Method and system for analyzing human error mode of complex system

Abstract

The invention discloses a complex system human error mode analysis method and system, and the method comprises the steps: carrying out the matching degree calculation based on all tasks in a complex system in combination with all capabilities in a pre-constructed operator capability library, and screening out tasks with unmatched capabilities based on the calculated matching degree; based on the tasks with mismatched capabilities, human error mode evaluation is carried out by utilizing a comprehensive evaluation method combining experts and a cloud model, and human error modes corresponding to the tasks with mismatched capabilities are obtained; on the basis of the human error mode corresponding to the task with the mismatched capability, calculation is carried out in combination with a mapping function, and the severity level caused by the task with the mismatched capability is obtained; according to the method, man-machine interaction can be optimized and the possibility of misoperation can be reduced by calculating the matching degree of the operator capability and the task, and the problem of scarcity of data samples is also solved; and the risk of human errors can be predicted and quantified more accurately through comprehensive evaluation of combination of experts and the cloud model.

Description

Technical Field

[0001] The present application relates to the field of human factors engineering technology, and in particular to an analysis method and system for human error patterns in complex systems. Background Art

[0002] In complex and high-risk systems such as nuclear power, navigation, and aerospace, human error is one of the main causes of system degradation or failure. Existing methods for analyzing human errors are usually limited by the scarcity of historical data samples and the complexity of human error patterns, resulting in relatively simple or ignored considerations of human errors in these complex systems. Previous analysis methods often rely on data-driven methods, do not provide a systematic approach, lack an in-depth understanding of human characteristics and the human-computer interaction process, and are unable to effectively identify and improve weak links in the human factors design of complex systems.

[0003] The patent application, filed on November 21, 2018, with application number 2018113942844 and titled "A human error modeling and analysis method for maintenance tasks in a space man-machine system", proposes a human error modeling and analysis method for maintenance tasks in a space man-machine system, analyzing maintenance tasks through crew cognitive behavior and common performance conditions. However, this method is highly dependent on data in practical applications, and does not provide sufficient analysis of the interaction between human behavior and the system.

[0004] The application date is March 8, 2019, the application number is 2019101773457, and the name is "A Human Reliability Analysis Method for Operators in Digital Control Rooms of Nuclear Power Plants". It provides a human reliability analysis method for operators in digital control rooms of nuclear power plants, focusing on the three cognitive stages of monitoring, decision-making and execution. However, it only focuses on the cognitive link, and is not comprehensive, generalizable or universal.

[0005] In February 2023, a paper titled "Research on the Causal Mechanism Model of Human Error in Nuclear Power Plants Driven by Data" was published in Nuclear Power Engineering. By analyzing the human event reports of nuclear power plants, correlation and factor analysis methods were used to identify the influencing factors and triggering scenarios of human errors, and a causal mechanism model of human errors was established. It relies on large sample data for correlation and factor analysis, but in complex systems, large sample data may be difficult to obtain. In the case of scarce data samples, the accuracy and applicability of the model are limited.

[0006] The patent application, filed on December 20, 2022, with application number 202211643271X and titled "A Nuclear Power Plant Human Error Prevention and Safety Monitoring System and Method", proposes a system that uses high-definition cameras and image recognition processing models to monitor and prevent human errors of nuclear power plant personnel during work. It mainly focuses on safety monitoring during and after the event, does not provide sufficient preventive measures, and lacks early warning of operator behavior, resulting in the inability to prevent errors in the design stage. Excessive focus on monitoring rather than improving design may not fundamentally reduce human errors and is not conducive to providing solutions in the design stage.

[0007] How to solve the shortcomings of human error analysis in complex systems, especially in high-risk industries such as nuclear power plants and spacecraft operations. Existing analysis methods often ignore the complexity of human-computer interaction and the scarcity of data samples. Summary of the invention

[0008] In order to solve the problem that the existing human error analysis in complex systems ignores the complexity of human-computer interaction and the scarcity of data samples, this application proposes an analysis method for human error patterns in complex systems, including:

[0009] Calculate the matching degree of each task in the complex system with each capability in the pre-built operator capability library, and filter out tasks that do not match the capabilities based on the calculated matching degree;

[0010] Based on the task of capability mismatch, a comprehensive evaluation method combining experts and cloud models is used to evaluate the human error pattern, and the human error pattern corresponding to the task of capability mismatch is obtained;

[0011] A calculation is performed based on the human error pattern corresponding to the task with the mismatched capabilities in combination with a mapping function to obtain a severity level caused by the task with the mismatched capabilities.

[0012] Preferably, the matching degree calculation is performed based on each task in the complex system in combination with each capability in the pre-built operator capability library, and tasks with incompatible capabilities are screened out based on the calculated matching degree, including:

[0013] Use the operational concept analysis method to identify and decompose various tasks to obtain the operational requirements of each task;

[0014] Based on each task, natural language processing technology is used to extract keywords for each task;

[0015] Calculate the correlation strength based on the keywords of each task and each capability in the operator capability library to determine the key capabilities required to complete each task;

[0016] Based on the capability baselines corresponding to the key capabilities required to complete each task and the operational requirements of each task, the matching degree of each key capability relative to the task is obtained;

[0017] Based on the comparison between the matching degree of each key capability relative to the task and the set range, the tasks beyond the set range are screened out as tasks with mismatched capabilities.

[0018] Preferably, the calculation of association strength based on the keywords of each task and each capability in the operator capability library to determine the key capabilities required to complete each task includes:

[0019] Calculate the correlation strength based on the keywords of each task and each ability to obtain a similarity score between the keywords and the ability;

[0020] Performing statistics based on the similarity scores to obtain the sum of the association strengths of all keywords;

[0021] Based on the comparison between the sum of the association strengths and a set threshold, the capabilities corresponding to the sum of the association strengths greater than the set threshold are screened out as key capabilities required to complete the task.

[0022] Preferably, the task based on capability mismatch uses a comprehensive evaluation method combining experts and cloud models to evaluate the human error pattern, and obtains the human error pattern corresponding to the task based on capability mismatch, including:

[0023] Convert the fuzzy terms given by experts to the tasks of the described ability mismatch in terms of both likelihood and confidence into quantitative scores;

[0024] Performing weighted calculation based on the quantified score to obtain a weighted score of the task with mismatched capabilities;

[0025] The weighted scores of the tasks based on the capability mismatch are calculated using a cloud model to obtain an expected value and an entropy value;

[0026] Calculating based on the expected value and the entropy value to obtain a comprehensive score of the task with mismatched capabilities;

[0027] A calculation is performed based on the comprehensive score in combination with a mapping function to obtain a human error pattern corresponding to the task with mismatched capabilities.

[0028] Preferably, the mapping function is as follows:

[0029]

[0030] Where, Human Error Mode (SC) is the human error mode mapping function; SC is the comprehensive score; T 未操作is the score interval corresponding to the human error mode that was not operated; T 部分操作 is the score range corresponding to the human error mode of some operations; T 低于标准 T is the score range corresponding to the human error mode of operation below the standard level; 高于标准 T is the score range corresponding to the human error mode that is higher than the standard level of operation; 快于标准 T is the score range corresponding to the human error mode that operates faster than the standard level; 慢于标准 T is the score range corresponding to the human error mode that is slower than the standard level of operation; 错序操作 This is the score range corresponding to the human error pattern of wrong-sequence operation.

[0031] Preferably, the human error modes include: a human error mode of no operation, a human error mode of partial operation, a human error mode of operation below the standard level, a human error mode of operation above the standard level, a human error mode of operation faster than the standard level, a human error mode of operation slower than the standard level and a human error mode of operation out of order.

[0032] Preferably, the method further includes: providing corresponding human factors design improvement suggestions according to the severity level caused by the task with the mismatched capabilities.

[0033] Based on the same application concept, this application also proposes an analysis system for human error patterns in complex systems, including:

[0034] The task screening module is used to calculate the matching degree based on various tasks in the complex system and various capabilities in the pre-built operator capability library, and screen out tasks with incompatible capabilities based on the calculated matching degree;

[0035] The human error pattern assessment module is used to assess the human error pattern based on the task of capability mismatch using a comprehensive assessment method combining experts and cloud models to obtain the human error pattern corresponding to the task of capability mismatch;

[0036] The severity level calculation module is used to calculate based on the human error pattern corresponding to the task with mismatched capabilities in combination with a mapping function to obtain the severity level caused by the task with mismatched capabilities.

[0037] Preferably, the task screening module includes:

[0038] The operation requirement calculation submodule is used to identify and decompose various tasks using the operation concept analysis method to obtain the operation requirements of each task;

[0039] The keyword extraction submodule is used to extract keywords for each task using natural language processing technology based on each task;

[0040] A key capability confirmation submodule is used to calculate the correlation strength based on the keywords of each task and each capability in the operator capability library to determine the key capabilities required to complete each task;

[0041] The matching degree calculation submodule is used to calculate the matching degree of each key capability with respect to the task based on the capability baseline corresponding to the key capability required to complete each task and the operation requirements of each task;

[0042] The task screening submodule is used to compare the matching degree of each key capability with the task with the set range, and screen out the tasks that are beyond the set range as tasks that do not match the capability.

[0043] Preferably, the key capability confirmation submodule is specifically used for:

[0044] Calculate the correlation strength based on the keywords of each task and each ability to obtain a similarity score between the keywords and the ability;

[0045] Performing statistics based on the similarity scores to obtain the sum of the association strengths of all keywords;

[0046] Based on the comparison between the sum of the association strengths and a set threshold, the capabilities corresponding to the sum of the association strengths greater than the set threshold are screened out as key capabilities required to complete the task.

[0047] Compared with the prior art, the beneficial effects of this application are:

[0048] A method and system for analyzing human error patterns in complex systems, comprising: calculating the matching degree based on various tasks in the complex system combined with various capabilities in a pre-built operator capability library, and screening out tasks with mismatched capabilities based on the calculated matching degree; evaluating human error patterns based on tasks with mismatched capabilities using a comprehensive evaluation method combining experts and cloud models to obtain human error patterns corresponding to tasks with mismatched capabilities; calculating based on the human error patterns corresponding to tasks with mismatched capabilities combined with a mapping function to obtain the severity level caused by tasks with mismatched capabilities; the application can optimize human-computer interaction and reduce the possibility of operational errors by calculating the matching degree of operator capabilities and tasks, and also solves the problem of scarce data samples; more accurately predicting and quantifying the risk of human errors through a comprehensive evaluation combining experts and cloud models;

[0049] This application provides different improvement suggestions for different human error modes, which helps decision makers make more scientific decisions in system design and operation processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A flow chart of a method for analyzing human error patterns in a complex system in this application;

[0051] Figure 2 This is a technical flow chart of the analysis and detection of human error patterns in complex systems for this application;

[0052] Figure 3 This is the operation task sequence diagram of this application;

[0053] Figure 4 This is the operational task parameter diagram for this application;

[0054] Figure 5 This is a structural diagram of a complex system human error mode analysis system for this application;

[0055] Figure 6 This is a schematic diagram of the human error analysis management software system architecture and data interaction logic for this application. DETAILED DESCRIPTION

[0056] This application provides a comprehensive and systematic method for analyzing human error patterns that can reflect the interaction process between people and systems and does not strictly rely on big data samples. It uses semantic modeling, natural language processing, cloud models and other technologies to improve the intelligence of the analysis method, and at the same time proposes relevant human risk management methods in a targeted manner to enhance the closure, feasibility and application value of the method. In order to better understand this application, the content of this application is further explained below in conjunction with the drawings and examples of the specification.

[0057] Embodiment 1:

[0058] A method for analyzing human error patterns in complex systems. The specific process is as follows: Figure 1 As shown, including:

[0059] Step 1: Calculate the matching degree based on various tasks in the complex system and various capabilities in the pre-built operator capability library, and filter out tasks with incompatible capabilities based on the calculated matching degree;

[0060] Step 2: Based on the task with capability mismatch, a comprehensive evaluation method combining experts and cloud models is used to evaluate the human error pattern, and the human error pattern corresponding to the task with capability mismatch is obtained;

[0061] Step 3: Calculate the human error pattern corresponding to the task with mismatched capabilities in combination with a mapping function to obtain a severity level caused by the task with mismatched capabilities.

[0062] Before step 1, it also includes building an operator capability library, including but not limited to cognitive and physical capability dimensions, as shown in the following table:

[0063]

[0064]

[0065] Table 1 Operator capability library

[0066] In step 1, the matching degree is calculated based on the various tasks in the complex system and the various capabilities in the pre-built operator capability library, and tasks with incompatible capabilities are screened out based on the calculated matching degree, including:

[0067] Define operational tasks and operational requirements:

[0068] Based on the task profile, the operational concept analysis method is adopted to identify and decompose the tasks involving human operation in complex systems to form an operation sequence diagram. At the same time, the requirements of each operation are clarified in natural language, including but not limited to: information, decision-making, response, communication, workload, task support, and workplace.

[0069] Use SysML system modeling language to model the operation sequence and attributes, output the parameter diagram related to the operation requirements and check the consistency and ambiguity of the parameters. If there is a problem, return to the previous step to modify the operation requirement description.

[0070] Identify key capabilities in operational tasks:

[0071] Use NLP natural language processing technology to extract keywords, build a correlation matrix between operator capabilities and keywords, determine the relevance of each capability to the task keywords, and output the key capabilities required to complete the task. The main steps are as follows:

[0072] Text preprocessing: Use NLP technology to preprocess the task description, including word segmentation, stop word removal, stemming, etc.

[0073] Keyword extraction. Extract keywords K = {K1, K2…K m}, these keywords represent the key aspects of the task. K is a set of keywords, K1 is the first keyword, K2 is the second keyword, K m is the mth keyword.

[0074] Capabilities are associated with keywords. Establish an association matrix M between capabilities and keywords, where M ij Express ability i With keyword k j The strength of the association between them.

[0075] Similarity calculation. For each capability c i With keyword k j , calculate the similarity. i and keyword k jRepresented as a feature vector and Calculate the dot product and magnitude of two vectors:

[0076]

[0077] Map the calculated similarity score to the range of [0,1]. Set the similarity threshold according to actual needs. For example, only when the similarity score is higher than 0.8, the ability c is considered i and keyword k j Related.

[0078] Capability importance score. For each capability c i Calculate an importance score S i , based on the sum of its association strengths with all keywords:

[0079] Ability ranking. Score according to the importance of ability i Rank all capabilities to determine the key capabilities required to complete the mission.

[0080] Threshold determination. Determine a threshold θ to distinguish key capabilities from other capabilities. i >θ, the capacity c i are considered key capabilities.

[0081] Key capabilities identification. Finally, all S i >θ capabilities serve as key capabilities required to complete the task.

[0082] Computing operation requirements-key capability matching:

[0083] Based on the operator capability library and domain data, the capability baseline of individuals or groups is determined and given in quantitative or qualitative form.

[0084] For a capability baseline represented in quantitative form, the following formula is used to calculate the fit of each key capability:

[0085] Matching degree = (capability baseline - operation requirements) / capability baseline

[0086] Tasks whose matching degree exceeds the acceptable range α are screened out and are considered as tasks with mismatched capabilities. α is generally defined as ±0.1.

[0087] For the ability baseline characterized by qualitative form, which involves multi-faceted integration, it is directly treated as pending ability.

[0088] Create an expert assessment list of all tasks with mismatched capabilities, including task description, capability dimensions, baseline capabilities, task requirements, matching, and gaps. See the table below:

[0089]

[0090] Table 2 Task list with mismatched capabilities

[0091] In step 2, based on the task with capability mismatch, a comprehensive evaluation method combining experts and cloud models is used to evaluate the human error pattern, and the human error pattern corresponding to the task with capability mismatch is obtained, which specifically includes:

[0092] For each operation task in the evaluation list, experts give an assessment value of the occurrence of human error from the two dimensions of possibility and confidence, expressed in fuzzy terms such as large, medium, and small.

[0093] The fuzzy terms "large", "medium", and "small" used by experts were converted into quantitative scores.

[0094] High probability: 5

[0095] Medium probability: 3

[0096] Small probability: 1

[0097] Maximum confidence: 5

[0098] Medium confidence: 3

[0099] Small confidence level: 1

[0100] Calculate the weighted score S for each expert l =P l ×C l ; where P l is the likelihood score given by expert l, C l is the confidence score. S l Expressed as an interval [l l ,u l ].

[0101] The expected value and entropy value are calculated by the following formula:

[0102]

[0103]

[0104] In the formula, E is the expected value; S l is the weighted score; N is the number of experts; l is the expert ranking; En is the entropy value.

[0105] Calculate the composite score:

[0106] The interval range of the comprehensive score SC is expressed as: [EK·En, E+K·En]

[0107] where K is a constant determined according to the desired confidence level. This approach combines the precision of interval algebra with the fuzziness of cloud models to provide a comprehensive approach to the assessment task.

[0108] In order to map the comprehensive score to the seven human error modes, a mapping function is defined, which maps the comprehensive score interval to a specific human error mode.

[0109] Human error patterns include:

[0110] No operation: The operator did not perform the required operation at all.

[0111] Partial operation: The operator performed the operation but did not fully complete it.

[0112] Subpar Performance: The operator's performance quality is below the expected standard.

[0113] Above-par performance: The operator's performance quality exceeds the expected standards and may lead to overreaction.

[0114] Operating faster than standard speed: The operator's operating speed exceeds the expected standard.

[0115] Slower than standard speed operation: The operator's operating speed is lower than the expected standard.

[0116] Out-of-sequence operations: The operator performed the operations, but in the wrong order.

[0117] The score interval thresholds for the seven human error modes are defined as follows:

[0118] T 未操作 =[0, t u1 ]

[0119] T 部分操作 =(t u1 , t u2 ]

[0120] T 低于标准 =(t u2 , t u3 ]

[0121] T 高于标准 =(t u3 , t u4 ]

[0122] T 快于标准 =(t u4 , t u5 ]

[0123] T 慢于标准 =(t u5 , t u6 ]

[0124] T 错序操作 =(t u6 ,∞)

[0125] Where, t u1 is the upper limit corresponding to the human error mode without operation; t u2 is the upper limit corresponding to the human error mode of some operations; t u3 is the upper limit of the human error mode corresponding to the operation below the standard level; t u4 is the upper limit of the human error mode corresponding to the operation above the standard level; t u5 is the upper limit of the human error mode corresponding to the operation faster than the standard level; t u6 is the upper limit of the human error mode corresponding to the slower than standard level operation; ∞ is the upper limit of the human error mode corresponding to the wrong sequence operation. u1 ,t u1 ,t u3 ,t u4 ,t u5 ,t u6 These are preset thresholds and need to be determined based on expert experience and task characteristics.

[0126] The mapping function Human Error Mode(S) is defined as follows:

[0127]

[0128] Where, Human Error Mode (SC) is the human error mode mapping function; SC is the comprehensive score; T 未操作 is the score interval corresponding to the human error mode that was not operated; T 部分操作 is the score range corresponding to the human error mode of some operations; T 低于标准 T is the score range corresponding to the human error mode of operation below the standard level; 高于标准 T is the score range corresponding to the human error mode that is higher than the standard level of operation; 快于标准 T is the score range corresponding to the human error mode that operates faster than the standard level; 慢于标准 T is the score range corresponding to the human error mode that is slower than the standard level of operation; 错序操作 This is the score range corresponding to the human error pattern of wrong-sequence operation.

[0129] By defining thresholds and mapping functions, the comprehensive score interval is mapped to a specific human error pattern. This method helps experts quickly identify and understand the potential human error risks in task execution and take corresponding preventive measures.

[0130] In step 3, the human error pattern corresponding to the task with mismatched capabilities is calculated in combination with a mapping function to obtain the severity level caused by the task with mismatched capabilities, which specifically includes:

[0131] Using graph theory and other methods, the worst consequences caused by different human error modes are calculated, namely, level I, II, and III, which correspond to catastrophic consequences, serious consequences, and general or minor consequences, respectively.

[0132] Assume that in the operation sequence diagram, node "A" represents normal operation and node "B1" represents an erroneous operation that may lead to serious consequences. Use the Bellman-Ford algorithm to calculate the shortest path from the source point "S" to the target point "T", while considering the following:

[0133] The edge weight from node "S" to "A" is the cost of normal operation.

[0134] The edge weight from node “A” to “B1” is the cost of human error, which may be negative, indicating that the erroneous operation reduces the cost.

[0135] The edge from node “B1” to “T” may represent the cost of recovering from the error, which may be high.

[0136] Calculate the path from "S" to "B1" and then to "T". Shorter paths mean worse consequences.

[0137] Let Wthreshold be a predefined weight threshold used to distinguish consequences of different severity. We can define the following mapping function:

[0138]

[0139] Generally speaking, according to the safety principle of complex systems, W 低 , W 中低 Generally, it is 1 or 2. You can also choose a value based on the actual situation.

[0140] Note that if there are negative weight cycles in the operation sequence graph, the consequence level is directly determined using human evaluation.

[0141] This embodiment also includes providing corresponding human factors design improvement suggestions according to the severity level, specifically including the following contents:

[0142] According to the rules of the following matrix table, corresponding suggestions for improvement of human factors design are given respectively, according to the effectiveness level of error prevention measures, including but not limited to intrinsic error prevention design level I, auxiliary measures error prevention level II, procedure error prevention level III, and personnel training level III. Specific suggestions can be further clarified in combination with the system human-machine interface, functional characteristics, engineering feasibility, and technical maturity.

[0143]

[0144]

[0145] Table 3 Human factors design improvement suggestions

[0146] Through the above analysis, human error patterns in complex systems can be identified in the early stages of design, so that relevant robustness designs can be carried out in a targeted manner to improve the safety and reliability of the system.

[0147] Combine the following Figure 2 The flowchart gives the specific implementation process:

[0148] Assume that the operation tasks and requirements are:

[0149] Task 1: After seeing the displayed information, the operator needs to complete parameter binding within 1.5 seconds.

[0150] Task 2: If the target appears, one real target must be selected from 10 objects within 2 seconds and tracked for 10 minutes, with a tracking error of no more than 0.5°.

[0151] Task 3: If a display unit failure occurs, you need to enter the 0.5m×0.4m cabinet space within 30 minutes to complete the replacement of the faulty unit.

[0152] Create sequence diagrams based on sysml language such as Figure 3 As shown, the steps from receiving display information to completing parameter binding are shown. DisplayUnit display unit; Operator operator; TargetSystem target system; TrackingSystem tracking system; FaultDetection fault detection; Cabinet cabinet.

[0153] Construct parameter diagram as Figure 4 As shown, key capability characteristics are identified and matching degree is calculated.

[0154] Invite 5 experts to conduct human error pattern assessment.

[0155] The expert scoring data is as follows:

[0156] Expert 1: Probability 5, Confidence 5;

[0157] Expert 2: Probability 4, Confidence 3;

[0158] Expert 3: probability 3, confidence 5;

[0159] Expert 4: Probability 5, Confidence 4;

[0160] Expert 5: Probability 3, confidence level 3.

[0161] Calculate the weighted score S for each expert l :

[0162] S1=5×5=25;

[0163] S2=4×3=12;

[0164] S3=3×5=15;

[0165] S4=5×4=20;

[0166] S5=3×3=9.

[0167] Calculate the expected value E and entropy value En:

[0168] E = (25 + 12 + 15 + 20 + 9) / 5 = 16;

[0169] En can be based on S l Distribution calculation, here assuming En=5.

[0170] Determine the interval of the comprehensive score SC:

[0171] K is set to 2, 95% confidence level;

[0172] The interval of SC is: [16-2·5,16+2·5]=[6,26][16-2·5,16+2·5]=[6,26].

[0173] Assume that the SC value calculated according to the cloud model is 16, and the following thresholds are set:

[0174] t u1 =10

[0175] t u2 =12

[0176] t u3 =14

[0177] t u4 =18

[0178] t u5 =20

[0179] t u6 =25

[0180] Since the SC value is 16, which falls within the interval (14,18), the mapped human error mode is “operation above the standard level”.

[0181] The following thresholds are set:

[0182] W 中低 =20

[0183] W 低 = 15

[0184] Based on the SC = 16 obtained from the previous calculation and the mapping function:

[0185] Since 15 < SC = 16 < 20, the consequence level is level II, severe.

[0186] For the level II consequence, the severe human error mode of "operating above the standard level", the following are some human factors design improvement suggestions:

[0187] Enhance the monitoring system, introduce or enhance the monitoring system to track the operator's behavior in real time and issue warnings when anomalies are detected;

[0188] Improve the operation guidance to ensure that the operator clearly understands the operation standards and limitations;

[0189] Design error tolerance, design the system to tolerate small errors and prevent them from escalating into severe consequences; Strengthen training and simulation, and improve the operator's awareness of potential risks and response capabilities by simulating possible error scenarios;

[0190] Optimize the human-machine interaction, improve the interface design, reduce the operation complexity, and ensure the clear display of key information and easily accessible controls.

[0191] This embodiment has the following beneficial effects:

[0192] Improve system safety: By identifying weak links in human factors design at an early stage, targeted improvement measures can be proposed, thereby enhancing the safety and reliability of the system.

[0193] Enhance risk assessment: The comprehensive assessment method combining expert assessment and cloud model provides a comprehensive risk assessment means to more accurately predict and quantify the risk of human error.

[0194] Optimize the human-machine interaction: By analyzing the matching degree between the operator's ability and task requirements, the human-machine interface design can be optimized to reduce the possibility of operation errors.

[0195] Enhance decision-making support: The improvement suggestions provided help decision-makers make more scientific decisions in system design and operation processes.

[0196] Strong adaptability: This technology is not only applicable to specific fields but also can be extended to other complex systems that need to consider human error.

[0197] Promote innovative design: By systematically identifying and solving potential human factors problems, it encourages designers to innovate and develop systems that better conform to the principles of human factors engineering.

[0198] Embodiment 2:

[0199] A complex system human error mode analysis system, the structure is as follows Figure 5 As shown, including:

[0200] The task screening module is used to calculate the matching degree based on various tasks in the complex system and various capabilities in the pre-built operator capability library, and screen out tasks with incompatible capabilities based on the calculated matching degree;

[0201] The human error pattern assessment module is used to assess the human error pattern based on the task of capability mismatch using a comprehensive assessment method combining experts and cloud models to obtain the human error pattern corresponding to the task of capability mismatch;

[0202] The severity level calculation module is used to calculate based on the human error pattern corresponding to the task with mismatched capabilities in combination with a mapping function to obtain the severity level caused by the task with mismatched capabilities.

[0203] Task screening module, including:

[0204] The operation requirement calculation submodule is used to identify and decompose various tasks using the operation concept analysis method to obtain the operation requirements of each task;

[0205] The keyword extraction submodule is used to extract keywords for each task using natural language processing technology based on each task;

[0206] A key capability confirmation submodule is used to calculate the correlation strength based on the keywords of each task and each capability in the operator capability library to determine the key capabilities required to complete each task;

[0207] The matching degree calculation submodule is used to calculate the matching degree of each key capability with respect to the task based on the capability baseline corresponding to the key capability required to complete each task and the operation requirements of each task;

[0208] The task screening submodule is used to compare the matching degree of each key capability with the task with the set range, and screen out the tasks that are beyond the set range as tasks that do not match the capability.

[0209] The key capability confirmation submodule is specifically used for:

[0210] Calculate the correlation strength based on the keywords of each task and each ability to obtain a similarity score between the keywords and the ability;

[0211] Performing statistics based on the similarity scores to obtain the sum of the association strengths of all keywords;

[0212] Based on the comparison between the sum of the association strengths and a set threshold, the capabilities corresponding to the sum of the association strengths greater than the set threshold are screened out as key capabilities required to complete the task.

[0213] Human error mode assessment module, specifically for:

[0214] Convert the fuzzy terms given by experts to the tasks of the described ability mismatch in terms of both likelihood and confidence into quantitative scores;

[0215] Performing weighted calculation based on the quantified score to obtain a weighted score of the task with mismatched capabilities;

[0216] The weighted scores of the tasks based on the capability mismatch are calculated using a cloud model to obtain an expected value and an entropy value;

[0217] Calculating based on the expected value and the entropy value to obtain a comprehensive score of the task with mismatched capabilities;

[0218] A calculation is performed based on the comprehensive score in combination with a mapping function to obtain a human error pattern corresponding to the task with mismatched capabilities.

[0219] The mapping function in the human error mode assessment module is as follows:

[0220]

[0221] Where, Human Error Mode (SC) is the human error mode mapping function; SC is the comprehensive score; T 未操作 is the score interval corresponding to the human error mode that was not operated; T 部分操作 is the score range corresponding to the human error mode of some operations; T 低于标准 T is the score range corresponding to the human error mode of operation below the standard level; 高于标准 T is the score range corresponding to the human error mode that is higher than the standard level of operation; 快于标准 T is the score range corresponding to the human error mode that operates faster than the standard level; 慢于标准 T is the score range corresponding to the human error mode that is slower than the standard level of operation; 错序操作 This is the score range corresponding to the human error pattern of wrong-sequence operation.

[0222] The human error patterns in the human error pattern assessment module include: human error pattern of no operation, human error pattern of partial operation, human error pattern of operation below the standard level, human error pattern of operation above the standard level, human error pattern of operation faster than the standard level, human error pattern of operation slower than the standard level and human error pattern of operation out of sequence.

[0223] It also includes an improvement suggestion module, specifically for:

[0224] Corresponding human factors design improvement suggestions are given according to the severity level caused by the task of capability mismatch.

[0225] Human error analysis management software system architecture and data interaction logic Figure 6 shown.

[0226] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0227] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0228] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0229] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0230] The above are merely embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application are included in the scope of the claims of the present application to be approved.

Claims

1. A method for analyzing human error patterns in complex systems, characterized in that: include: Calculate the matching degree of each task in the complex system with each capability in the pre-built operator capability library, and filter out tasks that do not match the capabilities based on the calculated matching degree; Based on the task of capability mismatch, a comprehensive evaluation method combining experts and cloud models is used to evaluate the human error pattern, and the human error pattern corresponding to the task of capability mismatch is obtained; A calculation is performed based on the human error pattern corresponding to the task with the mismatched capabilities in combination with a mapping function to obtain a severity level caused by the task with the mismatched capabilities.

2. The method according to claim 1, characterized in that The matching degree calculation is performed based on each task in the complex system in combination with each capability in the pre-built operator capability library, and tasks with incompatible capabilities are screened out based on the calculated matching degree, including: Use the operational concept analysis method to identify and decompose various tasks to obtain the operational requirements of each task; Based on each task, natural language processing technology is used to extract keywords for each task; Calculate the correlation strength based on the keywords of each task and each capability in the operator capability library to determine the key capabilities required to complete each task; Based on the capability baselines corresponding to the key capabilities required to complete each task and the operational requirements of each task, the matching degree of each key capability relative to the task is obtained; Based on the comparison between the matching degree of each key capability relative to the task and the set range, the tasks beyond the set range are screened out as tasks with mismatched capabilities.

3. The method according to claim 2, characterized in that The calculation of association strength based on the keywords of each task and each capability in the operator capability library to determine the key capabilities required to complete each task includes: Calculate the correlation strength based on the keywords of each task and each ability to obtain a similarity score between the keywords and the ability; Performing statistics based on the similarity scores to obtain the sum of the association strengths of all keywords; Based on the comparison between the sum of the association strengths and a set threshold, the capabilities corresponding to the sum of the association strengths greater than the set threshold are screened out as key capabilities required to complete the task.

4. The method according to claim 1, characterized in that: The task based on capability mismatch uses a comprehensive evaluation method combining experts and cloud models to evaluate the human error pattern, and obtains the human error pattern corresponding to the task based on capability mismatch, including: Convert the fuzzy terms given by experts to the tasks of the described ability mismatch in terms of both likelihood and confidence into quantitative scores; Performing weighted calculation based on the quantified score to obtain a weighted score of the task with mismatched capabilities; The weighted scores of the tasks based on the capability mismatch are calculated using a cloud model to obtain an expected value and an entropy value; Calculating based on the expected value and the entropy value to obtain a comprehensive score of the task with mismatched capabilities; A calculation is performed based on the comprehensive score in combination with a mapping function to obtain a human error pattern corresponding to the task with mismatched capabilities.

5. The method according to claim 4, characterized in that The mapping function is as follows: Where, Human Error Mode (SC) is the human error mode mapping function; SC is the comprehensive score; T 未操作 is the score interval corresponding to the human error mode that was not operated; T 部分操作 is the score range corresponding to the human error mode of some operations; T 低于标准 T is the score range corresponding to the human error mode of operation below the standard level; 高于标准 T is the score range corresponding to the human error mode that is higher than the standard level of operation; 快于标准 T is the score range corresponding to the human error mode that operates faster than the standard level; 慢于标准 T is the score range corresponding to the human error mode that is slower than the standard level of operation; 错序操作 This is the score range corresponding to the human error pattern of wrong-sequence operation.

6. The method according to claim 4, characterized in that The human error modes include: human error modes of no operation, human error modes of partial operation, human error modes of operation below the standard level, human error modes of operation above the standard level, human error modes of operation faster than the standard level, human error modes of operation slower than the standard level and human error modes of operation out of order.

7. The method according to claim 1, characterized in that Also includes: Corresponding human factors design improvement suggestions are given according to the severity level caused by the task of capability mismatch.

8. A complex system human error mode analysis system, characterized by: include: The task screening module is used to calculate the matching degree based on various tasks in the complex system and various capabilities in the pre-built operator capability library, and screen out tasks with incompatible capabilities based on the calculated matching degree; The human error pattern assessment module is used to assess the human error pattern based on the task of capability mismatch using a comprehensive assessment method combining experts and cloud models to obtain the human error pattern corresponding to the task of capability mismatch; The severity level calculation module is used to calculate based on the human error pattern corresponding to the task with mismatched capabilities in combination with a mapping function to obtain the severity level caused by the task with mismatched capabilities.

9. The system according to claim 8, characterized in that The task screening module includes: The operation requirement calculation submodule is used to identify and decompose various tasks using the operation concept analysis method to obtain the operation requirements of each task; The keyword extraction submodule is used to extract keywords for each task using natural language processing technology based on each task; A key capability confirmation submodule is used to calculate the correlation strength based on the keywords of each task and each capability in the operator capability library to determine the key capabilities required to complete each task; The matching degree calculation submodule is used to calculate the matching degree of each key capability with respect to the task based on the capability baseline corresponding to the key capability required to complete each task and the operation requirements of each task; The task screening submodule is used to compare the matching degree of each key capability with the task with the set range, and screen out the tasks that are beyond the set range as tasks that do not match the capability.

10. The system according to claim 9, characterized in that The key capability confirmation submodule is specifically used for: Calculate the correlation strength based on the keywords of each task and each ability to obtain a similarity score between the keywords and the ability; Performing statistics based on the similarity scores to obtain the sum of the association strengths of all keywords; Based on the comparison between the sum of the association strengths and a set threshold, the capabilities corresponding to the sum of the association strengths greater than the set threshold are screened out as key capabilities required to complete the task.

Images