Method for predicting human error probability of each operator in civil aircraft traction departure process

Abstract

The invention discloses a method for predicting the human error probability of each operation in the civil aircraft traction departure process, and the method comprises the steps: calculating the risk coefficient of each historical risk event type, screening a target risk event type, and screening a behavior formation factor according to the mapping relation with human error factors; obtaining a manual task and a general operation condition in an aircraft traction departure scene; and evaluating each behavior formation factor when each manual task is executed under the general operation condition through the expert group, obtaining the weight and the grade of each behavior formation factor, and calculating the human error probability of each manual task. The behavior formation factors are screened according to the mapping relation between the risk event types and the human error factors, and the human error probability of each human task is calculated through the expert group construction weight and level, so that the subjectivity problem can be avoided, the behaviors causing human errors are effectively identified, the risk assessment accuracy and management level are improved, and the risk assessment efficiency is improved. And the human reliability of the staff in the aspect of aircraft traction departure is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of aviation safety assessment, and in particular to a method for predicting the probability of human error in various operations during the towing and departure process of a civil aircraft. Background Art

[0002] The new aircraft taxi-out departure mode, with its significant advantages such as low fuel consumption, high departure efficiency, and environmental friendliness, has become the preferred method for civil aircraft departures at future smart airports. However, the aircraft taxi-out process requires close collaboration and good communication among multiple personnel, and accidents are often caused by human error. Therefore, human reliability analysis is particularly important. Summary of the Invention

[0003] The present invention provides a method for predicting the probability of human error in each operation during the towing and departure process of a civil aircraft, so as to solve the problem that the prediction of the probability of human error in the aircraft towing and departure scenario is subjective and relies on expert experience, and the resulting problems of low risk assessment accuracy, poor management level and low human reliability of staff in aircraft towing and departure.

[0004] According to one aspect of an embodiment of the present invention, a method for predicting the probability of human error in an aircraft towing departure scenario is provided, comprising:

[0005] Collect statistics on various risk events that have occurred in the past in aircraft towing departure scenarios, and calculate the risk coefficient of each risk event type;

[0006] According to each risk coefficient, multiple target risk event types are screened, and based on each target risk event type and the mapping relationship between the risk event type and the human error factor, multiple target human error factors are screened as behavior formation factors;

[0007] Analyze manual business operation information in the aircraft towing departure scenario, obtain multiple manual tasks in the aircraft towing departure scenario, and obtain common operation conditions that match the aircraft towing departure scenario;

[0008] An expert group evaluates the behavior formation factors when performing various manual tasks under general operating conditions, and obtains the weights and levels of each behavior formation factor under each manual task;

[0009] Based on the evaluation results, the probability of human error for each manual task in the aircraft towing departure scenario is calculated.

[0010] According to another aspect of an embodiment of the present invention, a device for predicting the probability of human error in an aircraft towing departure scenario is provided, comprising:

[0011] The coefficient calculation module is used to count various risk events that have occurred in the history of aircraft towing departure scenarios and calculate the risk coefficient of each risk event type;

[0012] A factor screening module is used to screen multiple target risk event types according to each risk coefficient, and to screen multiple target human error factors as behavior formation factors based on each target risk event type and the mapping relationship between the risk event type and the human error factor;

[0013] An operation parsing module is used to parse manual business operation information in the aircraft towing departure scenario, obtain multiple manual tasks in the aircraft towing departure scenario, and obtain common operation conditions that match the aircraft towing departure scenario;

[0014] The factor evaluation module is used to evaluate the behavior formation factors when performing various manual tasks under general operating conditions by an expert group, and obtain the weight and level of each behavior formation factor under each manual task;

[0015] The probability calculation module is used to calculate the human error probability of each manual task in the aircraft towing departure scenario based on the evaluation results.

[0016] According to another aspect of an embodiment of the present invention, an electronic device is provided, the electronic device comprising:

[0017] at least one processor; and

[0018] a memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for predicting the probability of human error in an aircraft towing departure scenario described in any embodiment of the present invention.

[0020] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for predicting the probability of human error in an aircraft towing departure scenario described in any embodiment of the present invention when executed.

[0021] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the steps of the method according to any embodiment of the present invention are implemented.

[0022] The technical solution of the embodiment of the present invention can identify the main factors in human error factors by calculating the risk coefficient of historical risk event types and screening target risk event types, combining the mapping relationship with human error factors to screen behavior formation factors. Obtaining manual tasks and general operating conditions in the aircraft towing departure scenario provides comprehensive and specific background information for analysis. The expert group evaluates the behavior formation factors of each manual task under general operating conditions, obtains the weights and levels reflecting the importance of each behavior formation factor, calculates the human error probability of each manual task, avoids the subjective bias of expert experience, can effectively identify the main behavior formation factors that lead to human errors, reduces the probability of human errors, improves the accuracy of risk assessment and management level, and enhances the human factors scientificity and reliability of staff in aircraft towing departure.

[0023] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 This is a flowchart of a method for predicting the probability of human error in an aircraft towing departure scenario provided in accordance with the first embodiment of the present invention;

[0026] Figure 2 This is a flowchart of another method for predicting the probability of human error in an aircraft towing departure scenario provided by Embodiment 2 of the present invention;

[0027] Figure 3 2 is a schematic diagram of a method for predicting the probability of human error in an aircraft towing departure scenario applicable to an embodiment of the present invention;

[0028] Figure 4 1 is a schematic diagram of data fusion in a method for predicting the probability of human error in an aircraft towing departure scenario applicable to an embodiment of the present invention;

[0029] Figure 5 is a schematic diagram of a probability distribution of human errors applicable to an embodiment of the present invention;

[0030] Figure 62 is a schematic structural diagram of a device for predicting the probability of human error in an aircraft towing departure scenario according to a third embodiment of the present invention;

[0031] Figure 7 The present invention is a schematic structural diagram of an electronic device for implementing the method for predicting the probability of human error in an aircraft towing departure scenario according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] Example 1

[0035] Among related technologies, the Success Likelihood Index Methodology (SLIM) is a commonly used human reliability analysis method. It mainly uses an expert panel to evaluate performance shaping factors (PSFs) and their weights and levels to calculate the probability of human error. This method is widely used in industries such as marine, railways, and aerospace, and is particularly practical in areas where human error data is difficult to obtain. However, the PSFs and parameters are determined by experts, resulting in a highly subjective evaluation result. In addition, there is a lack of a method to select the main behavior shaping factors that lead to human errors from each PSF, which reduces the scientific nature and reliability of human reliability analysis during aircraft towing and departure.

[0036] Figure 1This is a flowchart of a method for predicting the probability of human error in an aircraft towing departure scenario provided by the first embodiment of the present invention. This embodiment is applicable to predicting the probability of human error in an aircraft towing departure scenario. The method can be executed by a human error probability prediction device in an aircraft towing departure scenario. The human error probability prediction device in an aircraft towing departure scenario can be implemented in the form of hardware and / or software and can generally be configured in an electronic device. Accordingly, if Figure 1 As shown, the method includes:

[0037] S110. Count various risk events that have occurred historically in aircraft towing departure scenarios, and calculate the risk coefficient for each risk event type.

[0038] Specifically, data on various risk events that occurred during aircraft towing departures over a certain period of time (defined based on business needs, such as the past five years) is collected (public data can be obtained from sources such as civil aviation safety reports, accident investigation reports, or airline records). The collected events are categorized by type, such as aircraft-ground facility collisions (including aircraft scrapes and nose landing gear loss), tow truck-aircraft collisions, aircraft component failures (including nose landing gear loss and component breakage), simplified inspection procedures, neglect or misunderstanding of signal lights, inattention, and inaccurate dispatch command language. For each type of risk event, its historical probability of occurrence can be calculated by counting the number of occurrences of that type of event and dividing it by the total number of towing departures. The impact of each risk event can then be quantified, for example, using the economic losses caused by each risk event as an indicator. The risk coefficient is calculated by multiplying the risk probability of each risk event by its quantified impact.

[0039] S120. Filter multiple target risk event types according to each risk coefficient, and filter multiple target human error factors as behavior formation factors based on each target risk event type and a mapping relationship between the risk event type and the human error factor.

[0040] In an embodiment of the present invention, the target risk event type can be specifically understood as: a risk event type whose risk coefficient exceeds a preset risk threshold, or, by sorting various risk events and selecting a risk event type in descending order of risk coefficient. The target human error factor can be specifically understood as: a human error factor that appears repeatedly in multiple high-risk event types, indicating that it has an important impact on the safety of the aircraft towing departure process. It can be a human error factor whose occurrence frequency exceeds a preset threshold or a human error factor selected in descending order of occurrence frequency. The behavior formation factor (PSF) can be specifically understood as: a factor that can affect the operator's behavioral performance when performing the aircraft towing departure task, such as the staff's professional skills and experience level, health status, psychological state and working environment.

[0041] Specifically, based on the calculated risk coefficients, various risk events can be ranked. A preset number of target risk event types can be selected based on business needs. In descending order of risk coefficient, a preset number of risk event types can be selected as target risk event types. Each target risk event type is analyzed to determine which human error factors (such as task complexity, environmental conditions, time pressure, training and exercises, operational experience, and communication skills) it is associated with. From the associated human error factors, multiple target human error factors can be selected in descending order of frequency as behavior formation factors. To ensure the comprehensiveness and effectiveness of the analysis, four to six key human error factors can be selected as behavior formation factors.

[0042] S130: Analyze manual business operation information in the aircraft towing departure scenario, obtain multiple manual tasks in the aircraft towing departure scenario, and obtain general operation conditions matching the aircraft towing departure scenario.

[0043] In the embodiment of the present invention, the general operating conditions can be specifically understood as: by analyzing the manual business operation information in the aircraft towing departure scenario and obtaining multiple manual tasks in the aircraft towing departure scenario (such as can be obtained from the operation manual), the general conditions and standards that need to be followed to complete each manual task are determined to ensure operational consistency and safety. For example, the general operating conditions can be: the aircraft towing departure task begins in the morning, the weather is clear, and the road is clear and unobstructed; the pilot, driver, ground staff and air traffic control are fully involved in the execution of the task; the working environment, time pressure, employee experience, proficiency and communication and collaboration skills all meet the work requirements; and, due to sufficient rest, the pilot and driver are in a state of high spirits; during the towing taxiing period, the pilot and driver strictly taxi according to the prescribed speed and route, and comply with ground traffic rules and airport regulations; and the noise level and mental capacity are acceptable.

[0044] Specifically, hierarchical task analysis identifies the tasks personnel must complete during the aircraft towing and departure process. These tasks include at least one primary task (such as pushing out, towing, and taxiing the aircraft) and corresponding subtasks. Each primary task consists of at least one subtask. The subtasks describe each step in the towing and departure process in detail and clarify the specific responsibilities of personnel in each position. For example, for the primary task of pre-towing preparation, its subtasks may include: ground crew members clean and inspect the aircraft's exterior; ensure all cargo doors, passenger doors, and emergency exits are closed and locked; inspect the landing gear to ensure they are free of obvious damage or abnormalities; ensure that equipment such as seats and armrests are in their proper positions and that cabin equipment is functioning properly; ensure that the aircraft has sufficient fuel to support the planned flight, as well as for alternate landings and emergency situations; pilots conduct self-tests of the avionics system to ensure they are functioning properly; ensure the air pressure system is functioning properly; conduct communications tests to ensure the radio communication equipment is functioning properly; pilots calibrate the aircraft's compass to ensure accurate heading information; and pilots conduct final confirmation and assessment of route, weather conditions, and airport information.

[0045] Then, the general operating conditions that match the aircraft towing departure scenario are obtained, which may include: environmental conditions (such as weather conditions and ground conditions), equipment conditions (such as the performance of the towing vehicle and the performance of the aircraft's braking system), personnel conditions (such as the training level and health status of the operator), and operating specifications (such as taxiing speed and route).

[0046] S140. An expert group evaluates each behavior formation factor when each manual task is performed under general operating conditions to obtain the weight and level of each behavior formation factor under each manual task.

[0047] In the embodiments of the present invention, the expert panel can be specifically understood as a panel of experienced experts who can assess the importance (weight) and impact (level) of each behavior-forming factor based on industry standards, best practices, or professional knowledge. The expert panel can also be a pre-trained expert model that can automatically assess behavior-forming factors by learning patterns and regularities from historical data.

[0048] Specifically, the impact of each behavior-forming factor in different tasks can be scored based on preset evaluation criteria or pre-trained expert models. Alternatively, a pre-trained expert model can be used to conduct a preliminary evaluation of the staff's behavior-forming factors, which are then reviewed and adjusted by an expert group based on the preset evaluation criteria. Based on the scoring results, the weight of each behavior-forming factor is calculated. For example, the importance score of each PSF in all tasks is added together to obtain a total score, and then the score of each PSF is divided by the total score to obtain the corresponding weight. The weight reflects the relative importance of the factor in the task. Then, based on the weight, the level of each behavior-forming factor is determined in descending order, such as high, medium, and low, corresponding to levels 1, 2, and 3, respectively, or a more detailed grading can be performed based on business needs.

[0049] S150. Calculate the human error probability of each manual task in the aircraft towing departure scenario based on the evaluation results.

[0050] In the embodiment of the present invention, the human error probability (HEP) can be specifically understood as the probability of an error occurring due to human factors during the aircraft towing departure mission.

[0051] Specifically, the human error probability for each manual task in the aircraft towing departure scenario is calculated based on the weight and level of each behavioral factor under each manual task. Based on historical data or expert evaluation, the level of the behavioral factor can be converted into a quantifiable indicator. For example, a high level (indicating a high degree of impact or a high frequency of human error) can be assigned a corresponding error probability of 5%, a medium level can be assigned a corresponding error probability of 3%, and a low level can be assigned a corresponding error probability of 1%. Using a weighted summation method, the weight of each behavioral factor is multiplied by its corresponding level value, and then the sum is calculated to obtain the human error probability. For example, if the weights for the high, medium, and low levels are 0.5, 0.3, and 0.2, respectively, and the corresponding levels are 5%, 3%, and 1%, respectively, the human error probability = (0.5 × 5%) + (0.3 × 3%) + (0.2 × 1%) = 3.6%.

[0052] The technical solution of the embodiment of the present invention can identify the main factors in human error factors by calculating the risk coefficient of historical risk event types and screening target risk event types, combining the mapping relationship with human error factors to screen behavior formation factors. Obtaining manual tasks and general operating conditions in the aircraft towing departure scenario provides comprehensive and specific background information for analysis. The expert group evaluates the behavior formation factors of each manual task under general operating conditions, obtains the weights and levels reflecting the importance of each behavior formation factor, calculates the human error probability of each manual task, avoids the subjective bias of expert experience, can effectively identify the main behavior formation factors that lead to human errors, reduces the probability of human errors, improves the accuracy of risk assessment and management level, and enhances the human factors scientificity and reliability of staff in aircraft towing departure.

[0053] Optionally, based on the above embodiments, statistics of various risk events that have occurred in the past in the aircraft towing departure scenario and calculation of the risk coefficient of each risk event type may include:

[0054] In the civil aviation knowledge base, the number of historical risk events and the risk event rates of various risk events in aircraft towing departure scenarios are collected, and an original event sample that meets the event number and risk event rate is constructed;

[0055] Perform random sampling with replacement on the original event sample to obtain multiple expanded event samples;

[0056] Based on the original event sample and the expanded event sample, calculate the target statistical indicator value corresponding to the number of events of each risk event type, and calculate the probability distribution function and confidence interval corresponding to each target statistical indicator value;

[0057] The risk coefficient of each risk event type is calculated based on the probability distribution function and confidence interval corresponding to the target statistical indicator value of each risk event type.

[0058] In the embodiment of the present invention, the risk event rate can be specifically understood as: the frequency or probability of risk events occurring in a specific time period or a specific number of activities. For example, in an aircraft towing departure scenario, the risk event rate can be expressed as the number of risk events such as scratches or collisions that occur in every 100 towing departure operations. The original event sample can be specifically understood as: a set of event data collected and sorted from historical data for risk analysis, which can include various risk events that occurred in a specific scenario and the specific circumstances of normal events. For example, in an aircraft towing departure scenario, the original event sample can include records of all towing departure operations in the past year, including both successful operations and operations in which risk events such as scratches and nose landing gear detachment occurred.

[0059] Specifically, the historical number of various risk events that occurred in aircraft towing departure scenarios, as well as the risk event rates for aircraft towing departure scenarios, are collected from the civil aviation knowledge base. Based on the collected number of events and risk event rates, a raw event sample that meets statistical requirements is constructed. For example, if the number of events is 30 and the risk event rate is 5%, it can be inferred that the total amount of data in the raw event sample is 600 (30 risk events and 570 normal events). By setting the ratio of risk events to normal events in the constructed raw event sample to be consistent with the historical number of risk events and the risk event rate for aircraft towing departure scenarios, it helps to improve the accuracy of risk assessment.

[0060] Perform random sampling with replacement (e.g., Bootstrap) on the original event sample, i.e., each time a sample is drawn, it is returned to the dataset so that the sample can be selected again. Repeating random sampling with replacement multiple times (e.g., 1,000 times or more) can help estimate the sampling distribution of the statistic and generate multiple augmented event samples in the case of small samples. For the original event sample and each augmented event sample, calculate the target statistical indicator value corresponding to the number of events of each risk event type. The target statistical indicator value can be the mean, median, variance, or regression coefficient corresponding to the number of events of each risk event type.

[0061] Based on the calculated statistical indicator value, estimate its probability distribution function (for example, by selecting a normal distribution, binomial distribution, Poisson distribution, or exponential distribution, and performing parameter estimation and chi-square tests to verify the determination). Calculate the corresponding confidence interval based on the selected probability distribution function and the estimated statistic to quantify the uncertainty of the result. For example, if the data follows a normal distribution and you wish to estimate a 95% confidence interval for the population mean, you can use the t-distribution to calculate the confidence interval.

[0062] Finally, based on the probability distribution function and confidence interval of the target statistical indicator value of each risk event type, the risk coefficient of each risk event type is calculated, where the upper and lower limits of the confidence interval are used to adjust the calculation of the risk coefficient to reflect this uncertainty, for example, risk coefficient = target statistical indicator value × difference between the upper and lower limits of the confidence interval × 100%.

[0063] In a specific example, event A follows a normal distribution with a mean (target statistical indicator value) of 5, and the upper and lower limits of the confidence interval (95%) range from 0.03 to 0.07. Event B follows a binomial distribution with p (target statistical indicator value) of 3, and the upper and lower limits of the confidence interval (95%) range from 0.02 to 0.04. Therefore, the risk coefficient of event A = 5 × (0.07-0.03) × 100% = 20%. A larger risk coefficient indicates that this is a high-risk event, and a larger range of upper and lower limits corresponding to the confidence interval (95%) indicates a greater uncertainty in the risk coefficient. The risk coefficient of event B = 3 × (0.04-0.02) × 100% = 6%. A smaller risk coefficient indicates that this is a low-risk event, and a smaller range of upper and lower limits corresponding to the confidence interval (95%) indicates a smaller uncertainty in the risk coefficient.

[0064] By collecting historical risk event data from the civil aviation knowledge base and constructing original event samples, risk assessment can be carried out based on actual data, making the assessment results more objective and reliable; using random sampling with replacement (such as the Bootstrap method) on the original samples to generate multiple expanded samples, when the original sample size is small, the sample size is increased to provide data support for the prediction of human error probability in subsequent aircraft towing departure scenarios; by calculating the target statistical indicator value corresponding to each risk event type, its probability distribution function and confidence interval, and based on this, the risk coefficient of each risk event type is determined, the statistical characteristics and uncertainty of risk events can be fully understood, the stability of statistical analysis and the reliability of results can be improved, thereby improving the accuracy and management level of subsequent risk assessments, and enhancing the human factors science and reliability of staff in aircraft towing departure.

[0065] Optionally, based on the above embodiments, calculating the risk coefficient of each risk event type according to the probability distribution function and confidence interval corresponding to the target statistical indicator value of each risk event type may include:

[0066] A regression analysis model was established with the target statistical index value of each risk event type as the independent variable and the risk event rate in the aircraft towed departure scenario as the dependent variable;

[0067] According to the probability distribution function and confidence interval corresponding to the target statistical indicator values ​​of each risk event type, the regression analysis model is fitted with data to obtain the risk event rate impact coefficient corresponding to each risk event type as the risk coefficient.

[0068] Specifically, a regression analysis model (such as a linear regression model, a logistic regression model, or a Poisson regression model) is established, using the target statistical indicator value for each risk event type as the independent variable and the risk event rate in the aircraft towed departure scenario as the dependent variable. Using methods such as maximum likelihood estimation or least squares estimation, the parameters (such as regression coefficients) in the regression model are estimated based on the collected data and the probability distribution function. During the parameter estimation process, the confidence interval of the target statistical indicator value is considered, and a weight is assigned to each target statistical indicator value. For example, the narrower the confidence interval, the higher the weight, to reflect the uncertainty and variability of the data. Statistical tests (such as t-tests or F-tests) and goodness-of-fit indicators (such as chi-square tests) are used to evaluate the significance and goodness-of-fit of the regression model. Based on the verification results, the model is optimized, such as by adding or deleting variables, changing the form of variables, or adjusting the model type, until the preset convergence conditions are met or the preset upper limit of the number of optimizations is reached.

[0069] After the data fitting is completed, the regression coefficient corresponding to each risk event type is the risk coefficient. For example, the established regression model is a linear regression model: y = β0 + β1x A +β2x B +β3x C +∈, where y is the risk event rate, x A ,x B and x C The target statistical indicator values ​​corresponding to risk event types A, B, and C respectively, ∈ is the error term, and β0, β1, β2, and β3 are regression coefficients. Then, after the data fitting is completed, the risk coefficients of risk event types A, B, and C are β1, β2, and β3.

[0070] By establishing a regression analysis model and performing data fitting, the degree of influence of the risk event rate corresponding to each risk event type can be quantified, that is, the risk coefficient, to provide data support for the prediction of the probability of human error in the subsequent aircraft towing departure scenario. By considering the probability distribution function and confidence interval of the target statistical indicator value, the uncertainty and potential impact of the risk can be more comprehensively reflected, the stability of the statistical analysis and the reliability of the results can be improved, thereby improving the accuracy and management level of subsequent risk assessments, and enhancing the human factors scientificity and reliability of staff in aircraft towing departure.

[0071] Optionally, based on the above embodiments, obtaining general operating conditions that match the aircraft towing departure scenario may include:

[0072] Collect external environment information, work status information of each operator, and business capability information from multiple historical aircraft towing departure scenarios;

[0073] The information collection results are clustered, and based on the clustering results, the general operating conditions matching the aircraft towing departure scenario are obtained.

[0074] Specifically, external environmental information (which may include weather conditions (such as wind speed, wind direction, and visibility), ground conditions (such as wetness or snow accumulation), and airport facilities (such as the smoothness of runway and taxiway facilities)) from multiple historical aircraft towing departure scenarios is collected, as well as information on the working status of each operator (such as fatigue level, concentration, and health status), and business capability information (such as the operator's professional skills, training level, and experience). The collected information is clustered (such as K-means clustering or hierarchical clustering) to group similar scenarios or conditions together (for example, rainy days and wet ground are grouped together) to identify common patterns or types. The clustering results are used as common operating conditions for matching aircraft towing departure scenarios.

[0075] By identifying and summarizing common operating conditions and integrating multi-dimensional information, the one-sidedness of single-dimensional analysis is avoided, and common operating modes or types corresponding to manual tasks are provided for the subsequent prediction of human error probability in aircraft towing departure scenarios. This improves the adaptability of subsequent risk assessments to different manual tasks in aircraft towing departure scenarios and the comprehensiveness of predictive assessments. It can ensure that operators can operate according to safety standards in different scenarios, help operators adapt to different manual tasks more quickly, improve operational efficiency, thereby improving the accuracy and management level of subsequent risk assessments, and enhancing the human factors science and reliability of staff in aircraft towing departure.

[0076] Optionally, based on the above embodiments, calculating the probability of human error for each manual task in the aircraft towing departure scenario according to the evaluation results may include:

[0077] Obtaining target weights and target levels evaluated by each expert model in the expert group for each behavior formation factor under the target manual task;

[0078] Perform data fusion on each target weight and each target level of each behavior formation factor under the target manual task to obtain the fusion weight and fusion level corresponding to each behavior formation factor under the target manual task;

[0079] According to the formula: Calculate the success likelihood index SLI corresponding to the target manual task;

[0080] Where n is the total number of behavior-forming factors, R i is the fusion level of the i-th behavior formation factor, W i is the fusion weight of the i-th behavior formation factor;

[0081] The human error probability HEP corresponding to the target manual task is calculated according to the formula: Log(HEP)=aSLI+b, where a and b are preset empirical constants.

[0082] Specifically, the target weight and target level of each behavior formation factor (i.e., the weight and level of the current behavior formation factor) are determined by evaluating each behavior formation factor for the target manual task (i.e., the manual task currently being processed) by each expert model in the expert group. Data fusion is performed on the target weights and target levels of each behavior formation factor for the target manual task. For example, a weighted average method can be used, where each expert model is assigned a weight based on its reliability and accuracy, and then the evaluation results of each expert are weighted averaged.

[0083] Where n is the number of experts.

[0084] According to the formula: Calculate the success likelihood index SLI corresponding to the target manual task, where n is the total number of behavior formation factors, R i is the fusion level of the i-th behavior formation factor, W i is the fusion weight of the i-th behavior formation factor. According to the formula: Log(HEP) = aSLI + b, the human error probability HEP corresponding to the target manual task is calculated, where a and b are preset empirical constants. The constants a and b can be obtained from the two boundary points of the known human error probability HEP. For example, when the boundary conditions need to be set to SLI = 1, HEP = 0.9 and SLI = 9, HEP = 10 -4 When the corresponding a and b values ​​are calculated, the SLI can be converted into the corresponding HEP according to the formula.

[0085] Data fusion can combine the assessment results of multiple expert models to improve the accuracy and reliability of assessments. The fused weights and rankings more comprehensively reflect the importance and impact of each behavioral factor on the target manual task, thereby improving the accuracy of subsequent risk assessments and management capabilities, and enhancing the human factors science and reliability of personnel in aircraft towing departures.

[0086] Optionally, based on the above embodiments, data fusion is performed on the target weights of each behavior formation factor under the target manual task to obtain the fusion weights corresponding to each behavior formation factor under the target manual task, which may include:

[0087] Normalize the target weights evaluated by each expert model for each behavior formation factor under the target manual task;

[0088] The obtained normalized weights are formed into normalized weight rows corresponding to each expert model, and the normalized weight rows are arranged in rows to form a normalized weight matrix;

[0089] Get the transposed result of the first normalized weight row in the normalized weight matrix as the current iteration column, and get the second normalized weight row in the normalized weight matrix as the initialization fusion calculation row;

[0090] Calculate the matrix multiplication between the current iteration column and the fusion calculation row to obtain the fusion matrix, and form the new current iteration column with the main diagonal elements in the fusion matrix;

[0091] According to the sum of the non-main diagonal elements in the fusion matrix, the conflict degree factor is cumulatively updated, where the conflict degree factor is initialized to 0;

[0092] After obtaining the next row of the fusion calculation row in the normalized weight matrix as a new fusion calculation row, return to perform the operation of calculating the matrix multiplication between the current iteration column and the fusion calculation row until all normalized weight rows in the normalized weight matrix are completely processed;

[0093] After the iterative process is completed, the counterintuitive correction coefficient is calculated based on the currently updated conflict degree factor and the current iteration column;

[0094] According to the counter-intuitive correction coefficient, conflict correction is performed on each fusion element in the current iteration column, and based on the correction result, the fusion weight corresponding to each behavior formation factor under the target manual task is obtained.

[0095] Specifically, the fusion can be performed based on the Dempster-Shafer evidence theory. First, the target weights estimated by each expert model for each behavior formation factor under the target manual task are normalized (for example, using a linear normalization method to normalize the weight values ​​to the range [0, 1]). The conflict degree factor K is initialized to 0.

[0096] Assume that there are n experts evaluating the system, there are 6 behavior formation factors, and the basic probability distribution given by the experts is

[0097] Among them, any element m in the matrix M ij Represents the basic probability distribution (normalized weight) of the i-th expert to the j-th behavior formation factor, so the sum of the rows of this matrix is ​​1. Multiply the transpose of M1 (the transpose result of the first normalized weight row in the normalized weight matrix is ​​used as the current iteration column) with M2 (the second normalized weight row in the normalized weight matrix is ​​used as the initial fusion calculation row) to obtain the matrix R (fusion matrix):

[0098]

[0099] The conflict degree factor is cumulatively updated based on the sum of the non-main diagonal elements in the fusion matrix. That is, the sum of the non-main diagonal elements in the matrix R is the conflict degree factor K in the DS evidence theory combination rule. The main diagonal elements in the fusion matrix form a new current iteration column. After obtaining the next row of the fusion calculation row in the normalized weight matrix as the new fusion calculation row, the matrix multiplication operation between the current iteration column and the fusion calculation row is returned until all normalized weight rows in the normalized weight matrix are processed. That is, the column matrix composed of the main diagonal elements in the matrix R is multiplied by M3 to obtain a new matrix R:

[0100]

[0101] The conflict factor K is the sum of all non-diagonal elements of the original matrix R and the current matrix R. This process continues until the evaluation results of n experts are integrated. The sum of all non-diagonal elements of the matrix R during this process is the conflict factor K between the n pieces of evidence.

[0102] When the K value approaches 1, it indicates a very high level of conflict between evidence. When fusing evidence in this situation, the synthesis rule overemphasizes conflict, potentially producing counterintuitive results. This is because synthesis rules assume that evidence is independent. However, in practice, evidence can have correlations or dependencies, which are often overlooked in high-conflict situations. For example, even if individual pieces of evidence provide some support for certain propositions, due to severe conflict, the fusion result may show very low support for these propositions, even close to zero, reducing the legitimacy of the final decision or conclusion.

[0103] Therefore, after completing the iterative process, the counterintuitive correction coefficient is calculated based on the currently updated conflict degree factor K and the current iteration column. f(A) is the probability distribution function of evidence conflict, that is, the conflict degree factor K between evidences is distributed to each element of the recognition framework. i (A) is the basic probability distribution of the behavior formation factor A during the evaluation of the system by the i-th expert, reflecting the degree of influence of A. This probability distribution function satisfies Where A is a specific factor in the behavior formation factor set θ. Based on the counterintuitive correction coefficient, conflict correction is performed on each fusion element in the current iteration column, and based on the correction result, the fusion weight corresponding to each behavior formation factor is obtained. The specific correction method is:

[0104]

[0105] Among them, when the behavior formation factor A is empty, its corresponding basic probability distribution is also 0. i is the support set of the i-th expert for a certain behavior formation factor, B j is the support set of the jth expert for a certain behavior formation factor, C k is the support set of the kth expert for a certain behavior formation factor. In the process of evidence fusion, it is necessary to consider the support degree of all experts' evaluations for the behavior formation factor A. i ∩B j ∩C k ∩…=A means that the intersection of the support sets of the evaluations of different experts is exactly the behavior formation factor A, It means that for each combination that meets the conditions, the corresponding basic probability distributions are multiplied and summed, and then the adjustment term f(A) is added to obtain the final basic probability distribution m(A), which is the fusion weight corresponding to each behavior formation factor.

[0106] Similarly, the fusion level corresponding to each behavior formation factor can also be obtained by using the above method to fuse the level of each PSF using DS evidence theory.

[0107] By fusing the evaluation results of multiple expert models using DS evidence theory, the results can be synthesized to improve the accuracy and reliability of the assessment. The fusion weights and levels more comprehensively reflect the importance and impact of each behavior-forming factor in the target manual task. By using counterintuitive correction coefficients to correct conflicts in the fusion results, the accuracy and reliability of the behavior-forming factor assessment in the manual task are effectively improved. This, in turn, enhances the accuracy and management level of subsequent risk assessments, and improves the human factors science and reliability of personnel in aircraft towing departures.

[0108] Example 2

[0109] Figure 2A flowchart of another method for predicting human error probabilities in an aircraft towing departure scenario provided in a second embodiment of the present invention is provided. This embodiment is a refinement of the method for predicting human error probabilities in an aircraft towing departure scenario in the above-mentioned embodiment. After calculating the human error probabilities corresponding to each manual task in the aircraft towing departure scenario based on evaluation results, the method further includes: constructing a secondary task architecture for the aircraft towing departure scenario based on manual business operation information in the aircraft towing departure scenario; wherein the secondary task architecture includes a main task level and a subtask level, wherein the subtask level includes each of the manual tasks; summarizing the human error probabilities corresponding to each manual task to obtain a summarized human error probability corresponding to each main task; and when it is determined that the summarized human error probability of a target main task is greater than an error probability threshold, parsing each manual task included in the target main task, planning at least one target manual task that can be converted or decomposed into an automatically executed task, and generating a task improvement strategy that matches the target manual task.

[0110] Correspondingly, such as Figure 2 As shown, the method includes:

[0111] S210. Count various risk events that have occurred historically in aircraft towing departure scenarios, and calculate the risk coefficient of each risk event type.

[0112] S220: Filter multiple target risk event types according to each risk coefficient, and filter multiple target human error factors as behavior formation factors based on each target risk event type and a mapping relationship between the risk event type and the human error factor.

[0113] S230: Analyze manual business operation information in the aircraft towing departure scenario, obtain multiple manual tasks in the aircraft towing departure scenario, and obtain general operation conditions matching the aircraft towing departure scenario.

[0114] S240. The expert group evaluates each behavior formation factor when each manual task is performed under general operating conditions to obtain the weight and level of each behavior formation factor under each manual task.

[0115] S250. Calculate the human error probability of each manual task in the aircraft towing departure scenario based on the evaluation results.

[0116] S260: Construct a secondary task architecture for the aircraft towing departure scenario based on manual business operation information for the aircraft towing departure scenario.

[0117] The secondary task architecture includes a main task level and a subtask level, and the subtask level includes the manual tasks.

[0118] Specifically, the secondary task architecture includes a main task level and a subtask level. The main task level is divided according to the main stages or operation types of towing departure, such as aircraft pushout, towing, and taxiing. Each main task contains multiple subtasks, which are the specific operational steps to complete the main task. For example, under the main task of aircraft pushout, subtasks may include the commander issuing the pushout command, the tractor driver operating the tractor, and the supervisor monitoring the surrounding environment. By building a secondary task architecture, each task link can be analyzed and managed more meticulously, which helps to identify risk points and improve the accuracy of predicting the probability of human error in the aircraft towing departure scenario.

[0119] S270 , summarizing the human error probability corresponding to each manual task to obtain a summary human error probability corresponding to each main task.

[0120] Specifically, based on the two-level task architecture, each main task and its subordinate subtasks (manual tasks) are clearly defined. The human error probabilities of all subtasks under the same main task are aggregated to obtain the overall human error probability of the main task. For example, the main task "aircraft pushout" has three subtasks: "commander issues instructions," "tractor driver operates the tractor," and "supervisor monitors the surrounding environment." Their human error probabilities can be 0.02, 0.03, and 0.01, respectively. Therefore, the aggregated human error probability of the main task "aircraft pushout" is 0.06.

[0121] S280. When it is determined that the aggregated human error probability of the target main task is greater than the error probability threshold, each manual task included in the target main task is parsed, at least one target manual task that can be converted or decomposed into an automatically executed task is planned, and a task improvement strategy that matches the target manual task is generated.

[0122] Specifically, by summarizing the human error probability of each subtask, the overall error probability of each main task is obtained. If the overall error probability of a main task is greater than the set threshold, it will be determined as the target main task. Conduct a detailed analysis of each subtask under the target main task, and evaluate the execution process, risk points, and possible improvement space of each subtask. Among the subtasks, obtain tasks (target manual tasks) that can be performed through automation technology or equipment, such as operational tasks with high repetitiveness, clear rules, or high precision requirements. Develop specific improvement measures for each selected subtask, such as automated task implementation plans, personnel training, or task operation process optimization, to reduce the probability of human error.

[0123] For example, if the aggregated human error probability for the main task of "aircraft pushout" exceeds a threshold, it may be due to a high error rate in the subtask of "tractor driver operating the tractor." In this case, consider automating part or all of this subtask, such as introducing an automated guidance system to assist the tractor driver and developing a corresponding training plan to ensure the driver's proficiency in the new system.

[0124] Optionally, based on the above embodiments, a pre-trained model (such as a decision tree or neural network model) can be used to evaluate the automation potential of each subtask to predict whether it is suitable for execution through automation technology or equipment. The model can output a probability value indicating the degree to which the task is suitable for automation. Based on business needs and risk preferences, a threshold is set to determine which tasks are selected as target manual tasks. For example, tasks with an automation potential evaluation probability greater than 0.7 can be selected as target manual tasks.

[0125] For each selected target manual task, specific improvement measures are formulated based on the characteristics of the task and the results of the automation potential assessment. The preset knowledge base can be retrieved through the rule engine, and corresponding improvement strategies are generated for each task based on the preset rules and knowledge, such as automated task implementation plans, personnel training plans, and task operation process optimization suggestions.

[0126] The technical solution of the embodiments of the present invention screens behavioral factors by combining them with mapping relationships with human error factors, thereby identifying the primary factors in human error. An expert panel evaluates the behavioral factors of each manual task under common operating conditions and calculates the human error probability of each manual task. This avoids the subjective bias of expert experience and effectively identifies the primary behavioral factors that lead to human error, reducing the probability of human error, improving the accuracy of risk assessment, and enhancing the scientific and reliable human factors analysis of personnel in aircraft towing and departure. Based on manual business operation information in aircraft towing and departure scenarios, a two-level task architecture consisting of a main task layer and a subtask layer is constructed. The subtask layer includes each manual task, enabling more detailed task management and analysis. By calculating the human error probability of each manual task and summarizing it to obtain the aggregated human error probability of each main task, when the aggregated probability of a target main task exceeds an error probability threshold, each manual task is analyzed, and target manual tasks that can be converted or decomposed into automated tasks are planned. A matching task improvement strategy is generated, further improving the reliability and safety of task execution and enhancing the management level of aircraft towing and departure.

[0127] Specific application scenarios

[0128] For ease of understanding, the specific application scenarios applicable to each embodiment of the present invention are now described. The new aircraft towing taxi-out departure mode offers typical advantages such as low fuel consumption, high departure efficiency, and environmental friendliness. It has become the preferred method for civil aircraft departures at future smart airports. The aircraft towing departure process requires close collaboration and good communication among multiple staff members, and accidents are often caused by human error. Therefore, conducting a human reliability analysis is crucial.

[0129] SLIM, a commonly used human reliability analysis method in related technologies, primarily uses an expert panel to evaluate PSFs, their weights, and levels, and then calculate the probability of human error. However, the fact that PSFs and parameters are determined by experts results in a highly subjective evaluation. Furthermore, while related technologies have employed some methods to improve SLIM and reduce the subjectivity of expert judgment, they still lack a method for selecting the primary behavioral factors that contribute to human error from the various PSFs. This reduces the scientific nature and reliability of human reliability analysis during aircraft towing and departure.

[0130] To solve the above problems, the present invention proposes a method for predicting the probability of human error in an aircraft towing departure scenario. Figure 3 FIG. 1 is a schematic diagram of a method for predicting the probability of human error in an aircraft towing departure scenario applicable to an embodiment of the present invention. Figure 3 As shown, the method may specifically include:

[0131] 1. Obtain PSFs (equivalent to the behavior formation factors mentioned above)

[0132] The data is expanded using the Bootstrap method, and then deep learning is used to perform human risk analysis to map out a set of PSFs. Specifically, it can be:

[0133] (1) Prepare the original sample by consulting the public data, and take the number of unsafe events and the event rate during the aircraft towing departure process as the input sample X = {x1, x2, .., x n Unsafe incidents refer to risk events such as simplifying inspection procedures, ignoring or misunderstanding traffic lights, lack of concentration, and inaccurate dispatch command language.

[0134] (2) Import the dataset X = {x1, x2, ..., x n}, then generate a new Bootstrap sample data set X from the known sample data set by random sampling with replacement * ={x1,x2,...,x n}.

[0135] (3) For each Bootstrap sample X *, calculate the target statistic θ * , the target statistic can be the mean, median, variance or regression coefficient, etc.;

[0136] (4) Repeat steps (2) and (3) B-1 times (optionally, B is 1000 or larger) to obtain B Bootstrap statistics Includes statistics for B-1 expanded samples and 1 original sample.

[0137] (5) Based on B Bootstrap statistics, estimate the probability distribution function and confidence interval corresponding to each statistic.

[0138] (6) The obtained B Bootstrap statistics As training data.

[0139] (7) Establish a regression analysis model, take the risk event type as the independent variable X, and the risk event rate as the dependent variable Y to perform regression analysis model fitting analysis and construct the relationship between Y and X.

[0140] (8) Based on the probability distribution function and confidence interval, the risk coefficient of each risk event type is calculated, that is, the size of the coefficient of the fitting equation is used as the judgment of the degree of impact. The larger the coefficient, the higher the degree of impact, and the smaller the coefficient, the lower the degree of impact.

[0141] In a specific example, a set of PSFs can be mapped out by the above process, as shown in Table 1 as the behavioral formation factors.

[0142] Table 1

[0143] No PSF definition 1 Complex tasks Indicates the difficulty level of the task 2 Environmental conditions Environmental factors that affect traction and sliding operations, such as temperature, weather, and road conditions. 3 Time pressure The relationship between task completion time and available time 4 Training and performance Frequency of tow-out training and drills, and coverage of training and drill personnel 5 experience Length of work or accidents experienced 6 Communication skills Ability to clearly convey information / commands from one place to another during mission execution

[0144] Furthermore, hierarchical task analysis was used to identify the tasks that personnel must complete during the departure process. These tasks include primary tasks and subtasks, with primary tasks consisting of subtasks, as shown in Table 2. Electronic Flight Bag (EFB) software, installed on the electronic flight bag system, helps pilots manage various flight information and tasks electronically, replacing traditional paper documents and improving flight safety and efficiency.

[0145] Table 2

[0146]

[0147] 2. Obtain general operating conditions

[0148] Based on the mapped set of PSFs, the aircraft towing departure scenario is determined, the manual business operation information in the aircraft towing departure scenario is analyzed, and multiple manual tasks in the aircraft towing departure scenario are obtained. According to the analyzed tasks, a set of general operation conditions that match the aircraft towing departure scenario is determined.

[0149] In a specific example, after completing the mission analysis, a set of aircraft towing departure conditions was defined, and the aircraft towing departure scenario was determined. The aircraft towing departure mission began in the morning, with clear weather and a clear, unobstructed road. The pilot, driver, ground crew, and air traffic control personnel participated in the entire mission. The working environment, time pressure, employee experience, proficiency, and communication and collaboration skills all met the requirements. Furthermore, due to adequate rest, the pilot and driver were in a state of high alert. During the towing taxi, the pilot and driver strictly adhered to the prescribed speed and route, complying with ground traffic regulations and airport regulations. Furthermore, the noise level and mental capacity were acceptable.

[0150] 3. Data Fusion

[0151] Based on a set of common operating conditions, the level and weight of each PSF are directly derived from the expert evaluation of the importance of each PSF when performing each manual task, and the DS evidence theory is used for data fusion to overcome the subjectivity and uncertainty.

[0152] Using DS evidence theory to perform data fusion can be understood as follows:

[0153] Assume that there are n experts evaluating the system, there are 6 behavior formation factors, and the basic probability distribution given by the experts is

[0154] Among them, any element m in the matrix M ij Represents the basic probability distribution of the i-th expert to the j-th behavior formation factor, so the sum of the rows of this matrix is ​​1. Multiplying the transpose of M1 with M2 gives the matrix R:

[0155]

[0156] The sum of the main diagonal elements in the matrix R is the numerator in the DS evidence theory combination rule, and the sum of the non-main diagonal elements is the conflict degree factor K in the DS evidence theory combination rule. Multiply the column matrix composed of the main diagonal elements in the matrix R by M3 to obtain the new matrix R:

[0157]

[0158] The sum of the main diagonal elements in matrix R remains the numerator in the evidence theory combination rule, and the conflict factor K is the sum of all non-main diagonal elements of the original matrix R and the current matrix R. This process continues in this way until the evaluation results of n experts are fully integrated. During this process, the sum of all non-main diagonal elements of matrix R is the conflict factor K between the n pieces of evidence.

[0159] Furthermore, when the K value approaches 1, it indicates a very high degree of conflict between the evidence. When fusing evidence in this situation, the synthesis rule overemphasizes conflict, potentially producing counterintuitive results. This is because synthesis rules assume that evidence is independent. However, in practice, evidence can have correlations or dependencies, which are often overlooked in high-conflict situations. For example, even if individual pieces of evidence provide some support for certain propositions, due to severe conflict, the fusion result may show very low support for these propositions, even close to zero, reducing the legitimacy of the final decision or conclusion.

[0160] Therefore, after completing the iterative process, the counterintuitive correction coefficient is calculated based on the currently updated conflict degree factor K and the current iteration column. f(A) is the probability distribution function of evidence conflict, that is, the conflict degree factor K between evidences is distributed to each element of the recognition framework. i (A) is the basic probability distribution of the behavior formation factor A during the evaluation of the system by the i-th expert, reflecting the degree of influence of A. This probability distribution function satisfies Where A is a specific factor in the behavior formation factor set θ. Based on the counterintuitive correction coefficient, conflict correction is performed on each fusion element in the current iteration column, and based on the correction result, the fusion weight corresponding to each behavior formation factor is obtained. The specific correction method is:

[0161]

[0162] Among them, when the behavior formation factor A is empty, its corresponding basic probability distribution is also 0. i is the support set of the i-th expert for a certain behavior formation factor, B j is the support set of the jth expert for a certain behavior formation factor, C k is the support set of the kth expert for a certain behavior formation factor. In the process of evidence fusion, it is necessary to consider the support degree of all experts' evaluations for the behavior formation factor A. i ∩B j ∩C k ∩…=A means that the intersection of the support sets of the evaluations of different experts is exactly the behavior formation factor A, It means that for each combination that meets the conditions, the corresponding basic probability distributions are multiplied and summed, and then the adjustment term f(A) is added to obtain the final basic probability distribution m(A), which is the fusion weight corresponding to each behavior formation factor.

[0163] In a specific example, the expert group will give a grade and weight evaluation to each PSF. Further, the DS evidence theory is used to perform data fusion on the weights given above. Figure 4 FIG. 1 is a schematic diagram of data fusion in a method for predicting the probability of human error in an aircraft towing departure scenario applicable to an embodiment of the present invention, such as Figure 4 As shown, n experts participate in the evaluation, 6 behavior formation factors, and the conflict coefficient K is initialized to 0. K is used to measure the degree of conflict between different experts on the behavior formation factors. Matrix initialization assigns the first expert's weight matrix M1 to matrix A after transposition, and assigns the first expert's weight matrix M1 to matrix B. Initialize counter i to 2, indicating that the process starts from the second expert. Calculate the matrix A and the weight matrix M of the i-th expert i The product of the weight matrix M of the i-th expert is stored in the matrix R. i Add to matrix B. Calculate the sum of the non-main diagonal elements of matrix R and add it to the conflict coefficient K. Extract the main diagonal elements of matrix R to form a new column matrix and assign it to matrix A. Loop through the initialization operations until all experts' weight matrices have been processed (i>n), exit the loop, calculate and output the final fusion weights of the 6 PSFs. j represents the final fusion weight of the j-th PSF, R jj represents the jth main diagonal element of the matrix R, K is the conflict coefficient factor, B 1j is the jth element of matrix B, and n is the number of experts. The fused data is shown in Table 3.

[0164] Similarly, the fusion level corresponding to each behavior formation factor can also be obtained by using the above method to fuse the level of each PSF using DS evidence theory.

[0165] 4. Evaluate human reliability

[0166] Based on a set of determined PSFs and their levels and weights, the SLIM method is used to evaluate the human reliability of the aircraft towing departure system. Specifically, it can be:

[0167] (1) Calculate the success likelihood index (SLI) corresponding to the task

[0168] According to the formula Where n represents the number of PSFs, R irepresents the level of the i-th PSF, W i represents the weight of the i-th PSF.

[0169] Table 3

[0170]

[0171] (2) Convert the success likelihood index SLI into the corresponding human error probability HEP

[0172] The human error probability (HEP) corresponding to the task is calculated according to the formula: Log(HEP) = aSLI + b, where a and b are preset empirical constants. The constants a and b can be obtained from the two boundary points of the known human error probability (HEP). For example, when the boundary conditions need to be set to SLI = 1, HEP = 0.9 and SLI = 9, HEP = 10 -4 Calculate the corresponding a and b values, so that the SLI can be converted into the corresponding HEP according to the formula. Figure 5 is a schematic diagram of a probability distribution of human errors applicable to an embodiment of the present invention, such as Figure 5 As shown in the figure, the success likelihood index (SLI) and human error probability (HEP) of different tasks and subtasks in the aircraft towing departure scenario are shown. The figure contains multiple concentric rings, each of which represents a different level of information. The innermost layer represents the subtask number (such as 2.11), the middle layer represents the SLI value (such as 2.15), and the outermost layer represents the HEP value. Each color corresponds to a main task, and there are four main tasks in total, namely towing slideout, pushout, pre-towing preparation, and disengagement and evacuation. Subtasks are represented by different sectors, and the corresponding subtask sectors constitute the main task sector. The height of each sector represents the size of the HEP value. The higher the height, the larger the HEP value.

[0173] Optionally, based on the above embodiments, after calculating the human error probability (HEP) value, the following steps may be further performed:

[0174] Based on the manual business operation information in the aircraft towing departure scenario, a secondary task architecture for the aircraft towing departure scenario is constructed;

[0175] The secondary task structure includes a main task level and a subtask level, and the subtask level includes the manual tasks.

[0176] According to the human error probability corresponding to each manual task, the summary human error probability corresponding to each main task is obtained;

[0177] When it is determined that the aggregated human error probability of the target main task is greater than the error probability threshold, the various manual tasks contained in the target main task are analyzed, and at least one target manual task that can be converted or decomposed into an automatically executed task is planned, and a task improvement strategy matching the target manual task is generated. That is, when the aggregated probability of the target main task is greater than the error probability threshold, the various manual tasks therein are analyzed, and a target manual task that can be converted or decomposed into an automatically executed task is planned, and a task improvement strategy matching the target manual task is generated, thereby further improving the reliability and safety of task execution and improving the management level of aircraft towing and departure.

[0178] The method for predicting the probability of human error in the aircraft towing departure scenario proposed in the embodiment of the present invention combines the Bootstrap method with deep learning for qualitative analysis, and combines the DS evidence theory with the success likelihood index method for quantitative calculation to obtain the PSFs and human error probabilities that lead to human errors in the aircraft towing departure process. This effectively improves the subjectivity problem of experts determining PSFs in traditional SLIM, can effectively identify the main PSFs that lead to human errors, promote risk assessment and management, reduce the probability of human errors, and effectively solve the problem of lack of guidance in selecting the main PSF from PSFs, providing support for improving the human reliability of staff in aircraft towing departure.

[0179] Example 3

[0180] Figure 6 This is a schematic diagram of the structure of a device for predicting the probability of human error in an aircraft towing departure scenario provided by the third embodiment of the present invention. Figure 6 As shown, the apparatus includes: a coefficient calculation module 610, a factor screening module 620, an operation analysis module 630, a factor evaluation module 640 and a probability calculation module 650, wherein:

[0181] The coefficient calculation module 610 is used to collect statistics on various risk events that have occurred in the history of aircraft towing departure scenarios and calculate the risk coefficient of each risk event type;

[0182] A factor screening module 620 is configured to screen multiple target risk event types according to each risk coefficient, and screen multiple target human error factors as behavior formation factors based on each target risk event type and the mapping relationship between the risk event type and the human error factor;

[0183] Operation parsing module 630, used to parse manual business operation information in the aircraft towing departure scenario, obtain multiple manual tasks in the aircraft towing departure scenario, and obtain common operation conditions matching the aircraft towing departure scenario;

[0184] The factor evaluation module 640 is used to evaluate the behavior formation factors when performing each manual task under general operating conditions by an expert group, and obtain the weight and level of each behavior formation factor under each manual task;

[0185] The probability calculation module 650 is used to calculate the human error probability of each manual task in the aircraft towing departure scenario based on the evaluation results.

[0186] The technical solution of the embodiment of the present invention can identify the main factors in human error factors by calculating the risk coefficient of historical risk event types and screening target risk event types, combining the mapping relationship with human error factors to screen behavior formation factors. Obtaining manual tasks and general operating conditions in the aircraft towing departure scenario provides comprehensive and specific background information for analysis. The expert group evaluates the behavior formation factors of each manual task under general operating conditions, obtains the weights and levels reflecting the importance of each behavior formation factor, calculates the human error probability of each manual task, avoids the subjective bias of expert experience, can effectively identify the main behavior formation factors that lead to human errors, reduces the probability of human errors, improves the accuracy of risk assessment and management level, and enhances the human factors scientificity and reliability of staff in aircraft towing departure.

[0187] Based on the above embodiments, the coefficient calculation module 610 is specifically configured to:

[0188] In the civil aviation knowledge base, the number of historical risk events and the risk event rates of various risk events in aircraft towing departure scenarios are collected, and an original event sample that meets the event number and risk event rate is constructed;

[0189] Perform random sampling with replacement on the original event sample to obtain multiple expanded event samples;

[0190] Based on the original event sample and the expanded event sample, calculate the target statistical indicator value corresponding to the number of events of each risk event type, and calculate the probability distribution function and confidence interval corresponding to each target statistical indicator value;

[0191] The risk coefficient of each risk event type is calculated based on the probability distribution function and confidence interval corresponding to the target statistical indicator value of each risk event type.

[0192] Optionally, based on the above embodiments, the coefficient calculation module 610 may include: a model building unit and a data fitting unit, wherein:

[0193] A model building unit is used to build a regression analysis model with the target statistical indicator value of each risk event type as the independent variable and the risk event rate in the aircraft towed departure scenario as the dependent variable;

[0194] The data fitting unit is used to perform data fitting on the regression analysis model according to the probability distribution function and confidence interval corresponding to the target statistical indicator value of each risk event type, and obtain the risk event rate impact coefficient corresponding to each risk event type as the risk coefficient.

[0195] Based on the above embodiments, the operation analysis module 630 is specifically configured to:

[0196] Collect external environment information, work status information of each operator, and business capability information from multiple historical aircraft towing departure scenarios;

[0197] The information collection results are clustered, and based on the clustering results, the general operating conditions matching the aircraft towing departure scenario are obtained.

[0198] Based on the above embodiments, the probability calculation module 650 is specifically configured to:

[0199] Obtaining target weights and target levels evaluated by each expert model in the expert group for each behavior formation factor under the target manual task;

[0200] Perform data fusion on each target weight and each target level of each behavior formation factor under the target manual task to obtain the fusion weight and fusion level corresponding to each behavior formation factor under the target manual task;

[0201] According to the formula: Calculate the success likelihood index SLI corresponding to the target manual task;

[0202] Where n is the total number of behavior-forming factors, R i is the fusion level of the i-th behavior formation factor, W i is the fusion weight of the i-th behavior formation factor;

[0203] The human error probability HEP corresponding to the target manual task is calculated according to the formula: Log(HEP)=aSLI+b, where a and b are preset empirical constants.

[0204] Optionally, based on the above embodiments, the probability calculation module 650 may include: a normalization processing unit, a matrix construction unit, an initialization unit, an iterative calculation unit, an accumulation update unit, an operation return unit, a correction coefficient unit, and a fusion weight unit, wherein:

[0205] A normalization processing unit is used to normalize the target weights respectively evaluated by each expert model for each behavior formation factor under the target manual task;

[0206] A matrix construction unit is used to form normalized weight rows corresponding to each expert model according to the obtained normalized weights, and to arrange the normalized weight rows in rows to form a normalized weight matrix;

[0207] An initialization unit, used to obtain the transposed result of the first normalized weight row in the normalized weight matrix as the current iteration column, and obtain the second normalized weight row in the normalized weight matrix as the initialization fusion calculation row;

[0208] An iterative calculation unit is used to calculate the matrix multiplication between the current iteration column and the fusion calculation row to obtain a fusion matrix, and to form a new current iteration column from each main diagonal element in the fusion matrix;

[0209] A cumulative updating unit, configured to cumulatively update the conflict degree factor according to the sum of the non-main diagonal elements in the fusion matrix, wherein the conflict degree factor is initialized to 0;

[0210] An operation return unit, configured to obtain the next row of the fusion calculation row in the normalized weight matrix as a new fusion calculation row, and then return to perform the operation of calculating the matrix multiplication between the current iteration column and the fusion calculation row until all normalized weight rows in the normalized weight matrix are completely processed;

[0211] A correction coefficient unit is used to calculate a counter-intuitive correction coefficient based on the currently updated conflict degree factor and the current iteration column after completing the iterative processing;

[0212] The fusion weight unit is used to perform conflict correction on each fusion element in the current iteration column according to the counter-intuitive correction coefficient, and obtain the fusion weight corresponding to each behavior formation factor under the target manual task according to the correction result.

[0213] Furthermore, based on the above embodiments, the human error probability prediction device in the aircraft towing departure scenario may further include:

[0214] The secondary architecture module is used to construct a secondary task architecture for the aircraft towing departure scenario based on manual business operation information in the aircraft towing departure scenario;

[0215] The secondary task structure includes a main task level and a subtask level, and the subtask level includes the manual tasks.

[0216] A summary probability module is used to summarize the human error probability corresponding to each manual task to obtain the summary human error probability corresponding to each main task;

[0217] The task improvement module is used to analyze the manual tasks contained in the target main task when it is determined that the aggregated human error probability of the target main task is greater than the error probability threshold, plan at least one target manual task that can be converted or decomposed into an automatically executed task, and generate a task improvement strategy that matches the target manual task.

[0218] The human error probability prediction device in the aircraft towing departure scenario provided by the embodiment of the present invention can execute the human error probability prediction method in the aircraft towing departure scenario provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0219] In the technical solutions disclosed herein, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0220] Example 4

[0221] Figure 7 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0222] like Figure 7 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0223] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0224] The processor 11 may be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the method for predicting the probability of human error in an aircraft towing departure scenario, namely:

[0225] Collect statistics on various risk events that have occurred in the past in aircraft towing departure scenarios, and calculate the risk coefficient of each risk event type;

[0226] According to each risk coefficient, multiple target risk event types are screened, and based on each target risk event type and the mapping relationship between the risk event type and the human error factor, multiple target human error factors are screened as behavior formation factors;

[0227] Analyze manual business operation information in the aircraft towing departure scenario, obtain multiple manual tasks in the aircraft towing departure scenario, and obtain common operation conditions that match the aircraft towing departure scenario;

[0228] An expert group evaluates the behavior formation factors when performing various manual tasks under general operating conditions, and obtains the weights and levels of each behavior formation factor under each manual task;

[0229] Based on the evaluation results, the probability of human error for each manual task in the aircraft towing departure scenario is calculated.

[0230] In some embodiments, the method for predicting the probability of human error in an aircraft towing departure scenario can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for predicting the probability of human error in an aircraft towing departure scenario described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to execute the method for predicting the probability of human error in an aircraft towing departure scenario in any other appropriate manner (e.g., via firmware).

[0231] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system comprising at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0232] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0233] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0234] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0235] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0236] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0237] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0238] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for predicting the probability of human error in an aircraft towing departure scenario, characterized in that: include: Collect statistics on various risk events that have occurred in the past in aircraft towing departure scenarios, and calculate the risk coefficient of each risk event type; According to each risk coefficient, multiple target risk event types are screened, and based on each target risk event type and the mapping relationship between the risk event type and the human error factor, multiple target human error factors are screened as behavior formation factors; Analyze manual business operation information in the aircraft towing departure scenario, obtain multiple manual tasks in the aircraft towing departure scenario, and obtain common operation conditions that match the aircraft towing departure scenario; An expert group evaluates the behavior formation factors when performing various manual tasks under general operating conditions, and obtains the weights and levels of each behavior formation factor under each manual task; Based on the evaluation results, the probability of human error for each manual task in the aircraft towing departure scenario is calculated.

2. The method according to claim 1, characterized in that Statistics are collected on various risk events that have occurred in the past in aircraft towing departure scenarios, and the risk coefficient of each risk event type is calculated, including: In the civil aviation knowledge base, the number of historical risk events and the risk event rates of various risk events in aircraft towing departure scenarios are collected, and an original event sample that meets the event number and risk event rate is constructed; Perform random sampling with replacement on the original event sample to obtain multiple expanded event samples; Based on the original event sample and the expanded event sample, calculate the target statistical indicator value corresponding to the number of events of each risk event type, and calculate the probability distribution function and confidence interval corresponding to each target statistical indicator value; The risk coefficient of each risk event type is calculated based on the probability distribution function and confidence interval corresponding to the target statistical indicator value of each risk event type.

3. The method according to claim 2, characterized in that Calculate the risk coefficient of each risk event type based on the probability distribution function and confidence interval corresponding to the target statistical indicator value of each risk event type, including: A regression analysis model was established with the target statistical index value of each risk event type as the independent variable and the risk event rate in the aircraft towed departure scenario as the dependent variable; According to the probability distribution function and confidence interval corresponding to the target statistical indicator values ​​of each risk event type, the regression analysis model is fitted with data to obtain the risk event rate impact coefficient corresponding to each risk event type as the risk coefficient.

4. The method according to claim 1, wherein Obtain common operating conditions that match the aircraft towing departure scenario, including: Collect external environment information, work status information of each operator, and business capability information from multiple historical aircraft towing departure scenarios; The information collection results are clustered, and based on the clustering results, the general operating conditions matching the aircraft towing departure scenario are obtained.

5. The method according to claim 1, wherein Based on the evaluation results, the probability of human error for each manual task in the aircraft towing departure scenario is calculated, including: Obtaining target weights and target levels evaluated by each expert model in the expert group for each behavior formation factor under the target manual task; Perform data fusion on each target weight and each target level of each behavior formation factor under the target manual task to obtain the fusion weight and fusion level corresponding to each behavior formation factor under the target manual task; According to the formula: Calculate the success likelihood index SLI corresponding to the target manual task; Where n is the total number of behavior-forming factors, R i is the fusion level of the i-th behavior formation factor, W i is the fusion weight of the i-th behavior formation factor; The human error probability HEP corresponding to the target manual task is calculated according to the formula: Log(HEP)=aSLI+b, where a and b are preset empirical constants.

6. The method according to claim 5, characterized in that Data fusion is performed on the target weights of each behavior formation factor under the target manual task to obtain the fusion weights corresponding to each behavior formation factor under the target manual task, including: Normalize the target weights evaluated by each expert model for each behavior formation factor under the target manual task; The obtained normalized weights are formed into normalized weight rows corresponding to each expert model, and the normalized weight rows are arranged in rows to form a normalized weight matrix; Get the transposed result of the first normalized weight row in the normalized weight matrix as the current iteration column, and get the second normalized weight row in the normalized weight matrix as the initialization fusion calculation row; Calculate the matrix multiplication between the current iteration column and the fusion calculation row to obtain the fusion matrix, and form the new current iteration column with the main diagonal elements in the fusion matrix; According to the sum of the non-main diagonal elements in the fusion matrix, the conflict degree factor is cumulatively updated, where the conflict degree factor is initialized to 0; After obtaining the next row of the fusion calculation row in the normalized weight matrix as a new fusion calculation row, return to perform the operation of calculating the matrix multiplication between the current iteration column and the fusion calculation row until all normalized weight rows in the normalized weight matrix are completely processed; After the iterative process is completed, the counterintuitive correction coefficient is calculated based on the currently updated conflict degree factor and the current iteration column; According to the counter-intuitive correction coefficient, conflict correction is performed on each fusion element in the current iteration column, and based on the correction result, the fusion weight corresponding to each behavior formation factor under the target manual task is obtained.

7. The method according to any one of claims 1 to 6, characterized in that After calculating the probability of human error corresponding to each manual task in the aircraft towing departure scenario based on the evaluation results, the following steps are also included: Based on the manual business operation information in the aircraft towing departure scenario, a secondary task architecture for the aircraft towing departure scenario is constructed; The secondary task structure includes a main task level and a subtask level, and the subtask level includes the manual tasks. According to the human error probability corresponding to each manual task, the summary human error probability corresponding to each main task is obtained; When it is determined that the aggregated human error probability of the target main task is greater than the error probability threshold, the manual tasks included in the target main task are analyzed, at least one target manual task that can be converted or decomposed into an automatically executed task is planned, and a task improvement strategy matching the target manual task is generated.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for predicting the probability of human error in an aircraft towing departure scenario according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for predicting the probability of human error in an aircraft towing departure scenario according to any one of claims 1 to 7 when executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, which, when executed by a processor, implements the method for predicting the probability of human error in an aircraft towing departure scenario according to any one of claims 1 to 7.

Images