CN-120748261-B - Method and system for risk early warning and cause identification of airplane rushing out of runway event

CN120748261BCN 120748261 BCN120748261 BCN 120748261BCN-120748261-B

Abstract

The invention discloses a risk early warning and incentive recognition method and system for an airplane out-of-runway event, wherein the method comprises the following steps of S0, obtaining a key parameter set and a risk factor set of the out-of-runway event through a FOQA standard, SOP and QRH, S1, constructing a corresponding relation between the key parameter and the risk factor based on a Delphi method, S2, constructing a out-of-runway risk prediction model based on a Transformer, realizing online risk recognition, and S3, obtaining the most similar incentive based on a hypothesis test method, and completing risk factor recognition.

Inventors

GAO ZHENXING
LIU JIMING
WU HAO
Yue Jinchengrui

Assignees

南京航空航天大学

Dates

Publication Date: 20260512
Application Date: 20250625

Claims (9)

1. A method for risk early warning and predisposition identification of an aircraft rushing/out of runway event, the method comprising: S0, acquiring a key parameter set related to the take-off and landing of the aircraft through a flight quality monitoring standard, a flight standard operation program and a flight rapid reference manual; S1, constructing a corresponding relation between key parameters and risk factors based on an improved Delphi method; S2, constructing a collision/deviation runway risk prediction model based on a transducer, extracting collision/deviation runway event characteristics from flight data by using the collision/deviation runway risk prediction model based on the transducer, and predicting the distance of the aircraft from the central line of the runway and the distance of the rest runway at a certain moment in the future by using a historical flight data sequence so as to realize online risk identification; S3, obtaining key parameters of the difference between the early warning flights and the normal flights exceeding a preset threshold value based on a hypothesis testing method, and outputting the most similar incentive according to the obtained corresponding relation between the key parameters and the risk factors to finish the risk factor identification; In the S1, key parameters are established based on an improved Delphi method And risk factors Specifically, the corresponding relation of (a) includes: s1.1, constructing a interdisciplinary expert team and designing to issue an anonymous scoring questionnaire for evaluating any risk factor , And any key parameter , Collecting subjective comments and opinions of experts; S1.2 based on the obtained scoring questionnaire for each expert, for each risk factor , Constructing a parameter association matrix , wherein, For the number of experts to be present, Is the first Expert to key parameters And risk factors And then introducing the average value of the evaluation index Index of consensus degree Distance between four parts Anonymous scoring data, mean value of all experts Index of consensus degree Distance between four parts And subjective comments and comments are provided to each expert again, and each expert needs to pay attention to the subjective comments and comments again <3.5、 <0.7、 2 Scoring and indexes mentioned in subjective comment opinions, and improving and updating the original scoring result, wherein the process is repeated for 2-3 times; S1.3, according to the finally obtained scoring data, if 3.5, The risk factor is considered Subject to critical parameters The influence of the key parameter is larger than a preset threshold value, and finally the corresponding relation between the key parameter and the risk factor is obtained and is recorded as , 。
2. The method according to claim 1, wherein in S0, key parameters are And risk factors The method specifically comprises the following steps: key parameters , As a number of key parameters, Is the 1 st critical parameter Comprising And Two parts of the two-way valve are arranged on the two sides, For a set of key parameters acquired from the flight data, , Is that The number of elements in (a) is, Is that Element 1 of (a); for a key parameter set acquired by weather message data and airport pavement data, , Is that The number of elements in (1) and , Is that Element 1 of (a); Risk factors , As a function of the number of risk factors, Representing risk factor 1.
3. The method according to claim 1, wherein in S2, a collision/deviation runway risk prediction model based on a Transformer is constructed, collision/deviation runway event characteristics are extracted from flight data by using the collision/deviation runway risk prediction model based on the Transformer, and the off-runway center line distance and the remaining runway distance of the aircraft at a certain time in the future are obtained from a historical parameter sequence, and the method for implementing online risk identification comprises: S2.1 is directed at 0- Sequence of flight data for an aircraft over time Converting a flight parameter sequence into a high-dimensional feature matrix with time sequence dependence and multi-parameter association by using a 6-layer Encoder-layer structure Then input it into 4 layer Decoder layer, i.e. Decoder layer to decode, based on the global feature and history information provided by Encoder layer to gradually deduce runway state at future time, and finally output the future Time of day offset from runway centerline distance Distance from remaining runway Wherein On representative flight A sequence of flight data at a time of day is recorded, Representing the 1 st flight data in A recorded value of the time of day, For a high-dimensional time-series feature matrix, For the flight on The time-of-day timing feature vector, Hiding layer dimensions for the model; S2.2, based on the historical flight data, all flights are on Setting 95% quantile of the distance from the center line of the runway to the maximum value of the threshold interval of the distance from the center line Setting the minimum value of the threshold interval of the offset center line distance to 5% quantile All flights are on The 5% quantile of the remaining runway distance at the moment is set as the remaining runway distance threshold When (1) > Or (b) < When the aircraft is in At the moment, there is a risk of deviating from the runway < Indicating that the aircraft is in Recording the current early warning moment when the airplane has the risk of deviating or rushing out of the runway And Key parameters of time of day The risk status at the next moment is then continued to be monitored.
4. The method of claim 1, wherein in S3, based on the hypothesis testing method, obtaining key parameters that the difference between the early warning flight and the normal flight exceeds a preset threshold, and combining the obtained key parameters with the corresponding relationship between the risk factors to output the most similar causes, the method for completing the risk factor identification comprises: When (when) When the early warning is issued at any time, the device will Key parameters of time of day With other normal flights Key parameters of time of day A hypothesis test is performed, wherein, Representative of The number of different flights of a flight, Represents the first On individual flights The key parameter of the moment is obtained, and the key parameter that the difference between the early warning flight and the normal flight exceeds a preset threshold value is recorded as the incentive parameter , Subsequently calculate incentive parameters Corresponding relation with each , And finally outputting the risk factor with the maximum Jaccard similarity And (5) completing risk factor identification.
5. A risk early warning and cause identification system for an aircraft approach/departure runway event for implementing the method of any one of claims 1-4, comprising an acquisition module, a construction module, an online risk identification module, and a risk factor identification module; The acquisition module is used for acquiring a key parameter set related to the take-off and landing of the aircraft through a flight quality monitoring standard, a flight standard operation program and a flight rapid reference manual; the construction module is used for constructing the corresponding relation between the key parameters and the risk factors based on the improved Delphi method; The online risk identification module is used for constructing a collision/deviation runway risk prediction model based on a transducer, extracting collision/deviation runway event characteristics from flight data by using the collision/deviation runway risk prediction model based on the transducer, and predicting the distance of the aircraft from the central line of the runway and the distance of the remaining runway at a certain moment in the future by using a historical flight data sequence so as to realize online risk identification; the risk factor identification module is used for obtaining key parameters of the difference between the early warning flight and the normal flight exceeding a preset threshold value based on a hypothesis test method, and outputting the most similar incentive according to the obtained corresponding relation between the key parameters and the risk factors to complete the risk factor identification.
6. The system of claim 5, wherein the obtaining module obtains key parameters And risk factors The method specifically comprises the following steps: key parameters , As a number of key parameters, Is the 1 st critical parameter Comprising And Two parts of the two-way valve are arranged on the two sides, For a set of key parameters acquired from the flight data, , Is that The number of elements in (a) is, Is that Element 1 of (a); for a key parameter set acquired by weather message data and airport pavement data, , Is that The number of elements in (1) and , Is that Element 1 of (a); Risk factors , As a function of the number of risk factors, Representing risk factor 1.
7. The system of claim 5, wherein the construction module comprises a scoring questionnaire unit, a matrix construction unit, a comparison unit; The scoring questionnaire unit is used for building a cross-discipline expert team and designing and issuing an anonymous scoring questionnaire for evaluating any risk factor , And any key parameter , Collecting subjective comments and opinions of experts; The matrix construction unit is used for aiming at each risk factor based on the obtained scoring questionnaire of each expert , Constructing a parameter association matrix , wherein, For the number of experts to be present, Is the first Expert to key parameters And risk factors And then introducing the average value of the evaluation index Index of consensus degree Distance between four parts Anonymous scoring data, mean value of all experts Index of consensus degree Distance between four parts And subjective comments and comments are provided to each expert again, and each expert needs to pay attention to the subjective comments and comments again <3.5、 <0.7、 2 Scoring and indexes mentioned in subjective comment opinions, and improving and updating the original scoring result, wherein the process is repeated for 2-3 times; the comparison unit is used for comparing the obtained scoring data if 3.5, The risk factor is considered Subject to critical parameters The influence of the key parameter is larger than a preset threshold value, and finally the corresponding relation between the key parameter and the risk factor is obtained and is recorded as , 。
8. The system of claim 5, wherein the online risk identification module comprises a network construction unit, a risk monitoring unit; The network construction unit is used for aiming at 0- Sequence of flight data for an aircraft over time Converting a flight parameter sequence into a high-dimensional feature matrix with time sequence dependence and multi-parameter association by using a 6-layer Encoder-layer structure Then input it into 4 layer Decoder layer to decode, based on the global feature and history information provided by Encoder layer, gradually deduce runway state at future time, and finally output future Time of day offset from runway centerline distance Distance from remaining runway Wherein On representative flight A sequence of flight data at a time of day is recorded, Representing the 1 st flight data in A recorded value of the time of day, For a high-dimensional time-series feature matrix, For the flight on The time-of-day timing feature vector, Hiding layer dimensions for the model; The risk monitoring unit is used for enabling all flights to be on the basis of historical flight data Setting 95% quantile of the distance from the center line of the runway to the maximum value of the threshold interval of the distance from the center line Setting the minimum value of the threshold interval of the offset center line distance to 5% quantile All flights are on The 5% quantile of the remaining runway distance at the moment is set as the remaining runway distance threshold When (1) > Or (b) < When the aircraft is in At the moment, there is a risk of deviating from the runway < Indicating that the aircraft is in Recording the current early warning moment when the airplane has the risk of deviating or rushing out of the runway And Key parameters of time of day The risk status at the next moment is then continued to be monitored.
9. The system of claim 5, wherein the risk factor identification module obtains key parameters of the difference between the early-warning flight and the normal flight exceeding a preset threshold based on a hypothesis testing method, and outputs a most similar incentive in combination with the obtained corresponding relation between the key parameters and the risk factors, and the process of completing the risk factor identification comprises: When (when) When the early warning is issued at any time, the device will Key parameters of time of day With other normal flights Key parameters of time of day A hypothesis test is performed, wherein, Representative of The number of different flights of a flight, Represents the first On individual flights The key parameter of the moment is obtained, and the key parameter that the difference between the early warning flight and the normal flight exceeds a preset threshold value is recorded as the incentive parameter , Subsequently calculate incentive parameters Corresponding relation with each , And finally outputting the risk factor with the maximum Jaccard similarity And (5) completing risk factor identification.

Description

Method and system for risk early warning and cause identification of airplane rushing out of runway event Technical Field The invention belongs to the field of civil aviation safety technology and flight data application, and particularly relates to a risk early warning and incentive recognition method and system for an airplane rushing out of runway event. Background Runway events, in which aircraft are in the near landing stage (out or off), are one of the most significant causes of fatal accidents. The flight quality monitoring technology based on post-flight data analysis is a key means of cause analysis and risk assessment of airplane rushing out/deviating runway events, and has important significance for flight safety guarantee. However, current flight quality monitoring techniques remain inadequate in terms of real-time early warning and traceability of the aircraft for out-of-runway events. The main expression is as follows: firstly, the existing runway-flushing/runway-deviating detection mode mainly combines flight data to judge whether runway-flushing/runway-deviating events occur after the course is finished, and real-time monitoring of the state of an airplane and online risk early warning are difficult to achieve, so that flight crews cannot find potential risks in advance, and corresponding strategies are further made to prevent threats and errors from further evolving into runway-flushing/runway-deviating events. Second, there is a lack of an automated incentive recognition method for hedging/off-track events. The identification of the evoked cause of the current approach/departure runway event relies on manual analysis by a flight safety expert in combination with post-flight data to further determine the event evoked risk type. Automated identification of the cause of the washout/out runway event can assist flight safety professionals in better understanding risk patterns. And gives auxiliary decision information to flight crews to a certain extent, and takes corresponding precautions. Disclosure of Invention In order to solve the problems existing in the prior art, the invention provides a risk early warning and incentive recognition method and system for an airplane rushing out of runway event, aiming at the airplane rushing out of runway event, and starting from an aircraft running quality monitoring standard, a flight standard operating program and a flight rapid inspection list, obtaining key parameters related to the take-off and landing of the aircraft, and establishing a risk factor fault tree to obtain risk factors for triggering the runway of the aircraft. And then obtaining the corresponding relation between the key parameters and the risk factors through a Delphi method so as to realize objective analysis of the risk factors. On the basis, an anomaly prediction model is established to realize online risk monitoring and early warning, and an assumption inspection method is combined to complete automatic incentive recognition. In order to achieve the above object, the present invention provides the following solutions: A method of risk early warning and incentive recognition for an aircraft approach/departure runway event, the method comprising: S0, acquiring a key parameter set related to the take-off and landing of the aircraft through a flight quality monitoring standard, a flight standard operation program and a flight rapid reference manual; S1, constructing a corresponding relation between key parameters and risk factors based on an improved Delphi method; S2, constructing a collision/deviation runway risk prediction model based on a transducer, extracting collision/deviation runway event characteristics from flight data by using the collision/deviation runway risk prediction model based on the transducer, and predicting the distance of the aircraft from the central line of the runway and the distance of the rest runway at a certain moment in the future by using a historical flight data sequence so as to realize online risk identification; And S3, obtaining key parameters of the difference between the early warning flight and the normal flight exceeding a preset threshold value based on a hypothesis test method, and outputting the most similar incentive by combining the obtained corresponding relation between the key parameters and the risk factors to finish the risk factor identification. Preferably, in S0, the key parameters pv= { v 1,v2,...,vd } and the risk factor pr= { R 1,R2,...,Rm } specifically include: The key parameter PV comprises PX and PW, wherein PX is a key parameter set obtained through flight data, PX= { x 1,x2,...,xb }, b is the number of elements in PX, x 1 is the 1 st element in PX, PW is a key parameter set obtained through weather message data and airport pavement data, PW= { w 1,w2,...,wc }, c is the number of elements in PW, d=b+c, and w 1 is the 1 st element in PW; Risk factors pr= { R 1,R2,...,Rm }, m is the number of risk factors, and R 1 represent