Competency data acquisition and storage system and method based on observable behavior classification, electronic equipment and storage medium

Abstract

The invention provides a full life cycle competency data acquisition and storage system based on observable behavior classification, and belongs to the technical field of personnel training evaluation. The system comprises a competency-OB classification analysis module used for performing multi-dimensional analysis on competency and observable behaviors to generate a structured classification framework; the multi-source data acquisition module is used for acquiring data from multiple systems in real time and automatically marking the data according to a structured classification framework; the special classification database cluster adopts a three-layer architecture to support efficient data processing; the data quality guarantee module is used for implementing verification in multiple links to guarantee data quality; the offline-online integrated module supports full-function offline evaluation and automatic account checking; the block chain evidence storage interface ensures that the evaluation record cannot be tampered; and the standardized API gateway provides an external system interface. The technical problems that traditional competency assessment data acquisition is low in structuralization degree, poor in cross-system interoperability, insufficient in data credibility and the like are solved, and the method can be widely applied to full-life-cycle training and assessment of high-risk occupational posts such as civil aviation and nuclear power.

Description

Technical Field

[0001] This invention relates to data acquisition and storage systems, specifically to a full lifecycle competency data acquisition and storage system based on observable behavior classification. Background Technology

[0002] As the competency requirements for high-risk and highly complex professional positions continue to increase, competency assessment methods based on Observable Behavior (OB) have become an industry standard. As the foundation for personnel selection, training, and evaluation, the scientific validity and effectiveness of competency assessment systems directly impact the safe operation of organizations and the quality of talent development. Currently, specialized positions such as civil aviation pilots, air traffic controllers, and nuclear power plant operators all employ competency-based training and assessment (CBTA) methods, which assess personnel competency levels by observing and recording key behaviors.

[0003] In the prior art, CN114219302B discloses a method for profiling pilots' work style competence across the entire lifecycle of flight skills. This method acquires aviation management data, establishes a pilot work style assessment model, evaluates the competence of a sample set of pilot work style, and obtains a multivariate work style competence assessment index. CN119417429A proposes a competence extraction and assessment method based on employee professional behavior data. By collecting employee professional behavior data and extracting competence-related behavioral characteristics, it constructs a competence element model, defines the hierarchical structure of competence elements, and achieves real-time assessment of employee competence.

[0004] In the field of education, CN116663988A discloses an assessment method for general teacher competence and specific teacher competence. The method includes steps such as data collection, data processing, assessment of general teacher competence, and construction of assessment model for specific teacher competence. It can be applied to scenarios such as teacher competence evaluation and assessment.

[0005] In the power industry, CN120672190A proposes a competency assessment method for power enterprise personnel based on deep learning and support vector machines. This method collects multi-dimensional data on personnel, identifies eight indices affecting competency, and uses a deep learning model to extract deep-level features, combined with an SVM model to predict competency levels. Furthermore, CN120672207A discloses a multi-dimensional core competency evaluation system for air traffic controllers. This system includes a model management module, a data processing module, a weight calculation module, an evaluation result generation module, and a result output module, enabling quantitative evaluation of the competency of target controllers.

[0006] However, existing technologies still have the following problems in the consistent interpretation and data acquisition of competence and observable behavior (OB):

[0007] 1. The interpretation of OBs often treats them as isolated individuals, ignoring the correlation, gradient relationship and hierarchical relationship between OBs, resulting in a lack of systematicity and consistency in the evaluation results;

[0008] 2. Existing methods of interpreting OBs are highly subjective, making it difficult to achieve cognitive consensus among instructors and assessors, resulting in insufficient reliability and consistency of assessment results;

[0009] 3. Existing systems cannot effectively transform the number, frequency, and degree of OBs into training and assessment indicators that span the entire life cycle, and lack the ability to dynamically track the competency development process;

[0010] 4. The existing competency framework only presents OBs in a list format, failing to show their complex attributes and logical relationships, making it difficult to support refined competency analysis;

[0011] 5. There is a lack of methods for building competency and OB development gradients based on the entire career life cycle, which cannot effectively support the growth path planning of personnel from novices to experts;

[0012] 6. Existing systems lack unified standards and efficient architectures in data acquisition, storage, and processing, making it difficult to handle large-scale, multi-source, and heterogeneous competency assessment data;

[0013] 7. Existing technologies lack effective support for offline evaluation scenarios and cannot meet the evaluation needs in environments with limited network conditions;

[0014] 8. Existing systems cannot guarantee the authenticity and immutability of assessment data and lack a reliable mechanism for cross-institutional verification.

[0015] Therefore, there is an urgent need for a full lifecycle competency data collection and storage system that can systematically analyze the relationship between competencies and observable behaviors, support multi-source data collection, ensure data quality, adapt to offline assessment scenarios, and guarantee data credibility, so as to improve the scientificity, consistency and reliability of competency assessment. Summary of the Invention

[0016] To address the current problems in the consistent interpretation of competencies and observable behaviors (OBs), and to significantly reduce the difficulty and workload of interpreting competencies and OBs for high-risk and complex positions, thereby improving the efficiency of training and assessment, this invention provides a full lifecycle competency data collection and storage system based on observable behavior classification.

[0017] The technical solution adopted by this invention to solve its technical problem is: to provide a full life cycle competency data collection and storage system through observable behavior classification, including a competency-OB classification and parsing module, a multi-source data collection module, a dedicated classification database cluster, a data quality assurance module, an offline-online integrated module, a blockchain evidence storage interface, and a standardized API gateway.

[0018] Preferably, the competency-OB classification and analysis module is used to analyze competencies in two dimensions: technical / non-technical competency and crystallized / fluid memory, and to analyze observable behaviors (OB) in six dimensions: crystallized / fluid memory, system / element attributes, rigid / flexible performance, process / outcome performance, cognitive / motor skills, and attitude-value-personality traits, thereby generating a structured classification framework.

[0019] Preferably, the multi-source data acquisition module is used to collect training and assessment data in real time from simulators, LMS, and examination systems, and automatically label the raw data at the acquisition end with nine-dimensional competency labels and OB classification labels according to the structured classification framework, forming a highly structured data stream.

[0020] Preferably, the dedicated classification database cluster consists of a competency dimension library, an OB classification standard library, and a classification application library. It adopts a three-layer architecture of MySQL master-slave + Redis distributed cache + Elasticsearch retrieval, supporting TB-level data storage, millisecond-level query and second-level write.

[0021] Preferably, the data quality assurance module is used to implement format verification, classification rule verification, cross-validation and transaction consistency control in the three stages of collection, writing and reading to ensure that the data reliability is ≥0.8 and the validity is ≥0.75.

[0022] Preferably, the offline-online integrated module is used to generate encrypted and compressed offline classified data packets, supporting breakpoint resume, incremental synchronization and local SQLite caching, realizing full-function offline evaluation and automatic reconciliation after transmission when the network is interrupted.

[0023] Preferably, the blockchain evidence storage interface is used to write the key competency assessment results, OB classification labels and traceability paths into the consortium blockchain to ensure that the assessment records are tamper-proof and can be verified across institutions.

[0024] Preferably, the standardized API gateway is used to provide RESTful interfaces and OAuth2.0 security authentication to external systems, enabling plug-and-play functionality for external LMS, SMS, and QMS systems.

[0025] Preferably, the structured classification framework is distributed in JSON Schema form at the data collection end, defining the enumeration values, data types, and dependencies of 9 competency dimension fields and 6 OB dimension fields to ensure a unified output format from different data sources.

[0026] Preferably, the multi-source data acquisition module has a built-in dynamic mapping table that can automatically parse and map traditional unstructured simulator QAR messages and instructor handwritten comments to OB classification labels, with a parsing accuracy of ≥92%.

[0027] Preferably, the dedicated classification database cluster partitions the competency dimension data by "dimension + time" and the OB data by "behavior type + assessment stage", and uses covering indexes and materialized views to make the response time of complex association queries ≤200ms.

[0028] Preferably, the data quality assurance module includes a real-time verification submodule and a cross-validation submodule. The real-time verification submodule is used to execute the classification rule engine before data is written to reject non-compliant records. The cross-validation submodule is used to perform Kappa consistency checks on the simulator scores and theoretical scores of the same trainee.

[0029] Preferably, the offline-online integrated module deploys an embedded SQLite library on the local terminal, pre-stores a nine-dimensional competency standard table and an OB taxonomy table, and supports offline evaluation scoring, data caching, and difference merging after network recovery.

[0030] Preferably, the blockchain evidence storage interface adopts the Fabric consortium blockchain, which writes the trainee ID, competency score, OB tag hash, instructor digital signature and timestamp of each evaluation onto the chain, supports block generation within 3 seconds and a single block capacity of ≥1000 evaluation records.

[0031] Preferably, the standardized API gateway provides OpenAPI 3.0 specification documentation, supports export in multiple formats such as Excel, CSV, JSON, and XML, and has a built-in rate limiter to ensure an average response time of ≤50ms when there are ≥2000 concurrent requests / second.

[0032] Preferably, the system supports horizontal scaling: the number of classification and parsing Pods can be increased on demand through Kubernetes container orchestration, so that the peak concurrent classification processing capacity is ≥10,000 items / second, and the scaling process is completed with zero downtime.

[0033] Preferably, the system supports a multi-tenant mode, where each tenant can customize the competency dimension weights and OB grading colors, but the system-level classification fields must not be deleted to ensure cross-tenant data comparability.

[0034] By employing the above technical solutions, this invention significantly reduces the difficulty and workload of consistent interpretation of competencies and OBs for high-risk and highly complex positions by constructing a system based on the interrelationships, hierarchical structure, and classification logic of competencies and OBs. Through multi-source data collection and a structured classification framework, it improves the efficiency of training and assessment. Through a data quality assurance module and blockchain notarization interface, it guarantees the quality of training and assessment at the system level. Through a dedicated classification database cluster and standardized API gateway, it achieves consistency and coherence of professional talent training data across the entire career lifecycle. By supporting horizontal scaling and multi-tenant modes, it enables simultaneous response to the dynamic needs of professional talent training within the operating system, empowered by big data and AI. Furthermore, by systematically classifying and combining competencies and OBs, it accurately predicts key capability breakthroughs and growth bottlenecks at crucial growth stages, such as technical skills development, complex problem-solving, in-depth technological development, and experience asset transformation. Attached Figure Description

[0035] Figure 1 This is a system block diagram of the Competency-OB Classification Analysis Module; Detailed Implementation

[0036] This application will be described in further detail so that those skilled in the art can better understand and implement the invention, but the embodiments described are not intended to limit the invention.

[0037] Example 1: This system is a full lifecycle competency data acquisition and storage system that classifies observable behaviors. The system consists of multiple functional modules to achieve efficient and accurate competency data management.

[0038] The system's overall architecture includes a competency-OB classification and parsing module, a multi-source data acquisition module, a dedicated classification database cluster, a data quality assurance module, an offline-online integrated module, a blockchain evidence storage interface, and a standardized API gateway. These modules work together to form a complete competency data ecosystem.

[0039] As attached Figure 1As shown, the Competency-OB Classification and Analysis module is responsible for multi-dimensional analysis of competencies and observable behaviors. This module analyzes competencies using two dimensions: technical / non-technical competence and crystallized / fluid memory. It also analyzes observable behaviors (OBs) using six dimensions: crystallized / fluid memory, system / element attributes, rigid / flexible performance, process / outcome performance, cognitive / motor skills, and attitude-values-personality traits, generating a structured classification framework. This framework is distributed in JSON Schema format at the data acquisition end, defining the enumeration values, data types, and dependencies of the nine competency dimension fields and the six OB dimension fields, ensuring a consistent output format across different data sources. This structured classification method makes data acquisition more standardized, facilitating subsequent processing and analysis.

[0040] The multi-source data acquisition module collects training and assessment data in real time from simulators, LMS, and examination systems. At the acquisition end, the raw data is automatically labeled with nine competency dimensions and OB category labels according to a structured classification framework, forming a highly structured data stream. This module has a built-in dynamic mapping table that can automatically parse and map traditional unstructured simulator QAR messages and instructor handwritten comments to OB category labels, achieving an accuracy rate of over 92%. This automated data acquisition and labeling method significantly reduces manual intervention and improves data processing efficiency.

[0041] The dedicated classification database cluster consists of a competency dimension library, an OB classification standard library, and a classification application library. It employs a three-tier architecture of MySQL master-slave replication, Redis distributed caching, and Elasticsearch retrieval, supporting terabyte-level data storage, millisecond-level queries, and second-level writes. The cluster partitions competency dimension data by "dimension + time" and OB data by "behavioral type + evaluation stage," using covering indexes and materialized views to ensure complex join queries have a response time of no more than 200ms. This high-performance database architecture ensures the system maintains excellent responsiveness even when processing large amounts of data.

[0042] The data quality assurance module implements format validation, classification rule verification, cross-validation, and transaction consistency control in the three stages of data acquisition, writing, and reading to ensure that data reliability is no less than 0.8 and validity is no less than 0.75. This module includes three sub-modules: a real-time validation sub-module executes the classification rule engine before data writing to reject non-compliant records; a cross-validation sub-module performs Kappa consistency checks on the simulator scores and theoretical scores of the same trainee, triggering manual review if the Kappa value is less than 0.75; and a transaction control sub-module uses distributed transactions to ensure the atomicity of cross-database writes, automatically rolling back and tracing upwards to the first-level node in case of failure. These measures collectively ensure high data quality and consistency.

[0043] The offline-online integrated module generates encrypted and compressed offline classification data packets, supporting breakpoint resumption, incremental synchronization, and local SQLite caching. This enables full-featured offline assessment during network interruptions and automatic reconciliation after transmission. The module deploys an embedded SQLite library on the local terminal, pre-stores a nine-dimensional competency standard table and an OB taxonomy table, and supports assessment scoring, data caching, and difference merging after network recovery in offline mode. This design allows the system to operate reliably in unstable network environments, ensuring no data loss.

[0044] The blockchain evidence storage interface uses the Fabric consortium blockchain, writing the trainee ID, competency score, OB tag hash, instructor digital signature, and timestamp for each assessment onto the chain. It supports block generation within 3 seconds, with each block containing at least 1000 assessment records. This application of blockchain technology ensures that assessment records are immutable and verifiable across institutions, enhancing the system's credibility.

[0045] The standardized API gateway provides OpenAPI 3.0 specification documentation, supports exporting to multiple formats including Excel, CSV, JSON, and XML, and has a built-in rate limiter to ensure an average response time of no more than 50ms when there are no fewer than 2000 concurrent requests per second. This standardized interface design allows for plug-and-play use of external LMS, SMS, and QMS systems, facilitating system integration with other platforms.

[0046] In terms of scalability, this system supports horizontal scaling: by using Kubernetes container orchestration, the number of classification and parsing Pods can be increased on demand, ensuring a peak concurrent classification processing capacity of no less than 10,000 records per second, with zero downtime during the scaling process. This design allows the system to dynamically adjust resource allocation according to load conditions, ensuring good performance even during peak periods.

[0047] Regarding multi-tenancy support, this system supports a multi-tenant mode, where each tenant can customize competency dimension weights and OB grading colors, but system-level classification fields must not be deleted to ensure cross-tenant data comparability. This flexibility allows the system to adapt to the needs of different organizations while maintaining data standardization and comparability.

[0048] In terms of application equipment, this system can be deployed in electronic devices such as flight simulators, VR trainers, portable instructor terminals, or nuclear power full-range simulators, which have built-in local copies for caching a dedicated classification database. This design enables the system to operate seamlessly on a variety of professional training equipment, expanding its application scope.

[0049] This system is applicable to the full career lifecycle training and assessment of civil aviation pilots, air traffic controllers, aircraft maintenance personnel, nuclear power plant operators, and rail transit dispatchers. By collecting, storing, and analyzing the competency data of these high-risk professionals throughout their entire career lifecycle, this system can help relevant organizations better assess and improve personnel's professional capabilities, thereby enhancing the industry's safety level.

[0050] Example 2: A method for collecting full lifecycle competency data through observable behavior classification. This method is suitable for high-risk, highly complex occupational positions and is used to transform unstructured training data into quantitative structured competency and OB data, enabling reliable cross-organizational assessment. The method specifically includes the following steps:

[0051] S1: At the data acquisition end, a structured classification framework is deployed, defining the field enumerations and dependencies for 9-dimensional competencies and 6-dimensional observables (OBs). This step provides a standardized framework for subsequent data acquisition and classification, ensuring that all acquisition points use consistent evaluation criteria and classification methods. This structured classification framework includes detailed competency dimension definitions and observable behavior classification standards, providing a basis for labeling during the data acquisition process.

[0052] S2: Raw data is captured in real-time from simulators, LMS, and examination systems, and automatically labeled with competency dimension labels and OB category labels at edge nodes, forming a highly structured data stream. The automatic labeling process combines a configurable mapping table with a natural language processing model. Unstructured text is first semantically embedded, and then matched to the closest OB label based on cosine similarity. The system requires a cosine similarity of 0.85 or higher before label matching is performed to ensure labeling accuracy. This automatic labeling mechanism significantly improves data processing efficiency, reduces manual intervention, and ensures the consistency and accuracy of category labels.

[0053] S3: Writes structured data streams to a dedicated categorized database cluster consisting of MySQL, Redis, and Elasticsearch, supporting terabyte-level storage and millisecond-level query speeds. This database cluster employs a distributed architecture: MySQL handles relational data storage and transaction processing, Redis provides high-speed caching and real-time data processing, and Elasticsearch supports full-text search and complex multi-dimensional aggregation analysis. This combined architecture satisfies both the need for large-capacity data storage and ensures high query performance, making it particularly suitable for multi-dimensional analysis and rapid retrieval of competency data.

[0054] S4: During the write and read processes, classification rule validation, cross-validation, and transaction consistency control are performed to ensure data reliability ≥ 0.8 and validity ≥ 0.75. Cross-validation specifically includes: performing a Kappa test on the simulator scores and theoretical scores of the same trainee; if the Kappa value is less than 0.75, it is marked as an anomaly and manual review is triggered; performing a Pearson correlation test on OB frequency and competency scores; if the correlation coefficient is less than 0.4, it is determined as a classification failure and the batch of data is rolled back. These validation mechanisms ensure data quality and consistency, preventing erroneous data from entering the system.

[0055] S5: Generates encrypted offline data packets, supporting breakpoint resumption and incremental synchronization, enabling offline evaluation and reconciliation after data return. The offline data packets employ AES-256 encryption combined with Gzip compression technology, achieving a compression rate of 80% or higher, and support multi-dimensional segmentation export by trainee, course, and time range. This design allows data to be used in unstable network or offline environments while ensuring data security and integrity, making it particularly suitable for evaluation work in remote or network-constrained scenarios.

[0056] S6: Key assessment results and OB tag hashes are written to the blockchain to achieve cross-institutional immutable verification. The blockchain write operation employs an asynchronous pipelined processing method: the system first batches 1000 records locally, then completes hash calculation, digital signature, and on-chain submission within 3 seconds. Failed records are automatically transferred to a retry queue and can be retried up to 3 times. This mechanism ensures the credibility and immutability of the assessment data, enabling different institutions to establish mutually trusting assessment systems, which is particularly important in personnel assessments for high-risk positions.

[0057] S7: Provides data services to external systems through standardized APIs, enabling plug-and-play functionality. These APIs follow RESTful design, offering functions such as data querying, analysis, and export, and supporting multiple authentication methods and access control policies to ensure the security and availability of data services. Standardized APIs significantly improve the system's scalability and interoperability, allowing competency data to be easily integrated into other human resource management systems, training systems, or assessment systems.

[0058] This method constructs a complete competency data collection and assessment system through the above seven steps. It is particularly suitable for high-risk and highly complex occupational positions such as aviation, medical care, and nuclear power. It can transform unstructured data generated during training into quantifiable and comparable competency indicators and observable behavioral data, and ensure the credibility of the data through blockchain technology, thereby achieving standardized assessment across institutions.

[0059] Example 3: This example provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a life-cycle competency data acquisition method based on observable behavior classification.

[0060] The computer-readable storage medium can be a non-volatile storage device such as a hard disk, solid-state drive, optical disk, USB flash drive, or flash memory card, or a temporary storage device such as RAM. The computer program stored on the storage medium contains an instruction set, which, after being loaded and executed by the processor, enables the following steps:

[0061] S1: Deploy a structured classification framework at the acquisition end, defining the field enumeration and dependency relationships of 9-dimensional competencies and 6-dimensional OBs;

[0062] S2: Real-time capture of raw data from simulators, LMS, and examination systems, and automatic labeling of raw data as competency dimension labels and OB category labels at edge nodes to form a highly structured data stream;

[0063] In this step, automatic tagging employs a combination of configurable mapping tables and natural language processing models. For unstructured text, the program first performs semantic embedding, then calculates the matching degree with preset OB tags based on cosine similarity. When the cosine similarity is greater than or equal to 0.85, the text is matched to the closest OB tag. This automatic tagging mechanism significantly improves data processing efficiency and reduces the need for manual intervention.

[0064] S3: Writes structured data streams to a dedicated category database cluster consisting of MySQL, Redis, and Elasticsearch, supporting TB-level storage and millisecond-level queries;

[0065] S4: Perform taxonomic rule verification, cross-validation, and transaction consistency control during the write and read processes to ensure data reliability ≥ 0.8 and validity ≥ 0.75;

[0066] In this step, cross-validation includes two key validation processes: First, a Kappa test is performed on the simulator scores and theoretical scores of the same trainee. If the Kappa value is less than 0.75, the program will automatically mark the data as an anomaly and trigger a manual review process. Second, a Pearson correlation test is performed on the OB frequency and competency score. If the correlation coefficient is less than 0.4, it is determined to be a classification failure, and the program will automatically roll back the batch of data to ensure data quality.

[0067] S5: Generates encrypted offline data packets, supports breakpoint resume and incremental synchronization, and enables offline evaluation and reconciliation after data return.

[0068] In a preferred embodiment, offline data packets are processed using AES-256 encryption combined with Gzip compression technology, achieving a compression rate of 80% or higher, significantly reducing data transmission and storage burden. The program supports exporting data in multiple slices based on trainee, course, time range, etc., meeting the offline assessment needs of different scenarios. This design ensures the system maintains high efficiency even in environments with unstable network connections and guarantees data security.

[0069] S6: Write key assessment results and OB tag hashes into the blockchain to achieve cross-institutional immutable verification;

[0070] In this step, the blockchain write operation employs an asynchronous pipelined processing mechanism: the program first batches 1000 records locally, then completes the hash calculation, digital signature, and on-chain commit operation within 3 seconds. For failed records, the program automatically transfers them to a retry queue, with a maximum of 3 retries. This design ensures both the efficiency of data upload to the blockchain and the integrity and reliability of the data.

[0071] S7: Provides data services to external systems through standardized APIs, enabling plug-and-play functionality.

[0072] The program stored on the computer-readable storage medium, through the implementation of the above steps, constructs a complete competency data acquisition, processing, and verification system. During program execution, particular emphasis is placed on data accuracy and consistency, ensuring the quality of the acquired data through multiple verification mechanisms. Simultaneously, modern cryptography and distributed storage technologies are employed to guarantee data security and reliability.

[0073] In practical applications, this program can be deployed on servers, workstations, or edge computing devices, with flexible resource allocation configured according to actual needs. Its modular design gives it excellent scalability, enabling it to adapt to competency assessment scenarios of varying scales and complexities.

[0074] Example 4: This example provides an electronic device, which includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it can implement a method for collecting full lifecycle competency data through observable behavior classification.

[0075] The electronic device can be at least one of a flight simulator, VR trainer, portable instructor terminal, or nuclear power full-range simulator. The device has a built-in local copy for caching a dedicated classification database, ensuring efficient access to and use of the classification data even in environments with unstable network connections or offline conditions.

[0076] The method implemented by the processor when executing a program includes the following steps:

[0077] S1: A structured classification framework is deployed at the data acquisition end, defining the field enumerations and dependencies of 9-dimensional competencies and 6-dimensional OBs. This step establishes a unified standard and specification for subsequent data acquisition, ensuring the consistency of the data structure.

[0078] S2: Raw data is captured in real time from simulators, LMS, and examination systems, and automatically labeled with competency dimension labels and OB category labels at edge nodes, forming a highly structured data stream. This automatic labeling uses a configurable mapping table combined with a natural language processing model to first perform semantic embedding on unstructured text, and then match the closest OB label based on a cosine similarity greater than or equal to 0.85. This automatic labeling mechanism significantly improves data processing efficiency and reduces the workload of manual annotation.

[0079] S3: Writes structured data streams to a dedicated categorized database cluster consisting of MySQL, Redis, and Elasticsearch, supporting terabyte-level storage and millisecond-level queries. The database cluster adopts a distributed architecture, with MySQL serving as a relational database to store structured data, Redis providing high-speed caching, and Elasticsearch handling full-text search and complex queries. The three work together to ensure that the system maintains high performance even with large data volumes.

[0080] S4: During the write and read processes, classification rule validation, cross-validation, and transaction consistency control are performed to ensure that data reliability is greater than or equal to 0.8 and validity is greater than or equal to 0.75. Cross-validation specifically includes: performing a Kappa test on the simulator scores and theoretical scores of the same trainee; if the Kappa value is less than 0.75, it is marked as an anomaly and manual review is triggered; performing a Pearson correlation test on OB frequency and competency scores; if the correlation coefficient is less than 0.4, it is determined as a classification failure and the batch of data is rolled back. This multi-validation mechanism effectively ensures data quality.

[0081] S5: Generates encrypted offline data packets, supporting breakpoint resumption and incremental synchronization, enabling offline evaluation and reconciliation after data return. The offline data packets use AES-256 encryption combined with Gzip compression, achieving a compression rate of 80% or higher, and support multi-dimensional segmentation export by trainee, course, and time range. This feature allows the system to adapt to environments with unstable network conditions, ensuring the continuity of data collection.

[0082] S6: Key assessment results and OB tag hashes are written to the blockchain, enabling cross-institutional, tamper-proof verification. The blockchain write operation employs an asynchronous pipeline: 1000 records are first batched locally, then hash calculation, digital signature, and on-chain submission are completed within 3 seconds. Failed records are automatically moved to a retry queue and can be retried up to 3 times. Blockchain technology ensures the security and credibility of the assessment data.

[0083] S7: Provides data services to external systems through standardized APIs, enabling plug-and-play functionality. The APIs adopt a RESTful architecture, providing a unified data access interface, supporting multiple authentication methods and access control policies, and facilitating integration with other systems.

[0084] Inside electronic devices, the processor and memory are connected via a high-speed bus to ensure real-time data processing. The memory consists of two parts: ROM and RAM. ROM stores the operating system and basic programs, while RAM is used for runtime data processing. Electronic devices are also equipped with dedicated data processing units to accelerate natural language processing and data encryption / decryption operations.

[0085] When used as a flight simulator, the electronic device can collect pilot operation data in real time and classify and evaluate it according to preset competency standards; when used as a VR trainer, it can capture trainees' behavioral data in a virtual environment; when used as a portable instructor terminal, it facilitates instructors to record and evaluate trainees' performance in a mobile environment; and when used as a nuclear power full-range simulator, it can comprehensively monitor operators' ability to cope with various working conditions.

[0086] The dedicated local copy of the classification database built into electronic devices maintains consistency with the central database through an incremental synchronization mechanism, while also enabling localized data processing, reducing network dependencies, and improving system response speed and reliability. The local copy adopts the same structural design as the central database, but is optimized for local storage and access to adapt to the hardware conditions of different devices.

Claims

1. A life-cycle competency data acquisition and storage system based on observable behavior classification, characterized in that, The system includes: The Competency-OB Classification Analysis Module is used to analyze competencies in two dimensions: technical / non-technical competencies and crystallized / fluid memory. It also analyzes observable behaviors (OBs) in six dimensions: crystallized / fluid memory, system / element attributes, rigid / flexible performance, process / outcome performance, cognitive / motor skills, and attitude-value-personality traits, generating a structured classification framework. The multi-source data acquisition module is used to collect training and assessment data from simulators, LMS, and examination systems in real time, and automatically label the raw data at the acquisition end with the nine-dimensional competency labels and OB classification labels according to the structured classification framework, forming a highly structured data stream; The dedicated classification database cluster consists of a competency dimension library, an OB classification standard library, and a classification application library. It adopts a three-layer architecture of MySQL master-slave + Redis distributed cache + Elasticsearch retrieval, supporting TB-level data storage, millisecond-level query and second-level write. The data quality assurance module is used to implement format verification, classification rule verification, cross-validation, and transaction consistency control in the three stages of data collection, writing, and reading, ensuring data reliability ≥ 0.8 and validity ≥ 0.

75. The offline-online integrated module is used to generate encrypted and compressed offline classified data packets, supporting breakpoint resume, incremental synchronization and local SQLite caching, realizing full-featured offline evaluation and automatic reconciliation after transmission when the network is interrupted; The blockchain evidence storage interface is used to write key competency assessment results, OB classification labels and traceability paths into the consortium blockchain to ensure that the assessment records are tamper-proof and can be verified across institutions. A standardized API gateway is used to provide RESTful interfaces and OAuth2.0 security authentication to external systems, enabling plug-and-play functionality for external LMS, SMS, and QMS systems.

2. The life-cycle competency data acquisition and storage system based on observable behavior classification according to claim 1, characterized in that, The structured classification framework is distributed in JSON Schema format at the data collection end, defining the enumeration values, data types, and dependencies of 9 competency dimension fields and 6 OB dimension fields to ensure a unified output format from different data sources.

3. A life-cycle competency data acquisition and storage system based on observable behavior classification as described in claim 1 or 2, characterized in that, The multi-source data acquisition module has a built-in dynamic mapping table that can automatically parse and map traditional unstructured simulator QAR messages and instructor handwritten comments to OB classification labels, with a parsing accuracy of ≥92%.

4. A life-cycle competency data acquisition and storage system based on observable behavior classification according to any one of claims 1-3, characterized in that, The dedicated classification database cluster partitions competency dimension data by "dimension + time" and OB data by "behavior type + assessment stage", and uses covering indexes and materialized views to make the response time of complex relational queries ≤200ms.

5. A life-cycle competency data acquisition and storage system based on observable behavior classification according to any one of claims 1-4, characterized in that, The data quality assurance module includes a real-time verification submodule, which executes a taxonomy rule engine before data is written to reject non-compliant records; The cross-validation submodule is used to perform Kappa consistency checks on the simulator scores and theoretical scores of the same trainee. If Kappa < 0.75, manual review is triggered. The transaction control submodule is used to ensure the atomicity of cross-database writes using distributed transactions. In case of failure, it automatically rolls back and traces back to the first-level node.

6. A life-cycle competency data acquisition and storage system based on observable behavior classification according to any one of claims 1-5, characterized in that, The offline-online integrated module deploys an embedded SQLite library on the local terminal, pre-stores a nine-dimensional competency standard table and an OB taxonomy table, and supports offline evaluation scoring, data caching, and difference merging after network recovery.

7. A life-cycle competency data acquisition and storage system based on observable behavior classification according to any one of claims 1-6, characterized in that, The blockchain evidence storage interface uses the Fabric consortium blockchain to write the trainee ID, competency score, OB tag hash, instructor digital signature and timestamp of each evaluation onto the chain, supporting block generation within 3 seconds and a single block capacity of ≥1000 evaluation records.

8. A life-cycle competency data acquisition and storage system based on observable behavior classification according to any one of claims 1-7, characterized in that, The standardized API gateway provides OpenAPI 3.0 specification documentation, supports exporting in multiple formats such as Excel, CSV, JSON, and XML, and has a built-in rate limiter to ensure an average response time of ≤50ms when there are ≥2000 concurrent requests / second.

9. A life-cycle competency data acquisition and storage system based on observable behavior classification according to any one of claims 1-8, characterized in that, The system supports horizontal scaling: the number of classification and parsing Pods can be increased on demand through Kubernetes container orchestration, so that the peak concurrent classification processing capacity is ≥10,000 items / second, and the scaling process is completed with zero downtime.

10. A life-cycle competency data acquisition and storage system based on observable behavior classification according to any one of claims 1-9, characterized in that, The system supports a multi-tenant mode, where each tenant can customize the competency dimension weights and OB grading colors, but the system-level classification fields must not be deleted to ensure cross-tenant data comparability.

11. A method for collecting full life-cycle competency data through observable behavior classification, characterized in that, include: S1. Deploy a structured classification framework at the data acquisition end, defining the field enumeration and dependency relationships of 9-dimensional competencies and 6-dimensional OBs; S2. Real-time capture of raw data from simulators, LMS, and examination systems, and automatic labeling of raw data as competency dimension labels and OB category labels at edge nodes to form a highly structured data stream; S3. Write structured data streams to a dedicated classification database cluster consisting of MySQL, Redis, and Elasticsearch, supporting TB-level storage and millisecond-level queries; S4. Perform classification rule verification, cross-validation, and transaction consistency control during the writing and reading processes to ensure data reliability ≥ 0.8 and validity ≥ 0.75; S5. Generate encrypted offline data packets, supporting breakpoint resume and incremental synchronization, enabling offline evaluation and reconciliation after data return. S6. Write the key assessment results and OB tag hashes into the blockchain to achieve cross-institutional immutable verification; S7. Provide data services to external systems through standardized APIs, enabling plug-and-play functionality.

12. The method for collecting full life-cycle competency data through observable behavior classification according to claim 11, characterized in that, The automatic tagging described in S2 uses a configurable mapping table + natural language processing model to first perform semantic embedding on unstructured text, and then match the closest OB tag based on a cosine similarity of ≥0.

85.

13. A method for collecting full life-cycle competency data through observable behavior classification according to claim 11 or 12, characterized in that, The cross-validation described in S4 includes: performing a Kappa test on the simulator scores and theoretical scores of the same trainee; if Kappa < 0.75, it is marked as an anomaly and manual review is triggered; performing a Pearson correlation test on the OB frequency and competency scores; if the correlation coefficient < 0.4, it is determined as a classification failure and the batch of data is rolled back.

14. A method for collecting full life-cycle competency data through observable behavior classification according to any one of claims 11-13, characterized in that, The offline data package described in S5 uses AES-256 encryption and Gzip compression, with a compression rate of ≥80%, and supports multi-dimensional slicing and export by trainee, course, and time range.

15. A method for collecting full life-cycle competency data through observable behavior classification according to any one of claims 11-14, characterized in that, The blockchain write operation described in S6 adopts an asynchronous pipeline: first, 1,000 records are batched locally, and then the hash calculation, digital signature and on-chain submission are completed within 3 seconds. Failed records are automatically transferred to the retry queue and can be retried up to 3 times.

16. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method according to any one of claims 11-15.

17. An electronic device comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method of any one of claims 11-15.

18. The electronic device of claim 17, wherein the electronic device is at least one of a flight simulator, a VR trainer, a portable instructor terminal, or a nuclear power full-range simulator, and has a built-in local copy for caching the dedicated classification database of claim 1.

19. The use of the system according to any one of claims 1-10 in the full career life cycle training and assessment of civil aviation pilots, air traffic controllers, aircraft maintenance personnel, nuclear power plant operators, and rail transit dispatchers.

20. The method of any one of claims 11-15 is used in high-risk, highly complex occupational positions to transform unstructured training data into quantitative structured competency and OB data, and to achieve cross-organizational credible assessment.

Images