Detection personnel occupational portrait dynamic modeling method based on multi-dimensional data analysis

Abstract

The invention discloses a detection personnel occupational portrait dynamic modeling method based on multi-dimensional data analysis, and relates to the technical field of artificial intelligence and human resource management crossing. The detection personnel occupational portrait dynamic modeling method based on multi-dimensional data analysis specifically comprises the following steps: collecting multi-modal operation data of detection personnel; establishing a detection personnel capability quantitative evaluation model, and determining occupational roles of detection personnel; determining feature vectors and feature factors meeting vocational ability requirements, and obtaining vocational levels of the detection personnel; the detection tasks are classified to obtain demand vectors of different task types, a quantitative relation is established, and a visual occupational portrait is generated; establishing a cosine similarity matching model of the capability vector and the demand vector; compared with a traditional ordering method, the scheme has the advantages that the task matching accuracy is improved, the detection period is shortened, the laboratory human resource configuration efficiency is optimized, and the technical blank of human resource management of artificial intelligence in the detection field is filled.

Description

Technical Field

[0001] The present invention relates to the cross - technical field of artificial intelligence and human resource management, and specifically to a dynamic modeling method for the professional portraits of inspectors based on multi - dimensional data analysis. Background Art

[0002] Currently, the inspection and testing industry is in a crucial stage of intelligent transformation. The existing evaluation methods for inspectors have revealed three limitations: 1) First, although the existing ability evaluation system has constructed an ability map model, it generally uses a static label system and has not broken through the framework of "single - dimensional evaluation of technical ability", making it difficult to comprehensively reflect the dynamic evolution law of the professional portraits of inspectors in multi - task scenarios; 2) Second, in the existing laboratory management, the solutions involving intelligent task allocation mostly focus on the scheduling of instrument and equipment resources, while there are few solutions for the matching of inspectors - inspection tasks, and there has not yet been an intelligent task assignment solution based on professional portraits; 3) Finally, there is an over - matching error in the task distribution link of existing inspection agencies. The root cause lies in the lack of a multi - dimensional correlation mathematical model for the technical structure, inspection characteristics, and task demand matrix of the personnel in the inspection structure. More notably, although digital twin technology has been widely used in the field of equipment management, there are still three technical problems when expanding it to the dynamic modeling of personnel freezing ability: ① Normalization processing of multi - modal data structures; ② Quantitative characterization of multi - task scenarios in professional portraits; ③ Real - time two - way optimization mechanism for task requirements and personnel portraits. Summary of the Invention

[0003] Aiming at the deficiencies of the existing technology, the present invention provides a dynamic modeling method for the professional portraits of inspectors based on multi - dimensional data analysis, which solves the problem that the existing evaluation criteria for inspectors' abilities are single and it is difficult to comprehensively reflect the comprehensive abilities of inspectors.

[0004] To achieve the above objectives, the present invention is realized through the following technical solutions: A dynamic modeling method for the professional portraits of inspectors based on multi - dimensional data analysis, including the following steps: S1. Real - time collect multi - modal operation data of inspectors through digital twin technology; S2. Based on the data types in the multi - modal operation data, use machine learning algorithms to establish a quantitative evaluation model for inspectors' abilities and determine the professional roles of inspectors; S3. Based on the professional roles of inspectors and the quantitative evaluation model for abilities, determine the eigenvectors and eigen - factors that meet the professional ability requirements, and obtain the professional levels of inspectors; S4. Based on the professional roles of inspectors, classify the inspection tasks in the digital laboratory system to obtain demand vectors for different task types, establish a quantitative relationship based on the demand vectors and the ability vectors in the personnel ability evaluation model, and generate a visual professional portrait; S5. Based on the ability vector in the professional portrait of the inspector and the requirement vector of the inspection task, establish a cosine similarity matching model between the two, formulate the principle of intelligent allocation of inspection tasks, and realize the full-cycle intelligent adaptation of inspection tasks and the dynamic display of the professional portrait of inspectors.

[0005] Preferably, the multi-modal operation data of the inspector at least includes inspection report traceability data, instrument and equipment traceability logs, daily collaboration and interaction data of inspectors, and technical ability interaction data of inspectors.

[0006] Preferably, S2 specifically includes the following steps: S2.1 Obtain the workload base of the inspector based on the inspection report traceability data in the multi-modal operation data, obtain the work coordination ability of the inspector based on the inspection report traceability data and the daily collaboration and interaction data of the inspector in the multi-modal operation data, and obtain the work skills of the inspector based on the instrument and equipment traceability logs and the technical ability interaction data of the inspector in the multi-modal operation data; S2.2 Build a quantitative evaluation model for the inspector's ability based on the workload base, work coordination ability, and work skills of the inspector; S2.3 Confirm the professional role of the inspector based on the work skills of the inspector.

[0007] Preferably, S3 specifically includes the following steps: S3.1 Group the inspectors based on their professional roles to obtain inspection skill groups; S3.2 Evaluate based on the eigenvectors of each ability in the quantitative evaluation model of the inspector's ability to obtain the unit vector and characteristic factor that meet the ability index requirements, and perform feature engineering on the unit vector to obtain the ability score corresponding to each unit vector; S3.3 Obtain the scores of each ability index based on the unit vector and characteristic factor required by the quantitative evaluation model of the inspector's ability, take the average of the ability scores of each ability index and combine the preset score range to obtain the professional level of the inspector, and associate the professional level with the professional role.

[0008] Preferably, the feature engineering in S3.2 includes: Feature selection: Use feature selection algorithms (such as correlation analysis, principal component analysis, etc.) to screen out the features that have the greatest impact on ability evaluation; Feature dimensionality reduction: For high-dimensional feature vectors, use dimensionality reduction techniques (such as PCA) to reduce the feature dimension and improve the calculation efficiency; Feature normalization: Normalize the features to ensure that different features are compared on the same scale; Learn to perform feature engineering on unit vectors using machine learning algorithms, establish the relationship between the ability quantification evaluation model and unit vectors, and assign ability scores to each unit vector.

[0009] Preferably, S4 specifically includes the following steps: S4.1 Analyze the content of the detection task using machine learning algorithms, and use the professional roles of the detection personnel as the screening conditions to obtain the task types and demand vectors of the detection task. S4.2 Based on the professional roles of the detection personnel and the task types and demand vectors in the digital laboratory, establish a quantitative evaluation model for the abilities of the detection personnel and generate a visual professional portrait.

[0010] Preferably, S5 specifically includes the following steps: S5.1 Obtain the professional roles of the detection personnel, the unit vectors and feature factors in the ability evaluation model, and obtain the ability vectors in the professional portrait of the detection personnel. S5.2 Based on the ability vectors in the professional portrait of the detection personnel and the demand vectors of the detection tasks in the digital laboratory, establish a cosine similarity matching model, and intelligently assign detection tasks according to the principle of decreasing matching degree, so as to realize the full-cycle intelligent adaptation of detection tasks and the dynamic display of the professional portrait of detection personnel.

[0011] Preferably, the higher the matching degree between the ability vectors in the professional portrait of the detection personnel and the demand vectors of the detection tasks in the digital laboratory, the higher the reliability of the detection personnel to complete the detection tasks; after the cosine similarity matching is completed, establish a commission relationship, and sort according to the size of the feature factors of the detection personnel. The higher the value, the higher the reliability of completing the detection tasks.

[0012] Preferably, evaluate and sort the importance and time requirements of the detection tasks. When there are multiple detection tasks for which the detection personnel have the highest reliability of completion and there are time interferences among multiple detection tasks, according to the sorting of the importance of the detection tasks, give priority to assigning them to the detection personnel with the highest reliability. The detection tasks with the second highest importance are assigned to the detection personnel with the second highest reliability. If the detection tasks adapted by the detection personnel with the second highest reliability also have time interferences, adjust them again according to the above method; the task assignment adjustment method is determined using the weighted sum algorithm. Set weights for the importance of the detection tasks and the reliability of the detection personnel respectively, perform weighted calculations on the ranking numbers of the reliability of the detection personnel for different detection tasks and the ranking numbers of the importance of the detection tasks, and assign the detection task with the highest score to the corresponding detection personnel, and note the results after the professional portrait of the detection personnel.

[0013] Preferably, in order to ensure the real-time matching between the detection task list and the abilities of the detection personnel, a two-way dynamic intelligent linkage mechanism is adopted to update the professional portrait of the detection personnel in real time.

[0014] The present invention provides a dynamic modeling method for the occupational portrait of inspectors based on multi-dimensional data analysis. Compared with the prior art, it has the following beneficial effects: 1. For the dynamic modeling method of the occupational portrait of inspectors based on multi-dimensional data analysis, in order to construct a four-dimensional linked occupational ability evaluation system to comprehensively reflect the ability change trend and comprehensive inspection level of inspectors; this solution uses digital twin technology to collect multi-modal operation data of inspectors in real time, and then uses machine learning algorithms to perform feature engineering processing on the original data, establishes an ability quantification evaluation model based on workload base, work coordination ability and work skills. At the same time, it classifies the inspection task sheets in the digital laboratory system, and establishes a cosine similarity matching model between the demand vectors of different types of inspection task sheets and the feature vectors of the ability quantification evaluation model, thereby formulating a two-way dynamic intelligent dispatching mechanism for inspection task sheets and inspectors. From the portrait of the inspector drawn under this mechanism, generally speaking, this solution improves the accuracy of task matching, shortens the inspection cycle, optimizes the efficiency of laboratory human resource allocation, and fills the technical gap of artificial intelligence in the human resource management of the inspection field.

[0015] 2. For the dynamic modeling method of the occupational portrait of inspectors based on multi-dimensional data analysis, in order to construct a four-dimensional linked occupational ability evaluation system, it analyzes multi-modal operation data, obtains the workload base of inspectors based on the traceability data of inspection reports, and can monitor the progress of inspectors in completing tasks in real time, and collect information such as report timeliness rate, accuracy rate and customer satisfaction rate; obtains the work coordination ability of inspectors based on the traceability data of inspection reports and the daily collaborative interaction data of inspectors, which can reflect the overall coordination ability of inspectors in multi-task scenarios; obtains the work skills of inspectors based on the traceability log of instrument equipment and the technical ability interaction data of inspectors, which can comprehensively reflect the occupational role and comprehensive occupational level of inspectors.

[0016] 3. The dynamic modeling method for the professional portrait of testers based on multi-dimensional data analysis uses vectors in three ability indicators, namely workload base, work coordination ability, and work skills, to evaluate and determine the professional level of testers. For example, the vectors affecting the workload base include completed piecework, delayed piecework, completed order quantity, delayed order quantity, error order quantity, piecework timeliness rate, report accuracy rate, number of customer reminders, and customer satisfaction rate. Among them, the feature vector is the completed piecework, and the piecework timeliness rate, report accuracy rate, and customer satisfaction rate are the influencing factors of the feature factors. The unit vector is the product of the feature vector and the feature factor. An ability quantification evaluation model is established based on the unit vectors and feature factors of each ability indicator to obtain the ability scores of testers, determine the professional levels and professional roles of testers, so as to comprehensively reflect the ability change trend and comprehensive detection level of testers, facilitate the high matching of detection tasks and tester abilities, and improve the work efficiency and quality of the digital laboratory.

[0017] 4. The dynamic modeling method for the professional portrait of testers based on multi-dimensional data analysis uses the score of the ability quantification evaluation model as the basis for defining the professional level of testers. By scoring the ability indicators, the sum of the scores of each ability indicator is obtained as the average score. Based on the preset score range of the abilities of testers in each digital laboratory, the two are coordinated as the professional level score range, so as to provide targeted training for unqualified testers and improve the overall technical ability of the laboratory.

[0018] 5. The dynamic modeling method for the professional portrait of testers based on multi-dimensional data analysis establishes a cosine similarity matching model by obtaining the ability vectors and task requirement vectors of testers to achieve intelligent matching of tasks and personnel. When there are time interferences in multiple tasks, weights are set according to the importance of the detection tasks and the reliability of the testers, and the weighted sum algorithm is used to readjust the task allocation to ensure that important tasks are preferentially assigned to testers with high reliability. At the same time, the allocation results are noted after the professional portrait of the testers for subsequent tracking and evaluation, which improves the rationality and efficiency of task allocation and optimizes resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the step flow of the present invention; Figure 2 It is a specific detection type table diagram of the detection task form completed by the experimentalist of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0021] Referring to Figure 1 - Figure 2 , the present invention discloses a dynamic modeling method for the professional portrait of inspection personnel based on multi-dimensional data analysis, including the following steps: S1. Real-time collect the multi-modal operation data of inspection personnel through digital twin technology; the multi-modal operation data of inspection personnel at least includes inspection report traceability data, instrument and equipment traceability logs, daily cooperation and interaction data of inspection personnel, and technical ability interaction data of inspection personnel, and respectively evaluate the abilities of inspection personnel in three aspects: workload base, work coordination ability, and work skills; S2. Based on the data types in the multi-modal operation data, use machine learning algorithms to establish a quantitative evaluation model for the abilities of inspection personnel and determine the professional roles of inspection personnel; S3. Based on the professional roles of inspection personnel and the quantitative evaluation model of abilities, determine the feature vectors and feature factors that meet the professional ability requirements, and obtain the professional levels of inspection personnel; S4. Based on the professional roles of inspection personnel, classify the inspection tasks in the digital laboratory system to obtain the demand vectors of different task types, establish a quantitative relationship based on the demand vectors and the ability vectors in the personnel ability evaluation model, and generate a visual professional portrait; S5. Based on the ability vectors in the professional portrait of inspection personnel and the demand vectors of inspection tasks, establish a cosine similarity matching model between the two, formulate the principle of intelligent assignment of inspection tasks, and realize the full-cycle intelligent adaptation of inspection tasks and the dynamic display of the professional portrait of inspection personnel.

[0022] In this solution, in order to construct a four-dimensional linked professional ability evaluation system to comprehensively reflect the ability change trend and comprehensive detection level of inspectors; this solution uses digital twin technology to collect multi-modal operation data of inspectors in real time, and then uses machine learning algorithms to perform feature engineering processing on the original data, establishing an ability quantification evaluation model based on the workload base, work coordination ability, and work skills. At the same time, it classifies the inspection task sheets in the digital laboratory system, and establishes a cosine similarity matching model between the demand vectors of different types of inspection task sheets and the feature vectors of the ability quantification evaluation model, thereby formulating a two-way dynamic intelligent dispatching mechanism for inspection task sheets and inspectors. From the portrait drawn under this mechanism, generally speaking, this solution improves the accuracy of task matching, shortens the inspection cycle, optimizes the efficiency of laboratory human resource allocation, and fills the technical gap of artificial intelligence in the human resource management of the inspection field.

[0023] In this embodiment, S2 specifically includes the following steps: S2.1 Obtain the workload base of the inspector based on the inspection report traceability data in the multi-modal operation data, obtain the work coordination ability of the inspector based on the inspection report traceability data and the daily collaboration interaction data of the inspector, and obtain the work skills of the inspector based on the instrument equipment traceability log and the technical ability interaction data of the inspector. S2.2 Construct an ability quantification evaluation model for the inspector based on the workload base, work coordination ability, and work skills of the inspector. S2.3 Confirm the professional role of the inspector based on the work skills of the inspector.

[0024] In this solution, in order to construct a four-dimensional linked professional ability evaluation system, analyze the multi-modal operation data, obtain the workload base of the inspector based on the inspection report traceability data, and can monitor the progress of the inspector's task completion in real time, collect information such as the report timely rate, correct rate, and customer satisfaction rate; obtain the work coordination ability of the inspector based on the inspection report traceability data and the daily collaboration interaction data of the inspector, which can reflect the overall coordination ability of the inspector in multi-task scenarios; obtain the work skills of the inspector based on the instrument equipment traceability log and the technical ability interaction data of the inspector, which can comprehensively reflect the professional role and comprehensive professional level of the inspector.

[0025] In this embodiment, S3 specifically includes the following steps: S3.1 Group the inspectors based on their professional roles to obtain inspection skill groups. S3.2 Based on the eigenvectors of each ability in the detection personnel ability quantification evaluation model, conduct an evaluation to obtain the unit vectors and characteristic factors that meet the requirements of the ability indicators. Perform feature engineering on the unit vectors to obtain the ability scores corresponding to each unit vector. S3.3 Based on the unit vectors and characteristic factors required by the detection personnel ability quantification evaluation model, obtain the scores of each ability indicator. Take the average of the ability scores of each ability indicator and combine it with the preset score range to obtain the professional level of the detection personnel, and associate the professional level with the professional role.

[0026] In this solution, to evaluate the professional level of detection personnel, vectors in three ability indicators of workload base, work coordination ability, and work skills are used for evaluation. For example, the vectors affecting the workload base include the number of piecework completed, delayed piecework, number of orders completed, delayed orders, number of wrong orders, piecework timeliness rate, report accuracy rate, number of customer reminders, and customer satisfaction rate. Among them, the eigenvector is the number of piecework completed, and the piecework timeliness rate, report accuracy rate, and customer satisfaction rate are the influencing factors of the characteristic factors. The unit vector is the product of the eigenvector and the characteristic factor. Based on the unit vectors and characteristic factors of each ability indicator, establish an ability quantification evaluation model to obtain the ability scores of the detection personnel, determine the professional level and professional role of the detection personnel, so as to comprehensively reflect the ability change trend and comprehensive detection level of the detection personnel, so as to highly match the detection tasks with the abilities of the detection personnel and improve the work efficiency and quality of the digital laboratory.

[0027] Feature extraction includes: Extraction of workload base eigenvectors: Extract key indicators such as the number of piecework completed, delayed piecework, number of orders completed, delayed orders, and number of wrong orders from the traceability data of the detection report. Calculate derivative indicators such as piecework timeliness rate, report accuracy rate, and customer satisfaction rate. Combine these indicators into the workload base eigenvector W. Extraction of work coordination ability eigenvectors: Extract the overall coordination information in the multi-task scenario from the traceability data of the detection report and the daily collaboration interaction data. Analyze the time allocation, task switching frequency, collaboration efficiency, etc. of the detection personnel in multi-task processing. Convert this information into the work coordination ability eigenvector C, including dimensions such as task switching efficiency and collaboration satisfaction. Extraction of work skill eigenvectors: Extract information such as the operation skills, equipment proficiency, and problem-solving ability of the detection personnel from the traceability log of the instrument and equipment and the technical ability interaction data. Analyze the performance of the detection personnel in handling complex tasks and dealing with emergencies. Convert this information into a working base feature vector T, including dimensions such as operation accuracy rate, equipment failure handling speed, problem-solving rate, etc. Feature factor determination includes: Unit vector extraction: Extract key unit vectors from each feature vector, such as the number of completed pieces and piecework timeliness rate in the workload base, task switching efficiency in work coordination ability, operation accuracy rate in work skills, etc. Feature factor determination: Assign feature factors to each unit vector, and these factors reflect the contribution degree of the unit vector to the overall ability evaluation. Feature engineering includes: Feature selection: Use feature selection algorithms (such as correlation analysis, principal component analysis, etc.) to screen out the features that have the greatest impact on ability evaluation. Feature dimensionality reduction: For high-dimensional feature vectors, use dimensionality reduction techniques (such as PCA) to reduce the feature dimensions and improve the calculation efficiency. Feature normalization: Normalize the features to ensure that different features are compared on the same scale.

[0028] In this solution, the more feature vectors W, C, and T in the ability quantification evaluation model of inspectors, the more scoring items. To simplify the evaluation model, extract the unit vectors X, Y, Z, H, J, K, L and feature factors b, c, d, e, m, n from the feature vectors. Among them, the unit vector is the demand vector x of the inspection task list i Sub-item type, feature factor a i is the sub-item coefficient, which is specifically as follows: ; ; ; ; Among them: E represents the ability score, W represents the workload base feature vector, C represents the work coordination ability feature vector, T represents the work base feature vector, X represents the workload base unit vector, Y and Z represent the work coordination ability unit vectors, H, J, K, L represent the work base unit vectors, and x i represents the demand vector, and a i , b, c, d, e, m, n all represent feature factors.

[0029] Use machine learning algorithms to learn the feature engineering of unit vectors, establish the relationship between the ability quantification evaluation model and unit vectors, and assign ability scores to each unit vector.

[0030] In this solution, the score of the ability quantification evaluation model is used as the basis for defining the professional level of the inspectors. By scoring the ability indicators, the sum of the scores of each ability indicator is obtained as the average score. Based on the preset score range of the ability of the inspectors in each digital laboratory, the two are coordinated as the professional level score range, so as to provide targeted training for unqualified inspectors and improve the overall technical ability of the laboratory.

[0031] In this embodiment, S4 specifically includes the following steps: S4.1 Use machine learning algorithms to analyze the content of the inspection tasks, and use the professional roles of the inspectors as the screening conditions to obtain the task types and demand vectors of the inspection tasks. S4.2 Based on the professional roles of the inspectors and the task types and demand vectors in the digital laboratory, establish a quantitative model for evaluating the ability of the inspectors and generate a visual professional portrait.

[0032] In this solution, in order to improve the matching degree between the inspection task sheets and the inspectors, machine learning algorithms are used to classify the content of the inspection tasks, and at the same time, taking the professional perspective of the inspectors as the screening condition, the demand vector x of the inspection task sheets is obtained. i and the task types, and calculate the cosine similarity with the workload base feature vector W in the ability evaluation model. When the similarity of one of the polynomials in the result is 1, the task relationship can be established; when the inspector completes the task, the ability can be obtained, and then a visual professional portrait can be generated. ; Among them, sim(x i , W) is the function for calculating the cosine similarity of W, and I is the total number of the demand vector x i .

[0033] In this embodiment, S5 specifically includes the following steps: S5.1 Obtain the professional roles of the inspectors and the unit vectors and feature factors in the ability evaluation model to obtain the ability vectors in the professional portraits of the inspectors. S5.2 Based on the ability vectors in the professional portraits of the inspectors and the demand vectors of the inspection tasks in the digital laboratory, establish a cosine similarity matching model, and intelligently assign inspection tasks according to the principle of decreasing matching degree, so as to realize the full-cycle intelligent adaptation of inspection tasks and the dynamic display of the professional portraits of inspectors. The higher the matching degree between the ability vectors in the professional portraits of the inspectors and the demand vectors of the inspection tasks in the digital laboratory, the higher the reliability of the inspectors to complete the inspection tasks; after the cosine similarity matching is completed, a commission relationship is established and sorted according to the size of the feature factors of the inspectors. The higher the value, the higher the reliability of completing the inspection tasks.

[0034] Evaluate and rank the importance and time requirements of inspection tasks. When the reliability of a tester for multiple inspection tasks is the highest and there are time interferences among multiple inspection tasks, according to the ranking of the importance of inspection tasks, assign them to the tester with the highest reliability first, and assign the inspection tasks with the second highest importance to the tester with the second highest reliability. If there are also time interferences in the inspection tasks suitable for the tester with the second highest reliability, adjust them again according to the above method; the task assignment adjustment method is determined by using the weighted sum algorithm. Set weights for the importance of inspection tasks and the reliability of testers respectively, perform weighted calculations on the ranking numbers of the reliability of testers for different inspection tasks and the ranking numbers of the importance of inspection tasks, assign the inspection task with the highest score to the corresponding tester, and note the results after the professional portrait of the tester.

[0035] Parameter settings: Weight ω1 of the importance of inspection tasks: This weight reflects the importance degree of inspection tasks in the overall work process; it can be set according to factors such as the urgency of tasks, the impact on business, and customer requirements; for example, it can be set to 0.6 (indicating that the importance accounts for 60% of the weight). Weight ω2 of the reliability of testers: This weight reflects the reliability degree of testers in completing inspection tasks; it can be determined according to the cosine similarity between the ability vector in the professional portrait of the tester and the requirement vector of the inspection task; for example, it can be set to 0.4 (indicating that the reliability accounts for 40% of the weight). Ranking number R1 of the importance of inspection tasks: Sort all inspection tasks to be assigned according to their importance and assign ranking numbers; the smaller the number, the more important the task.

[0036] Ranking number R2 of the reliability of testers for different inspection tasks: Sort the reliability of each tester for different inspection tasks and assign ranking numbers; the smaller the number, the higher the reliability.

[0037] For each tester i and each inspection task j, we can calculate a weighted score S ij , which is used to determine the priority of task assignment. The calculation formula of the weighted score is as follows: ; Where: N1 is the total number of inspection tasks.

[0038] R 1，j is the ranking number of the importance of inspection task j.

[0039] N2 is the total number of rankings of the reliability of testers for all inspection tasks (for a specific tester i, it may be the number of tasks he participates in the evaluation).

[0040] R 2，ij is the reliability ranking serial number of tester i for test task j.

[0041] is to convert the importance ranking serial number of the task into a positive score. The smaller the serial number (the more important the task), the higher the score.

[0042] is to convert the reliability ranking serial number of the tester into a positive score. The smaller the serial number (the higher the reliability), the higher the score.

[0043] Task assignment decision: For each test task, select the tester with the highest weighted score S ij If there are multiple testers with the same weighted score for the same task, other factors (such as the current workload and historical performance of the tester) can be further considered for decision-making. The assignment result is noted after the tester's professional portrait for subsequent tracking and evaluation.

[0044] This solution realizes the technical effects of intelligent assignment and dynamic adaptation of test tasks. By obtaining the tester ability vector and task requirement vector, a cosine similarity matching model is established to achieve intelligent matching of tasks and personnel. When there are time interferences among multiple tasks, weights are set according to the importance of the test tasks and the reliability of the testers, and the weighted sum algorithm is used to readjust the task assignment to ensure that important tasks are preferentially assigned to testers with high reliability. At the same time, the assignment result is noted after the tester's professional portrait for subsequent tracking and evaluation, improving the rationality and efficiency of task assignment and optimizing resource allocation.

[0045] The tester's professional portrait collects the multi-modal operation data of the tester in real time through digital twin technology. And in order to ensure the real-time matching of the test task sheet and the tester's ability, a two-way dynamic intelligent linkage mechanism is adopted to update the tester's professional portrait in real time.

[0046] This invention breaks the static system of "single-dimensional evaluation of technical ability", establishes the dynamic evolution law of the professional portrait in a multi-task scenario; proposes an intelligent dispatching algorithm based on the dynamic professional portrait of the tester; and establishes a quantitative evaluation model of the tester's ability based on multi-dimensional data, which is more objective, fair and real than the previous expert evaluation method.

[0047] Specific implementation case 1: In order to comprehensively reflect the changing trend of the ability and comprehensive testing level of the testing personnel, the multimodal operation data of the testing personnel were collected in the digital laboratory system, mainly including the data on workload base, work coordination ability and work skills. The workload base mainly includes completed piecework, delayed piecework, number of completed orders, number of delayed orders, number of error orders, piecework timeliness rate, correctness rate, number of customer reminders, and customer satisfaction rate. The work coordination ability mainly includes the number of business trip days and the number of business trip orders. The work skills mainly include the number of reviews, review parameters, value traceability confirmation, number of standard updates, and external comparisons. The following is an example of the experimenter using Zhou as an example to calculate the workload base. The piece rate he completed was 564.1. In the table assignment, the occupational role was 1 point for 12.5 working hours for the experimenter. However, during the experiment, some inspection tasks had timeliness issues (piece rate timeliness was 78.02%), and it was found that there were 124.0 delayed working hours. It was necessary to convert them. The table found that the characteristic factor of the inspection personnel was 0.7. Then Zhou's score in terms of workload base was: Workload base = ((completed piece rate - delayed piece rate) + delayed piece rate × characteristic factor) × customer satisfaction × accuracy rate = ((564.1-124.0) + 124.0 × 0.7) × 100% × 100% / 12.5 = 42.2 points.

[0048] Specific implementation case 2: In order to realize the intelligent distribution of inspection tasks, a cosine similarity matching model between the inspection task demand vector and the personnel capability vector was established; the inspection tasks to be completed in the digital laboratory system need to be automatically assigned to the experimenters with corresponding inspection capabilities; taking the metal bellows as an example, there are 6 experimental tasks to be completed, and the workload / piecework completion status of the inspection personnel is queried. Figure 2 The specific completion status is shown. It was found that the work report of experimenter Lin showed that he had completed 2 metal bellows. If it met the requirements, a detection task relationship could be established. Then the task was assigned to experimenter Lin, and he would receive corresponding piece-rate working hours after completing the experiment.

[0049] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0050] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0051] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A dynamic modeling method for the occupational portrait of inspectors based on multi-dimensional data analysis, characterized in that: Including the following steps: S1. Real-time collect the multi-modal operation data of the inspectors through digital twin technology; S2. Based on the data types in the multi-modal operation data, use machine learning algorithms to establish a quantitative evaluation model for the capabilities of the inspectors and determine the professional roles of the inspectors; S3. Based on the professional roles of the inspectors and the quantitative evaluation model of capabilities, determine the feature vectors and feature factors that meet the professional ability requirements, and obtain the professional levels of the inspectors; S4. Based on the professional roles of the inspectors, classify the inspection tasks in the digital laboratory system to obtain the demand vectors of different task types, establish a quantitative relationship based on the demand vectors and the ability vectors in the personnel ability evaluation model, and generate a visual professional portrait; S5. Based on the ability vectors in the professional portrait of the inspectors and the demand vectors of the inspection tasks, establish a cosine similarity matching model between the two, formulate the principle of intelligent assignment of inspection tasks, and realize the full-cycle intelligent adaptation of inspection tasks and the dynamic display of the professional portraits of inspectors.

2. The method for dynamically modeling the occupational portrait of detection personnel based on multi-dimensional data analysis according to claim 1, characterized in that: The multi-modal operation data of the inspectors includes inspection report traceability data, instrument and equipment traceability logs, daily collaboration and interaction data of inspectors, and technical ability interaction data of inspectors.

3. The dynamic modeling method for the occupational portrait of detection personnel based on multi-dimensional data analysis according to claim 1, characterized in that: The specific steps of S2 include the following: S2.1 Obtain the workload base of the inspectors based on the inspection report traceability data in the multi-modal operation data, obtain the work coordination ability of the inspectors based on the inspection report traceability data and the daily collaboration and interaction data of the inspectors in the multi-modal operation data, and obtain the work skills of the inspectors based on the instrument and equipment traceability logs and the technical ability interaction data of the inspectors in the multi-modal operation data; S2.2 Build a quantitative evaluation model for the capabilities of the inspectors based on the workload base, work coordination ability, and work skills of the inspectors; S2.3 Confirm the professional roles of the inspectors based on the work skills of the inspectors.

4. The method for dynamically modeling the occupational portrait of a tester based on multi-dimensional data analysis according to claim 1, wherein: The specific steps of S3 include the following: S3.1 Group the inspectors based on their professional roles to obtain inspection skill groups; S3.2 Evaluate based on the feature vectors of various capabilities in the quantitative evaluation model of inspector capabilities, obtain the unit vectors and feature factors that meet the requirements of the ability indicators, perform feature engineering on the unit vectors, and obtain the ability scores corresponding to each unit vector; S3.3 Obtain the scores of each ability indicator based on the unit vectors and feature factors required by the quantitative evaluation model of inspector capabilities, take the average of the ability scores of each ability indicator and combine it with the preset score range to obtain the professional level of the inspector, and associate the professional level with the professional role.

5. The method for dynamically modeling the professional portrait of detection personnel based on multi-dimensional data analysis according to claim 4, characterized in that: The feature engineering in S3.2 includes: Feature selection: Use feature selection algorithms to screen out the features that have the greatest impact on ability evaluation; Feature dimensionality reduction: For high-dimensional feature vectors, use dimensionality reduction techniques to reduce the feature dimensions and improve the calculation efficiency; Feature normalization: Normalize the features to ensure that different features are compared on the same scale; Use machine learning algorithms to learn the feature engineering of the unit vectors, establish the relationship between the quantitative evaluation model of capabilities and the unit vectors, and assign ability scores to each unit vector.

6. The method for dynamically modeling the occupational portrait of a tester based on multi-dimensional data analysis according to claim 1, characterized in that: The specific steps of S4 include the following: S4.1 Analyze the content of the detection task using machine learning algorithms, and take the professional role of the detector as the screening condition to obtain the task type and requirement vector of the detection task; S4.2 Based on the professional role of the detector and the task type and requirement vector in the digital laboratory, establish a quantitative model for evaluating the detector's ability and generate a visualized professional portrait.

7. The method for dynamically modeling the professional portrait of a detected person based on multi-dimensional data analysis according to claim 1, characterized in that: The specific steps of S5 are as follows: S5.1 Obtain the unit vector and characteristic factor in the professional role and ability evaluation model of the detector to get the ability vector in the detector's professional portrait; S5.2 Based on the ability vector in the detector's professional portrait and the requirement vector of the detection task in the digital laboratory, establish a cosine similarity matching model, and intelligently assign detection tasks according to the principle of decreasing matching degree, so as to realize the full-cycle intelligent adaptation of detection tasks and the dynamic display of the detector's professional portrait.

8. The method for dynamically modeling the occupational portrait of detection personnel based on multi-dimensional data analysis according to claim 7, characterized in that: The higher the matching degree between the ability vector in the detector's professional portrait and the requirement vector of the detection task in the digital laboratory, the higher the reliability of the detector to complete the detection task; after the cosine similarity matching is completed, establish a commission relationship and sort according to the size of the detector's characteristic factor, the higher the value, the higher the reliability of completing the detection task.

9. The method for dynamically modeling the occupational portrait of a detected person based on multi-dimensional data analysis according to claim 8, characterized in that: Evaluate and sort the importance and time requirements of the detection tasks. When there are multiple detectors with the highest reliability for completing multiple detection tasks and there are time interferences among multiple detection tasks, according to the sorting of the importance of the detection tasks, give priority to assigning the detection tasks to the detector with the highest reliability, and assign the detection tasks with the second highest importance to the detector with the second highest reliability. If the detection tasks adapted by the detector with the second highest reliability also have time interferences, adjust them again according to the above method; The task assignment adjustment method is determined by using the weighted sum algorithm. Set weights for the importance of the detection task and the reliability of the detector respectively, perform weighted calculations on the sorting serial numbers of the reliability of the detector for different detection tasks and the sorting serial numbers of the importance of the detection tasks, assign the detection task with the highest score to the corresponding detector, and note the result after the detector's professional portrait.

10. The method for dynamically modeling the professional portrait of a detected person based on multi-dimensional data analysis according to claim 1, characterized in that: In order to ensure the real-time matching between the detection task list and the detector's ability, a two-way dynamic intelligent linkage mechanism is adopted to update the detector's professional portrait in real time.

Images