Intelligent AI personnel skill portrait management system and method for team management

Abstract

The invention discloses an intelligent AI personnel skill portrait management system and method for team management. The method comprises the following steps: collecting and preprocessing multi-source business data associated with team personnel skills; performing feature engineering processing to obtain a skill feature vector corresponding to each team personnel; performing quantitative calculation in a preset multi-dimensional skill quantitative model to generate an initial quantitative score of each skill dimension; performing score correction and normalization processing to generate a standardized skill score of each skill dimension; generating team personnel skill portrait data; generating a skill portrait visualization result; and executing portrait application processing, and generating a training resource push result, thereby realizing multi-dimensional quantitative evaluation and standardized comparison of team personnel skills, and improving the accuracy and efficiency of personnel matching.

Description

Technical Field

[0001] This invention relates to the field of intelligent work team management, and in particular to an intelligent AI personnel skill profiling management system and method for work team management. Background Technology

[0002] As enterprises improve their digitalization levels in safe production, equipment operation and maintenance, and on-site operations, team management is gradually shifting from traditional experience-based management to data-driven management. As the smallest organizational unit for production execution, the skill level of team members directly impacts work quality, safety risks, and task execution efficiency. To enhance the precision of team management, existing technologies typically use information platforms such as training management systems, hazard identification systems, safety monitoring systems, and innovation management systems to record personnel training, hazard identification, violations, and innovations. Based on this data, they attempt to assess and apply personnel capabilities, for example, in management scenarios such as job assignment, task dispatch, and training plan development.

[0003] Existing technical solutions for team personnel competency management generally suffer from the following characteristics: First, data sources are diverse and systems are scattered, with significant differences in data fields, coding rules, time granularity, and personnel identification rules across various business systems. Second, competency assessment methods largely rely on manual aggregation and statistics or simple rule calculations, making it difficult to create structured, computable, and comparable skill profiles. Third, there is a lack of a unified data structure and process connection between competency assessment results and subsequent business applications, making it difficult for assessment results to reliably support business requests such as personnel matching and training delivery. Therefore, given the continuous growth of multi-source business data and the increasing demands for team management, how to construct a personnel skill profile management method and system for team management within a computer system has become an urgent technical problem to be solved. Summary of the Invention

[0004] One objective of this invention is to propose a smart AI personnel skill profile management system and method for team management. This invention fully utilizes multi-source business data collection and preprocessing, feature engineering processing, and multi-dimensional skill quantification models to quantitatively assess team members' skills in preset dimensions such as safety awareness, hazard identification ability, anti-violation ability, and innovation and improvement ability. Standardized skill scores are generated through score correction and normalization, thereby constructing team member skill profile data and supporting personnel matching recommendations and training resource push. It has the advantages of strong data fusion capabilities, objective and comparable skill assessments, traceable profile updates, and accurate application decisions.

[0005] A smart AI-powered personnel skill profile management method for team management according to an embodiment of the present invention includes the following steps: Collect and preprocess multi-source business data related to the skills of team members; Feature engineering is performed on the preprocessed multi-source business data to obtain the skill feature vector corresponding to each team member. The skill feature vectors are input into a preset multidimensional skill quantification model to quantify the ability level of team members in each preset skill dimension and generate initial quantification scores for each skill dimension. The initial quantitative scores for each skill dimension are corrected and normalized to generate standardized skill scores for each skill dimension. Skill profile data for team members is generated based on standardized skill scores across various skill dimensions. Generate a visualization of skill profiles based on the skill profile data of team members; When a team management business request is received, the system performs profile application processing based on the team members' skill profile data and generates training resource push results.

[0006] Optionally, the multi-source business data includes personnel identification information, time information, training data, hazard data, violation data, and innovation data. The preprocessing includes data cleaning, field standardization, time alignment, and unified processing of personnel identification.

[0007] Optionally, obtaining the skill feature vector specifically includes: The pre-processed multi-source business data is collected and divided according to the identification of team members to obtain safety training data, hidden danger investigation data, violation data and innovation and improvement data. Extracting safety training features from safety training data; Extracting hazard identification features from hazard identification data; Extracting characteristics of violations based on violation data; Based on innovation and improvement data, extract innovation and improvement features; The characteristics of safety training, hazard investigation, violations, and innovation and improvement are vectorized, encoded, and concatenated to obtain the skill characteristic vector corresponding to each team member.

[0008] Optionally, the generation of the initial quantitative scores for each skill dimension specifically includes: The skill feature vector is input into a preset multidimensional skill quantification model, which includes a safety awareness sub-model, a hidden danger investigation capability sub-model, an anti-violation capability sub-model, and an innovation and improvement capability sub-model. In the safety awareness sub-model, by setting a first weighting coefficient and a second weighting coefficient, and their sum being one, an initial quantitative score for safety awareness is calculated based on the average score of the safety exam and the individual violation rate. The individual violation rate is obtained by the ratio of the number of individual violations to the historical average number of violations of the work group, with an upper limit constraint. The initial quantitative score for safety awareness is obtained by weighting the average score of the safety exam with the first weighting coefficient, and weighting the result of subtracting the individual violation rate from one and multiplying it by one hundred points with the second weighting coefficient, and then adding the two weighted results. In the sub-model of hazard identification capability, by setting a first weight coefficient, a second weight coefficient, and a third weight coefficient, and summing them to one, the initial quantitative score of hazard identification capability is calculated based on the individual hazard registration rate, the effective hazard registration rate, and the individual hazard rectification rate. The individual hazard registration rate is obtained by the ratio of the number of individual hazard registrations to the historical average number of hazard registrations of the work group, with an upper limit constraint. The effective hazard registration rate is obtained by the ratio of the number of effective hazard registrations to the number of individual hazard registrations. The individual hazard rectification rate is obtained by the ratio of the number of hazards rectified by the individual to the total number of hazards rectified by the individual. The initial quantitative score of hazard identification capability is obtained by weighting the individual hazard registration rate, the effective hazard registration rate, and the individual hazard rectification rate according to the first weight coefficient, the second weight coefficient, and the third weight coefficient, and then multiplying by one hundred points. In the anti-violation capability sub-model, by setting a first weight coefficient and a second weight coefficient, and summing them to one, the initial quantitative score of anti-violation capability is calculated based on the individual violation rate and the individual violation rectification rate. The individual violation rate is obtained by the ratio of the number of individual violations to the historical average number of violations of the team, with an upper limit constraint. The individual violation rectification rate is obtained by the ratio of the number of violations rectified by the individual to the total number of violations rectified by the individual. The initial quantitative score of anti-violation capability is obtained by weighting the result of subtracting the individual violation rate from the first weight coefficient, weighting the individual violation rectification rate by the second weight coefficient, summing the two weighted results, and multiplying by one hundred points. In the innovation and improvement capability sub-model, by setting a first weighting coefficient and a second weighting coefficient, the initial quantitative score of innovation and improvement capability is calculated based on the proposal submission rate and the proposal adoption rate. The proposal submission rate is obtained by the ratio of the number of proposals to a preset benchmark value and subject to an upper limit constraint. The proposal adoption rate is obtained by the ratio of the number of approved proposals to the total number of proposals. The initial quantitative score of innovation and improvement capability is obtained by weighting the proposal submission rate and the proposal adoption rate according to the first weighting coefficient and the second weighting coefficient and multiplying by 100.

[0009] Optionally, the generation of standardized skill scores for each skill dimension specifically includes: The initial quantitative scores for each preset skill dimension are grouped according to the team member identifier and the preset skill dimension identifier; For each preset skill dimension, the group statistical baseline mean and the group statistical baseline standard deviation are calculated based on the initial quantitative scores after grouping. The group statistical baseline mean is the sum of the initial quantitative scores of all members in the group on the preset skill dimension divided by the total number of members in the group. The group statistical baseline standard deviation is the square root of the sum of the squares of the differences between the initial quantitative scores of all members in the group on the preset skill dimension and the group statistical baseline mean divided by the total number of members in the group. For each preset skill dimension, the initial quantitative score of each team member on the preset skill dimension is corrected based on the team statistical benchmark to obtain the corrected score. The corrected score is obtained by subtracting the mean of the team statistical benchmark from the initial quantitative score and then dividing by the standard deviation of the team statistical benchmark. For each preset skill dimension, the corrected score is input into the standard normal cumulative distribution function to obtain the cumulative probability value, and the cumulative probability value is multiplied by 100 to obtain the normalized score; Standardized skill scores are obtained by applying score interval constraint processing to the normalized scores. The score interval constraint processing refers to limiting normalized scores less than zero to zero and normalized scores greater than one hundred to one hundred, thus forming standardized skill scores for each preset skill dimension.

[0010] Optionally, the generation of the team member skill profile data specifically includes: The standardized skill scores are aggregated using the team member identifier as an index to form a set of standardized skill scores for each preset skill dimension corresponding to each team member identifier. For each shift's personnel, generate personnel identification information, which includes an employee ID field; For each shift, personnel are identified, and statistical period information is generated, which includes a statistical period start time field and a statistical period end time field. For each shift's personnel, a standardized set of skill scores is combined according to a preset set of skill dimensions to form a skill score vector. By concatenating personnel identification information, statistical period information, and skill score vectors, a skill profile data of team members is formed.

[0011] Optionally, the generation of the skill profile visualization results specifically includes: Standardize skill scores in the skill profile data of team members and create an index based on team member identification, statistical period information, and preset skill dimension identification; For each shift's personnel identification and statistical period information, the identifiers of each preset skill dimension are fixedly sorted according to the preset skill dimension set, and the circumferential angle is equally divided based on the number of preset skill dimensions to obtain the angle value corresponding to each preset skill dimension identifier. The standardized skill score is used as the radial value and paired with the angle value to generate a polar coordinate point set, and the polar coordinate point set is encapsulated as radar chart data. For each shift member identifier and each preset skill dimension identifier, obtain a statistical periodic sequence arranged in chronological order, and use the statistical periodic sequence as the horizontal axis and the standardized skill score of the corresponding preset skill dimension as the vertical axis to generate a time series point set, and encapsulate the time series point set into trend chart data; By linking radar chart data and trend chart data with personnel identification information and statistical period information respectively, a skill profile visualization result is formed.

[0012] Optionally, the generation of the training resource push result specifically includes: The system receives a team management business request and parses it to obtain the job or task identifier and the job or task skill requirements. The job or task skill requirements include the skill threshold, dimension weight coefficient, and recommended number of people corresponding to each preset skill dimension in the preset skill dimension set. The sum of the dimension weight coefficients is one. Based on the skill profile data of team members, the standardized skill scores of each team member in each preset skill dimension are extracted under the statistical period information corresponding to the job or task identifier, and the standardized skill scores are indexed according to the team member identifier and the preset skill dimension identifier. Based on preset skill thresholds, candidate personnel are screened to obtain a set of candidates. For each candidate in the set of candidates, a personnel matching score is calculated. The personnel matching score is obtained by multiplying the standardized skill scores of the candidate in each preset skill dimension by the corresponding dimension weight coefficient and then summing the results. The candidate set is sorted according to the personnel matching score. The candidates corresponding to the number of recommended persons in the sorting result are selected to form a recommended personnel set, and the recommendation result is output. The recommendation result includes the personnel identification information and personnel matching score of each recommended person in the recommended personnel set. For each member of a work group, the training difference for each preset skill dimension is calculated based on preset skill thresholds and standardized skill scores, and the training priority is calculated. Based on the training difference, a set of preset skill dimensions to be pushed is determined, and training resources corresponding to the set of preset skill dimensions to be pushed are retrieved from the training resource library to generate a training resource push result. The training resource push result includes personnel identification information, the set of preset skill dimensions to be pushed, the training priority, and the corresponding training resource identification list.

[0013] According to an embodiment of the present invention, a smart AI personnel skill profile management system for team management includes: The data acquisition and preprocessing module is used to collect and preprocess multi-source business data related to the skills of team members. The feature engineering module is used to collect and classify the preprocessed multi-source business data to obtain safety training data, hidden danger investigation data, violation behavior data and innovation and improvement data, and extract safety training features, hidden danger investigation features, violation behavior features and innovation and improvement features respectively to obtain the skill feature vector corresponding to each team member. The skill quantization module is used to input skill feature vectors into a preset multi-dimensional skill quantization model to generate initial quantization scores for each preset skill dimension. The scoring correction and normalization module is used to perform scoring correction and normalization processing on the initial quantitative scores of each preset skill dimension, and generate standardized skill scores for each preset skill dimension. The skills profile generation module is used to generate skills profile data for team members based on standardized skills scores on each preset skills dimension. The visualization module is used to generate skill profile visualization results based on the skill profile data of the team members; The profile application module is used to perform profile application processing based on the skill profile data of the team members when a team management business request is received. The profile application processing includes matching personnel according to job or task skill requirements and outputting recommendation results, and generating training resource push results.

[0014] The beneficial effects of this invention are: This invention unifies the collection and preprocessing of multi-source business data related to the skills of team members. It integrates training data, hazard data, violation data, and innovation data scattered across training management systems, hazard investigation systems, safety supervision systems, and innovation management systems into a single computer processing link. This achieves data cleaning, field standardization, time alignment, and unified personnel identification, thereby solving the problems of inconsistent data standards, difficulty in correlation, and difficulty in reuse across systems in existing technologies. It provides a stable and sustainable data foundation for subsequent skills assessment, transforming the management of team members' skills from relying on manual aggregation to an automated, data-driven process.

[0015] This invention further uses feature engineering to collect and classify preprocessed multi-source business data into safety training data, hazard investigation data, violation data, and innovation and improvement data. It then extracts safety training features, hazard investigation features, violation features, and innovation and improvement features respectively, and vectorizes and concatenates these features to generate skill feature vectors. This achieves a structured expression of the skill information of team members, avoiding the one-sidedness caused by using only a single indicator or scattered statistics to describe personnel capabilities in existing technologies. It provides a unified data structure and computability for skill information across different teams, supporting stable input for a multi-dimensional skill quantification model.

[0016] This invention quantifies skill feature vectors based on a pre-defined multi-dimensional skill quantification model, forming initial quantified scores for each pre-defined skill dimension. Furthermore, it performs score correction and normalization processing based on team statistical benchmarks on the initial quantified scores, generating standardized skill scores for each pre-defined skill dimension. This ensures that skill scores are comparable and consistent across different statistical periods and individuals, reducing evaluation bias caused by differences in team size, data volume, or distribution shifts. This improves the objectivity, stability, and interpretability of skill profiling results, meeting the technical requirements of team management for skill evaluation that allows for "horizontal comparison and vertical tracking."

[0017] Based on the standardized skill scores, this invention generates skill profile data for team members, which can be further used for the output of skill profile visualization results and the execution of profile application processing. This enables the skill profiles to not only present the distribution of team members' abilities across multiple skill dimensions and their trends over time, but also, upon receiving team management business requests, to complete personnel matching and output recommendation results based on job or task skill requirements, while simultaneously generating training resource push results. This achieves direct linkage between skill assessment results and team management application scenarios, improves the accuracy of job assignment and task dispatch matching, enhances the targeting of training resource allocation and the rationality of priority decisions, and overall improves the intelligence level and execution efficiency of team management. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of a smart AI personnel skill profile management system and method for team management proposed in this invention; Figure 2 This is a schematic diagram illustrating the construction of skill feature vectors in a smart AI personnel skill profile management system and method for team management proposed in this invention. Figure 3This is a schematic diagram of the initial quantitative scores for each skill dimension of the intelligent AI personnel skill profile management system and method for team management proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-3 A smart AI-powered personnel skills profiling management method for team management includes the following steps: Collect and preprocess multi-source business data related to the skills of team members; Feature engineering is performed on the preprocessed multi-source business data to obtain the skill feature vector corresponding to each team member. The skill feature vectors are input into a preset multidimensional skill quantification model to quantify the ability level of team members in each preset skill dimension and generate initial quantification scores for each skill dimension. The initial quantitative scores for each skill dimension are corrected and normalized to generate standardized skill scores for each skill dimension. Skill profile data for team members is generated based on standardized skill scores across various skill dimensions. Generate a visualization of skill profiles based on the skill profile data of team members; When a team management business request is received, the system performs profile application processing based on the team members' skill profile data and generates training resource push results.

[0021] In this embodiment, the multi-source business data includes personnel identification information, time information, training data, hidden danger data, violation data, and innovation data. The preprocessing includes data cleaning, field standardization, time alignment, and unified personnel identification processing.

[0022] In this embodiment, obtaining the skill feature vector specifically includes: The pre-processed multi-source business data is collected and divided according to the identification of team members to obtain safety training data, hidden danger investigation data, violation data and innovation and improvement data. The specific process of classification is as follows: using the team member identifier as the association key, the preprocessed multi-source business data is collected by personnel, and then the collected data is classified and mapped according to the data source system and business data type fields. Data from the training management system that includes fields for training courses, exam scores and exam times is classified into safety training data; data from the hazard investigation system that includes fields for hazard registration, validity judgment, rectification status and rectification completion time is classified into hazard investigation data; data from the safety supervision system that includes fields for violation occurrence time, violation type, handling result and rectification status is classified into violation behavior data; and data from the innovation management system that includes fields for rationalization suggestions, innovation proposals, submission time and review status is classified into innovation improvement data. Extracting safety training features from safety training data; The extraction process is as follows: After the statistical period, all safety examination records falling within the statistical period are screened from the safety training data corresponding to the team members. The number of records is counted to obtain the number of safety examinations. The safety examination records are sorted by examination time and the most recent N records are selected. The examination scores in the most recent N records are summed and divided by N to obtain the average score of the most recent N safety examinations. The examination score corresponding to the last record after sorting by time is taken as the score of the most recent safety examination, thus forming a safety training feature. Extracting hazard identification features from hazard identification data; The extraction process is as follows: After the statistical period, all hazard records within the statistical period are screened from the hazard investigation data corresponding to the team members. The number of records is taken as the number of individual hazard registrations. The number of hazard records that have passed the evaluation based on the validity judgment results is taken as the number of valid hazard registrations. The number of hazard records with the team members as the hazard responsible persons is screened based on the rectification responsibility field as the total number of hazard rectifications for each individual. Among the hazard records rectified by the individual, the number of records that have been rectified is screened based on the rectification status as the number of hazard rectifications completed for the individual. The individual hazard registration rate, the valid hazard registration rate, and the individual hazard rectification rate are calculated to form hazard investigation characteristics. Extracting characteristics of violations based on violation data; The extraction process is as follows: After the statistical period, all violation records within the statistical period are filtered from the violation data corresponding to the team members, and the number of violation records is taken as the number of individual violations; based on the rectification responsibility field in the violation records, the number of violation records for which the rectification responsibility is a team member is selected as the total number of violations rectified by the individual, and the number of records that have been rectified is selected according to the rectification status as the number of violations that have been rectified by the individual; the individual violation rate is calculated by the ratio of the number of individual violations to the historical average number of violations of the team and by applying an upper limit constraint; the individual violation rectification rate is calculated by the ratio of the number of violations that have been rectified by the individual to the total number of violations rectified by the individual, thus forming the violation behavior characteristics; Based on innovation and improvement data, extract innovation and improvement features; The extraction process is as follows: After the statistical period, all rationalization suggestion records within the statistical period are selected from the innovation and improvement data corresponding to the team members, and the number of records is counted as the number of rationalization suggestions. All innovation proposal records within the statistical period are selected, and the number of records is counted as the number of innovation proposals. Based on the review status field in the rationalization suggestion records and the innovation proposal records, the number of records that have passed the review is selected as the number of approved proposals. The number of rationalization suggestions and the number of innovation proposals are summed to obtain the number of proposals. The proposal submission rate is calculated by the ratio of the number of proposals to a preset benchmark value and an upper limit constraint is applied. The proposal adoption rate is calculated by the ratio of the number of approved proposals to the number of proposals, thus forming the innovation and improvement characteristics. The characteristics of safety training, hazard investigation, violations, and innovation and improvement are vectorized, encoded, and concatenated to obtain the skill characteristic vector corresponding to each team member.

[0023] In this embodiment, the generation of the initial quantitative scores for each skill dimension specifically includes: The skill feature vector is input into a preset multidimensional skill quantification model, which includes a safety awareness sub-model, a hidden danger investigation capability sub-model, an anti-violation capability sub-model, and an innovation and improvement capability sub-model. In the safety awareness sub-model, by setting a first weighting coefficient and a second weighting coefficient, and their sum being one, an initial quantitative score for safety awareness is calculated based on the average score of the safety exam and the individual violation rate. The individual violation rate is obtained by the ratio of the number of individual violations to the historical average number of violations of the work group, with an upper limit constraint. The initial quantitative score for safety awareness is obtained by weighting the average score of the safety exam with the first weighting coefficient, and weighting the result of subtracting the individual violation rate from one and multiplying it by one hundred points with the second weighting coefficient, and then adding the two weighted results. In the sub-model of hazard identification capability, by setting a first weight coefficient, a second weight coefficient, and a third weight coefficient, and summing them to one, the initial quantitative score of hazard identification capability is calculated based on the individual hazard registration rate, the effective hazard registration rate, and the individual hazard rectification rate. The individual hazard registration rate is obtained by the ratio of the number of individual hazard registrations to the historical average number of hazard registrations of the work group, with an upper limit constraint. The effective hazard registration rate is obtained by the ratio of the number of effective hazard registrations to the number of individual hazard registrations. The individual hazard rectification rate is obtained by the ratio of the number of hazards rectified by the individual to the total number of hazards rectified by the individual. The initial quantitative score of hazard identification capability is obtained by weighting the individual hazard registration rate, the effective hazard registration rate, and the individual hazard rectification rate according to the first weight coefficient, the second weight coefficient, and the third weight coefficient, and then multiplying by one hundred points. In the anti-violation capability sub-model, by setting a first weight coefficient and a second weight coefficient, and summing them to one, the initial quantitative score of anti-violation capability is calculated based on the individual violation rate and the individual violation rectification rate. The individual violation rate is obtained by the ratio of the number of individual violations to the historical average number of violations of the team, with an upper limit constraint. The individual violation rectification rate is obtained by the ratio of the number of violations rectified by the individual to the total number of violations rectified by the individual. The initial quantitative score of anti-violation capability is obtained by weighting the result of subtracting the individual violation rate from the first weight coefficient, weighting the individual violation rectification rate by the second weight coefficient, summing the two weighted results, and multiplying by one hundred points. In the innovation and improvement capability sub-model, by setting a first weighting coefficient and a second weighting coefficient, the initial quantitative score of innovation and improvement capability is calculated based on the proposal submission rate and the proposal adoption rate. The proposal submission rate is obtained by the ratio of the number of proposals to a preset benchmark value and subject to an upper limit constraint. The proposal adoption rate is obtained by the ratio of the number of approved proposals to the total number of proposals. The initial quantitative score of innovation and improvement capability is obtained by weighting the proposal submission rate and the proposal adoption rate according to the first weighting coefficient and the second weighting coefficient and multiplying by 100.

[0024] In this embodiment, the generation of standardized skill scores for each skill dimension specifically includes: The initial quantitative scores for each preset skill dimension are grouped according to the team member identifier and the preset skill dimension identifier; For each preset skill dimension, the group statistical baseline mean and the group statistical baseline standard deviation are calculated based on the initial quantitative scores after grouping. The group statistical baseline mean is the sum of the initial quantitative scores of all members in the group on the preset skill dimension divided by the total number of members in the group. The group statistical baseline standard deviation is the square root of the sum of the squares of the differences between the initial quantitative scores of all members in the group on the preset skill dimension and the group statistical baseline mean divided by the total number of members in the group. For each preset skill dimension, the initial quantitative score of each team member on the preset skill dimension is corrected based on the team statistical benchmark to obtain the corrected score. The corrected score is obtained by subtracting the mean of the team statistical benchmark from the initial quantitative score and then dividing by the standard deviation of the team statistical benchmark. For each preset skill dimension, the corrected score is input into the standard normal cumulative distribution function to obtain the cumulative probability value, and the cumulative probability value is multiplied by 100 to obtain the normalized score; Standardized skill scores are obtained by applying score interval constraint processing to the normalized scores. The score interval constraint processing refers to limiting normalized scores less than zero to zero and normalized scores greater than one hundred to one hundred, thus forming standardized skill scores for each preset skill dimension.

[0025] In this embodiment, the generation of the team member skill profile data specifically includes: The standardized skill scores are aggregated using the team member identifier as an index to form a set of standardized skill scores for each preset skill dimension corresponding to each team member identifier. For each shift's personnel, generate personnel identification information, which includes an employee ID field; For each shift, personnel are identified, and statistical period information is generated, which includes a statistical period start time field and a statistical period end time field. For each shift's personnel, a standardized set of skill scores is combined according to a preset set of skill dimensions to form a skill score vector. By concatenating personnel identification information, statistical period information, and skill score vectors, a skill profile data of team members is formed.

[0026] In this embodiment, the generation of the skill profile visualization result specifically includes: Standardize skill scores in the skill profile data of team members and create an index based on team member identification, statistical period information, and preset skill dimension identification; For each shift's personnel identification and statistical period information, the identifiers of each preset skill dimension are fixedly sorted according to the preset skill dimension set, and the circumferential angle is equally divided based on the number of preset skill dimensions to obtain the angle value corresponding to each preset skill dimension identifier. The standardized skill score is used as the radial value and paired with the angle value to generate a polar coordinate point set, and the polar coordinate point set is encapsulated as radar chart data. For each shift member identifier and each preset skill dimension identifier, obtain a statistical periodic sequence arranged in chronological order, and use the statistical periodic sequence as the horizontal axis and the standardized skill score of the corresponding preset skill dimension as the vertical axis to generate a time series point set, and encapsulate the time series point set into trend chart data; By linking radar chart data and trend chart data with personnel identification information and statistical period information respectively, a skill profile visualization result is formed.

[0027] In this embodiment, the generation of the training resource push result specifically includes: The system receives a team management business request and parses it to obtain the job or task identifier and the job or task skill requirements. The job or task skill requirements include the skill threshold, dimension weight coefficient, and recommended number of people corresponding to each preset skill dimension in the preset skill dimension set. The sum of the dimension weight coefficients is one. Based on the skill profile data of team members, the standardized skill scores of each team member in each preset skill dimension are extracted under the statistical period information corresponding to the job or task identifier, and the standardized skill scores are indexed according to the team member identifier and the preset skill dimension identifier. Based on preset skill thresholds, candidate personnel are screened to obtain a set of candidates. For each candidate in the set of candidates, a personnel matching score is calculated. The personnel matching score is obtained by multiplying the standardized skill scores of the candidate in each preset skill dimension by the corresponding dimension weight coefficient and then summing the results. The candidate set is sorted according to the personnel matching score. The candidates corresponding to the number of recommended persons in the sorting result are selected to form a recommended personnel set, and the recommendation result is output. The recommendation result includes the personnel identification information and personnel matching score of each recommended person in the recommended personnel set. For each member of a work group, the training difference for each preset skill dimension is calculated based on preset skill thresholds and standardized skill scores, and the training priority is calculated. Based on the training difference, a set of preset skill dimensions to be pushed is determined, and training resources corresponding to the set of preset skill dimensions to be pushed are retrieved from the training resource library to generate a training resource push result. The training resource push result includes personnel identification information, the set of preset skill dimensions to be pushed, the training priority, and the corresponding training resource identification list.

[0028] A smart AI-powered personnel skills profiling management system for work team management includes: The data acquisition and preprocessing module is used to collect and preprocess multi-source business data related to the skills of team members. The feature engineering module is used to collect and classify the preprocessed multi-source business data to obtain safety training data, hidden danger investigation data, violation behavior data and innovation and improvement data, and extract safety training features, hidden danger investigation features, violation behavior features and innovation and improvement features respectively to obtain the skill feature vector corresponding to each team member. The skill quantization module is used to input skill feature vectors into a preset multi-dimensional skill quantization model to generate initial quantization scores for each preset skill dimension. The scoring correction and normalization module is used to perform scoring correction and normalization processing on the initial quantitative scores of each preset skill dimension, and generate standardized skill scores for each preset skill dimension. The skills profile generation module is used to generate skills profile data for team members based on standardized skills scores on each preset skills dimension. The visualization module is used to generate skill profile visualization results based on the skill profile data of the team members; The profile application module is used to perform profile application processing based on the skill profile data of the team members when a team management business request is received. The profile application processing includes matching personnel according to job or task skill requirements and outputting recommendation results, and generating training resource push results. Example

[0029] This study uses a maintenance and repair team from a large energy company as an example. This team, consisting of eighteen members, is responsible for equipment inspection, defect handling, on-site troubleshooting, and emergency response. Daily management involves the training management system recording training courses and exam results, the hazard identification system recording hazard registration and rectification status, the safety supervision system recording violations and rectification, and the innovation management system recording rationalization suggestions and innovation proposals. For some time, the team's personnel allocation relied heavily on the team leader's experience, and training was primarily distributed uniformly. This resulted in problems such as "fragmented and difficult-to-connect data, subjective skills assessments, unstable job assignments, and inaccurate training delivery." Specifically, there were significant differences in personnel matching for the same position or task when different team leaders were on duty, leading to large fluctuations in the first-time pass rate; hazard and violation rectification were delayed, resulting in "forgetting to follow up during busy periods and repeating training during slow periods"; while the training content was rich, it was weakly related to personnel weaknesses, resulting in little improvement in exam scores after training and low utilization of training resources.

[0030] When applying this invention in this scenario, the first step is to uniformly process multi-source business data through data acquisition and preprocessing. The system retrieves personnel identification information, time information, training courses, exam scores, and exam time fields from the training management system; hazard registration, validity judgment, rectification status, and rectification completion time fields from the hazard investigation system; violation occurrence time, violation type, processing result, and rectification status fields from the safety supervision system; and rationalization suggestions, innovation proposals, submission time, and review status fields from the innovation management system. Since personnel identification in various systems exists in multiple forms such as employee ID, account, and mobile phone number, the preprocessing stage performs unified processing of personnel identification, mapping different identifications to the employee ID field; simultaneously, it performs field standardization, unifying different expressions such as "completed / completed / closed loop" into a standard value for the rectification status field; then, it performs time alignment, uniformly mapping the time granularity of different systems to statistical period information, ensuring that the statistical period start time field and the statistical period end time field are consistent and traceable; finally, it performs data cleaning, removing duplicate records and records missing key fields, and performing rule-based correction of outliers. Through the above processing, multi-source business data can be stably collected by team member identification within the same statistical period, providing a consistent data input standard for subsequent skill calculations, and solving the problem from the source that existing technologies "have difficulty in directly linking cross-system data, and manual aggregation is time-consuming and prone to errors".

[0031] After preprocessing, the system performs feature engineering on the preprocessed multi-source business data to obtain the skill feature vector corresponding to each team member. Using the team member identifier as the association key, the system aggregates and divides the data into safety training data, hazard identification data, violation data, and innovation and improvement data. Subsequently, safety training features are extracted from the safety training data, including the number of safety exams, the average score of the last N safety exams, and the score of the most recent safety exam; hazard identification features are extracted from the hazard identification data, including the number of individual hazard registrations, the number of valid hazard registrations, the total number of hazards rectified by the individual, and the number of hazards rectified by the individual, and the individual hazard registration rate, valid hazard registration rate, and individual hazard rectification rate are calculated; violation data is extracted from the violation data, including the number of individual violations, the total number of violations rectified by the individual, and the number of violations rectified by the individual, and the individual violation rate and individual violation rectification rate are calculated; innovation and improvement features are extracted from the innovation and improvement data, including the number of rationalization suggestions, the number of innovation proposals, and the number of approved proposals, and the number of proposals, proposal submission rate, and proposal adoption rate are calculated. Finally, the characteristics of safety training, hazard identification, violations, and innovation and improvement are vectorized, encoded, and concatenated to form a skill feature vector. Through the skill feature vector, the originally scattered and incomparable recorded data is transformed into a unified and computable expression, solving the problem that existing technologies "cannot characterize multidimensional capabilities with only a single indicator or scattered statistics."

[0032] In the quantitative assessment phase, the system inputs skill feature vectors into a pre-defined multi-dimensional skill quantification model, performs quantification calculations on each pre-defined skill dimension of the pre-defined skill dimension set, and outputs an initial quantification score. The pre-defined multi-dimensional skill quantification model includes sub-models for safety awareness, hazard identification capability, violation prevention capability, and innovation and improvement capability. Each sub-model uses calculation rules with consistent weight coefficient constraints and indicator calibers to ensure that the initial quantification scores within the same statistical period have a uniform scale. Subsequently, in the score correction and normalization phase, the system performs correction processing based on the team's statistical benchmark for the initial quantification scores of each pre-defined skill dimension and completes normalization processing, outputting standardized skill scores for each pre-defined skill dimension. Because this team experiences situations where "the base number of violations increases in certain periods due to phased changes in task types," the standardized skill scores can eliminate distribution drift between statistical periods, making the comparison between team members more stable and solving the problem of "incomparable scores for different periods and different personnel" in existing technologies.

[0033] After generating standardized skill scores, the system generates skill profile data for team members based on these scores across various preset skill dimensions, and then creates a visual representation of these skill profiles. In this visualization, radar charts display the score distribution for safety awareness, hazard identification capabilities, violation prevention capabilities, and innovation and improvement capabilities within the same statistical period, while trend charts show the score changes across multiple statistical periods. When viewing the skill profile visualization, team leaders can intuitively see if a person has a high safety awareness score but a low hazard identification capability score, or if a person's violation prevention capability score fluctuates greatly while their innovation and improvement capability score remains stable, thus transforming "experience-based judgment" into "data-driven evidence."

[0034] The most critical implementation step in this embodiment is the profile application processing. When a work team receives a work team management request, such as when two people need to perform a "high-risk operation + on-site defect elimination" task for a maintenance window, the job or task skill requirements are parsed into skill thresholds, dimension weight coefficients, and recommended number of people under a preset skill dimension set. The system extracts the standardized skill scores of each person under the corresponding statistical period information from the work team's personnel skill profile data. First, it filters the candidate personnel set based on the skill thresholds, then calculates the personnel matching score and outputs the recommended personnel set and recommendation results. At the same time, the system calculates the training difference and training priority for each person in the work team, forms a preset skill dimension set to be pushed based on the training difference, and retrieves the corresponding training resources from the training resource library to generate training resource push results. In this way, work assignment and training run in a closed loop on the same set of work team personnel skill profile data, solving the problems of "disconnect between evaluation results and business applications, and mismatch between training push and personnel shortcomings" in existing technologies.

[0035] To verify the beneficial effects, this embodiment compares two consecutive statistical periods: two months before and two months after the implementation. During both periods, the workload and personnel size of the work teams remained largely consistent. Before implementation, the work teams experienced a high number of daily task adjustments and significant fluctuations in the first-time pass rate, primarily due to personnel matching discrepancies and insufficient follow-up on potential issues. After implementation, personnel matching was driven by the candidate pool and matching scores, while training resource delivery was driven by training differences and training priorities. Under the same workload, the work teams exhibited more stable delivery quality and lower compliance risks.

[0036] Table 1 Comparison of Key Indicators for Work Teams Before and After the Implementation of the Image Processing Application Table 1 Comparison of Key Indicators for Work Teams Before and After the Image Processing Application Goes Online

[0037] As shown in Table 1, with the number of team members remaining at 18 and the total number of team management business requests slightly changing from 386 to 402, the average time for personnel matching decreased from 14.6 minutes / time to 5.2 minutes / time, a reduction of 9.4 minutes / time, or approximately 64.4%. This indicates that personnel matching based on team member skill profile data can significantly reduce the cost of manual comparison and communication. At the same time, the first-time pass rate of tasks increased from 86.1% to 93.4%, an increase of 7.3 percentage points, indicating that the matching degree between the recommended personnel set and the job or task skill requirements has improved, thereby improving the quality and stability of task delivery.

[0038] Regarding compliance and risk control, the total number of individual violations decreased from 27 to 15 during the statistical period, a reduction of 12 violations, or approximately 44.4%. The percentage of violations rectified by individuals that were completed increased from 78.0% to 92.0%, an increase of 14.0 percentage points. This indicates that the profiling application has a stronger process constraint and tracking effect on the violation rectification loop, thus suppressing the occurrence of violations and delays in rectification. In terms of potential hazards, the total number of valid hazard registrations increased from 112 to 139, an increase of 27, or approximately 24.1%. The percentage of hazard rectification completed by individuals increased from 81.5% to 94.1%, an increase of 12.6 percentage points. This demonstrates that standardized skill scores related to hazard identification capabilities and the visualization results of skill profiles can promote the simultaneous improvement of "effective hazard discovery" and "rectification loop," avoiding the deviation of only increasing the number of registrations without improving effectiveness or rectification rate.

[0039] Regarding training effectiveness, the number of people reached by the training resources was 18 in both statistical periods, indicating consistent coverage. However, the average score of the safety exams conducted after the training resources were delivered increased from 2.1 points to 6.8 points, an increase of 4.7 points, or approximately 223.8%. This demonstrates that the training resources delivered more accurately address the shortcomings reflected in the preset set of skills to be delivered, thereby significantly improving the training conversion rate and verifiability within the same coverage area.

[0040] In summary, without changing the size of the work team and keeping the total number of business requests basically the same, this invention generates comparable standardized skill scores by preprocessing multi-source business data, constructing skill feature vectors, outputting multi-dimensional skill quantification models, and normalizing and correcting scores. It also uses the skill profile data of work team members to support personnel matching and training resource delivery, achieving quantifiable beneficial effects such as a significant decrease in efficiency indicators, a significant increase in quality indicators, simultaneous improvement in compliance indicators, simultaneous improvement in risk governance indicators, and a significant enhancement in training effectiveness.

[0041] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A smart AI-powered personnel skill profiling management method for team management, characterized in that, Includes the following steps: Collect and preprocess multi-source business data related to the skills of team members; Feature engineering is performed on the preprocessed multi-source business data to obtain the skill feature vector corresponding to each team member. The skill feature vectors are input into a preset multidimensional skill quantification model to quantify the ability level of team members in each preset skill dimension and generate initial quantification scores for each skill dimension. The initial quantitative scores for each skill dimension are corrected and normalized to generate standardized skill scores for each skill dimension. Skill profile data for team members is generated based on standardized skill scores across various skill dimensions. Generate a visualization of skill profiles based on the skill profile data of team members; When a team management business request is received, the system performs profile application processing based on the team members' skill profile data and generates training resource push results.

2. The intelligent AI personnel skill profile management method for team management according to claim 1, characterized in that, The multi-source business data includes personnel identification information, time information, training data, hazard data, violation data, and innovation data. The preprocessing includes data cleaning, field standardization, time alignment, and unified processing of personnel identification.

3. The intelligent AI personnel skill profile management method for team management according to claim 1, characterized in that, The acquisition of the skill feature vector specifically includes: The pre-processed multi-source business data is collected and divided according to the identification of team members to obtain safety training data, hidden danger investigation data, violation data and innovation and improvement data. Extracting safety training features from safety training data; Extracting hazard identification features from hazard identification data; Extracting characteristics of violations based on violation data; Based on innovation and improvement data, extract innovation and improvement features; The characteristics of safety training, hazard investigation, violations, and innovation and improvement are vectorized, encoded, and concatenated to obtain the skill characteristic vector corresponding to each team member.

4. The intelligent AI personnel skill profiling management method for team management according to claim 1, characterized in that, The generation of the initial quantitative scores for each skill dimension specifically includes: The skill feature vector is input into a preset multidimensional skill quantification model, which includes a safety awareness sub-model, a hidden danger investigation capability sub-model, an anti-violation capability sub-model, and an innovation and improvement capability sub-model. In the safety awareness sub-model, by setting a first weighting coefficient and a second weighting coefficient, and their sum being one, an initial quantitative score for safety awareness is calculated based on the average score of the safety exam and the individual violation rate. The individual violation rate is obtained by the ratio of the number of individual violations to the historical average number of violations of the work group, with an upper limit constraint. The initial quantitative score for safety awareness is obtained by weighting the average score of the safety exam with the first weighting coefficient, and weighting the result of subtracting the individual violation rate from one and multiplying it by one hundred points with the second weighting coefficient, and then adding the two weighted results. In the sub-model of hazard identification capability, by setting a first weight coefficient, a second weight coefficient, and a third weight coefficient, and summing them to one, the initial quantitative score of hazard identification capability is calculated based on the individual hazard registration rate, the effective hazard registration rate, and the individual hazard rectification rate. The individual hazard registration rate is obtained by the ratio of the number of individual hazard registrations to the historical average number of hazard registrations of the work group, with an upper limit constraint. The effective hazard registration rate is obtained by the ratio of the number of effective hazard registrations to the number of individual hazard registrations. The individual hazard rectification rate is obtained by the ratio of the number of hazards rectified by the individual to the total number of hazards rectified by the individual. The initial quantitative score of hazard identification capability is obtained by weighting the individual hazard registration rate, the effective hazard registration rate, and the individual hazard rectification rate according to the first weight coefficient, the second weight coefficient, and the third weight coefficient, and then multiplying by one hundred points. In the anti-violation capability sub-model, by setting a first weight coefficient and a second weight coefficient, and summing them to one, the initial quantitative score of anti-violation capability is calculated based on the individual violation rate and the individual violation rectification rate. The individual violation rate is obtained by the ratio of the number of individual violations to the historical average number of violations of the team, with an upper limit constraint. The individual violation rectification rate is obtained by the ratio of the number of violations rectified by the individual to the total number of violations rectified by the individual. The initial quantitative score of anti-violation capability is obtained by weighting the result of subtracting the individual violation rate from the first weight coefficient, weighting the individual violation rectification rate by the second weight coefficient, summing the two weighted results, and multiplying by one hundred points. In the innovation and improvement capability sub-model, by setting a first weighting coefficient and a second weighting coefficient, the initial quantitative score of innovation and improvement capability is calculated based on the proposal submission rate and the proposal adoption rate. The proposal submission rate is obtained by the ratio of the number of proposals to a preset benchmark value and subject to an upper limit constraint. The proposal adoption rate is obtained by the ratio of the number of approved proposals to the total number of proposals. The initial quantitative score of innovation and improvement capability is obtained by weighting the proposal submission rate and the proposal adoption rate according to the first weighting coefficient and the second weighting coefficient and multiplying by 100.

5. The intelligent AI personnel skill profile management method for team management according to claim 1, characterized in that, The generation of standardized skill scores for each skill dimension specifically includes: The initial quantitative scores for each preset skill dimension are grouped according to the team member identifier and the preset skill dimension identifier; For each preset skill dimension, the group statistical baseline mean and the group statistical baseline standard deviation are calculated based on the initial quantitative scores after grouping. The group statistical baseline mean is the sum of the initial quantitative scores of all members in the group on the preset skill dimension divided by the total number of members in the group. The group statistical baseline standard deviation is the square root of the sum of the squares of the differences between the initial quantitative scores of all members in the group on the preset skill dimension and the group statistical baseline mean divided by the total number of members in the group. For each preset skill dimension, the initial quantitative score of each team member on the preset skill dimension is corrected based on the team statistical benchmark to obtain the corrected score. The corrected score is obtained by subtracting the mean of the team statistical benchmark from the initial quantitative score and then dividing by the standard deviation of the team statistical benchmark. For each preset skill dimension, the corrected score is input into the standard normal cumulative distribution function to obtain the cumulative probability value, and the cumulative probability value is multiplied by 100 to obtain the normalized score; Standardized skill scores are obtained by applying score interval constraint processing to the normalized scores. The score interval constraint processing refers to limiting normalized scores less than zero to zero and normalized scores greater than one hundred to one hundred, thus forming standardized skill scores for each preset skill dimension.

6. The intelligent AI personnel skill profiling management method for team management according to claim 1, characterized in that, The generation of the skill profile data for the team members specifically includes: The standardized skill scores are aggregated using the team member identifier as an index to form a set of standardized skill scores for each preset skill dimension corresponding to each team member identifier. For each shift's personnel, generate personnel identification information, which includes an employee ID field; For each shift, personnel are identified, and statistical period information is generated, which includes a statistical period start time field and a statistical period end time field. For each shift's personnel, a standardized set of skill scores is combined according to a preset set of skill dimensions to form a skill score vector. By concatenating personnel identification information, statistical period information, and skill score vectors, a skill profile data of team members is formed.

7. The intelligent AI personnel skill profile management method for team management according to claim 1, characterized in that, The generation of the skill profile visualization results specifically includes: Standardize skill scores in the skill profile data of team members and create an index based on team member identification, statistical period information, and preset skill dimension identification; For each shift's personnel identification and statistical period information, the identifiers of each preset skill dimension are fixedly sorted according to the preset skill dimension set, and the circumferential angle is equally divided based on the number of preset skill dimensions to obtain the angle value corresponding to each preset skill dimension identifier. The standardized skill score is used as the radial value and paired with the angle value to generate a polar coordinate point set, and the polar coordinate point set is encapsulated as radar chart data. For each shift member identifier and each preset skill dimension identifier, obtain a statistical periodic sequence arranged in chronological order, and use the statistical periodic sequence as the horizontal axis and the standardized skill score of the corresponding preset skill dimension as the vertical axis to generate a time series point set, and encapsulate the time series point set into trend chart data; By linking radar chart data and trend chart data with personnel identification information and statistical period information respectively, a skill profile visualization result is formed.

8. The intelligent AI personnel skill profiling management method for team management according to claim 1, characterized in that, The generation of the training resource push results specifically includes: The system receives a team management business request and parses it to obtain the job or task identifier and the job or task skill requirements. The job or task skill requirements include the skill threshold, dimension weight coefficient, and recommended number of people corresponding to each preset skill dimension in the preset skill dimension set. The sum of the dimension weight coefficients is one. Based on the skill profile data of team members, the standardized skill scores of each team member in each preset skill dimension are extracted under the statistical period information corresponding to the job or task identifier, and the standardized skill scores are indexed according to the team member identifier and the preset skill dimension identifier. Based on preset skill thresholds, candidate personnel are screened to obtain a set of candidates. For each candidate in the set of candidates, a personnel matching score is calculated. The personnel matching score is obtained by multiplying the standardized skill scores of the candidate in each preset skill dimension by the corresponding dimension weight coefficient and then summing the results. The candidate set is sorted according to the personnel matching score. The candidates corresponding to the number of recommended persons in the sorting result are selected to form a recommended personnel set, and the recommendation result is output. The recommendation result includes the personnel identification information and personnel matching score of each recommended person in the recommended personnel set. For each member of a work group, the training difference for each preset skill dimension is calculated based on preset skill thresholds and standardized skill scores, and the training priority is calculated. Based on the training difference, a set of preset skill dimensions to be pushed is determined, and training resources corresponding to the set of preset skill dimensions to be pushed are retrieved from the training resource library to generate a training resource push result. The training resource push result includes personnel identification information, the set of preset skill dimensions to be pushed, the training priority, and the corresponding training resource identification list.

9. A smart AI personnel skill profile management system for team management, comprising the smart AI personnel skill profile management method for team management as described in any one of claims 1 to 9, characterized in that, include: The data acquisition and preprocessing module is used to collect and preprocess multi-source business data related to the skills of team members. The feature engineering module is used to collect and classify the preprocessed multi-source business data to obtain safety training data, hidden danger investigation data, violation behavior data and innovation and improvement data, and extract safety training features, hidden danger investigation features, violation behavior features and innovation and improvement features respectively to obtain the skill feature vector corresponding to each team member. The skill quantization module is used to input skill feature vectors into a preset multi-dimensional skill quantization model to generate initial quantization scores for each preset skill dimension. The scoring correction and normalization module is used to perform scoring correction and normalization processing on the initial quantitative scores of each preset skill dimension, and generate standardized skill scores for each preset skill dimension. The skills profile generation module is used to generate skills profile data for team members based on standardized skills scores on each preset skills dimension. The visualization module is used to generate skill profile visualization results based on the skill profile data of the team members; The profile application module is used to perform profile application processing based on the skill profile data of the team members when a team management business request is received. The profile application processing includes matching personnel according to job or task skill requirements and outputting recommendation results, and generating training resource push results.

Images