Seaman comprehensive evaluation method based on big data

Abstract

The invention discloses a crew comprehensive evaluation method based on big data, and relates to the field of big data processing, and the method comprises the steps: collecting static qualification, dynamic performance, ship operation and external data through encryption to form an original data set, and carrying out the classification cleaning and structural processing to generate standardized data; fusing subjective and objective weights by adopting an analytic hierarchy process and an entropy weight method, dynamically adjusting index weights in combination with post characteristics, respectively performing normalization and fuzzy quantization processing on quantitative / qualitative indexes, and performing weighted calculation on a criterion layer score and a comprehensive evaluation total score; and data reliability is verified through anomaly detection and is dynamically corrected, and finally a grading evaluation result and personalized improvement suggestions are output. The method has the advantages that based on the big data technology, through a hierarchical evaluation system, dynamic weight fusion and an anomaly verification mechanism, a comprehensive evaluation mode which is accurate, adapts to a shipping scene and can be dynamically iterated is constructed, and the problems that traditional evaluation is high in subjectivity and single in dimension are solved.

Description

Technical Field

[0001] This invention relates to the field of big data processing, and in particular to a comprehensive evaluation method for crew members based on big data. Background Technology

[0002] As competition intensifies in the global shipping industry and safety and environmental requirements become increasingly stringent, the International Maritime Organization's International Convention on Standards of Training, Certification and Watchkeeping for Seafarers is constantly being revised, prompting ship management to shift from single-skill assessments to comprehensive competency evaluations. This system typically encompasses dimensions such as professional skills, safety awareness, emergency response capabilities, psychological qualities, teamwork, and cross-cultural communication, combining theoretical examinations, practical assessments, simulator tests, and peer reviews, and incorporating technologies such as artificial intelligence and big data analytics to achieve dynamic evaluation.

[0003] Current methods for evaluating seafarers are ill-suited to the demands of digital transformation in the shipping industry. Most methods rely on a single data source, primarily static qualification documents and manual records, lacking comprehensive collection of dynamic performance and vessel operation-related data. Furthermore, data transmission lacks encryption and standardization, exhibiting weak unstructured data parsing capabilities, making them susceptible to redundancy and missing values ​​that can affect evaluation accuracy. Evaluation systems are rigid, with indicators failing to adequately cover core dimensions such as safety performance and overall competence. Weight calculations rely solely on subjective experience or single objective data points, lacking dynamic adjustment mechanisms and failing to adapt to different positions and shipping routes. Simultaneously, the lack of effective anomaly detection processes makes it difficult to identify data errors and weight biases. Evaluation results are mostly static outputs, lacking time decay coefficients to highlight recent performance, and cannot generate personalized improvement suggestions or dynamic updates. Overall, these methods are highly subjective, inaccurate, and poorly adaptable, failing to meet the needs of scientific management and seafarer competence enhancement. Summary of the Invention

[0004] To improve existing methods, a comprehensive evaluation method for seafarers based on big data is proposed. This method relies on big data technology and constructs a comprehensive evaluation model that is accurate, adaptable to shipping scenarios, and dynamically iterative through a hierarchical evaluation system, dynamic weight fusion, and anomaly verification mechanism.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A comprehensive evaluation method for crew members based on big data includes:

[0007] The system collects static basic data, dynamic performance data, ship operation-related data and external auxiliary data of crew members through IoT terminals. The data is transmitted in encrypted form during collection, and the source and timestamp are marked to form the original dataset.

[0008] Based on the classification and processing of structured, semi-structured and unstructured data, the system removes redundancy, fills in missing values ​​and unifies the format of structured data, transforms semi-structured data, parses unstructured data and extracts core information, and generates a standardized dataset.

[0009] A hierarchical comprehensive evaluation system for crew members is constructed. The target layer is the comprehensive evaluation result, the criteria layer includes static qualifications, dynamic performance, safety performance, and comprehensive quality, and the indicator layer refines the corresponding indicators and obtains the data docking relationship.

[0010] The subjective weights of each criterion and indicator layer are calculated using the analytic hierarchy process (AHP), and the objective weights of each indicator are calculated using the entropy weight method. A dynamic adjustment coefficient is introduced, and the subjective and objective weights are integrated based on the characteristics of crew positions and shipping routes to generate the final comprehensive weights of each indicator.

[0011] For different types of evaluation indicators, corresponding quantitative methods are adopted. Quantitative indicators are normalized, and qualitative indicators are quantified by level and fuzziness is corrected. A time decay coefficient is introduced to highlight recent performance.

[0012] Based on the normalized indicator data and the comprehensive weight of each indicator, a weighted summation algorithm is used to calculate the score of each criterion layer, and the total comprehensive evaluation score of the crew is calculated according to the weight of the criterion layer.

[0013] The evaluation data is verified by anomaly detection algorithms to identify abnormal fluctuations, collection errors and weight deviations. The abnormal items are traced back to their source and data is collected again, parameters are adjusted or weights are corrected and recalculated.

[0014] Evaluation results are output in the form of tiered reports, personalized improvement suggestions are generated for weak indicators, and evaluation results are updated regularly through a dynamic update mechanism, adjusting indicators and weights.

[0015] Preferably, the step of collecting static basic data, dynamic performance data, ship operation-related data, and external auxiliary data of crew members through IoT terminals, encrypting the data transmission during collection, and labeling the source and timestamp to form the original dataset specifically includes:

[0016] Through ship IoT terminals, shipping management platforms, and manual data entry channels, static basic data, dynamic performance data, ship operation-related data, and external auxiliary data of crew members are collected simultaneously.

[0017] During the data collection process, cross-platform data transmission is carried out through a data encryption transmission protocol, and the collected data is marked with source identification and timestamp to form the original data set.

[0018] Preferably, the step of classifying and processing structured, semi-structured, and unstructured data, including removing redundancy, filling in missing values, and unifying the format of structured data, converting it into semi-structured data, parsing unstructured data and extracting core information, and generating a standardized dataset, specifically includes:

[0019] The original dataset is processed using corresponding preprocessing strategies for structured, semi-structured, and unstructured data.

[0020] For structured data, duplicate records are deleted, key field gaps are filled, and data of different formats are uniformly converted into a preset format through standardization processing;

[0021] For semi-structured data, core information is extracted using tag parsing technology, a unified data field mapping relationship is constructed, and the data is transformed into structured data.

[0022] For unstructured data, text records are segmented and keywords are extracted; image certificate data is subjected to optical character recognition and information verification; and audio records are subjected to speech-to-text conversion and semantic analysis.

[0023] After preprocessing, a standardized crew evaluation dataset is generated.

[0024] Preferably, the construction of a hierarchical comprehensive evaluation system for crew members includes the following: the target layer is the comprehensive evaluation result; the criteria layer includes static qualifications, dynamic performance, safety performance, and comprehensive quality; and the indicator layer refines the corresponding indicators and obtains data docking relationships.

[0025] A hierarchical comprehensive evaluation system for seafarers is constructed, which is divided into three levels: the target level, the criteria level, and the indicator level.

[0026] The target layer is the comprehensive evaluation result of the crew members, and the criteria layer includes static qualification dimension, dynamic duty performance dimension, safety performance dimension and comprehensive quality dimension. Each criterion layer is independent of each other and covers all dimensions of crew member evaluation.

[0027] The indicator layer further refines the criteria layer. Under the static qualification dimension, indicators include competency certificate level, years of service, training compliance rate, and educational background matching. Under the dynamic performance dimension, indicators include duty compliance, equipment operation accuracy, work efficiency, and emergency response speed. Under the safety performance dimension, indicators include the number of safety violations, accident-related liability, safety inspection pass rate, and quality of hazard identification and rectification. Under the comprehensive quality dimension, indicators include teamwork score, communication and coordination ability, emergency psychological quality, and professional ethics evaluation.

[0028] Each indicator clearly defines the corresponding data source and reserves interfaces for indicator expansion, allowing for updates and adjustments to indicator items based on shipping industry standards.

[0029] Preferably, the step of calculating the subjective weights of each criterion layer and indicator layer using the analytic hierarchy process (AHP), calculating the objective weights of each indicator using the entropy weight method, introducing a dynamic adjustment coefficient, and integrating the subjective and objective weights based on the characteristics of the crew member's position and the shipping route to generate the final comprehensive weights of each indicator specifically includes:

[0030] The Analytic Hierarchy Process (AHP) was used to construct a judgment matrix by combining the experience of shipping experts, and the subjective weights of each criterion layer and indicator layer were calculated.

[0031] Based on the preprocessed evaluation dataset, the objective weight of each indicator is calculated using the entropy weight method, and the degree of influence of the indicator on the evaluation results is determined by analyzing the dispersion of the indicator data.

[0032] By introducing dynamic adjustment coefficients, subjective and objective weights are adaptively integrated based on crew member job type and navigation route characteristics to generate the final comprehensive weight of each indicator.

[0033] Preferably, the method of using corresponding quantification for different types of evaluation indicators, the normalization of quantitative indicators, the quantification of qualitative indicators by level and the correction of fuzziness, and the introduction of a time decay coefficient to highlight recent performance specifically include:

[0034] For different types of evaluation indicators, corresponding quantitative methods are adopted. Among them, quantitative indicators are directly based on the preprocessed data values, and the data are normalized to the [0,1] interval through linear transformation.

[0035] Qualitative indicators are quantified and assigned values ​​according to preset evaluation level standards, converting qualitative descriptions into corresponding numerical values, and the fuzziness of qualitative indicators is handled by combining fuzzy comprehensive evaluation method.

[0036] For data with differences in time dimension, a time decay coefficient is introduced to give high weight to recent data and low weight to long-term data, highlighting the impact of crew members’ recent performance on the evaluation results.

[0037] Preferably, the step of calculating the score of each criterion layer based on the normalized indicator data and the comprehensive weight of each indicator, and calculating the total score of the crew member's comprehensive evaluation based on the weight of each criterion layer specifically includes:

[0038] Based on the normalized indicator data and the comprehensive weight of each indicator, a weighted summation operation is performed on the criteria layer. The normalized value of each indicator is multiplied by the corresponding weight, and the summation is used to obtain the individual score of each criteria layer.

[0039] Based on the weight allocation results of the criteria layer, the scores of the criteria layer are weighted and summarized again to calculate the total score of the crew's comprehensive evaluation.

[0040] The system simultaneously generates a total score and scores for each criterion level item, records the contribution of each indicator to the criterion level score and the total score, and filters out the core positive and negative indicators that affect the evaluation results by sorting them.

[0041] Preferably, the step of verifying the evaluation data through anomaly detection algorithms, identifying abnormal fluctuations, acquisition errors, and weight deviations, tracing the source of anomalies, re-acquiring data, adjusting parameters, or correcting weights for secondary calculation specifically includes:

[0042] Anomaly detection algorithms are used to verify the data and intermediate results during the evaluation process, and to identify abnormal evaluation results caused by abnormal fluctuations in data, data acquisition errors, and weight allocation deviations.

[0043] For detected anomalies, trace the corresponding data source and processing steps, and perform secondary calculations by re-collecting data, adjusting preprocessing parameters, or correcting weighting coefficients;

[0044] Industry benchmarks and the average evaluation of crew members in the same position are used as references to correct for deviations in the preliminary evaluation results.

[0045] Preferably, the step of outputting evaluation results in the form of a tiered report, generating personalized improvement suggestions for the weakest indicators, and periodically updating the evaluation results through a dynamic update mechanism, adjusting the indicators and weights, specifically includes:

[0046] The revised comprehensive evaluation results of the crew members will be output in the form of a graded report, which will clearly indicate the crew member's evaluation level, performance ranking in each dimension, strength indicators and weakness indicators.

[0047] Personalized improvement suggestions are generated for the weaker indicators, a dynamic update mechanism for evaluation results is established, new data is collected regularly, the evaluation results are iteratively updated, and the evaluation indicator system and weight coefficients are adjusted according to shipping needs.

[0048] Compared with the prior art, the advantages of the present invention are:

[0049] A comprehensive and scientific evaluation system is built around big data, balancing comprehensiveness, accuracy, and dynamic adaptability. Through multi-dimensional data collection, encrypted transmission, and standardized processing of all data types, the integrity, security, and standardization of evaluation data sources are ensured. The hierarchical evaluation system covers core dimensions of crew members, with detailed indicators and reserved expansion interfaces to adapt to the needs of the shipping industry. A combination of subjective and objective weights with dynamically adjusted coefficients adapts to the characteristics of different positions and routes, while a time decay coefficient highlights recent performance, enhancing the relevance of the evaluation. Anomaly detection and secondary correction mechanisms mitigate data errors and weight biases, ensuring the reliability of results. Dynamic updates and personalized suggestions enable iterative optimization of evaluation results and precise improvement of weaknesses, breaking through the limitations of traditional evaluations that are highly subjective, have single dimensions, and are static. This provides data support for crew management, job allocation, and capability enhancement, adapting to the digital development needs of the shipping industry. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the method proposed in this invention;

[0051] Figure 2 This is a schematic diagram illustrating the comprehensive data collection of crew members proposed in this invention;

[0052] Figure 3 This is a schematic diagram of the data differentiation preprocessing proposed in this invention;

[0053] Figure 4 This is a schematic diagram illustrating the hierarchical comprehensive evaluation system for seafarers proposed in this invention;

[0054] Figure 5 This is a schematic diagram of the dynamic weight determination proposed in this invention;

[0055] Figure 6 This is a quantitative diagram of the indicator data proposed in this invention;

[0056] Figure 7 This is a schematic diagram of the comprehensive scoring calculation proposed in this invention;

[0057] Figure 8 This is a schematic diagram illustrating the correction of the evaluation results proposed in this invention;

[0058] Figure 9 This is a schematic diagram of the comprehensive evaluation results proposed in this invention. Detailed Implementation

[0059] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0060] See Figure 1 As shown, a comprehensive evaluation method for crew members based on big data includes:

[0061] Step 1: Collect static basic data, dynamic performance data, ship operation related data and external auxiliary data of crew members through IoT terminals. During the collection process, the data is transmitted in encrypted form, and the source and timestamp are marked to form the original dataset.

[0062] Step 2: Based on the classification and processing of structured, semi-structured and unstructured data, redundancy is removed, missing values ​​are filled in and format is unified for structured data, semi-structured data is converted, unstructured data is parsed and core information is extracted to generate a standardized dataset.

[0063] Step 3: Construct a hierarchical comprehensive evaluation system for crew members. The target layer is the comprehensive evaluation result, the criteria layer includes static qualifications, dynamic performance, safety performance, and comprehensive quality, and the indicator layer refines the corresponding indicators and obtains the data docking relationship.

[0064] Step 4: Calculate the subjective weights of each criterion layer and indicator layer using the analytic hierarchy process (AHP), calculate the objective weights of each indicator using the entropy weight method, introduce a dynamic adjustment coefficient, and integrate the subjective and objective weights based on the characteristics of the crew member's position and the shipping route to generate the final comprehensive weights of each indicator.

[0065] Step 5: For different types of evaluation indicators, adopt corresponding quantitative methods. Quantitative indicators are normalized, and qualitative indicators are quantified by level and fuzziness is corrected. A time decay coefficient is introduced to highlight recent performance.

[0066] Step Six: Based on the normalized indicator data and the comprehensive weight of each indicator, calculate the score of each criterion layer using a weighted summation algorithm, and calculate the total comprehensive evaluation score of the crew members according to the weight of each criterion layer;

[0067] Step 7: Verify the evaluation data through anomaly detection algorithms, identify abnormal fluctuations, collection errors, and weight deviations, trace the source of anomalies, recollect data, adjust parameters, or correct weights for secondary calculation;

[0068] Step 8: Output the evaluation results in the form of a tiered report, generate personalized improvement suggestions for the weakest indicators, and update the evaluation results regularly through a dynamic update mechanism, adjusting the indicators and weights.

[0069] See Figure 2 As shown, static basic data of crew members, dynamic performance data, ship operation-related data, and external auxiliary data are collected through IoT terminals. Data is encrypted during collection and transmitted, with source and timestamp annotations, forming the original dataset, which specifically includes:

[0070] Through ship IoT terminals, shipping management platforms, and manual data entry channels, static basic data, dynamic performance data, ship operation-related data, and external auxiliary data of crew members are collected simultaneously.

[0071] During the data collection process, cross-platform data transmission is carried out through a data encryption transmission protocol, and the collected data is marked with source identification and timestamp to form the original data set.

[0072] See Figure 3 As shown, based on the classification and processing of structured, semi-structured, and unstructured data, the following steps are taken: Redundancy is removed from structured data, missing values ​​are filled in, and the format is standardized; semi-structured data is converted; and unstructured data is parsed and core information is extracted to generate a standardized dataset. Specifically, this includes:

[0073] The original dataset is processed using corresponding preprocessing strategies for structured, semi-structured, and unstructured data.

[0074] For structured data, duplicate records are deleted, key field gaps are filled, and data of different formats are uniformly converted into a preset format through standardization processing;

[0075] For semi-structured data, core information is extracted using tag parsing technology, a unified data field mapping relationship is constructed, and the data is transformed into structured data.

[0076] For unstructured data, text records are segmented and keywords are extracted; image certificate data is subjected to optical character recognition and information verification; and audio records are subjected to speech-to-text conversion and semantic analysis.

[0077] After preprocessing, a standardized crew evaluation dataset is generated.

[0078] Specifically, the structured data preprocessing focuses on quality optimization. First, redundant data is removed through key field comparison algorithms. Using crew member ID numbers and competency certificate numbers as core primary keys, duplicate qualification information and performance records are checked and deleted, while the latest version of the data is retained. For missing values, the shipping industry rule completion method is used to complete key business fields such as crew member training results and watchkeeping hours by combining the average data of crew members in the same position and on the same route with historical trends. Non-critical fields are marked as "not collected" and then retained.

[0079] Semi-structured data preprocessing focuses on format conversion. For ship logs and port operation records in XML and JSON formats, the core fields are extracted through a tag parsing engine, a unified field mapping table is built, heterogeneous fields such as "operation start time", "equipment number" and "operator" are mapped to standard field names, and field attribute descriptions are completed simultaneously to complete the conversion to structured data. After conversion, the field integrity is verified again.

[0080] Unstructured data preprocessing focuses on information extraction. For text data, natural language processing technology is used to first segment and remove stop words, and then extract core keywords and semantic features such as "operational compliance" and "timely response" to transform them into quantifiable text tags. For image data, optical character recognition technology is used to extract text information, which is then cross-validated with structured qualification data to ensure the authenticity and validity of certificate information. For audio data, after speech-to-text processing, semantic analysis is performed to extract descriptions of job performance and key evaluation points.

[0081] See Figure 4 As shown, a hierarchical comprehensive evaluation system for crew members is constructed. The target layer represents the comprehensive evaluation results; the criteria layer includes static qualifications, dynamic performance, safety performance, and comprehensive qualities; and the indicator layer details the corresponding indicators. The data integration relationships are also obtained, specifically including:

[0082] A hierarchical comprehensive evaluation system for seafarers is constructed, which is divided into three levels: the target level, the criteria level, and the indicator level.

[0083] The target layer is the comprehensive evaluation result of the crew members, and the criteria layer includes static qualification dimension, dynamic duty performance dimension, safety performance dimension and comprehensive quality dimension. Each criterion layer is independent of each other and covers all dimensions of crew member evaluation.

[0084] The indicator layer further refines the criteria layer. Under the static qualification dimension, indicators include competency certificate level, years of service, training compliance rate, and educational background matching. Under the dynamic performance dimension, indicators include duty compliance, equipment operation accuracy, work efficiency, and emergency response speed. Under the safety performance dimension, indicators include the number of safety violations, accident-related liability, safety inspection pass rate, and quality of hazard identification and rectification. Under the comprehensive quality dimension, indicators include teamwork score, communication and coordination ability, emergency psychological quality, and professional ethics evaluation.

[0085] Each indicator clearly defines the corresponding data source and reserves interfaces for indicator expansion, allowing for updates and adjustments to indicator items based on shipping industry standards.

[0086] Specifically, the target layer is positioned as the comprehensive evaluation results of crew members. Its core function is to integrate the data of various indicators to form a comprehensive evaluation conclusion that can reflect the crew members' job suitability, performance ability, safety awareness and professional potential, and provide core basis for shipping companies' crew allocation, training and improvement, and salary assessment.

[0087] The criteria layer establishes four core dimensions, each independent of the others and covering all scenarios of crew evaluation, achieving multi-dimensional and comprehensive evaluation: The static qualification dimension focuses on crew members' basic abilities and qualification compliance, with pre-processed static basic data as its core support; the dynamic performance dimension emphasizes crew members' performance and efficiency in actual operations, connecting to pre-processed dynamic performance data and ship operation-related data; the safety performance dimension revolves around core navigation safety needs, linking to external auxiliary data such as rewards and punishments, safety inspection records, and ship safety data; and the comprehensive quality dimension supplements soft skills indicators such as teamwork and professional ethics, connecting to data such as peer reviews and assessment opinions.

[0088] The indicator layer further refines the criteria layer, with each indicator clearly defining its data source, definition, and evaluation direction to ensure accurate alignment with standardized datasets: Under the static qualification dimension, four indicators are set: competency certificate level and compliance, years of experience and route matching, training compliance rate, and educational background and job suitability. The competency certificate level is defined according to the maritime department's grading standards, and the training compliance rate is based on the results of specialized training assessments. Under the dynamic performance dimension, four indicators are set: duty compliance, equipment operation accuracy, operational efficiency, and emergency response speed. Operational efficiency is calculated based on the duration and compliance of cargo loading / unloading and equipment operation. Under the safety performance dimension, four indicators are set: number of safety violations, accident-related liability level, safety inspection pass rate, and quality of hazard identification and rectification. The quality of hazard rectification is assessed based on the timeliness of rectification completion and the results of review. Under the comprehensive quality dimension, four indicators are set: teamwork score, communication and coordination ability, emergency psychological qualities, and professional ethics evaluation. The teamwork score combines peer evaluation and administrator assessment results.

[0089] See Figure 5 As shown, the subjective weights of each criterion and indicator layer are calculated using the analytic hierarchy process (AHP), and the objective weights of each indicator are calculated using the entropy weight method. A dynamic adjustment coefficient is introduced, and the subjective and objective weights are integrated based on the characteristics of the crew member's position and the shipping route to generate the final comprehensive weights of each indicator. Specifically, these weights include:

[0090] The Analytic Hierarchy Process (AHP) was used to construct a judgment matrix by combining the experience of shipping experts, and the subjective weights of each criterion layer and indicator layer were calculated.

[0091] Based on the preprocessed evaluation dataset, the objective weight of each indicator is calculated using the entropy weight method, and the degree of influence of the indicator on the evaluation results is determined by analyzing the dispersion of the indicator data.

[0092] By introducing dynamic adjustment coefficients, subjective and objective weights are adaptively integrated based on crew member job type and navigation route characteristics to generate the final comprehensive weight of each indicator.

[0093] Specifically, subjective weights are calculated using the analytic hierarchy process (AHP). An evaluation team composed of shipping experts is formed to compare the importance of each dimension of the criteria layer and each indicator of the indicator layer, taking into account maritime regulatory standards, actual ship operation, and job competency requirements. A judgment matrix is ​​constructed based on the comparison results, and the rationality of the matrix is ​​verified through a consistency test. If the test fails, the results are fed back to the evaluation team to readjust the comparison results until the matrix meets the consistency requirements. Finally, the subjective weights of each indicator are calculated.

[0094] Objective weights are calculated using the entropy weight method. Based on the generated standardized crew evaluation dataset, valid data samples corresponding to each indicator are selected, and the influence weight of the indicator data on the evaluation results is determined by analyzing the dispersion of the indicator data. The higher the dispersion of the data, the more effectively the indicator can distinguish the differences in ability among different crew members, the higher the information value, and the greater the corresponding weight. The lower the dispersion, the weaker the indicator's distinguishability, the lower the information value, and the smaller the corresponding weight. During the calculation process, abnormal data samples that interfere with the weight results are removed simultaneously.

[0095] A dynamic adjustment coefficient is introduced to achieve adaptive optimization of weights. The adjustment coefficient is set according to the type of crew member and the characteristics of the navigation route. For example, the weight of dynamic performance and safety performance is emphasized for the driving position, the weight of emergency response-related indicators is increased for ocean routes, and the weight of operational efficiency indicators is strengthened for domestic trade routes. The final comprehensive weight of each indicator is generated by balancing the proportion of subjective and objective weights through the dynamic adjustment coefficient.

[0096] See Figure 6 As shown, corresponding quantification methods are adopted for different types of evaluation indicators. Quantitative indicators are normalized, and qualitative indicators are quantified by level and fuzziness is corrected. A time decay coefficient is introduced to highlight recent performance. Specifically, this includes:

[0097] For different types of evaluation indicators, corresponding quantitative methods are adopted. Among them, quantitative indicators are directly based on the preprocessed data values, and the data are normalized to the [0,1] interval through linear transformation.

[0098] Qualitative indicators are quantified and assigned values ​​according to preset evaluation level standards, converting qualitative descriptions into corresponding numerical values, and the fuzziness of qualitative indicators is handled by combining fuzzy comprehensive evaluation method.

[0099] For data with differences in time dimension, a time decay coefficient is introduced to give high weight to recent data and low weight to long-term data, highlighting the impact of crew members’ recent performance on the evaluation results.

[0100] Specifically, the quantification and normalization of quantitative indicators focuses on data standardization. These indicators cover directly quantifiable metrics such as the duration of compliance with competency certificates, years of service, shift duration, equipment operation accuracy, number of safety violations, and timeliness of hazard investigation and rectification. The data all originate from pre-processed standardized datasets. During processing, the effective data range of each quantitative indicator is first screened, and extreme outliers exceeding the reasonable range are removed. Then, a linear transformation is used to normalize all quantitative indicator data to the [0,1] range, enabling indicators of different magnitudes and units to have horizontal comparison conditions. For example, "equipment operation accuracy" is converted from a percentage format and "shift duration" is converted from an hourly format to standardized values ​​within the corresponding range.

[0101] The quantification of qualitative indicators focuses on correcting ambiguity. These indicators include those that cannot be directly quantified, such as teamwork ability, communication and coordination ability, emergency psychological qualities, and professional competence evaluation. The data comes from peer evaluation, administrator assessment, and third-party evaluation opinions. In the process, a unified evaluation level standard is first set, and the qualitative description is divided into four levels: "excellent, good, qualified, and unqualified". A basic quantitative score is assigned to each level. Then, the fuzzy comprehensive evaluation method is introduced to correct the ambiguity of the qualitative description. By extracting the core semantic features in the evaluation text, the basic score is fine-tuned. For example, for "emergency psychological qualities", the basic score is adjusted upward based on the specific performance description in the emergency drill. If there are descriptions such as "panic and mistakes, poor coordination", the score is appropriately lowered.

[0102] The formula for correcting the time decay coefficient is:

[0103]

[0104] in, The value of the i-th indicator after time decay correction is given. Let i be the normalized value of the i-th indicator. This is the time decay coefficient, with a value range of [0,1]. The larger the value, the greater the degree of data decay in the long term. This refers to the time interval between the data collection time and the evaluation benchmark time.

[0105] See Figure 7 As shown, based on the normalized indicator data and the comprehensive weight of each indicator, a weighted summation algorithm is used to calculate the score of each criterion layer. The calculation of the total comprehensive evaluation score of the crew members according to the weight of the criterion layer specifically includes:

[0106] Based on the normalized indicator data and the comprehensive weight of each indicator, a weighted summation operation is performed on the criteria layer. The normalized value of each indicator is multiplied by the corresponding weight, and the summation is used to obtain the individual score of each criteria layer.

[0107] Based on the weight allocation results of the criteria layer, the scores of the criteria layer are weighted and summarized again to calculate the total score of the crew's comprehensive evaluation.

[0108] The system simultaneously generates a total score and scores for each criterion level item, records the contribution of each indicator to the criterion level score and the total score, and filters out the core positive and negative indicators that affect the evaluation results by sorting them.

[0109] Specifically, the criterion-level score calculation is based on the indicator-level. For the four criterion-level dimensions of static qualification, dynamic performance, safety performance, and comprehensive quality, the quantitative data and corresponding comprehensive weights of all indicator-level dimensions are called up respectively, and the criterion-level score is calculated by weighted summation. During the calculation, the matching relationship between the quantitative value of each indicator and its corresponding weight is checked one by one to ensure that there are no issues of weight misalignment or data misuse. For example, for the dynamic performance dimension score, the quantitative values ​​of the four indicators of duty compliance, equipment operation accuracy, work efficiency, and emergency response speed are multiplied by their respective comprehensive weights and then summed to obtain the final score for that dimension. The scores of the four criterion-level dimensions are recorded separately to form the basic data for the sub-item scores.

[0110] The overall score is calculated based on the criteria-level scores and the comprehensive weights of the criteria-level scores. It also adopts a weighted summation method, multiplying the scores of the four criteria-level scores by their corresponding criteria-level weights and summing them to obtain the overall evaluation score of the crew. The overall score retains a fixed number of decimal places. Cross-validation is performed simultaneously during the calculation process. A sample of crew members is selected and the calculation is repeated. The deviation between the two calculation results is compared. If the deviation exceeds the preset range, the quantitative data of the indicators, the weight allocation and the calculation process are immediately traced back to investigate and correct the error.

[0111] See Figure 8 As shown, the evaluation data is verified through anomaly detection algorithms to identify abnormal fluctuations, collection errors, and weight biases. The process involves tracing the source of anomalies, re-collecting data, adjusting parameters, or correcting weights for secondary calculations. Specifically, this includes:

[0112] Anomaly detection algorithms are used to verify the data and intermediate results during the evaluation process, and to identify abnormal evaluation results caused by abnormal fluctuations in data, data acquisition errors, and weight allocation deviations.

[0113] For detected anomalies, trace the corresponding data source and processing steps, and perform secondary calculations by re-collecting data, adjusting preprocessing parameters, or correcting weighting coefficients;

[0114] Industry benchmarks and the average evaluation of crew members in the same position are used as references to correct for deviations in the preliminary evaluation results.

[0115] Specifically, anomaly detection employs a layered screening strategy. First, anomaly identification is conducted at the data level: for standardized data, reasonable thresholds for the shipping industry are set, and clustering algorithms are used to identify extreme values ​​deviating from the data cluster; for quantified and normalized results, it is verified whether there are values ​​exceeding the [0,1] range, or inconsistencies between the quantified qualitative indicators and the original evaluation opinions; then, anomaly identification is conducted at the result level: the logical correlation between the scores of each criterion level and the overall score of the same crew member is compared. If a score in a certain dimension is extremely low but the overall score is high, or the score ranking differs significantly from that of crew members in the same position, it is determined to be an anomaly; simultaneously, the scoring process is verified through repeated calculations to investigate issues such as incorrect weight allocation and weighted accumulation errors.

[0116] Anomaly tracing and correction require precise identification of the problem's source and the development of solutions for each issue: If the anomaly originates from the data, trace it back to the data collection or preprocessing stage, re-collect and supplement erroneous data, adjust parameters for preprocessing deviations, and re-process the data before updating the standardized dataset and re-calculating the comprehensive score; if the anomaly originates from the weights, verify the consistency of the subjective weight judgment matrix, the rationality of the objective weight data samples, and the dynamic adjustment coefficient settings, correct the weights, and recalculate the score; if the anomaly originates from the calculation process, correct the erroneous items and perform a second weighted calculation.

[0117] The calibration process introduces dual reference benchmarks to improve the rationality of the evaluation: the industry benchmark value is selected from the evaluation standards for crew members in the same position published by the maritime authorities and the average safety performance of the shipping industry. The average value for the same position is selected from the evaluation average of a sample group that is consistent with the crew member's job type and route characteristics. The preliminary corrected results are compared with the dual benchmarks. If the deviation exceeds the preset reasonable range, the score is fine-tuned in combination with the crew member's actual performance record.

[0118] See Figure 9 As shown, the evaluation results are output in the form of a tiered report, personalized improvement suggestions are generated for the weakest indicators, and the evaluation results are updated regularly through a dynamic update mechanism. The adjustment of indicators and weights specifically includes:

[0119] The revised comprehensive evaluation results of the crew members will be output in the form of a graded report, which will clearly indicate the crew member's evaluation level, performance ranking in each dimension, strength indicators and weakness indicators.

[0120] Personalized improvement suggestions are generated for the weaker indicators, a dynamic update mechanism for evaluation results is established, new data is collected regularly, the evaluation results are iteratively updated, and the evaluation indicator system and weight coefficients are adjusted according to shipping needs.

[0121] Specifically, the dynamic update mechanism adopts a combination of "regular incremental updates + event-driven updates" to ensure the timeliness of evaluation results and the adaptability of methods. Regular updates are executed cyclically, with monthly incremental data collection and fine-tuning of evaluations, updating scores only for crew members' recent performance data; annual comprehensive updates are conducted, re-executing the entire process and simultaneously verifying the applicability of the evaluation indicator system; event-driven updates are triggered by key scenarios. When crew members experience qualification updates, job changes, route adjustments, or significant changes in industry policies or corporate management needs, a special update is immediately initiated, readjusting indicator items, dynamic weights, and evaluation standards.

[0122] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0123] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0124] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A comprehensive evaluation method for crew members based on big data, characterized in that, include: The system collects static basic data, dynamic performance data, ship operation-related data and external auxiliary data of crew members through IoT terminals. The data is transmitted in encrypted form during collection, and the source and timestamp are marked to form the original dataset. Based on the classification and processing of structured, semi-structured and unstructured data, the system removes redundancy, fills in missing values ​​and unifies the format of structured data, transforms semi-structured data, parses unstructured data and extracts core information, and generates a standardized dataset. A hierarchical comprehensive evaluation system for crew members is constructed. The target layer is the comprehensive evaluation result, the criteria layer includes static qualifications, dynamic performance, safety performance, and comprehensive quality, and the indicator layer refines the corresponding indicators and obtains the data docking relationship. The subjective weights of each criterion and indicator layer are calculated using the analytic hierarchy process (AHP), and the objective weights of each indicator are calculated using the entropy weight method. A dynamic adjustment coefficient is introduced, and the subjective and objective weights are integrated based on the characteristics of crew positions and shipping routes to generate the final comprehensive weights of each indicator. For different types of evaluation indicators, corresponding quantitative methods are adopted. Quantitative indicators are normalized, and qualitative indicators are quantified by level and fuzziness is corrected. A time decay coefficient is introduced to highlight recent performance. Based on the normalized indicator data and the comprehensive weight of each indicator, a weighted summation algorithm is used to calculate the score of each criterion layer, and the total comprehensive evaluation score of the crew is calculated according to the weight of the criterion layer. The evaluation data is verified by anomaly detection algorithms to identify abnormal fluctuations, collection errors and weight deviations. The abnormal items are traced back to their source and data is collected again, parameters are adjusted or weights are corrected and recalculated. Evaluation results are output in the form of tiered reports, personalized improvement suggestions are generated for weak indicators, and evaluation results are updated regularly through a dynamic update mechanism, adjusting indicators and weights.

2. The comprehensive evaluation method for crew members based on big data according to claim 1, characterized in that, The process of collecting static basic data of crew members, dynamic performance data, ship operation-related data, and external auxiliary data through IoT terminals, encrypting data transmission during collection, and labeling the source and timestamp to form the original dataset specifically includes: Through ship IoT terminals, shipping management platforms, and manual data entry channels, static basic data, dynamic performance data, ship operation-related data, and external auxiliary data of crew members are collected simultaneously. During the data collection process, cross-platform data transmission is carried out through a data encryption transmission protocol, and the collected data is marked with source identification and timestamp to form the original data set.

3. The comprehensive evaluation method for crew members based on big data according to claim 1, characterized in that, The process of classifying and processing structured, semi-structured, and unstructured data, including removing redundancy, filling in missing values, and standardizing the format of structured data, converting it into semi-structured data, parsing unstructured data and extracting core information, and generating a standardized dataset, specifically includes: The original dataset is processed using corresponding preprocessing strategies for structured, semi-structured, and unstructured data. For structured data, duplicate records are deleted, key field gaps are filled, and data of different formats are uniformly converted into a preset format through standardization processing; For semi-structured data, core information is extracted using tag parsing technology, a unified data field mapping relationship is constructed, and the data is transformed into structured data. For unstructured data, text records are segmented and keywords are extracted; image certificate data is subjected to optical character recognition and information verification; and audio records are subjected to speech-to-text conversion and semantic analysis. After preprocessing, a standardized crew evaluation dataset is generated.

4. The comprehensive evaluation method for crew members based on big data according to claim 1, characterized in that, The construction of a hierarchical comprehensive evaluation system for crew members includes: a target layer representing the comprehensive evaluation results; a criteria layer encompassing static qualifications, dynamic performance, safety performance, and overall competence; and an indicator layer detailing corresponding indicators. The acquisition of data integration relationships specifically includes: A hierarchical comprehensive evaluation system for seafarers is constructed, which is divided into three levels: the target level, the criteria level, and the indicator level. The target layer is the comprehensive evaluation result of the crew members, and the criteria layer includes static qualification dimension, dynamic duty performance dimension, safety performance dimension and comprehensive quality dimension. Each criterion layer is independent of each other and covers all dimensions of crew member evaluation. The indicator layer further refines the criteria layer. Under the static qualification dimension, indicators include competency certificate level, years of service, training compliance rate, and educational background matching. Under the dynamic performance dimension, indicators include duty compliance, equipment operation accuracy, work efficiency, and emergency response speed. Under the safety performance dimension, indicators include the number of safety violations, accident-related liability, safety inspection pass rate, and quality of hazard identification and rectification. Under the comprehensive quality dimension, indicators include teamwork score, communication and coordination ability, emergency psychological quality, and professional ethics evaluation. Each indicator clearly defines the corresponding data source and reserves interfaces for indicator expansion, allowing for updates and adjustments to indicator items based on shipping industry standards.

5. The comprehensive evaluation method for crew members based on big data according to claim 1, characterized in that, The process of calculating the subjective weights of each criterion and indicator layer using the analytic hierarchy process (AHP), calculating the objective weights of each indicator using the entropy weight method, introducing a dynamic adjustment coefficient, and integrating the subjective and objective weights based on crew positions and route characteristics to generate the final comprehensive weights for each indicator specifically includes: The Analytic Hierarchy Process (AHP) was used to construct a judgment matrix by combining the experience of shipping experts, and the subjective weights of each criterion layer and indicator layer were calculated. Based on the preprocessed evaluation dataset, the objective weight of each indicator is calculated using the entropy weight method, and the degree of influence of the indicator on the evaluation results is determined by analyzing the dispersion of the indicator data. By introducing dynamic adjustment coefficients, subjective and objective weights are adaptively integrated based on crew member job type and navigation route characteristics to generate the final comprehensive weight of each indicator.

6. The comprehensive evaluation method for crew members based on big data according to claim 1, characterized in that, The method employs corresponding quantification approaches for different types of evaluation indicators. Quantitative indicators are normalized, and qualitative indicators are quantified by level and fuzziness is corrected. A time decay coefficient is introduced to highlight recent performance. Specifically, this includes: For different types of evaluation indicators, corresponding quantitative methods are adopted. Among them, quantitative indicators are directly based on the preprocessed data values, and the data are normalized to the [0,1] interval through linear transformation. Qualitative indicators are quantified and assigned values ​​according to preset evaluation level standards, converting qualitative descriptions into corresponding numerical values, and the fuzziness of qualitative indicators is handled by combining fuzzy comprehensive evaluation method. For data with differences in time dimension, a time decay coefficient is introduced to give high weight to recent data and low weight to long-term data, highlighting the impact of crew members’ recent performance on the evaluation results.

7. The comprehensive evaluation method for crew members based on big data according to claim 1, characterized in that, The calculation of the score for each criterion layer based on the normalized indicator data and the comprehensive weight of each indicator, using a weighted summation algorithm, and the calculation of the total comprehensive evaluation score for the crew members based on the weight of each criterion layer specifically includes: Based on the normalized indicator data and the comprehensive weight of each indicator, a weighted summation operation is performed on the criteria layer. The normalized value of each indicator is multiplied by the corresponding weight, and the summation is used to obtain the individual score of each criteria layer. Based on the weight allocation results of the criteria layer, the scores of the criteria layer are weighted and summarized again to calculate the total score of the crew's comprehensive evaluation. The system simultaneously generates a total score and scores for each criterion level item, records the contribution of each indicator to the criterion level score and the total score, and filters out the core positive and negative indicators that affect the evaluation results by sorting them.

8. The comprehensive evaluation method for crew members based on big data according to claim 1, characterized in that, The process of verifying and evaluating data through anomaly detection algorithms, identifying abnormal fluctuations, acquisition errors, and weight biases, tracing the source of anomalies, re-acquiring data, adjusting parameters, or correcting weights for secondary calculations specifically includes: Anomaly detection algorithms are used to verify the data and intermediate results during the evaluation process, and to identify abnormal evaluation results caused by abnormal fluctuations in data, data acquisition errors, and weight allocation deviations. For detected anomalies, trace the corresponding data source and processing steps, and perform secondary calculations by re-collecting data, adjusting preprocessing parameters, or correcting weighting coefficients; Industry benchmarks and the average evaluation of crew members in the same position are used as references to correct for deviations in the preliminary evaluation results.

9. The comprehensive evaluation method for crew members based on big data according to claim 1, characterized in that, The process of outputting evaluation results in the form of tiered reports, generating personalized improvement suggestions for weak indicators, and regularly updating evaluation results through a dynamic update mechanism, including adjusting indicators and weights, specifically includes: The revised comprehensive evaluation results of the crew members will be output in the form of a graded report, which will clearly indicate the crew member's evaluation level, performance ranking in each dimension, strength indicators and weakness indicators. Personalized improvement suggestions are generated for the weaker indicators, a dynamic update mechanism for evaluation results is established, new data is collected regularly, the evaluation results are iteratively updated, and the evaluation indicator system and weight coefficients are adjusted according to shipping needs.

Images