Big Data-Based Competency Assessment System for Flight Attendants (CBTA)

Abstract

This invention relates to the field of big data assessment technology, specifically to a big data-based cabin crew CBTA (Cross-Board Assessment) competency assessment system. The system includes: a full-scenario data perception module, which collects data from the entire cabin crew performance process through interfaces connecting cabin sensors, crew terminals, passenger evaluation platforms, and training systems; performs real-time data cleaning, cross-terminal data synchronization, and data validity verification; and a CBTA dimensional modeling module, based on CBTA competency standards, performs dimensional hierarchical modeling, feature quantification mapping, and dimensional correlation verification to generate a standardized competency assessment dimensional system and feature dataset that conforms to the characteristics of the cabin crew position. This invention, through full-scenario data perception, achieves the linked collection and real-time processing of multi-terminal data throughout the entire cabin crew performance process, providing a comprehensive and accurate data source for competency assessment and ensuring that the assessment results truly reflect the cabin crew's performance level.

Description

Technical Field

[0001] This invention relates to the field of big data assessment technology, specifically to a big data-based CBTA (Cocktail Trainee) job competency assessment system. Background Technology

[0002] Currently, the competency assessment of cabin crew positions in the aviation industry relies heavily on traditional manual review models, which combine simple training and assessment results for evaluation. This approach suffers from several core problems, including assessment dimensions that do not align with CBTA standards, incomplete data collection, and a lack of dynamism in the assessment process.

[0003] Generally, existing assessment technologies have not achieved the linkage and collection of data across all scenarios of flight attendant performance, focusing only on a single training or service segment, resulting in assessment results that cannot fully reflect the true competence level of flight attendants. At the same time, the assessment dimensions are mostly designed in a general way, without combining the core competency requirements of CBTA positions to create hierarchical models, and the dimension weights are fixed and cannot be adapted to the assessment needs of different performance scenarios.

[0004] Furthermore, the existing assessment system lacks a collaborative mechanism for assessment, optimization, and training. The assessment results are only used as a basis for evaluation and cannot provide customized training and dynamic optimization for competency gaps. Moreover, the assessment models are mostly single and fixed models that cannot be adaptively adjusted according to changes in scenarios. The assessment accuracy and adaptability are insufficient, making it difficult to meet the aviation industry's demand for refined, standardized, and collaborative competency assessments for cabin crew positions, and unable to support the improvement of cabin crew capabilities and the optimization of job allocation.

[0005] In summary, a big data-based competency assessment system for flight attendants under the CBTA (Business Advisory and Taxation) program is needed to address the aforementioned issues. Summary of the Invention

[0006] The purpose of this invention is to provide a big data-based CBTA (Certified Cabin Crew and Trainee) job competency assessment system to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: This invention proposes a big data-based CBTA (Certified Cabin Crew Association) competency assessment system, comprising: The full-scene data perception module collects data on the entire process of flight attendant job performance through the linkage interface of cabin sensors, crew terminals, passenger evaluation platform and training system, and performs real-time data cleaning, cross-terminal data synchronization and data validity verification operations. The CBTA dimensional modeling module, based on the CBTA job competency standard, performs dimensional hierarchical modeling, feature quantization mapping, and dimensional correlation verification operations to generate a standardized competency assessment dimensional system and feature dataset that conforms to the characteristics of the flight attendant job. The dynamic collaborative evaluation module uses a scenario adaptation model library and a multi-dimensional weight collaborative algorithm to perform evaluation scenario matching, multi-model collaborative operation and dynamic adjustment of the evaluation process, and completes real-time evaluation of cabin crew competence based on feature datasets. The result adaptation output module is used to perform hierarchical verification of evaluation results, multi-scenario result adaptation, and evaluation report customization operations, and output multi-format evaluation results and linkage interfaces adapted to aviation human resource management, training optimization, and job allocation.

[0008] Preferably, it further includes: The closed-loop optimization module is used to collect data from the entire evaluation process, job suitability feedback, and training effectiveness data, and to perform evaluation dimension optimization, model parameter iteration, and dynamic calibration of evaluation standards to achieve continuous improvement in system evaluation performance.

[0009] Preferably, it further includes: The training empowerment module receives competency gap data output by the dynamic collaborative assessment module, performs CBTA training course matching corresponding to the gaps, dynamically tracks the training process, and conducts post-training competency reassessment, thereby achieving a seamless collaborative closed loop between assessment and training.

[0010] This invention also proposes an assessment method for a big data-based CBTA (Certified Cabin Crew Association) job competency assessment system, comprising the following steps: S1. By linking multiple terminals through the full-scene data perception module, data on the entire process of flight attendant performance is collected. The collected data is cleaned in real time, synchronized across terminals, and validated for validity. Invalid, duplicate, and abnormal data are removed to generate a standardized performance dataset. S2. Based on the CBTA job competency standard, construct a hierarchical dimension system through the CBTA dimension modeling module, quantify and map the standardized job performance dataset generated in step S1 to each evaluation dimension, complete the dimension association verification, and generate a standardized competency feature dataset. S3. Based on the evaluation scenario type, the optimal model combination is matched from the scenario adaptation model library through the dynamic collaborative evaluation module, and the model weights are allocated using a multi-dimensional weight collaborative algorithm. Based on the standardized competency feature dataset generated in S2, the scenario-based collaborative evaluation model is constructed by combining the scenario type. S4. Input the standardized competency feature dataset generated in S2 into the scenario-based collaborative assessment model, perform multi-model collaborative operation, and dynamically adjust the assessment parameters by combining the real-time performance data of cabin crew with the real-time performance data collected in step S1 to generate hierarchical and scenario-based cabin crew CBTA job competency assessment results. S5. The cabin crew CBTA job competency assessment results generated in step S4 are graded and verified through the result adaptation output module. Based on the different needs of human resource management, training optimization and job allocation, the corresponding assessment results are adapted and generated, and the linkage interface is output. S6. Collect data and feedback information from the entire evaluation process through the closed-loop optimization module, covering all data from steps S1-S5, including performance data, feature datasets, evaluation results, and verification data. Iterate and optimize the evaluation dimension system and model parameters. At the same time, conduct customized training and re-evaluation for the competency gaps generated in step S4 through the training and empowerment module, so as to achieve a collaborative closed loop of evaluation, optimization, and training.

[0011] Preferably, the implementation process of step S1 is as follows: It integrates cabin infrared sensors, voice collection equipment, crew collaboration terminals, passenger online evaluation platforms, and cabin crew training systems to collect data on passenger cabin service behavior, emergency response operations, crew collaboration communication, passenger satisfaction, training assessment, and job performance compliance. Edge computing technology is used to clean the collected data in real time, removing invalid data that was collected outside of the duty period, whose data values ​​are outside the reasonable range, or whose data format is not standardized. By using a cross-terminal data synchronization protocol, timestamp alignment and data identification are achieved for data collected from multiple terminals, ensuring data consistency. A data validity verification algorithm is used to verify the cleaned synchronous data. After the verification is passed, a standardized performance dataset is generated. The dataset contains four core fields: flight attendant personal identifier, performance period, data type, and quantitative value.

[0012] Preferably, the implementation process of step S2 is as follows: Based on the CBTA job competency standards, a three-tiered evaluation system is constructed. The first-level dimension includes three dimensions: CBTA core competencies, performance compliance, and service adaptability. CBTA's core capabilities are broken down into three secondary dimensions: emergency response practical ability, unit coordination ability, and risk identification ability. Each secondary dimension is further broken down into two tertiary indicators. Performance compliance is broken down into two secondary dimensions: the ability to execute operational norms and the ability to follow job procedures. Each secondary dimension is further broken down into two tertiary indicators. Service adaptability is broken down into two secondary dimensions: passenger demand adaptability and special scenario service capability. Each secondary dimension is further broken down into two tertiary indicators. Entropy weighting method is used to assign weights to dimensions and indicators at each level, and feature quantization algorithm is used to map standardized performance data to each third-level indicator to generate an indicator-data mapping matrix. The mapping matrix is ​​validated for dimensional correlation, and index data with a correlation degree lower than 0.6 are removed, finally generating a standardized competency feature dataset.

[0013] Preferably, the implementation process of step S3 is as follows: Build a scenario adaptation model library, including a lightweight neural network model for daily cabin service scenario evaluation, a convolutional neural network model for emergency response scenario evaluation, a logistic regression model for training and assessment scenario evaluation, and a multimodal collaborative model for comprehensive performance scenario evaluation. Based on the assessment scenario type, including daily service, emergency response, training and assessment, and comprehensive performance, the system automatically matches the corresponding model combination. The daily service scenario matches a lightweight neural network model, the emergency response scenario matches a combination of a convolutional neural network model and a multimodal collaborative model, the training and assessment scenario matches a logistic regression model, and the comprehensive performance scenario matches all four model combinations. The weights of each model are calculated using a multi-dimensional weight collaboration algorithm, as shown in equation (1): ; In the formula, Let be the final weight of the i-th model, and n be the total number of models participating in the collaboration. Let be the evaluation accuracy of the i-th model in the corresponding scenario over the past 6 months. The fit coefficient between the i-th model and the current evaluation dimension is determined based on the correlation between the model output and the dimension index, and its value ranges from [0.7, 1]. Construct a model collaborative computing layer to integrate the output results of various models, generate a scenario-based collaborative evaluation model, and store the model structure, weight parameters, and scenario adaptation rules.

[0014] Preferably, the implementation process of step S4 is as follows: The standardized competency feature dataset is divided into training, validation and test sets in an 8:1:1 ratio. The training set is used for training the parameters of the scenario-based collaborative evaluation model, the validation set is used for model hyperparameter tuning, and the test set is used for preliminary verification of the model evaluation accuracy. Set the model training termination condition: stop training and save the optimal model parameters when the evaluation error of each of the three-level indicators on the validation set decreases by ≤0.001 for 8 consecutive rounds of iteration. Establish a real-time performance data triggering mechanism. When flight attendants complete their duties for a single flight, emergency simulation training, or training assessment, real-time performance data will be automatically collected, triggering a scenario-based collaborative evaluation model for real-time evaluation. Generate individual flight attendant scores for three-level indicators, two-level dimensions, one-level dimensions, and overall competence, label the scenario suitability corresponding to each score, and generate a list of competence gap indicators, including three-level indicators with scores below 0.6.

[0015] Preferably, the implementation process of step S5 is as follows: The evaluation results are graded and verified using multiple indicators. The evaluation error of each third-level indicator, the consistency error of the first-level dimension, and the comprehensive error of the overall competence are calculated. The comprehensive error of the overall competence is required to be ≤5%. The evaluation results are adapted to different application scenarios: the human resource management scenario outputs a flight attendant competency ranking table; the training optimization scenario outputs shortcomings indicators and training suggestions; and the job allocation scenario outputs a job suitability score. Output multi-format evaluation results, including PDF comprehensive evaluation report, Excel score details, and visual competency radar chart. It also provides RESTful API interface to support seamless integration with aviation human resources management system and cabin crew training system, with an interface response time of ≤200ms. The stability of the output evaluation results and interfaces is verified, and the final output is completed after the verification is passed.

[0016] Preferably, the implementation process of step S6 is as follows: The closed-loop optimization module records data throughout the entire evaluation process, including data collection sources, dimension mapping parameters, model combination schemes, weight allocation results, evaluation errors, and application feedback in various scenarios. Monthly data on flight attendant performance, evaluation results, and application feedback are added to construct a closed-loop optimized sample library. An incremental iterative algorithm is used to update the evaluation dimension weights and model parameters, as shown in equation (2): ; In the formula, For the updated model parameters, Here are the model parameters before the update, λ is the iterative learning rate (0.008), and M is the number of new samples added in the current month. This represents the actual competence score for the m-th sample. The model prediction score for the m-th sample. Let m be the feature vector of the m-th sample; The evaluation dimension system is re-verified quarterly based on the sample library, removing indicators with a correlation of less than 0.6 and adding indicators that are suitable for job requirements. Through the training empowerment module, corresponding CBTA training courses are matched according to the list of competency gap indicators, the training progress is tracked, and a targeted re-evaluation is triggered after the training is completed. If the re-evaluation is successful, the competency assessment results are updated; if it is unsuccessful, the training plan is adjusted and retraining is carried out, forming a closed loop of assessment-training-re-evaluation.

[0017] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention achieves multi-terminal data linkage collection and real-time processing throughout the entire process of flight attendant performance through full-scene data perception, providing a comprehensive and accurate data source for competency assessment, ensuring that the assessment results truly reflect the flight attendant's performance level. Through CBTA dimensional modeling, a three-layer assessment dimension system conforming to CBTA standards is constructed. Combining entropy weighting and feature quantization mapping techniques, precise correlation between data and assessment dimensions is achieved, improving the scientific rigor and adaptability of the assessment dimensions. Through dynamic collaborative assessment, a scenario-adaptive model library and a multi-dimensional weighted collaborative algorithm are constructed to achieve assessment model... By employing scenario-based matching and dynamic adjustment, the accuracy of assessments in different scenarios is significantly improved. Through closed-loop optimization and training empowerment, a collaborative closed loop of assessment-optimization-training is constructed. In summary, this invention can specifically address shortcomings in cabin crew competency, continuously improve system assessment performance, and achieve full-process automation, scenario-based adaptation, and collaborative optimization of cabin crew CBTA job competency assessment. This reduces manual review costs, improves assessment efficiency and accuracy, and provides scientific and reliable technical support for the selection, training, assessment, and job allocation of cabin crew in the aviation industry. It also meets the refined needs of the aviation industry's high-quality development for cabin crew job competency assessment. Attached Figure Description

[0018] Figure 1 This is a topology diagram of the cabin crew CBTA job competency assessment system based on big data, as described in this invention. Figure 2 This is a flowchart of the big data-based CBTA job competency assessment method for flight attendants according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1, please refer to Figure 1 This invention proposes a big data-based CBTA (Certified Cabin Crew and Trainee) job competency assessment system, comprising: The full-scene data perception module collects data on the entire process of flight attendant job performance through the linkage interface of cabin sensors, crew terminals, passenger evaluation platform and training system, and performs real-time data cleaning, cross-terminal data synchronization and data validity verification operations. The full-scenario data perception module consists of cabin sensors, voice acquisition equipment, crew collaboration terminal, data transmission module, and supporting software. The hardware uses FLIR AX8 infrared thermal imaging sensor, Respeaker 4-Mic Array directional microphone array, Huawei MatePad Pro 12.6-inch 5G industrial-grade tablet, and Quectel EC200S 5G industrial module. The software uses data acquisition software developed with Python 3.9, EdgeX Foundry edge computing cleaning software, MQTT 3.1.1 synchronization protocol, and validity verification software developed with Java.

[0021] The CBTA dimensional modeling module, based on the CBTA job competency standard, performs dimensional hierarchical modeling, feature quantization mapping, and dimensional correlation verification operations to generate a standardized competency assessment dimensional system and feature dataset that conforms to the characteristics of the flight attendant job. The CBTA dimensional modeling module consists of a server and supporting modeling, mapping, and verification software. The hardware uses a Dell PowerEdge R750 server, and the software uses MATLAB 2023b dimensional modeling software, Python 3.9 feature quantization mapping software, and R language 4.2.0 correlation verification software.

[0022] The dynamic collaborative evaluation module uses a scenario adaptation model library and a multi-dimensional weight collaborative algorithm to perform evaluation scenario matching, multi-model collaborative operation and dynamic adjustment of the evaluation process, and completes real-time evaluation of cabin crew competence based on feature datasets. The dynamic collaborative evaluation module consists of a GPU server, a model library, and collaborative computing software. The hardware uses an NVIDIA DGX A100 GPU server, and the software uses Python 3.9 scene matching software, Docker model library management software, and PyTorch 1.13 collaborative computing software. The model library includes MobileNetV3, CNN-LSTM hybrid model, logistic regression model, and ViT-B / 16 multimodal collaborative model.

[0023] The result adaptation output module is used to perform hierarchical verification of evaluation results, multi-scenario result adaptation, and evaluation report customization operations, and output multi-format evaluation results and linkage interfaces adapted to aviation human resource management, training optimization, and job allocation.

[0024] The result adaptation and output module consists of an application server and supporting verification, reporting, visualization, and interface software. The hardware uses Huawei FusionServer Pro 2288H V5 application server, and the software uses Python 3.9 hierarchical verification software, ReportLab report customization software, ECharts 5.4 visualization software, and Spring Boot 2.7 interface development software.

[0025] In this embodiment, it should also be noted that it further includes: The closed-loop optimization module is used to collect data from the entire evaluation process, job suitability feedback, and training effectiveness data, and to perform evaluation dimension optimization, model parameter iteration, and dynamic calibration of evaluation standards to achieve continuous improvement in system evaluation performance.

[0026] The closed-loop optimization module and the CBTA dimensional modeling module share a Dell PowerEdge R750 server. The software is an incremental iterative optimization software developed with Python 3.9 to achieve dynamic updates and calibration of dimensions and parameters.

[0027] In this embodiment, it should also be noted that it further includes: The training empowerment module receives competency gap data output by the dynamic collaborative assessment module, performs CBTA training course matching corresponding to the gaps, dynamically tracks the training process, and conducts post-training competency reassessment, thereby achieving a seamless collaborative closed loop between assessment and training.

[0028] The training empowerment module and the result adaptation and output module share the Huawei FusionServer Pro 2288H V5 application server. The software is developed in Java and supports training course matching, progress tracking and re-evaluation result synchronization.

[0029] Example 2, please refer to Figure 2 In practical applications, the evaluation method of the CBTA (Certified Cabin Crew Association) job competency assessment system based on big data specifically includes the following steps: S1. By linking multiple terminals through the full-scene data perception module, data on the entire process of flight attendant performance is collected. The collected data is cleaned in real time, synchronized across terminals, and validated for validity. Invalid, duplicate, and abnormal data are removed to generate a standardized performance dataset. In this embodiment, it should also be noted that the implementation process of step S1 is as follows: It integrates cabin infrared sensors, voice collection equipment, crew collaboration terminals, passenger online evaluation platforms, and cabin crew training systems to collect data on passenger cabin service behavior, emergency response operations, crew collaboration communication, passenger satisfaction, training assessment, and job performance compliance. Edge computing technology is used to clean the collected data in real time, removing invalid data that was collected outside of the duty period, whose data values ​​are outside the reasonable range, or whose data format is not standardized. By using a cross-terminal data synchronization protocol, timestamp alignment and data identification are achieved for data collected from multiple terminals, ensuring data consistency. A data validity verification algorithm is used to verify the cleaned synchronous data. After the verification is passed, a standardized performance dataset is generated. The dataset contains four core fields: flight attendant personal identifier, performance period, data type, and quantitative value.

[0030] S2. Based on the CBTA job competency standard, construct a hierarchical dimension system through the CBTA dimension modeling module, quantify and map the standardized job performance dataset generated in step S1 to each evaluation dimension, complete the dimension association verification, and generate a standardized competency feature dataset. In this embodiment, it should also be noted that the implementation process of step S2 is as follows: Based on the CBTA job competency standards, a three-tiered evaluation system is constructed. The first-level dimension includes three dimensions: CBTA core competencies, performance compliance, and service adaptability. CBTA's core capabilities are broken down into three secondary dimensions: emergency response practical ability, unit coordination ability, and risk identification ability. Each secondary dimension is further broken down into two tertiary indicators. Performance compliance is broken down into two secondary dimensions: the ability to execute operational norms and the ability to follow job procedures. Each secondary dimension is further broken down into two tertiary indicators. Service adaptability is broken down into two secondary dimensions: passenger demand adaptability and special scenario service capability. Each secondary dimension is further broken down into two tertiary indicators. Entropy weighting method is used to assign weights to dimensions and indicators at each level, and feature quantization algorithm is used to map standardized performance data to each third-level indicator to generate an indicator-data mapping matrix. The mapping matrix is ​​validated for dimensional correlation, and index data with a correlation degree lower than 0.6 are removed, finally generating a standardized competency feature dataset.

[0031] S3. Based on the evaluation scenario type, the optimal model combination is matched from the scenario adaptation model library through the dynamic collaborative evaluation module, and the model weights are allocated using a multi-dimensional weight collaborative algorithm. Based on the standardized competency feature dataset generated in S2, the scenario-based collaborative evaluation model is constructed by combining the scenario type. In this embodiment, it should also be noted that the implementation process of step S3 is as follows: Build a scenario adaptation model library, including a lightweight neural network model for daily cabin service scenario evaluation, a convolutional neural network model for emergency response scenario evaluation, a logistic regression model for training and assessment scenario evaluation, and a multimodal collaborative model for comprehensive performance scenario evaluation. Based on the assessment scenario type, including daily service, emergency response, training and assessment, and comprehensive performance, the system automatically matches the corresponding model combination. The daily service scenario matches a lightweight neural network model, the emergency response scenario matches a combination of a convolutional neural network model and a multimodal collaborative model, the training and assessment scenario matches a logistic regression model, and the comprehensive performance scenario matches all four model combinations. The weights of each model are calculated using a multi-dimensional weight collaboration algorithm, as shown in equation (1): ; In the formula, Let be the final weight of the i-th model, and n be the total number of models participating in the collaboration. Let be the evaluation accuracy of the i-th model in the corresponding scenario over the past 6 months. The fit coefficient between the i-th model and the current evaluation dimension is determined based on the correlation between the model output and the dimension index, and its value ranges from [0.7, 1]. Construct a model collaborative computing layer to integrate the output results of various models, generate a scenario-based collaborative evaluation model, and store the model structure, weight parameters, and scenario adaptation rules.

[0032] S4. Input the standardized competency feature dataset generated in S2 into the scenario-based collaborative assessment model, perform multi-model collaborative operation, and dynamically adjust the assessment parameters by combining the real-time performance data of cabin crew with the real-time performance data collected in step S1 to generate hierarchical and scenario-based cabin crew CBTA job competency assessment results. In this embodiment, it should also be noted that the implementation process of step S4 is as follows: The standardized competency feature dataset is divided into training, validation and test sets in an 8:1:1 ratio. The training set is used for training the parameters of the scenario-based collaborative evaluation model, the validation set is used for model hyperparameter tuning, and the test set is used for preliminary verification of the model evaluation accuracy. Set the model training termination condition: stop training and save the optimal model parameters when the evaluation error of each of the three-level indicators on the validation set decreases by ≤0.001 for 8 consecutive rounds of iteration. Establish a real-time performance data triggering mechanism. When flight attendants complete their duties for a single flight, emergency simulation training, or training assessment, real-time performance data will be automatically collected, triggering a scenario-based collaborative evaluation model for real-time evaluation. Generate individual flight attendant scores for three-level indicators, two-level dimensions, one-level dimensions, and overall competence, label the scenario suitability corresponding to each score, and generate a list of competence gap indicators, including three-level indicators with scores below 0.6.

[0033] S5. The cabin crew CBTA job competency assessment results generated in step S4 are graded and verified through the result adaptation output module. Based on the different needs of human resource management, training optimization and job allocation, the corresponding assessment results are adapted and generated, and the linkage interface is output. In this embodiment, it should also be noted that the implementation process of step S5 is as follows: The evaluation results are graded and verified using multiple indicators. The evaluation error of each third-level indicator, the consistency error of the first-level dimension, and the comprehensive error of the overall competence are calculated. The comprehensive error of the overall competence is required to be ≤5%. The evaluation results are adapted to different application scenarios: the human resource management scenario outputs a flight attendant competency ranking table; the training optimization scenario outputs shortcomings indicators and training suggestions; and the job allocation scenario outputs a job suitability score. Output multi-format evaluation results, including PDF comprehensive evaluation report, Excel score details, and visual competency radar chart. It also provides RESTful API interface to support seamless integration with aviation human resources management system and cabin crew training system, with an interface response time of ≤200ms. The stability of the output evaluation results and interfaces is verified, and the final output is completed after the verification is passed.

[0034] S6. Collect data and feedback information from the entire evaluation process through the closed-loop optimization module, covering all data from steps S1-S5, including performance data, feature datasets, evaluation results, and verification data. Iterate and optimize the evaluation dimension system and model parameters. At the same time, conduct customized training and re-evaluation for the competency gaps generated in step S4 through the training and empowerment module, so as to achieve a collaborative closed loop of evaluation, optimization, and training.

[0035] In this embodiment, it should also be noted that the implementation process of step S6 is as follows: The closed-loop optimization module records data throughout the entire evaluation process, including data collection sources, dimension mapping parameters, model combination schemes, weight allocation results, evaluation errors, and application feedback in various scenarios. Monthly data on flight attendant performance, evaluation results, and application feedback are added to construct a closed-loop optimized sample library. An incremental iterative algorithm is used to update the evaluation dimension weights and model parameters, as shown in equation (2): ; In the formula, For the updated model parameters, Here are the model parameters before the update, λ is the iterative learning rate (0.008), and M is the number of new samples added in the current month. This represents the actual competence score for the m-th sample. The model prediction score for the m-th sample. Let m be the feature vector of the m-th sample; The evaluation dimension system is re-verified quarterly based on the sample library, removing indicators with a correlation of less than 0.6 and adding indicators that are suitable for job requirements. Through the training empowerment module, corresponding CBTA training courses are matched according to the list of competency gap indicators, the training progress is tracked, and a targeted re-evaluation is triggered after the training is completed. If the re-evaluation is successful, the competency assessment results are updated; if it is unsuccessful, the training plan is adjusted and retraining is carried out, forming a closed loop of assessment-training-re-evaluation.

[0036] Through the above steps, this invention achieves the coordinated collection and real-time processing of multi-terminal data throughout the entire process of flight attendant performance through full-scenario data perception. This provides a comprehensive and accurate data source for competency assessment, ensuring that the assessment results truly reflect the flight attendant's performance level. By using CBTA dimensional modeling, a three-layer assessment dimension system conforming to CBTA standards is constructed. Combining entropy weighting and feature quantization mapping technology, precise correlation between data and assessment dimensions is achieved, enhancing the scientific rigor and adaptability of the assessment dimensions. Through dynamic collaborative assessment, a scenario-adaptive model library and a multi-dimensional weighted collaborative algorithm are constructed to realize the scenario-based application of the assessment model. By matching and dynamically adjusting, the accuracy of assessments in different scenarios is significantly improved. Through closed-loop optimization and training empowerment, a collaborative closed loop of assessment-optimization-training is constructed. In summary, this invention can specifically address the shortcomings in cabin crew competency, continuously improve the system's assessment performance, and achieve full-process automation, scenario-based adaptation, and collaborative optimization of cabin crew CBTA job competency assessment. This reduces the cost of manual review, improves assessment efficiency and accuracy, and provides scientific and reliable technical support for the selection, training, assessment, and job allocation of cabin crew in the aviation industry. It also meets the refined needs of the aviation industry's high-quality development for cabin crew job competency assessment.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A big data-based CBTA (Certified Cabin Crew and Trainee) job competency assessment system, characterized in that, include: The full-scene data perception module collects data on the entire process of flight attendant job performance through the linkage interface of cabin sensors, crew terminals, passenger evaluation platform and training system, and performs real-time data cleaning, cross-terminal data synchronization and data validity verification operations. The CBTA dimensional modeling module, based on the CBTA job competency standard, performs dimensional hierarchical modeling, feature quantization mapping, and dimensional correlation verification operations to generate a standardized competency assessment dimensional system and feature dataset that conforms to the characteristics of the flight attendant job. The dynamic collaborative evaluation module uses a scenario adaptation model library and a multi-dimensional weight collaborative algorithm to perform evaluation scenario matching, multi-model collaborative operation and dynamic adjustment of the evaluation process, and completes real-time evaluation of cabin crew competence based on feature datasets. The result adaptation output module is used to perform hierarchical verification of evaluation results, multi-scenario result adaptation, and evaluation report customization operations, and output multi-format evaluation results and linkage interfaces adapted to aviation human resource management, training optimization, and job allocation.

2. The cabin crew CBTA job competency assessment system based on big data as described in claim 1, characterized in that, Also includes: The closed-loop optimization module is used to collect data from the entire evaluation process, job suitability feedback, and training effectiveness data, and to perform evaluation dimension optimization, model parameter iteration, and dynamic calibration of evaluation standards to achieve continuous improvement in system evaluation performance.

3. The cabin crew CBTA job competency assessment system based on big data as described in claim 2, characterized in that, Also includes: The training empowerment module receives competency gap data output by the dynamic collaborative assessment module, performs CBTA training course matching corresponding to the gaps, dynamically tracks the training process, and conducts post-training competency reassessment, thereby achieving a seamless collaborative closed loop between assessment and training.

4. The assessment method of the cabin crew CBTA job competency assessment system based on big data according to any one of claims 1-3, characterized in that, Includes the following steps: S1. By linking multiple terminals through the full-scene data perception module, data on the entire process of flight attendant performance is collected. The collected data is cleaned in real time, synchronized across terminals, and validated for validity. Invalid, duplicate, and abnormal data are removed to generate a standardized performance dataset. S2. Based on the CBTA job competency standard, construct a hierarchical dimension system through the CBTA dimension modeling module, quantify and map the standardized job performance dataset generated in step S1 to each evaluation dimension, complete the dimension association verification, and generate a standardized competency feature dataset. S3. Based on the evaluation scenario type, the optimal model combination is matched from the scenario adaptation model library through the dynamic collaborative evaluation module, and the model weights are allocated using a multi-dimensional weight collaborative algorithm. Based on the standardized competency feature dataset generated in S2, the scenario-based collaborative evaluation model is constructed by combining the scenario type. S4. Input the standardized competency feature dataset generated in step S2 into the scenario-based collaborative assessment model, perform multi-model collaborative operation, and combine the real-time performance data of flight attendants with the real-time performance data collected in step S1 to dynamically adjust the assessment parameters and generate hierarchical and scenario-based flight attendant CBTA job competency assessment results. S5. The cabin crew CBTA job competency assessment results generated in step S4 are graded and verified through the result adaptation output module. Based on the different needs of human resource management, training optimization and job allocation, the corresponding assessment results are adapted and generated, and the linkage interface is output. S6. Collect data and feedback information from the entire evaluation process through the closed-loop optimization module, covering all data from steps S1-S5, including performance data, feature datasets, evaluation results, and verification data. Iterate and optimize the evaluation dimension system and model parameters. At the same time, conduct customized training and re-evaluation for the competency gaps generated in step S4 through the training and empowerment module, so as to achieve a collaborative closed loop of evaluation, optimization, and training.

5. The assessment method of the cabin crew CBTA job competency assessment system based on big data as described in claim 4, characterized in that, The implementation process of step S1 is as follows: It integrates cabin infrared sensors, voice collection equipment, crew collaboration terminals, passenger online evaluation platforms, and cabin crew training systems to collect data on passenger cabin service behavior, emergency response operations, crew collaboration communication, passenger satisfaction, training assessment, and job performance compliance. Edge computing technology is used to clean the collected data in real time, removing invalid data that was collected outside of the duty period, whose data values ​​are outside the reasonable range, or whose data format is not standardized. By using a cross-terminal data synchronization protocol, timestamp alignment and data identification are achieved for data collected from multiple terminals, ensuring data consistency. A data validity verification algorithm is used to verify the cleaned synchronous data. After the verification is passed, a standardized performance dataset is generated. The dataset contains four core fields: flight attendant personal identifier, performance period, data type, and quantitative value.

6. The assessment method of the cabin crew CBTA job competency assessment system based on big data as described in claim 5, characterized in that, The implementation process of step S2 is as follows: Based on the CBTA job competency standards, a three-tiered evaluation system is constructed. The first-level dimension includes three dimensions: CBTA core competencies, performance compliance, and service adaptability. CBTA's core capabilities are broken down into three secondary dimensions: emergency response practical ability, unit coordination ability, and risk identification ability. Each secondary dimension is further broken down into two tertiary indicators. Performance compliance is broken down into two secondary dimensions: the ability to execute operational norms and the ability to follow job procedures. Each secondary dimension is further broken down into two tertiary indicators. Service adaptability is broken down into two secondary dimensions: passenger demand adaptability and special scenario service capability. Each secondary dimension is further broken down into two tertiary indicators. Entropy weighting method is used to assign weights to dimensions and indicators at each level, and feature quantization algorithm is used to map standardized performance data to each third-level indicator to generate an indicator-data mapping matrix. The mapping matrix is ​​validated for dimensional correlation, and index data with a correlation degree lower than 0.6 are removed, finally generating a standardized competency feature dataset.

7. The assessment method of the cabin crew CBTA job competency assessment system based on big data as described in claim 6, characterized in that, The implementation process of step S3 is as follows: Build a scenario adaptation model library, including a lightweight neural network model for daily cabin service scenario evaluation, a convolutional neural network model for emergency response scenario evaluation, a logistic regression model for training and assessment scenario evaluation, and a multimodal collaborative model for comprehensive performance scenario evaluation. Based on the assessment scenario type, including daily service, emergency response, training and assessment, and comprehensive performance, the system automatically matches the corresponding model combination. The daily service scenario matches a lightweight neural network model, the emergency response scenario matches a combination of a convolutional neural network model and a multimodal collaborative model, the training and assessment scenario matches a logistic regression model, and the comprehensive performance scenario matches all four model combinations. The weights of each model are calculated using a multi-dimensional weight collaboration algorithm, as shown in equation (1): ； In the formula, Let be the final weight of the i-th model, and n be the total number of models participating in the collaboration. Let be the evaluation accuracy of the i-th model in the corresponding scenario over the past 6 months. The fit coefficient between the i-th model and the current evaluation dimension is determined based on the correlation between the model output and the dimension index, and its value ranges from [0.7, 1]. Construct a model collaborative computing layer to integrate the output results of various models, generate a scenario-based collaborative evaluation model, and store the model structure, weight parameters, and scenario adaptation rules.

8. The assessment method of the cabin crew CBTA job competency assessment system based on big data as described in claim 7, characterized in that, The implementation process of step S4 is as follows: The standardized competency feature dataset is divided into training, validation and test sets in an 8:1:1 ratio. The training set is used for training the parameters of the scenario-based collaborative evaluation model, the validation set is used for model hyperparameter tuning, and the test set is used for preliminary verification of the model evaluation accuracy. Set the model training termination condition: stop training and save the optimal model parameters when the evaluation error of each of the three-level indicators on the validation set decreases by ≤0.001 for 8 consecutive rounds of iteration. Establish a real-time performance data triggering mechanism. When flight attendants complete their duties for a single flight, emergency simulation training, or training assessment, real-time performance data will be automatically collected, triggering a scenario-based collaborative evaluation model for real-time evaluation. Generate individual flight attendant scores for three-level indicators, two-level dimensions, one-level dimensions, and overall competence, label the scenario suitability corresponding to each score, and generate a list of competence gap indicators, including three-level indicators with scores below 0.

6.

9. The assessment method of the cabin crew CBTA job competency assessment system based on big data as described in claim 8, characterized in that, The implementation process of step S5 is as follows: The evaluation results are graded and verified using multiple indicators. The evaluation error of each third-level indicator, the consistency error of the first-level dimension, and the comprehensive error of the overall competence are calculated. The comprehensive error of the overall competence is required to be ≤5%. The evaluation results are adapted to different application scenarios: the human resource management scenario outputs a flight attendant competency ranking table; the training optimization scenario outputs shortcomings indicators and training suggestions; and the job allocation scenario outputs a job suitability score. Output multi-format evaluation results, including PDF comprehensive evaluation report, Excel score details, and visual competency radar chart. It also provides RESTful API interface to support seamless integration with aviation human resources management system and cabin crew training system, with an interface response time of ≤200ms. The stability of the output evaluation results and interfaces is verified, and the final output is completed after the verification is passed.

10. The assessment method of the cabin crew CBTA job competency assessment system based on big data as described in claim 9, characterized in that, The implementation process of step S6 is as follows: The closed-loop optimization module records data throughout the entire evaluation process, including data collection sources, dimension mapping parameters, model combination schemes, weight allocation results, evaluation errors, and application feedback in various scenarios. Monthly data on flight attendant performance, evaluation results, and application feedback are added to construct a closed-loop optimized sample library. An incremental iterative algorithm is used to update the evaluation dimension weights and model parameters, as shown in equation (2): ； In the formula, For the updated model parameters, Here are the model parameters before the update, λ is the iterative learning rate (0.008), and M is the number of new samples added in the current month. This represents the actual competence score for the m-th sample. The model prediction score for the m-th sample. Let m be the feature vector of the m-th sample; The evaluation dimension system is re-verified quarterly based on the sample library, removing indicators with a correlation of less than 0.6 and adding indicators that are suitable for job requirements. Through the training empowerment module, corresponding CBTA training courses are matched according to the list of competency gap indicators, the training progress is tracked, and a targeted re-evaluation is triggered after the training is completed. If the re-evaluation is successful, the competency assessment results are updated; if it is unsuccessful, the training plan is adjusted and retraining is carried out, forming a closed loop of assessment-training-re-evaluation.

Images