, identify the exam categories in the current learning behavior data of all workers whose exam pass rate is below a threshold, define the category as a weak link to be strengthened, and set improving the overall pass rate of the exam in this category as the optimization goal;

[0017] The integral reward intensity of the preset learning behavior for the weak link is constructed as a decision variable to be optimized; a particle swarm is initialized, where the position vector of each particle represents a set of reward intensity values ​​set for different learning sub-behaviors.

[0018] The fitness value of each particle is determined by predicting the overall incentive efficiency improvement effect on the target worker group after the reward intensity scheme represented by its position vector is executed; the particle swarm algorithm iteratively updates the velocity and position of the particles to find the optimal reward intensity scheme that maximizes the overall incentive efficiency evaluation value.

[0019] The optimal reward intensity scheme obtained through iterative optimization is updated to the points rule base and activated for the target worker group.

[0020] Preferably, as an implementation scheme, the target worker group is obtained by cluster analysis of the workers' historical behavior data and attribute data; for the weak link, the workers are divided into different cluster groups, and the workers in each group have similar behavioral characteristics and incentive effectiveness.

[0021] Preferably, as one possible implementation, the overall incentive effectiveness evaluation value is used to quantify the improvement of overall incentive effectiveness; the overall incentive effectiveness evaluation value integrates the incentive effectiveness values ​​of workers in each cluster group, as well as the integral cost factor consumed in implementing the reward intensity scheme.

[0022] Preferably, as one possible implementation, the overall incentive effectiveness evaluation value is calculated using the following vectorized formula:

[0023] ;

[0024] in, This represents the overall incentive effectiveness assessment value; Indicates the number of cluster groups; Indicates the first Cluster grouping The size weight is the proportion of the group's members to the total target number; Indicates the first Groups The number of workers assessed within the organization; Indicates the first The first group The historical incentive efficacy coefficient of an individual worker for the intensity of a reward is calculated based on their past participation and effectiveness with similar incentives. The time decay constant; Indicates the first The first group The time interval between each worker's most recent participation in relevant learning; Indicating the scheme regarding the first The intensity of the point-based reward for each learning sub-behavior; This represents the total number of learning sub-behaviors included in the scheme; and This is the parameter for adjusting integral cost.

[0025] Preferably, as an implementation scheme, a long-term incentive correction factor is further introduced when calculating the particle fitness during the particle swarm optimization process.

[0026] The calculation method for the long-term incentive correction factor is as follows: for the weak link targeted by the current reward intensity scheme, extract the relevant behavioral sequence of the target worker group in the past period, and then query the relationship mapping table;

[0027] Analyze the continuous learning duration of positive behaviors in the behavior sequence; the correction factor is positively correlated with the continuous learning duration presented in the behavior sequence;

[0028] During iterative optimization, the final fitness value of a particle is the product of its original predicted overall incentive performance evaluation value and the long-term incentive correction factor.

[0029] Preferably, as one possible implementation, the overall incentive effectiveness evaluation value is calculated using the following vectorized formula:

[0030] ;

[0031] in, This represents the overall motivational efficacy assessment value after incorporating the neuroplasticity model;

[0032] in, Indicates the number of cluster groups; Indicates the first Cluster grouping The size weight is the proportion of the group's members to the total target number; Indicates the first Groups The number of workers assessed within the organization; Indicates the first The first group The historical incentive efficacy coefficient of an individual worker for the intensity of a reward is calculated based on their past participation and effectiveness with similar incentives. The time decay constant; Indicates the first The first group The time interval between each worker's most recent participation in relevant learning; Indicating the scheme regarding the first The intensity of the point-based reward for each learning sub-behavior; This represents the total number of learning sub-behaviors included in the scheme; and This is the parameter for adjusting the integral cost;

[0033] in, Indicates the first The optimization cycle, the first The first group Individual worker's personal reward sensitivity coefficient.

[0034] Preferably, as one feasible implementation; the individual reward sensitivity coefficient The update mechanism is used to further understand the internal driving energy regulation of the work, specifically:

[0035] Implementing a set of reward intensity schemes During the subsequent evaluation cycle, record the workers' information. Relevant target behavior data, including: behavior trigger delay Quality of performance and the frequency of spontaneous repetition of the behavior without immediate reward. ;

[0036] Based on the behavioral feedback, an internal driving energy adjustment factor is calculated. : ;

[0037] in, To adjust the gain constant; The average incentive intensity applied to this worker in the current period; It is the hyperbolic tangent function;

[0038] Update the reward sensitivity coefficient based on the adjustment factor: .

[0039] Preferably, as one possible implementation, the personalized recommendation method includes the following steps:

[0040] Plotting the worker's historical reward intensities Its corresponding immediate behavioral incentive effectiveness Scatter plot;

[0041] Using nonlinear regression to fit and Relationship curve;

[0042] The reward intensity range corresponding to the peak of the curve is recommended as the optimal performance incentive range for the worker in the current cycle.

[0043] Preferably, as a feasible implementation, in the redemption process, the system uses the individual reward sensitivity coefficient of the redeeming worker. The display logic of the list of redeemable goods will be dynamically adjusted; for workers with a continuously rising sensitivity coefficient, high-value goods that require long-term accumulation will be displayed first when they have enough points; for workers with a low or fluctuating sensitivity coefficient, a lower points threshold will be added.

[0044] Compared with the prior art, the embodiments of this application have at least the following technical effects:

[0045] This invention provides a method for linking self-service redemption of construction workers' points with online learning based on facial recognition. The method includes: deploying facial recognition devices at construction site entrances, key work areas, and self-service redemption terminals; performing facial recognition on construction workers in online and / or offline modes; and generating a facial recognition event signal after confirming the current worker's identity. The facial recognition event signal includes at least attendance entry events and redemption verification events, and the event data is uploaded in real time.

[0046] The system linkage module receives the face recognition event signal and determines the event type according to the preset event-rule mapping table. For behavior events that meet the conditions for obtaining points, the linkage module calculates and distributes points according to the currently effective points rule library. At the same time, the learning behavior data of all workers on the online learning platform is obtained synchronously as the basis for distributing points. The learning behavior data includes the course completion status of all workers, the exam pass rate of all workers and their corresponding exam categories.

[0047] The system analyzes the behavioral and learning data of all workers in real time to identify group weaknesses; based on the identification results, it dynamically adjusts the intensity of point rewards for specific behaviors in the point rule base, and pushes the adjusted incentive strategy to the target worker group through the message push module associated with the facial recognition device.

[0048] When a redemption request is initiated at a self-service redemption terminal or on a mobile device, facial recognition is used for secondary identity verification. After successful verification, the system's linked modules query real-time points and display a list of redeemable goods. After the worker confirms the redemption, the points are deducted, and the terminal is driven to issue the goods or generate a collection voucher, thus completing the incentive loop.

[0049] This invention provides a method for linking construction workers' points redemption and online learning based on facial recognition. By constructing a dynamic feedback closed-loop control system, it solves the technical defects of existing systems, such as rigid rules, delayed incentives, and inability to specifically improve the weak links of the group. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a schematic diagram of the main process of the self-service redemption and online learning linkage method for construction workers based on facial recognition provided by the present invention.

[0052] Figure 2 This is a schematic diagram of a specific operation process in the self-service redemption and online learning linkage method for construction workers based on facial recognition provided by the present invention.

[0053] Figure 3 This is a schematic diagram of another specific operation process of the construction worker's self-service points redemption and online learning linkage method based on facial recognition provided by the present invention.

[0054] Figure 4 This is a schematic diagram illustrating another specific operation process of the self-service redemption and online learning linkage method for construction workers based on facial recognition provided by the present invention.

[0055] Figure 5 The schematic diagram of the self-service points redemption and online learning linkage system for construction workers based on facial recognition provided by this invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Example 1

[0058] See Figure 1 Embodiment 1 of the present invention provides a method for linking construction workers' points self-service redemption and online learning based on facial recognition, including:

[0059] S10. Deploy facial recognition devices at construction site entrances, key work areas, and self-service redemption terminals to perform facial recognition on construction workers in online and / or offline modes. After confirming the worker's identity, a facial recognition event signal is generated. The facial recognition event signal includes at least attendance entry events and redemption verification events, and the event data is uploaded in real time. The above process is used to initially achieve identity and behavior data collection.

[0060] S20. The system linkage module receives the facial recognition event signal and determines the event type according to the preset event-rule mapping table. For behavioral events that meet the conditions for obtaining points, the linkage module calculates and distributes points based on the currently effective points rule library. Simultaneously, it acquires the learning behavior data of all workers on the online learning platform as the basis for points distribution. (The learning behavior data includes the course completion status of all workers, the exam pass rate of all workers, and their corresponding exam categories.) The above steps are used to realize the linkage processing of behavioral events and points rules. The preset points rule library is an important basis for points calculation, but the points rule library of this application is dynamically adjusted.

[0061] S30. The system analyzes the behavioral and learning data of all workers in real time to identify group weaknesses. Based on the identification results, it dynamically adjusts the intensity of point rewards for specific behaviors in the point rule base and pushes the adjusted incentive strategy (i.e., dynamic rule adjustment) to the target worker group through the message push module associated with the facial recognition device.

[0062] S40. When a redemption request is initiated at a self-service redemption terminal or mobile device, facial recognition is used for secondary identity verification. After successful verification, the system linkage module queries real-time points and displays a list of redeemable goods. After the worker confirms the redemption, the points are deducted and the terminal is driven to ship the goods or generate a receipt, thus completing the incentive closed loop (i.e., the points verification and self-service redemption closed loop).

[0063] The present invention provides a method for linking construction workers' points redemption and online learning based on facial recognition. Its core effect lies in breaking the rigidity of traditional static points systems and solving the fundamental defects of "disconnect between incentives and needs" and "decoupling of rules and goals." Through algorithms, it achieves dynamic and precise incentives, identifying weaknesses in a group (such as a lack of certain safety knowledge) and dynamically adjusting relevant points rewards, thus achieving precise allocation of incentive resources and "precise incentives," greatly improving the efficiency of training and behavioral guidance. Not only considering the group as a whole, it also assesses the incentive sensitivity of different workers through cluster analysis and individual historical response coefficients, enabling rule adjustments to anticipate differentiated responses from different groups. The system connects the entire process of identity authentication, attendance, learning, and redemption, achieving unmanned and self-service management, reducing management costs, and improving the worker experience.

[0064] This invention provides a method for linking construction workers' points-based self-service redemption and online learning based on facial recognition. It is tailored to the characteristics of water conservancy construction—"complex environment, dispersed personnel, and special risks"—bringing multiple efficiency improvements to safety management. Regarding personnel access, facial recognition devices in key areas support online / offline recognition, quickly verifying identity, qualifications, and training status, preventing unqualified workers from working, and improving verification efficiency by over 80%, thus strengthening the safety defense line. In terms of safety incentives, water conservancy-specific safety behaviors such as wearing life-saving equipment and reporting hazards are linked to points, which are distributed in real-time and can be redeemed for supplies. This is linked to targeted safety learning to address shortcomings and stimulate proactive safety awareness. The management end automatically compiles attendance and safety data, and self-service redemption reduces manual intervention. Rules can be dynamically adapted to the needs of different construction phases, reducing statistical workload by 70%. Simultaneously, through data closure, management weaknesses are accurately identified, resource allocation is optimized, and a dynamic feedback safety control system is built, solving the pain points of traditional management.

[0065] Preferably, as one possible implementation, the "dynamically adjusting the intensity of point rewards for specific behaviors in the point rule base" in S30 specifically includes:

[0066] S31. Based on the learning data within a preset period, identify the exam categories in the current learning behavior data of all workers whose exam pass rate is lower than the threshold, define the category as a weak link to be strengthened, and set improving the overall pass rate of the exam in this category as the optimization goal.

[0067] S32. Assign an integral reward intensity to the preset learning behaviors (preset learning behaviors include completing relevant courses, passing mock exams, and other preset learning behaviors that are conducive to improving the pass rate) for the weak links, and then solve the problem; sub-learning behaviors are branches of the preset learning behaviors and specific learning behavior operations) as a decision variable to be optimized; initialize a particle swarm, where the position vector of each particle represents a set of reward intensity values ​​set for different learning sub-behaviors;

[0068] S33. The iterative optimization process includes: the fitness value of each particle is determined by predicting the overall incentive efficiency improvement effect of the target worker group after the reward intensity scheme represented by its position vector is executed; the particle swarm algorithm iteratively updates the velocity and position of the particles to find the optimal reward intensity scheme that maximizes the overall incentive efficiency evaluation value.

[0069] S34. The rules for dynamically adjusting the points rule base include updating the optimal reward intensity scheme obtained through iterative optimization to the points rule base and activating it for the target worker group.

[0070] Analysis of the above scheme shows that step S32 models the reward intensity as a mathematical object called a "particle position vector," transforming the complex multi-behavior joint optimization problem (e.g., whether to emphasize course integrals or exam integrals) into a standard optimization problem. Step S33 mainly implements process-oriented optimization, that is, it introduces the mature optimization algorithm, Particle Swarm Optimization (PSO), to automatically search for the optimal reward intensity scheme. The effect is that the system can find the best strategy combination that maximizes the incentive effect (overall response) under the expected cost through simulation and iteration.

[0071] Preferably, as one feasible implementation, the "target worker group" in S33 is obtained through cluster analysis of workers' historical behavioral and attribute data; targeting the weak link, workers are divided into different cluster groups, with workers in each group having similar behavioral characteristics and incentive effectiveness. The "overall incentive effectiveness improvement effect" in S33 is quantified by an overall incentive effectiveness evaluation value; this evaluation value integrates the incentive effectiveness values ​​of workers in each cluster group and the integral cost factor consumed in implementing the reward intensity scheme.

[0072] Analysis of the above scheme reveals that this operation explicitly states that workers can be clustered based on "historical behavior and attributes." This means the system recognizes that different groups, such as veteran workers and new workers, high-risk individuals and safety role models, may react differently to the same incentive. Group-based policy implementation allows the optimization algorithm to implicitly consider "which group is more sensitive to this reward" when adjusting rewards, potentially leading to an optimal solution with varying degrees of bias towards different groups, thus avoiding waste of incentive resources.

[0073] Preferably, as one possible implementation, the overall incentive effectiveness evaluation value is calculated using the following vectorized formula:

[0074] ;

[0075] in, This represents the overall incentive effectiveness assessment value; Indicates the number of cluster groups; Indicates the first Cluster grouping The size weight is the proportion of the group's members to the total target number; Indicates the first Groups The number of workers assessed within the organization; Indicates the first The first group The historical incentive efficacy coefficient of an individual worker for the intensity of a reward is calculated based on their past participation and effectiveness with similar incentives. The time decay constant; Indicates the first The first group The time interval between each worker's most recent participation in relevant learning; Indicating the scheme regarding the first The intensity of the point-based reward for each learning sub-behavior; This represents the total number of learning sub-behaviors included in the scheme; and This is the parameter for adjusting integral cost.

[0076] The above scheme comprehensively evaluates and utilizes computability, and incorporates nuanced factors such as time decay and individual historical characteristics, making the evaluation model extremely precise and objective. Numerator: This reflects the weighting of groups, with the response of larger groups being more important. Individual historical response coefficient is a highly personalized parameter. It represents the response of a worker who has historically been positive about incentives. (High), under the new scheme, it may continue to maintain a high response, with a larger contribution weight. This gives the model predictive capabilities. Time decay factor. For those who haven't studied for a long time ( For workers with high activity levels, the expected response will be discounted. This guides the optimization algorithm to focus more on workers who have been active recently or have shown signs of recent apathy, making the incentive strategy more timely.

[0077] The denominator is explicitly stated: cost is directly related to all reward intensity values. The weighted and linked values, α and β, provide levers for adjusting cost sensitivity. Overall effect: This complex formula integrates group differences, individual traits, time factors, and cost constraints. The E value it calculates is an extremely fine and forward-looking simulation prediction of the effect of an incentive scheme, providing a high-quality optimization objective function for particle swarm optimization.

[0078] Preferably, as an implementation scheme, the particle swarm iterative optimization process in S33 further introduces a long-term incentive correction factor when calculating particle fitness, so as to overcome the defect of simply pursuing short-term incentive efficiency improvement and possibly ignoring the long-term formation of behavioral habits;

[0079] The calculation method for the long-term incentive correction factor is as follows: for the weak link targeted by the current reward intensity scheme, extract the relevant behavioral sequence of the target worker group in the past period, and then query the relationship mapping table;

[0080] Analyze the continuous learning duration of positive behaviors (such as active learning) in the behavior sequence; the correction factor (i.e., the correction factor) is positively correlated with the continuous learning duration presented in the behavior sequence. Therefore, after obtaining the continuous learning duration, the corresponding correction factor can be obtained according to the preset mapping table of the relationship between the correction factor and the continuous learning duration.

[0081] During iterative optimization, the final fitness value of a particle is the product of its original predicted overall incentive performance evaluation value and the long-term incentive correction factor.

[0082] It should be noted that in the above technical solution, the "overall incentive effectiveness improvement effect" in S33 is quantified by an overall incentive effectiveness evaluation value. This evaluation value not only integrates the incentive effectiveness values ​​of workers in each cluster group and the integral cost factor consumed by implementing the reward intensity scheme, but more importantly, it introduces an individual reward sensitivity coefficient based on the theory of neuroplasticity. This coefficient is used to simulate and predict the differentiated impact of different reward intensities on the intrinsic behavioral drive of individual workers, so that the optimization goal shifts from simply triggering short-term behavior to evaluating the effectiveness of long-term habit formation.

[0083] Example 2

[0084] See Figure 1 Embodiment 1 of the present invention provides a method for linking construction workers' points self-service redemption and online learning based on facial recognition, including:

[0085] S10. Deploy facial recognition devices at construction site entrances, key work areas, and self-service redemption terminals to perform facial recognition on construction workers in online and / or offline modes. After confirming the worker's identity, a facial recognition event signal is generated. The facial recognition event signal includes at least attendance entry events and redemption verification events, and the event data is uploaded in real time. The above process is used to initially achieve identity and behavior data collection.

[0086] S20. The system linkage module receives the facial recognition event signal and determines the event type according to the preset event-rule mapping table. For behavioral events that meet the conditions for obtaining points, the linkage module calculates and distributes points based on the currently effective points rule library. Simultaneously, it acquires the learning behavior data of all workers on the online learning platform as the basis for points distribution. (The learning behavior data includes the course completion status of all workers, the exam pass rate of all workers, and their corresponding exam categories.) The above steps are used to realize the linkage processing of behavioral events and points rules. The preset points rule library is an important basis for points calculation, but the points rule library of this application is dynamically adjusted.

[0087] S30. The system analyzes the behavioral and learning data of all workers in real time to identify group weaknesses. Based on the identification results, it dynamically adjusts the intensity of point rewards for specific behaviors in the point rule base and pushes the adjusted incentive strategy (i.e., dynamic rule adjustment) to the target worker group through the message push module associated with the facial recognition device.

[0088] S40. When a redemption request is initiated at a self-service redemption terminal or mobile device, facial recognition is used for secondary identity verification. After successful verification, the system linkage module queries real-time points and displays a list of redeemable goods. After the worker confirms the redemption, the points are deducted and the terminal is driven to ship the goods or generate a receipt, thus completing the incentive closed loop (i.e., the points verification and self-service redemption closed loop).

[0089] During step S20, as the project enters its "high-temperature construction season," the system administrator issues a dynamic rule: "Within the next month, workers who complete the 'High-Temperature Work Prevention of Heatstroke' course will have their base score increased from 30 to 50 points, and all workers will receive a 1.2x points bonus." The event occurs: Worker Wang completes the "High-Temperature Work Prevention of Heatstroke" course and passes the exam by scanning his face at the self-service learning machine in the living area. The learning platform sends the event to the system linkage module.

[0090] Specific process: The system linkage module receives the event, encapsulates it into a standard object, and executes the following program:

[0091] { "worker_id": "W123456",

[0092] "event_type": "COURSE_COMPLETE",

[0093] "event_time": "2023-07-15 14:30:00",

[0094] "event_data": {

[0095] "course_id": "C007",

[0096] "course_name": "Preventing Heatstroke in High-Temperature Work",

[0097] "score": 95,

[0098] "duration_minutes": 25},

[0099] "rule_version_id": "V2.1"}.

[0100] The points acquisition linkage module (rule engine) starts working: Matching rule 1 (static basic rule): There is a rule in the rule base: "When event_type is COURSE_COMPLETE and course_id is in the standard course list, issue the preset basic points for that course." In the preset basic points table, the points for course C007 have been dynamically updated from 30 to 50. Matching rule 2 (dynamic seasonal rule): There is a dynamic rule in the rule base with an effective period of "July 1st to July 31st": "When event_type is COURSE_COMPLETE and course_name contains the keywords 'high temperature' or 'heatstroke prevention', apply a coefficient of 1.2 to the points distribution." Matching rule 3 (excellent reward rule): There is another rule in the rule base: "When event_type is COURSE_COMPLETE and score >= 90, award an additional 10 points."

[0101] Points Calculation: Base Points = 50 (from the dynamically updated preset table); Coefficient Bonus = 1.2 (from the dynamic seasonal rules); Extra Reward = 10 (from the excellent reward rules); Actual Points Issued = 50 × 1.2 + 10 = 70;

[0102] Finally, the distribution and recording were completed: worker Wang's account points changed from 500 to 570.

[0103] The transaction log entry reads: "W123456, COURSE_COMPLETE, +70, Balance 570, Rule V2.1, Time 2023-07-15 14:30:05".

[0104] A notification popped up on Wang's learning machine screen: "Congratulations on completing the course! Points earned: 60 (course) + 10 (excellent reward) = 70 points. Current total points: 570 points."

[0105] Analysis of the above cases shows that the entire process is highly automated, modular, and flexibly configurable. The core lies in the rule engine's ability to process standardized events and configurable rule bases, enabling the system to easily implement the "dynamic adjustment of points rules" and "intelligent push" described in claim 1, forming a sensitive incentive feedback loop.

[0106] See Figure 2 Preferably, as one possible implementation, the "dynamically adjusting the intensity of point rewards for specific behaviors in the point rule base" in S30 specifically includes:

[0107] S31. Based on the learning data within a preset period, identify the exam categories in the current learning behavior data of all workers whose exam pass rate is lower than the threshold, define the category as a weak link to be strengthened, and set improving the overall pass rate of the exam in this category as the optimization goal.

[0108] S32. Assign an integral reward intensity to the preset learning behaviors (preset learning behaviors include completing relevant courses, passing mock exams, and other preset learning behaviors that are conducive to improving the pass rate) for the weak links, and then solve the problem; sub-learning behaviors are branches of the preset learning behaviors and specific learning behavior operations) as a decision variable to be optimized; initialize a particle swarm, where the position vector of each particle represents a set of reward intensity values ​​set for different learning sub-behaviors;

[0109] S33. The iterative optimization process includes: the fitness value of each particle is determined by predicting the overall incentive efficiency improvement effect of the target worker group after the reward intensity scheme represented by its position vector is executed; the particle swarm algorithm iteratively updates the velocity and position of the particles to find the optimal reward intensity scheme that maximizes the overall incentive efficiency evaluation value.

[0110] S34. The rules for dynamically adjusting the points rule base include updating the optimal reward intensity scheme obtained through iterative optimization to the points rule base and activating it for the target worker group.

[0111] Preferably, as an implementation scheme, the "target worker group" in S33 is obtained by cluster analysis of the workers' historical behavior data and attribute data; in response to the weak link, the workers are divided into different cluster groups, and the workers in each group have similar behavioral characteristics and incentive effectiveness.

[0112] Based on the above scheme, the calculation of the overall incentive effectiveness evaluation value reintroduces the individual reward sensitivity coefficient and adopts the following vectorized formula:

[0113] ;

[0114] in, This represents the overall motivational efficacy assessment value after incorporating the neuroplasticity model; The meaning is the same as the parameter in Embodiment 1; Indicates the first The optimization cycle, the first The first group The individual reward sensitivity coefficient of each worker; this coefficient is a dynamic variable whose initial value is determined based on the worker's historical stability and learning ability assessment, and is updated after each system cycle based on the worker's actual behavioral feedback (such as the speed of incentive effectiveness and behavioral persistence of the previous round of incentives).

[0115] See Figure 3 Preferably, as one feasible implementation; the individual reward sensitivity coefficient The update mechanism is used to further understand the internal driving energy regulation of the work, specifically:

[0116] S61. Implementing a set of reward intensity schemes During the subsequent evaluation cycle, record the workers' information. Relevant target behavior data, including: behavior trigger delay Quality of performance and the frequency of spontaneous repetition of the behavior without immediate reward. ;

[0117] S62. The calculation of the internal driving energy adjustment factor includes calculating an internal driving energy adjustment factor based on the behavioral feedback. : ;

[0118] in, To adjust the gain constant; The average incentive intensity applied to this worker in the current period; It is a hyperbolic tangent function used to normalize the feedback difference;

[0119] S63. Update the reward sensitivity coefficient based on the adjustment factor:

[0120] ;

[0121] When the reward intensity matches an individual's current sensitivity and the behavioral feedback is positive, sensitivity increases, and future incentives can be maintained with more economical rewards. When the reward is too high or the feedback is negative, sensitivity decreases, preventing the waste of incentive resources and the development of tolerance.

[0122] It should be noted that in the particle swarm optimization process of S33, the fitness value of each particle is directly determined by the reward intensity scheme represented by its position vector. The overall incentive effectiveness assessment value is predicted and calculated. Confirmed; Particle swarm optimization algorithm maximizes The system aims to find the optimal reward intensity. It maintains a dynamic incentive profile for each worker, the core of which is the individual reward sensitivity coefficient that evolves over time. The system uses this profile to personalize the reward intensity of each worker or cluster group within their "optimal performance range," which guides personalized message pushes and incentive prompts after rule updates in S34.

[0123] See Figure 4 Preferably, as one feasible implementation, the personalized recommendation method includes the following steps:

[0124] S91. Plot the different reward intensities for this worker throughout history. Its corresponding immediate behavioral incentive effectiveness Scatter plot;

[0125] S92, Using nonlinear regression to fit the results and Relationship curve;

[0126] S93. The reward intensity range corresponding to the peak of the curve is recommended as the optimal performance incentive range for the worker in the current cycle.

[0127] Preferably, as one possible implementation; in the redemption stage of S40, the system determines the redemption worker's individual reward sensitivity coefficient. The display logic of the list of redeemable goods is dynamically adjusted; for workers with a continuously high sensitivity coefficient, high-value goods that require long-term accumulation are displayed first when they have enough points, so as to strengthen their goal-oriented behavior; for workers with a low or fluctuating sensitivity coefficient, the proportion of goods with low point thresholds and quick redemption is increased to rebuild their incentive feedback loop.

[0128] Example 3

[0129] See Figure 5 The present invention provides a construction worker points self-service redemption and online learning linkage system based on facial recognition, comprising: an information collection module 10, a system linkage module 20, an adjustment processing module 30, and a redemption processing module 40;

[0130] Information collection module 10 deploys facial recognition devices at construction site entrances, key work areas, and self-service redemption terminals. It performs facial recognition on construction workers in online and / or offline modes, and generates a facial recognition event signal after confirming the current worker's identity. The facial recognition event signal includes at least attendance entry events and redemption verification events, and uploads the event data in real time.

[0131] The system linkage module 20 is used to receive the face recognition event signal and determine the event type according to the preset event-rule mapping table. For behavior events that meet the conditions for obtaining points, the system linkage module calculates and distributes points according to the currently effective points rule library. At the same time, the system synchronously obtains the learning behavior data of all workers on the online learning platform as the basis for distributing points. The learning behavior data includes the course completion status of all workers, the exam pass rate of all workers and their corresponding exam categories.

[0132] The adjustment processing module 30 analyzes the behavioral and learning data of all workers in real time to identify group weaknesses; based on the identification results, it dynamically adjusts the intensity of point rewards for specific behaviors in the point rule base, and pushes the adjusted incentive strategy to the target worker group through the message push module associated with the facial recognition device.

[0133] The redemption processing module 40 calls facial recognition for secondary identity verification when a redemption request is initiated at a self-service redemption terminal or mobile terminal. After successful verification, the system linkage module queries real-time points and displays a list of redeemable goods. After the worker confirms the redemption, the points are deducted and the terminal is driven to ship the goods or generate a collection voucher, thus completing the incentive closed loop.

[0134] In summary, this invention provides a method for linking construction workers' points redemption and online learning based on facial recognition. Its core effect lies in breaking the rigidity of traditional static points systems and solving the fundamental defects of "disconnect between incentives and needs" and "decoupling of rules and goals." Through algorithms, it achieves dynamic and precise incentives, identifying weaknesses in a group (such as a lack of certain safety knowledge) and dynamically adjusting relevant points rewards, thus achieving precise allocation of incentive resources and "precise incentives," greatly improving the efficiency of training and behavioral guidance. Not only considering the group as a whole, it also assesses the incentive sensitivity of different workers through cluster analysis and individual historical response coefficients, enabling rule adjustments to anticipate differentiated responses from different groups. This system connects the entire process of identity authentication, attendance, learning, and redemption, achieving unmanned and self-service management, reducing management costs, and improving the worker experience.

[0135] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications or additions should fall within the protection scope of the present invention.

Claims

1. A method for linking self-service points redemption and online learning for construction workers based on facial recognition, characterized in that, The method includes the following steps: Facial recognition devices are deployed at construction site entrances, key work areas, and self-service redemption terminals to perform facial recognition on construction workers in online and / or offline modes. After confirming the current worker's identity, a facial recognition event signal is generated. The facial recognition event signal includes at least attendance entry events and redemption verification events, and the event data is uploaded in real time. The system linkage module receives the face recognition event signal and determines the event type according to the preset event-rule mapping table. For behavior events that meet the conditions for obtaining points, the linkage module calculates and distributes points according to the currently effective points rule library. At the same time, the learning behavior data of all workers on the online learning platform is obtained synchronously as the basis for distributing points. The learning behavior data includes the course completion status of all workers, the exam pass rate of all workers and their corresponding exam categories. The system analyzes the behavioral and learning data of all workers in real time to identify group weaknesses; based on the identification results, it dynamically adjusts the intensity of point rewards for specific behaviors in the point rule base, and pushes the adjusted incentive strategy to the target worker group through the message push module associated with the facial recognition device. When a redemption request is initiated at a self-service redemption terminal or on a mobile device, facial recognition is used for secondary identity verification. After successful verification, the system's linked modules query real-time points and display a list of redeemable goods. After the worker confirms the redemption, the points are deducted, and the terminal is driven to issue the goods or generate a collection voucher, thus completing the incentive loop.

2. The method according to claim 1, characterized in that, The dynamic adjustment of the points reward intensity for specific behaviors in the points rule base specifically includes: Based on the learning data within a preset period, identify the exam categories in the current learning behavior data of all workers whose exam pass rate is below a threshold, define the category as a weak link to be strengthened, and set improving the overall pass rate of the exam in this category as the optimization goal; The integral reward intensity of the preset learning behavior for the weak link is constructed as a decision variable to be optimized; a particle swarm is initialized, where the position vector of each particle represents a set of reward intensity values ​​set for different learning sub-behaviors. The fitness value of each particle is determined by predicting the overall incentive efficiency improvement effect on the target worker group after the reward intensity scheme represented by its position vector is executed; the particle swarm algorithm iteratively updates the velocity and position of the particles to find the optimal reward intensity scheme that maximizes the overall incentive efficiency evaluation value. The optimal reward intensity scheme obtained through iterative optimization is updated to the points rule base and activated for the target worker group.

3. The method according to claim 2, characterized in that, The target worker group was obtained by clustering analysis of workers' historical behavioral and attribute data; in response to the weak links, workers were divided into different cluster groups, and workers in each group had similar behavioral characteristics and motivational effectiveness.

4. The method according to claim 3, characterized in that, The overall incentive effectiveness evaluation value is used to quantify the improvement of overall incentive effectiveness; the overall incentive effectiveness evaluation value integrates the incentive effectiveness value of workers in each cluster group, as well as the integral cost factor consumed in implementing the reward intensity scheme.

5. The method according to claim 4, characterized in that, The overall incentive effectiveness evaluation value is calculated using the following vectorized formula: ； in, This indicates the overall incentive effectiveness assessment value; Indicates the number of cluster groups; Indicates the first Cluster grouping The size weight is the proportion of the group's members to the total target number; Indicates the first Groups The number of workers assessed within the organization; Indicates the first The first group The historical incentive efficacy coefficient of an individual worker for the intensity of a reward is calculated based on their past participation and effectiveness with similar incentives. The time decay constant; Indicates the first The first group The time interval between each worker's most recent participation in relevant learning; Indicating the scheme regarding the first The intensity of the point-based reward for each learning sub-behavior; This represents the total number of learning sub-behaviors included in the scheme; and This is the parameter for adjusting integral cost.

6. The method according to claim 2, characterized in that, In the iterative optimization process of particle swarm optimization, a long-term incentive correction factor is further introduced when calculating the particle fitness. The calculation method for the long-term incentive correction factor is as follows: for the weak link targeted by the current reward intensity scheme, extract the relevant behavioral sequence of the target worker group in the past period, and then query the relationship mapping table; Analyze the continuous learning duration of positive behaviors in the behavior sequence; the correction factor is positively correlated with the continuous learning duration presented in the behavior sequence; During iterative optimization, the final fitness value of a particle is the product of its original predicted overall incentive performance evaluation value and the long-term incentive correction factor.

7. The method according to claim 4, characterized in that, The overall incentive effectiveness evaluation value is calculated using the following vectorized formula: ； in, This represents the overall motivational efficacy assessment value after incorporating the neuroplasticity model; in, Indicates the number of cluster groups; Indicates the first Cluster grouping The size weight is the proportion of the group's members to the total target number; Indicates the first Groups The number of workers assessed within the organization; Indicates the first The first group The historical incentive efficacy coefficient of an individual worker for the intensity of a reward is calculated based on their past participation and effectiveness with similar incentives. The time decay constant; Indicates the first The first group The time interval between each worker's most recent participation in relevant learning; Indicating the scheme regarding the first The intensity of the point-based reward for each learning sub-behavior; This represents the total number of learning sub-behaviors included in the scheme; and This is the parameter for adjusting integral cost; in, Indicates the first The optimization cycle, the first The first group Individual worker's personal reward sensitivity coefficient.

8. The method according to claim 7, characterized in that, The individual reward sensitivity coefficient The update mechanism is used to further understand the internal driving energy regulation of the work, specifically: Implementing a set of reward intensity schemes During the subsequent evaluation cycle, record the workers' information. Relevant target behavior data, including: behavior trigger delay Quality of performance and the frequency of spontaneous repetition of the behavior without immediate reward. ; Based on the behavioral feedback, an internal driving energy adjustment factor is calculated. : ; in, To adjust the gain constant; The average incentive intensity applied to this worker in the current period; It is the hyperbolic tangent function; Update the reward sensitivity coefficient based on the adjustment factor: .

9. The method according to claim 8, characterized in that, The personalized recommendation method includes the following steps: Plotting the worker's historical reward intensities Its corresponding immediate behavioral incentive effectiveness Scatter plot; Using nonlinear regression to fit and Relationship curve; The reward intensity range corresponding to the peak of the curve is recommended as the optimal performance incentive range for the worker in the current cycle.

10. The method according to claim 1, characterized in that, During the redemption process, the system uses the individual reward sensitivity coefficient of the redeeming worker. The display logic of the list of redeemable goods will be dynamically adjusted; for workers with a continuously high sensitivity coefficient, high-value goods that require long-term accumulation will be displayed first when they have enough points; for workers with a low or fluctuating sensitivity coefficient, a low points threshold will be added.

Images