Quantification method and device for monitoring, preventing and controlling flight core risk based on SMS concept

Abstract

The invention provides a quantification method and device for monitoring, preventing and controlling flight core risks based on an SMS concept, and is applied to the technical field of data processing. According to the method, the flight risk related data is quantized, and the quantized risk value data is generated through the risk value calculation formula; constructing a core risk index system, setting different weights for safety performance monitoring indexes of each preposed low-level event, and generating an index template reflecting typical characteristics of flight core risks; calculating an annual average value and a standard deviation value of the core risk values, forming early warning values with different colors, and generating similarity distance parameters of new risk data and historical risk features; on the basis of single-variable time dependence analysis, the association degree between each safety performance index and the core risk is evaluated, and input variables of multivariable risk analysis are generated; and processing the quantitative risk value data, the index template, the similarity distance parameter and the input variable of the multivariable risk analysis to generate target risk scoring information.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a quantitative method and device for monitoring and preventing core flight risks based on the SMS concept. Background Technology

[0002] In the field of data processing technology, the monitoring and prevention of core flight risks face many challenges, and existing technologies have the following shortcomings: Existing methods for quantifying flight risk-related data (such as flight quality monitoring data like aircraft speed, bank angle, pitch attitude, and crew report information) are relatively simple and lack scientific grading mechanisms and weight allocation. This makes it difficult to accurately reflect the actual impact of different risk events, resulting in quantified risk values ​​failing to precisely characterize core flight risks. Furthermore, the lack of a systematic indicator system built around core risks and antecedent events prevents comprehensive coverage of various risk scenarios during flight.

[0003] Traditional early warning methods do not utilize the standard deviation method to calculate the annual average and standard deviation of core risk values, thus failing to establish scientific early warning thresholds and resulting in inaccurate and unreliable early warning results. The correlation analysis between various safety performance indicators and core risks is insufficient, making it impossible to identify target variables that significantly impact core risks and hindering multivariate risk analysis. Existing technologies cannot comprehensively quantify risk value data, indicator templates, similarity distance parameters, and input variables for multivariate risk analysis, making it difficult to generate comprehensive and accurate target risk scoring information.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore includes information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this application is to provide a quantitative method and device for monitoring and preventing core flight risks based on the SMS (Signal Management System) concept. This method overcomes, to some extent, the problems of existing technologies by quantifying flight risk data, assigning coefficients to the number of instances of exceeding limits according to levels, and generating quantitative risk values ​​using formulas by combining event weights and flight segment numbers. An indicator system is constructed and weights are assigned, generating indicator templates. The annual mean and standard deviation of core risk values ​​are calculated using the standard deviation method, generating different color-coded warning values ​​and similarity distance parameters. The correlation is assessed through univariate time-dependent analysis, target variables are selected to generate input variables for multivariate analysis, and finally, the data is integrated to generate a target risk score, thereby improving the monitoring and prevention capabilities of core flight risks.

[0006] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0007] According to one aspect of this application, a quantitative method for monitoring and preventing core flight risks based on the SMS (Signal Management System) concept is provided, comprising: acquiring flight risk-related data, including flight quality monitoring data and crew report information; the flight quality monitoring data including aircraft speed, bank angle, pitch attitude, and touchdown distance; quantifying the flight risk-related data by assigning corresponding coefficients to the number of exceedances in the flight quality monitoring data according to different levels, and generating quantitative risk value data by combining event weights and flight segment numbers through a risk value calculation formula; and constructing a core risk indicator system around 5 core risks and 77 precursor events, and assigning a risk value to each precursor event. The safety performance monitoring indicators for graded events are assigned different weights to generate indicator templates reflecting typical characteristics of core flight risks. Using the standard deviation method, the annual average and standard deviation of core risk values ​​are calculated, resulting in warning values ​​with different colors. A similarity distance parameter between new risk data and historical risk characteristics is generated. Based on univariate time-dependent analysis, the correlation between each safety performance indicator and core risks is assessed, target variables are selected, and input variables for multivariate risk analysis are generated. The quantitative risk value data, indicator templates, similarity distance parameters, and input variables for multivariate risk analysis are processed to generate target risk scoring information.

[0008] Another aspect of this application discloses a quantitative device for monitoring and preventing core flight risks based on the SMS (Survey Management System) concept. The device comprises: an acquisition module for acquiring flight risk-related data, including flight quality monitoring data and crew report information; the flight quality monitoring data includes aircraft speed, bank angle, pitch attitude, and touchdown distance; and a processing module for quantifying the flight risk-related data, assigning corresponding coefficients to the number of exceedances in the flight quality monitoring data according to different levels, and generating quantified risk value data through a risk value calculation formula, combining event weights and flight segment numbers; and constructing core risk indicators around 5 core risks and 77 precursor events. The system assigns different weights to safety performance monitoring indicators for each preceding low-level event, generating indicator templates that reflect typical characteristics of core flight risks. Using the standard deviation method, it calculates the annual average and standard deviation of core risk values, creating warning values ​​with different colors and generating similarity distance parameters between new risk data and historical risk characteristics. Based on univariate time-dependent analysis, it assesses the correlation between each safety performance indicator and core risks, selects target variables, and generates input variables for multivariate risk analysis. Finally, it processes the quantified risk value data, indicator templates, similarity distance parameters, and input variables for multivariate risk analysis to generate target risk scoring information.

[0009] According to another aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a second processor, implements the above-described method for quantifying and monitoring core flight risks based on the SMS concept.

[0010] This application provides a quantitative method and device for monitoring and preventing core flight risks based on the SMS (Signal Management System) concept. The server quantifies flight risk data, assigns coefficients to the number of exceedances according to level, and generates quantitative risk values ​​using a formula by combining event weights and flight segment numbers. An indicator system is constructed and weighted around 5 core risks and 77 precursor events, generating indicator templates. The annual mean and standard deviation of the core risk values ​​are calculated using the standard deviation method, generating different color-coded warning values ​​and similarity distance parameters. The correlation is assessed through univariate time-dependency analysis, target variables are selected to generate multivariate analysis input variables, and finally, the data is integrated to generate a target risk score, improving the monitoring and prevention capabilities of core flight risks.

[0011] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0012] Figure 1 A flowchart illustrating a quantitative method for monitoring and preventing core flight risks based on the SMS concept, provided in an embodiment of this application; Figure 2 This paper presents a schematic diagram of the structure of a quantitative device for monitoring and preventing core flight risks based on the SMS concept, provided in an embodiment of this application. Detailed Implementation

[0013] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0014] The following is combined Figure 1 This application describes a quantitative method for monitoring and preventing core flight risks based on the SMS (Signal Management System) concept, according to exemplary embodiments of the present application. In one embodiment, the present application also proposes a quantitative method and apparatus for monitoring and preventing core flight risks based on the SMS concept. Figure 1 As shown, this method is applied to a server and includes: S101, to acquire flight risk-related data.

[0015] In one implementation, flight quality monitoring data consists of operational parameter data collected in real time by various aircraft sensors and monitoring systems, used to reflect the operational status and potential risks during flight. Specifically, this includes aircraft speeds such as takeoff speed, cruise speed, and landing speed; for example, a flight in March 2023 had a takeoff speed of 180 knots, exceeding the standard value by 10 knots. Bank angle includes the left and right tilt angles during flight; for example, a flight's bank angle reached 30 degrees during a turn, exceeding the safety threshold by 5 degrees. The aircraft's nose angle relative to the horizontal plane is also considered; for example, a flight's pitch angle during landing was 15 degrees, within the normal range of 10-18 degrees. The distance between the aircraft's main landing gear and the runway threshold upon touchdown is also considered; for example, a flight in April 2023 had a touchdown distance of 2000 meters, exceeding the standard value by 500 meters.

[0016] The crew report includes written records from the crew of any abnormalities, equipment malfunctions, or operational problems encountered during or after the flight. Specific details include: Equipment abnormalities: e.g., "May 10, 2023, B-737-800 aircraft, flight number HX1234, autopilot abnormally disengaged during cruise." Operational problems: e.g., "June 15, 2023, B-737-800 aircraft, flight number HX5678, asymmetrical engine thrust during takeoff." Environmental feedback: e.g., "July 20, 2023, B-737-800 aircraft, flight number HX9012, encountered strong crosswinds (25 knots) during approach."

[0017] Flight quality monitoring data is received in real time or downloaded afterward through aircraft data recording systems (such as Fast Access Recorders, FAR) and ground monitoring systems. During flight, the flight data recording system collects parameters such as aircraft speed and bank angle once per second and stores them in onboard storage devices. After landing, ground maintenance personnel download the data through a dedicated data interface and import it into the company's flight quality monitoring system. For example, Hebei Airlines obtained 10,390 flight quality monitoring data points in this way in 2023.

[0018] Crew report information is obtained through submissions by crew members via paper reports or electronic systems. After the flight, the captain or responsible crew member fills out a crew report, describing the problems encountered during the flight, such as "March 20, 2023, B-737-800 aircraft, flight number HX3456, abnormal rudder control felt during landing." The report is submitted to the airline's operations management department, where it is entered into the information management system by designated personnel to form structured data. By analyzing flight quality monitoring data in March 2023, it was found that a certain flight had a long touchdown distance (2000 meters). Combined with the record of "obstructed visibility during landing" in the crew report, it was determined that this event was strongly correlated with the core risk of "runway overrun," triggering a risk warning. Comparing the number of aircraft speed exceedances from January to December 2023, it was found that the number of takeoff speed exceedances increased by 30% in July compared to other months. Combined with the meteorological data for that month, high temperature weather was identified as the main influencing factor, and a targeted training plan was developed.

[0019] S102 quantifies flight risk-related data by assigning corresponding coefficients to the number of times limits are exceeded in the flight quality monitoring data according to different levels, and generating quantitative risk value data by combining event weights and flight segment numbers through the risk value calculation formula.

[0020] In one implementation, the number of times exceeding limits in the flight quality monitoring data is assigned corresponding coefficients according to different levels to generate graded exceedance count data. Specifically, the number of times exceeding limits in the flight quality monitoring data is assigned corresponding coefficients (Level 1 × 0.1, Level 2 × 0.3, Level 3 × 0.4) according to different levels (Level 1, Level 2, Level 3) to generate graded exceedance count data. Taking the "long grounding distance" event as an example, flight quality monitoring data for a certain flight in March 2023 shows: Level 1 exceedances: 66 times (corresponding coefficient 0.1); Level 2 exceedances: 2 times (corresponding coefficient 0.3); Level 3 exceedances: 0 times (corresponding coefficient 0.4); the graded exceedance count data is calculated as follows: Level 1 grade value: 66 × 0.1 = 6.6; Level 2 grade value: 2 × 0.3 = 0.6; Level 3 grade value: 0 × 0.4 = 0; the generated graded exceedance count data is [6.6, 0.6, 0].

[0021] By combining event weights and the number of flight segments, the data on the number of times the limit is exceeded at different levels is weighted and normalized to generate weighted normalized data. The weighted calculation is then performed by dividing by (number of flight segments / 1000) to eliminate the influence of the number of flight segments. Continuing the previous example, the event weight for "long grounding distance" is 5.439% (i.e., 0.05439), and the number of flight segments in March 2023 is 3266. The weighted processing is as follows: Weighted value of the number of times the limit is exceeded by each level = (6.6 + 0.6 + 0) × 0.05439 = 7.2 × 0.05439 ≈ 0.3916. Normalization is performed as follows: Normalization factor = 3266 / 1000 = 3.266; Weighted normalized data = 0.3916 / 3.266 ≈ 0.120.

[0022] Based on weighted normalized data, a summation process is performed to generate quantified risk value data using the risk value calculation formula. The weighted normalized data of all preceding events under the same core risk are summed, and a quantified risk value is generated using the risk value calculation formula: .

[0023] The core risks of runway deviance in March 2023 included 77 preceding events. Among them, the risk value of "long touchdown distance" was 0.120. The risk value of another preceding event, "unstable landing roll direction," was calculated as follows: Level 1 violations: 10 times; Level 2 violations: 1 time; Level 3 violations: 0 times. Event weight: 5.45% (0.0545); Number of flight segments: 3266.

[0024] Weighted average of the number of times the limit was exceeded: (10×0.1+1×0.3+0×0.4)×0.0545=1.3×0.0545≈0.0709. Weighted normalized data: 0.0709 / (3266 / 1000)≈0.0709 / 3.266≈0.0217. Total risk value of running off the track: sum of the weighted normalized data of all preceding events. For example, the sum of the risk values ​​of each event under this core risk in March 2023 is 0.315.

[0025] S103 constructs a core risk indicator system around 5 core risks and 77 precursor events, sets different weights for the safety performance monitoring indicators of each precursor low-level event, and generates indicator templates that reflect the typical characteristics of core flight risks.

[0026] In one implementation, five core risks and 77 antecedent events are extracted and grouped to generate risk event grouping information. From ten initial core risks, five quantifiable and monitorable core risks (runway overrun, loss of control in mid-air, controlled flight into the ground, mid-air conflict, and taxiway deviation) are selected. For each core risk, associated low-level antecedent events are extracted to form grouping information. The extraction of antecedent events is based on 98 flight quality monitoring standards (such as exceeding thresholds for altitude angle and touchdown distance); high-frequency abnormal events in crew reports (such as GPWS warnings and abnormal approach bank angles); and flight expert experience (combined with retrospective analysis of historical unsafe events, such as the high altitude angle event associated with the skid-grabbing incident in April 2023).

[0027] Typical examples of the correlation between core risks and preceding events are as follows: Runway overrun: long touchdown distance (directly affecting landing distance); unstable landing direction (causing runway deviation); aborted takeoff (abnormal operation during takeoff); high final approach speed (increasing touchdown impact and landing distance). Controlled flight impact: high takeoff angle (potentially causing skid strike or stall); steep approach slope (deviation from the normal glide path, increasing the risk of impact); GPWSTERRAINPULLUP warning (terrain warning system triggered, indicating near-ground risk); high final approach descent rate (excessive descent rate, making timely attitude adjustment difficult).

[0028] The logic for generating risk event grouping information is as follows: clustering by core risk category to ensure that each preceding event belongs to only one core risk group; prioritizing the inclusion of events with a direct causal relationship to the core risks (e.g., "long touchdown distance" directly related to "runway deviation"). An example of the grouping results is as follows: 1. Runway deviation (20 preceding events): long touchdown distance, unstable landing roll direction, aborted takeoff, high final approach speed, secondary touchdown, high takeoff speed, large approach slope (200-50ft), ILS glide slope deviation (above / below standard), abnormal ground speed after automatic braking release, excessively high runway exit speed (>40 knots), ... (the remaining 10 items are omitted).

[0029] 2. Controllable flight crash (18 precursor events): large pitch angle, large approach bank angle, GPWSTERRAIN PULLUP warning, GPWSCAUTIONTERRAIN warning, large descent rate during final approach (500-50ft), large descent rate below 50 feet, low landing configuration establishment (flaps / landing gear not properly deployed), stall warning during flight, low nose wheel lift-off speed, ... (remaining 10 items omitted).

[0030] 3. In-flight loss of control (15 preceding events): bank angle exceeding limits (turning angle > 30 degrees), abnormal pitch attitude (pitch angle > 18 degrees or < -5 degrees), abnormal autopilot disengagement, asymmetrical engine thrust, Dutch roll during flight, ... (remaining 10 items omitted) 4. Airborne conflict (12 preceding events): deviation from designated altitude layer, failure to maintain safe separation, TCAS alarm (TrafficAlert), transponder coding error, ... (remaining 8 items omitted).

[0031] 5. Deviation from the taxiway (12 prerequisite events): Exceeding taxiing speed limits, deviating from the centerline, failing to turn as instructed, accidentally entering a closed runway while taxiing, ... (remaining 8 items omitted) Each preceding event was assigned a weight (e.g., "long touchdown distance" weighted at 5.439%), and a core risk value was calculated using a risk value formula (e.g., runway miss risk value of 0.315 in March 2023). When a preceding event exceeded its limit multiple times, triggering an early warning (e.g., a yellow warning for "large approach slope"), it was directly linked to the core risk of "controllable flight into the ground," driving targeted measures (as shown in the FX-2024-9-AQJX-08 preventive and corrective action order). Following the "large final approach descent rate" warning in March 2024, the event actually occurred on March 28th, verifying the accuracy of the grouping correlation.

[0032] The weights of safety performance monitoring indicators for 77 pre-race events were calculated and statistically analyzed to generate weight distribution characteristics. The Analytic Hierarchy Process (AHP) was used to calculate the weights of the 77 pre-race events. By constructing a judgment matrix, calculating eigenvectors, and performing consistency checks, the weight values ​​for each event were obtained, and the weight distribution characteristics were analyzed. A judgment matrix was constructed using the 10 pre-race events related to "runway deviation" as an example, as shown below:

[0033] The weights for "long touchdown distance" are calculated as 5.439%, "unstable landing roll direction" as 5.45%, and "aborted takeoff" as 5.77%. High-weight events (>5%) include aborted takeoff (5.77%) and long touchdown distance (5.439%); medium-weight events (3%-5%) include "unstable landing roll direction" (5.45%) and "high final approach speed" (3.10%); low-weight events (<3%) include some minor operational deviation events.

[0034] The risk event grouping information and weight distribution characteristics are processed to generate the average weight distribution information of the core risk indicator system. Precursor events are grouped according to core risks, and the average weight and distribution range of events within each group are calculated to form the average weight distribution information for each group. An example is shown below: Runway deviation core risk: Total weight of precursor events: 5.439% + 5.45% + 5.77% + ... = 100%; Average weight: 100% / 20 items ≈ 5%; Distribution range: 2.88% (low land morphology) ~ 5.77% (abrupt takeoff).

[0035] Generate weighted average distribution information, as shown in the following examples: Runway deviation: average weight: 5.0%; weight range: 2.88%-5.77%; Controlled flight impact: average weight: 4.8%; weight range: 3.00%-6.47% (GPWS warning events).

[0036] Based on the weighted average distribution information of the core risk indicator system, an indicator template reflecting the typical characteristics of core flight risks is generated. The construction basis of the indicator template is as follows: grouping information of 5 core risks and 77 preceding events (e.g., runway deviation includes 20 preceding events); weight values ​​calculated by the analytic hierarchy process (e.g., "long touchdown distance" has a weight of 5.439%); and integrating the over-limit thresholds in the flight quality monitoring standards with historical data characteristics (e.g., touchdown distance > 2000 meters is a level 1 over-limit).

[0037] The core elements of the template include the following: Core Risk: The clearly defined risk category (e.g., runway overrun, controlled flight into the ground); Precursor Event: The name of the specific low-level event that triggers the risk (e.g., long touchdown distance, large approach slope); Weight: The contribution of this event to the core risk (calculated by AHP, ranging from 2.88% to 6.47%); Typical Feature Description: Quantified exceedance standards, frequency of occurrence, or operational deviation threshold.

[0038] The following is an example of an indicator template:

[0039] The standards for classifying over-limit levels are as follows: Level 1 over-limit: slight deviation from the standard, such as a touchdown distance > 2000 meters (standard value 1500 meters); Level 2 over-limit: moderate deviation, such as a runway deviation angle > 3 degrees (standard value 2 degrees); Level 3 over-limit: severe deviation, such as aborting takeoff (triggering emergency procedures).

[0040] When the number of Level 2 exceedances of the "Severe Approach Inclination" event reaches 8 times per month (e.g., 9 times in May 2024), a yellow alert is triggered (average + 1σ), which is associated with the core risk of controllable flight impacting the ground. Prevention measures are formulated, including a specific training plan for the "GPWSTERRAIN PULLUP Warning" (weight 6.47%), requiring pilots to conduct simulation training for handling this warning twice per month. If a flight team experiences ≥50 exceedances of the "Long Touchdown Distance" event for two consecutive months, 5 points will be deducted from their performance evaluation, and a rectification report will be required.

[0041] Taking the 10 pre-runway overrun events as an example, the CR of the judgment matrix is ​​0.035 < 0.1, indicating a reasonable weight allocation (e.g., "aborted takeoff" 5.77% > "long touchdown distance" 5.439%, consistent with their impact on runway overrun). The template is dynamically adjusted, with weights updated annually based on new data. For example, if the "high final approach speed" event leads to an increase in the proportion of runway overruns in 2024, its weight can be adjusted from 3.10% to 3.5%.

[0042] S104 uses the standard deviation method to calculate the annual average and standard deviation of the core risk value, forming warning values ​​with different colors, and generating a similarity distance parameter between new risk data and historical risk characteristics.

[0043] In one implementation, the core risk value is processed by calculating its annual average and standard deviation to generate mean and standard deviation features. The monthly risk values ​​of the core risk are statistically analyzed throughout the year, and their average (μ) and standard deviation (σ) are calculated to form mean and standard deviation features reflecting historical risk levels. Taking the core risk of "runway slippage" as an example, the risk values ​​for January to December 2023 are as follows: 0.272, 0.216, 0.315, 0.324, 0.287, 0.337, 0.383, 0.282, 0.218, 0.250, 0.255, 0.216. Average value calculation: Standard deviation calculation: Calculate the sum of squared deviations of each data point from the mean. Standard deviation: The final generated features are the mean feature μ=0.280 and the standard deviation feature. =0.053.

[0044] Based on mean and standard deviation characteristics, and combined with warning level coefficients, different colored warning values ​​are generated. Based on the mean (μ) and standard deviation (σ), and combined with warning level coefficients (1σ, 2σ, 3σ), yellow, orange, and red warning values ​​are generated, forming risk warning lines. Specifically, the yellow warning value (mild risk): μ + 1σ = 0.280 + 0.053 = 0.333; the orange warning value (moderate risk): μ + 2σ = 0.280 + 2 × 0.053 = 0.386; and the red warning value (severe risk): μ + 3σ = 0.280 + 3 × 0.053 = 0.439. When the risk value in a certain month exceeds 0.333, a yellow warning is triggered, indicating the need for attention; exceeding 0.386 triggers an orange warning, initiating special rectification; and exceeding 0.439 triggers a red warning, implementing emergency prevention and control measures.

[0045] Based on warning values ​​and new risk data, a similarity distance parameter is generated between the new risk data and historical risk characteristics. The new risk data is compared with historical risk characteristics (mean, standard deviation, warning value), and the distance to the historical risk pattern is calculated to generate the similarity distance parameter. The new risk data is as follows: the runway deviation risk value in March 2024 was 0.269. Similarity distance calculation (Euclidean distance): .

[0046] Similarity assessment: the smaller the distance, the more similar the new risk is to historical characteristics. For example, if the distance is <0.1σ (0.0053), it is considered highly similar, and historical prevention and control measures should be referenced; if the distance is >0.5σ (0.0265), it is considered significantly different, and the root cause needs to be re-analyzed. The similarity distance parameter is generated as follows: the similarity distance of the runway deviation risk value in March 2024 is 0.054, which is considered moderately similar, indicating that historical data should be considered, but targeted adjustments to prevention and control measures are necessary.

[0047] Taking the core risk of "runway deviating" as an example: the risk value in July 2023 was 0.383, exceeding the orange warning value of 0.386 (difference 0.003), close to the warning threshold, and its similarity distance to the risk value of 0.315 in March 2023 was 0.068 (calculated using Euclidean distance). =0.068), judged as "moderate historical similarity, requiring vigilance against worsening trend". Coordinated measures: Retrieve the prevention and control plan from March 2023 (such as strengthening training on landing runway direction control), and formulate a combination of measures based on current risk characteristics (increasing proportion of events with long ground contact distances).

[0048] In March 2024, a major descent rate warning was issued during the final approach, with a risk value of 0.269 triggering a yellow alert. Similarity analysis showed a high degree of similarity to the risk pattern in November 2023 (0.255, similarity distance 0.014). The approach attitude-specific training program from November 2023 was continued, with the addition of simulator training subjects specifically for descent rate control. In May 2024, a major approach slope event occurred, with an initial risk value of 0.001568 triggering a yellow alert, and a similarity distance of 0.002 (similar to the data from May 2023). If the risk value rises to 0.0019 in the following two weeks (approaching the orange alert value of 0.001976), the similarity distance will be updated to 0.0007, which is considered a "sudden change in risk pattern," requiring a special investigation (e.g., insufficient crew training or abnormal instrument parameters).

[0049] Following the large approach slope control in May 2024, a preventative and corrective action order (FX-2024-9-AQJX-08) was issued to strengthen visual and instrument-based training. In June, the risk value decreased to 0.000647, and the similarity distance increased to 0.0007, similar to historical safety cycle data (June 2023), validating the effectiveness of the measures. Regarding the similarity correlation across core risks, when the "large approach slope" event of the "controllable flight into the ground" core risk triggers an orange alert, if similarity analysis shows an indirect correlation with the "long touchdown distance" event of the "runway overrun" event (e.g., both due to weather-related obstructed visibility), the allocation of prevention and control resources for both core risks will be adjusted accordingly.

[0050] S105, based on univariate time-dependent analysis, assesses the correlation between various safety performance indicators and core risks, selects target variables, and generates input variables for multivariate risk analysis.

[0051] In one implementation, a univariate time-dependence analysis is performed on the correlation between each safety performance indicator and the core risks to generate correlation analysis results. Each safety performance indicator includes safety performance monitoring data for each preceding low-level event, and the core risks include runway miss and controlled flight impact risk values. Time series analysis methods (such as Pearson correlation coefficient and autoregressive models) are used to calculate the time-dependent correlation between each preceding event indicator and the core risk value, identifying strongly correlated indicators. Taking the core risk of "runway miss" as an example, its correlation with 20 preceding events is analyzed: Precursor events include long touchdown distance (X1), unstable landing runway direction (X2), and aborted takeoff (X3); Core risk value: Y (runway deviation monthly risk value).

[0052] Correlation calculation (Pearson correlation coefficient): Calculate the correlation coefficient between X1 and Y: The calculation results are as follows: the correlation coefficient between the long ground contact distance (X1) and Y is r=0.78 (strong positive correlation); the correlation coefficient between the unstable landing roll direction (X2) and Y is r=0.65; the correlation coefficient between the aborted takeoff (X3) and Y is r=0.42; and the correlation coefficient between other events is r<0.3.

[0053] The correlation analysis results show that the correlation of the core risks of running off the runway is ranked as follows: 1. Long touchdown distance (0.78); 2. Unstable landing roll direction (0.65); 3. High final approach speed (0.59)... (Other events omitted).

[0054] The correlation analysis results were filtered to generate a target variable list. The template variable list included antecedent event indicators such as long touchdown distance, large approach slope, and large final approach descent rate. A correlation threshold (e.g., r>0.5) was set to filter out antecedent events strongly correlated with core risks, forming the target variable list. The threshold was set to retain events with r>0.5. The filtering results are as follows: Runway deviation: long touchdown distance (0.78), unstable landing roll direction (0.65), large final approach speed (0.59); Controlled flight impact: large approach slope (0.82), large final approach descent rate (0.75), large ground elevation angle (0.68); The target variable list is as follows: long touchdown distance, unstable landing roll direction, large final approach speed, large approach slope, large final approach descent rate, large ground elevation angle.

[0055] Based on the target variable list, input variables for multivariate risk analysis are generated. These input variables characterize key features of safety performance indicators associated with core risks. Based on the target variable list (such as antecedent event indicators like long grounding distance and steep approach slope), key features associated with core risks are extracted to form the input dataset for multivariate analysis. Feature extraction follows these principles: quantifiable data: based on structured records from flight quality monitoring data or crew reports; time-dependent: reflecting the changing trend of events over time; risk-related: directly affecting the calculation logic of core risk values.

[0056] Taking the core risks of runway deviation and controlled flight impact as an example: (I) Long touchdown distance events. Original data: Flight quality monitoring data in March 2023 showed that this event had 66 Level 1 violations and 2 Level 2 violations, with 3266 flight segments. The features extracted are as follows: Violation frequency features: Level 1 violations (X1): 66 times (directly involved in the risk value formula calculation); Level 2 violations (X2): 2 times (corresponding coefficient 0.3); Normalization features, violation frequency (X3): (66×0.1+2×0.3) / 3266≈0.0022 (i.e., the contribution value of each flight segment to the violation); Weighted correlation features: Event weight product (X4): 0.0022×5.439%≈0.00012 (intermediate result of the risk value calculation formula).

[0057] (II) Major Approach Slope Events. Raw data: In May 2024, there were 9 instances of Level 2 overruns, involving 2850 flight segments, with an event weight of 4.49%. Feature extraction: Overrun level characteristics: Number of Level 2 overruns (Y1): 9 times (corresponding coefficient 0.3); Trend characteristics: Number of consecutive overruns in months (Y2): 2 months (continuous overruns in April-May 2024); Risk contribution characteristics: Risk contribution value (Y3): (0×0.1+9×0.3+0×0.4)×4.49% / 2850≈0.00042 (directly included in the controllable flight collision risk value).

[0058] The following is an example of the input variable matrix structure:

[0059] When the X2_over-limit frequency of "long touchdown distance" exceeds 0.003 (historical average + 1σ), the runway overrun risk value increases by an average of 22%; for every 0.0001 increase in the Z2_risk contribution value of "large approach slope," the controllable flight collision risk value increases by 15%. In March 2024, the input variable (number of over-limits × coefficient × weight / segment) for "large final approach descent rate" was detected to be 0.00058, triggering a yellow alert. The event actually occurred on March 28, verifying the accuracy of variable prediction. Weight integration: The "event weight product" (such as X4, Z2) in the input variables directly uses the results calculated by the analytic hierarchy process (e.g., long touchdown distance weight 5.439%, large approach slope weight 4.49%); Consistency check: The weight association features in the input variables must meet the analytic hierarchy process CR < 0.1 standard (e.g., the judgment matrix for runway overrun events CR = 0.035).

[0060] S106 processes the quantitative risk value data, indicator templates, similarity distance parameters, and input variables for multivariate risk analysis to generate target risk score information.

[0061] In one implementation, the quantified risk value data undergoes quantitative analysis to generate risk quantification factors. The quantified risk value data is then standardized, transforming it into a dimensionless factor comparable across core risks. The original risk values ​​are: runway overrun risk of 0.315 in March 2023, and controlled flight impact risk of 0.287. Standardization processing (taking runway overrun as an example): This factor indicates the degree to which the current risk value deviates from the historical mean (0.66 means 0.66 standard deviations above the mean).

[0062] The average weight distribution information in the indicator template is quantitatively analyzed to generate indicator weight factors, which characterize the contribution of each preceding event to the core risk. Based on the average weight distribution information in the indicator template, the contribution weight percentage of each preceding event to the core risk is calculated. Example: Runway deviation core risk: includes 20 preceding events, total weight 100%, average weight 5%. Long ground contact distance event: weight 5.439%, then its weight factor is: Indicator Weight This event has a weighting higher than the average of 8.78%, and contributes significantly to the core risks.

[0063] The similarity distance parameter between new risk data and historical risk characteristics is quantitatively analyzed to generate a risk similarity factor. The similarity distance between new risk data and historical risk characteristics is converted into a similarity factor in the 0-1 range, as shown in the example below: New risk data: Runway deviation risk value in March 2024 is 0.269, and the Euclidean distance from the historical mean is 0.054. Similarity factor calculation: The similarity assessment shows a factor of 0.36, indicating that the new risk has a low similarity to historical patterns, and the differences need to be closely monitored.

[0064] The target variable in the input variables of multivariate risk analysis is quantitatively analyzed to generate key indicator impact factors. These key indicator impact factors characterize the degree of influence of key antecedent events on core risk. Key features of the target variable are extracted, and their dynamic impact on core risk is quantified. An example is as follows: the target variable is long grounding distance, 70 instances of Level 1 overruns in May 2024, 3000 flight segments, and a weight of 5.439%. Impact factor calculation: Impact level: This event contributed 0.0127 to the runway deviation risk value, accounting for 15% of the total risk value for the month.

[0065] Based on a risk assessment model framework that integrates risk quantification factors, indicator weighting factors, risk similarity factors, and key indicator impact factors, core risks are fused to generate comprehensive core risk assessment features. These comprehensive core risk assessment features, by fusing four types of quantification factors, form a multi-dimensional vector, achieving a three-dimensional characterization of core flight risks. The construction principles are as follows: multi-dimensional coverage: taking into account current risk levels, historical similarity, event weight contributions, and immediate impact, avoiding bias from single indicators; data standardization: converting factors with different dimensions (such as risk values, weight percentages, and distance parameters) into comparable dimensionless values; dynamic correlation: real-time integration of new risk data with historical models to reflect risk evolution trends.

[0066] Taking the runway deviation risk in March 2023 as an example, its comprehensive assessment feature vector is: V = [Risk quantification factor = 0.66, weight factor = 1.0878, similarity factor = 0.36, impact factor = 0.0127]. The calculation process of the risk quantification factor (0.66) is as follows: (Data source: 2023 runway deviation risk value mean μ=0.280, standard deviation σ=0.053, March risk value 0.315). The current risk value is 0.66 standard deviations higher than the historical average, which is considered a medium to high risk level.

[0067] The weighting factor (1.0878) is calculated as follows: (Runway deviation includes 20 preceding events, with an average weight of 5%; "long touchdown distance" has a weight of 5.439%, making it a high-contribution event). This event's weight is 8.78% higher than the average, significantly contributing to the core risk.

[0068] The similarity factor (0.36) is calculated as follows: (The Euclidean distance between the new risk data and historical features is 0.054; the distance is converted to similarity using an exponential function.) The new risk pattern has a similarity of 36% with historical data, requiring caution regarding its compatibility with historical prevention and control measures. The impact factor (0.0127) is calculated as follows. (In March 2023, there were 66 instances of Level 1 exceedance of the "long touchdown distance" rule, involving 3266 flight segments and a weight of 5.439%). This event directly contributed 0.0127 to the runway deviation risk value, accounting for 4% of the total risk value for the month (0.0127 / 0.315≈4%).

[0069] If the risk quantification factor is high but the similarity factor is low (e.g., V=[0.8,1.1,0.2,0.015]), it indicates that the current risk is caused by a novel event (e.g., a long grounding distance due to rare weather), and a targeted prevention and control plan needs to be developed. If the similarity factor is high (e.g., >0.7), historical prevention and control measures from the same period can be directly retrieved (e.g., the runway deviance risk pattern in July 2022 is similar, so the "enhanced landing visibility training" plan can be reused).

[0070] Analysis of the linkage between weighting factor and impact factor: If the weighting factor of an event is greater than 1.2 and the impact factor is greater than 0.02, it is judged as a "high contribution-high impact" event, and training resources are allocated in priority (such as conducting simulator-specific training for the "abrupt takeoff" event).

[0071] The runway deviation risk assessment in May 2024 was as follows: Comprehensive assessment vector: V=[0.72,1.15,0.68,0.018]. Risk quantification factor 0.72: Risk value 0.32 (0.28+0.72×0.053≈0.32), reaching 96% of the yellow warning threshold of 0.333; Weight factor 1.15: "Landing runway instability" weight 5.75% (higher than the average of 5%); Similarity factor 0.68: Highly similar to the risk pattern in May 2023 (V=[0.65,1.12,0.71,0.017]); Impact factor 0.018: "Landing runway instability" Level II exceedance 12 times, contributing 5.6% of the risk value for the month (0.018 / 0.32≈5.6%). The prevention and control measures are as follows, and the plan from May 2023 will be reused: conduct training on the taxiing direction control simulator; strengthen targeted measures and increase the monthly monitoring frequency for events with a weight factor > 1.1 (from once / week to twice / week).

[0072] Based on the risk assessment model, the comprehensive assessment characteristics of core risks are analyzed and processed to generate target risk score information. Taking the core risk of running off the track in March 2023 as an example, its comprehensive assessment feature vector is V=[risk quantification factor=0.66, weight factor=1.0878, similarity factor=0.36, impact factor=0.0127]. The target risk score is generated through the following two typical models: (I) Linear weighted model scoring model formula: target risk score = ×Risk Quantification Factor+ ×weight factor+ ×Similarity Factor+ × Impact Factor.

[0073] (The weighting coefficients are calibrated based on historical data.) =0.4, =0.3, =0.2, =0.1) The calculation process is as follows: Score = 0.4×0.66+0.3×1.0878+0.2×0.36+0.1×0.0127≈0.264+0.326+0.072+0.0013≈0.663. The score mapping is as follows, risk level classification (0-1 point system): Low risk: <0.4 Medium risk: 0.4-0.7 High risk: >0.7 Result: 0.663 belongs to medium risk, triggering a yellow warning (corresponding warning value 0.333).

[0074] The risk assessment model architecture is as follows: Input layer: 4 neurons (corresponding to 4 types of factors), Hidden layer: 1 layer (8 neurons, activation function ReLU), Output layer: 1 neuron (risk score 0-1). Training data is as follows: Historical data: comprehensive assessment features and actual risk levels of runway deviation risk in 2022-2023 (e.g., a score of 0.81 in July 2023 corresponds to high risk). Input vector V=[0.66,1.0878,0.36,0.0127]. Hidden layer calculation: f(0.66× +1.0878× +0.36× +0.0127× + ). ) is the weight, (where f is the ReLU function for bias). Output layer result: 0.68 (mapped to 0-1 via the Sigmoid function). Score interpretation: 0.68 is close to the upper limit of medium risk, indicating that attention should be paid to high contribution events (such as long grounding distance) corresponding to "weight factor = 1.0878", as the number of times it exceeds the limit may continue to increase.

[0075] The scoring system is linked to the early warning threshold for prevention and control measures: A linear model score of 0.663 corresponds to a risk value of 0.315, which is close to the yellow warning value of 0.333, triggering the following measures: issuing a "Risk Warning Notice" to the flight team, requiring them to focus on monitoring long grounding distance events; organizing specialized training: simulating landing operations under high grounding distance scenarios. Historical case reuse: If the score is ≥0.7 and the similarity factor is ≥0.7 (e.g., similar to a scenario with a score of 0.75 in November 2022), historical prevention and control plans are directly reused: adjusting flight quality monitoring parameter thresholds (e.g., tightening the first-level over-limit standard for long grounding distance from 2000 meters to 1800 meters); increasing the frequency of captain qualification retraining (from once a year to twice a year).

[0076] In July 2023, the risk score was 0.82 (high risk), and one runway slip occurred. The model's prediction accuracy was 85% (out of 12 historical high-risk scores, 10 corresponded to actual risk events). Dynamic optimization mechanism: Model parameters are updated monthly. If the predictive contribution of a certain factor is less than 5% for two consecutive months (e.g., the impact factor), its weight coefficient is adjusted (e.g., from 0.1 to 0.05) to avoid redundant calculations.

[0077] This application aims to address the problems of crude risk quantification and incomplete indicator systems in existing technologies. First, flight quality monitoring data (including aircraft speed, grounding distance, etc.) and crew report information are acquired. The number of instances exceeding limits is assigned coefficients according to level, and combined with event weights and flight segment numbers, a quantitative risk value is generated through a formula. An indicator system is constructed using the analytic hierarchy process (AHP) around 5 core risks and 77 preceding events, generating weighted indicator templates. The annual mean and standard deviation of the core risk values ​​are calculated using the standard deviation method, generating yellow, orange, and red warning values, as well as similarity distance parameters between new and historical risks. The correlation is assessed through univariate time dependence analysis, and target variables are selected to generate input variables for multivariate analysis. Finally, the quantitative risk values ​​and indicator templates are integrated to generate a target risk score. The resulting quantitative risk values ​​are accurate, the indicator templates are comprehensive, the warnings are scientific, the correlation analysis is in-depth, and the comprehensive assessment is thorough, enhancing risk monitoring and control capabilities and making it suitable for the aviation field.

[0078] In one implementation, such as Figure 2 As shown, this application also provides a quantitative device for monitoring and preventing core flight risks based on the SMS concept, comprising: The acquisition module 201 is used to acquire flight risk-related data, including flight quality monitoring data and crew report information. The flight quality monitoring data includes aircraft speed, bank angle, pitch attitude, and touchdown distance. Processing module 202 is used to quantify flight risk-related data. It assigns corresponding coefficients to the number of exceedances in flight quality monitoring data according to different levels, and generates quantified risk value data by combining event weights and flight segment numbers through a risk value calculation formula. It constructs a core risk indicator system around 5 core risks and 77 precursor events, setting different weights for the safety performance monitoring indicators of each precursor low-level event, and generating indicator templates reflecting the typical characteristics of core flight risks. Using the standard deviation method, it calculates the annual average and standard deviation of core risk values, forming warning values ​​with different colors, and generating similarity distance parameters between new risk data and historical risk characteristics. Based on univariate time dependence analysis, it evaluates the correlation between each safety performance indicator and core risks, selects target variables, and generates input variables for multivariate risk analysis. Finally, it processes the quantified risk value data, indicator templates, similarity distance parameters, and input variables for multivariate risk analysis to generate target risk score information.

[0079] The various embodiments in this application are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for the quantitative method, electronic device, electronic device, and readable storage medium for assessing the monitoring and prevention of core flight risks based on the SMS concept are basically similar to the above-described embodiments of the quantitative method for monitoring and prevention of core flight risks based on the SMS concept, and therefore are described relatively simply. Relevant parts can be referred to the descriptions of the above-described embodiments of the quantitative method for monitoring and prevention of core flight risks based on the SMS concept.

Claims

1. A quantitative method for monitoring and preventing core flight risks based on the SMS concept, characterized in that, include: Acquire flight risk-related data, including flight quality monitoring data and crew report information. Flight quality monitoring data includes aircraft speed, bank angle, pitch attitude, and touchdown distance. The flight risk-related data is quantified. The number of times the flight quality monitoring data exceeds the limit is assigned a corresponding coefficient according to different levels. Combined with the event weight and the number of flight segments, quantitative risk value data is generated through the risk value calculation formula. A core risk indicator system was constructed around 5 core risks and 77 precursor events. Different weights were assigned to the safety performance monitoring indicators of each precursor low-level event, and indicator templates reflecting the typical characteristics of core flight risks were generated. Using the standard deviation method, the annual average and standard deviation of the core risk value are calculated to form warning values ​​with different colors, and a similarity distance parameter between new risk data and historical risk characteristics is generated. Based on univariate time dependency analysis, the correlation between various safety performance indicators and core risks is evaluated, target variables are selected, and input variables for multivariate risk analysis are generated. The system processes quantitative risk value data, indicator templates, similarity distance parameters, and input variables for multivariate risk analysis to generate target risk score information.

2. The method as described in claim 1, characterized in that, The flight risk-related data is quantified by assigning corresponding coefficients to the number of exceedances in the flight quality monitoring data according to different levels. Combining event weights and flight segment numbers, quantitative risk value data is generated using a risk value calculation formula, including: The number of times exceeding limits in flight quality monitoring data is assigned corresponding coefficients according to different levels to generate graded number of times exceeding limits data. By combining event weights and flight segment counts, the data on the number of times the limits were exceeded at different levels are weighted and normalized to generate weighted normalized data. Based on the summation of weighted normalized data, quantified risk value data is generated using the risk value calculation formula, where the risk value calculation formula is as follows: .

3. The method as described in claim 1, characterized in that, A core risk indicator system was constructed based on 5 core risks and 77 potential events. Different weights were assigned to the safety performance monitoring indicators for each potential low-level event, generating indicator templates that reflect the typical characteristics of core flight risks, including: Five core risks and 77 precursor events were extracted and grouped to generate risk event grouping information. The weights of 77 pre-event safety performance monitoring indicators were calculated and statistically analyzed to generate weight distribution characteristic information. The risk event grouping information and weight distribution characteristics are processed to generate the average weight distribution information of the core risk indicator system; Based on the weight average distribution information of the core risk indicator system, an indicator template reflecting the typical characteristics of core flight risks is generated.

4. The method as described in claim 1, characterized in that, Using the standard deviation method, the annual average and standard deviation of the core risk value are calculated, generating warning values ​​with different colors. A similarity distance parameter between new risk data and historical risk characteristics is also generated, including: The core risk values ​​are processed by calculating the annual average and standard deviation values ​​to generate mean and standard deviation features. Based on mean and standard deviation characteristics, warning values ​​of different colors are generated by combining warning level coefficients. Based on the warning value and new risk data, a similarity distance parameter between the new risk data and historical risk characteristics is generated.

5. The method as described in claim 1, characterized in that, Based on univariate time-dependent analysis, the correlation between various safety performance indicators and core risks is assessed, target variables are selected, and input variables for multivariate risk analysis are generated, including: A univariate time-dependent analysis was conducted to assess the correlation between each safety performance indicator and the core risks, generating the correlation analysis results. Each safety performance indicator includes safety performance monitoring data for each preceding low-level event, and the core risks include runway deviance and the risk values ​​for controlled flight impact. The correlation analysis results were filtered to generate a list of target variables. The template variable list included antecedent event indicators such as long grounding distance, large approach slope, and large final approach descent rate. Based on the list of target variables, input variables for multivariate risk analysis are generated. These input variables are used to characterize the key features of safety performance indicators that are associated with core risks.

6. The method as described in claim 5, characterized in that, The quantitative risk value data, indicator templates, similarity distance parameters, and input variables for multivariate risk analysis are processed to generate target risk score information, including: Quantitative analysis and processing are performed on the quantitative risk value data to generate risk quantification factors; The average distribution information of weights in the indicator template is quantitatively analyzed and processed to generate indicator weight factors, which are used to characterize the contribution of each preceding event to the core risk. The similarity distance between new risk data and historical risk characteristics in the similarity distance parameter is quantitatively analyzed to generate a risk similarity factor. The target variable in the input variables of multivariate risk analysis is quantitatively analyzed to generate key indicator impact factors, which are used to characterize the degree of impact of key antecedent events on core risks. Based on the framework of risk quantification factors, indicator weight factors, risk similarity factors, and key indicator impact factors combined with the risk assessment model, the core risks are integrated and processed to generate comprehensive core risk assessment features. Based on the risk assessment model, the core risk comprehensive assessment characteristics are analyzed and processed to generate target risk score information.

7. A quantitative device for monitoring and preventing core flight risks based on the SMS concept, characterized in that, The device includes: The acquisition module is used to acquire flight risk-related data, including flight quality monitoring data and crew report information. Flight quality monitoring data includes aircraft speed, bank angle, pitch attitude, and touchdown distance. The processing module quantifies flight risk-related data, assigning coefficients to different levels of exceedances in flight quality monitoring data, and generating quantified risk value data through a risk value calculation formula, combining event weights and flight segment numbers. It constructs a core risk indicator system around 5 core risks and 77 precursor events, setting different weights for the safety performance monitoring indicators of each precursor low-level event, and generating indicator templates reflecting typical characteristics of core flight risks. Using the standard deviation method, it calculates the annual average and standard deviation of core risk values, forming warning values ​​with different colors, and generating similarity distance parameters between new risk data and historical risk characteristics. Based on univariate time dependence analysis, it assesses the correlation between each safety performance indicator and core risks, selects target variables, and generates input variables for multivariate risk analysis. Finally, it processes the quantified risk value data, indicator templates, similarity distance parameters, and input variables for multivariate risk analysis to generate target risk score information.

8. An electronic device, characterized in that, include: First processor; and memory for storing executable instructions of the first processor; The first processor is configured to execute the quantitative method for monitoring and preventing core flight risks based on the SMS concept, as described in any one of claims 1 to 6, by executing the executable instructions.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the second processor, it implements the quantitative method for monitoring and preventing core flight risks based on the SMS concept as described in any one of claims 1 to 6.

Images