CN114742458A - Airport flight operation toughness assessment and prediction method in severe weather - Google Patents

Airport flight operation toughness assessment and prediction method in severe weather Download PDF

Info

Publication number
CN114742458A
CN114742458A CN202210492086.9A CN202210492086A CN114742458A CN 114742458 A CN114742458 A CN 114742458A CN 202210492086 A CN202210492086 A CN 202210492086A CN 114742458 A CN114742458 A CN 114742458A
Authority
CN
China
Prior art keywords
performance
toughness
time
morp
airport
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210492086.9A
Other languages
Chinese (zh)
Inventor
王兴隆
陈仔燕
赵俊妮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Civil Aviation University of China
Original Assignee
Civil Aviation University of China
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Civil Aviation University of China filed Critical Civil Aviation University of China
Priority to CN202210492086.9A priority Critical patent/CN114742458A/en
Publication of CN114742458A publication Critical patent/CN114742458A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Data Mining & Analysis (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Operations Research (AREA)
  • Computational Linguistics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Educational Administration (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Game Theory and Decision Science (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Software Systems (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides an airport flight operation toughness assessment and prediction method in severe weather, which comprises the following steps of 1: constructing an airport flight operation toughness evaluation model, and evaluating the airport flight operation performance through different measurement indexes; step 2: and constructing an airport flight operation toughness prediction model based on the LSTM, and predicting the airport performance in severe weather. The method can predict the operation toughness of the airport flight according to meteorological factors, thereby lightening disasters and improving emergency response capability.

Description

Airport flight operation toughness assessment and prediction method in severe weather
Technical Field
The invention belongs to the technical field of aviation, and particularly relates to an airport flight operation toughness assessment and prediction method in severe weather.
Background
In recent years, the air transportation industry has continuously kept a high-speed growth situation, and airport traffic has rapidly increased year by year. Due to the fact that software and hardware are updated and iterated continuously, the performance of an aircraft, a control technology, ground guarantee and an operation program are improved continuously, and scheduling of an airline company is scientific day by day, the influence of factors such as equipment reasons and human factors on the safety and efficiency of airport operation is reduced continuously, and severe weather becomes a main factor restricting normal flight of a flight. In order to resist the serious influence caused by severe weather, the research on the operation toughness of the airport flights needs to be carried out.
Unlike a large body of literature on the influence of extreme weather on travel activities, traffic system capacity, reliability, robustness and vulnerability, research on toughness of traffic systems in post-disaster stages is still rare, and most of the existing research on traffic toughness adopts a topological method to quantify the toughness of traffic systems. The commonly used approach is to simulate an emergency scenario by continually removing nodes/links in the traffic network and to simulate an emergency gradual dissipation and system recovery scenario by recovering nodes/links. The recovery capability of the system is quantified each time a node/link is removed or restored. Although these studies provide a preliminary search for the measurement and assessment of traffic toughness, there are few studies that have focused primarily on the network level to quantify the toughness of traffic network nodes or infrastructures, especially lacking methods to assess the toughness of major traffic hubs. Furthermore, most of them are based on assumed emergency scenarios and simulation data, and lack of using real case data to evaluate the degree of influence of different factors on airport performance, which has limitations in reflecting airport flight toughness.
In the aspect of researching the influence of severe weather on the airport flight delay prediction, deep learning shows strong potential in general prediction research due to the flexible model structure and strong learning capacity. For data with strong time relation, the RNN method of the recurrent neural network shows competitive capacity. However, traffic in the traffic system has complex non-euclidean correlations and directivities, exhibiting strong topological properties, rather than the general euclidean spatial correlations. For these data, the original RNN is not applicable, and Long short-term memory networks (LSTM) may perform better.
In general, there are certain achievements in research on influence of severe weather on flight operation of an airport, but most models are based on certain assumptions and are disjointed from actual flight operation conditions. Most researches are not deep enough for researching the coupling relation between severe weather and flight operation, few researches are carried out on the change of the flight operation performance of an airport in the whole severe weather process, and the conventional researches mostly adopt the conventional statistical prediction method and are lack of the combination with deep learning.
Disclosure of Invention
In view of the above, the present invention is directed to a method for evaluating and predicting the operation toughness of an airport flight in severe weather.
In order to achieve the purpose, the technical scheme of the invention is realized as follows:
an airport flight operation toughness assessment and prediction method in severe weather comprises the following steps
Step 1: constructing an airport flight operation toughness evaluation model, and evaluating the airport flight operation toughness through different measurement indexes;
step 2: and constructing an airport flight operation toughness prediction model based on the LSTM, and predicting the performance of the airport in severe weather.
Further, in the step 1, an Departure Rate is used as an evaluation index of an airport flight operation toughness model in severe weather, and the Departure Rate (Departure Rate) is defined as follows:
Figure BDA0003631939250000031
wherein:
Figure BDA0003631939250000032
is [ T ]1,T2]Total number of scheduled departed flights in the time period;
Figure BDA0003631939250000033
representing planned takeoff time at [ T1,T2]Flight number with simultaneous departure delay less than 30 minutes in a time period;
Figure BDA0003631939250000034
is represented at T1Delayed before time to [ T ]1,T2]Number of flights leaving the field in the time period.
Further, in step 1, measuring the performance of the evaluation index at different stages by using different toughness stage metrics, where the metrics include response time, failure rate, robustness, recovery time, recovery rate, recovery capability, performance loss, and average performance loss.
Further, each metric is as follows
Response Time (RST): from the start of the interference event to the time t at which the system performance begins to declined'<t<tdIndicating the capability of the system to resist external interference;
RST＝td-td'
wherein, td' is the beginning of bad weather, tdThe time when the system is interfered by the outside world and the performance begins to decline;
destruction Time (DSS): time period t from initial decline of system performance to lowest performanced<t<trIndicating the length of the system performance degradation time;
DSS＝tr-td
wherein, trThe time when the system is interfered by the outside and the performance is reduced to the lowest value;
rate of destruction (RAPI)DP): in the destruction phase td<t<trRepresenting the speed at which the system degrades from initial performance to minimum performance;
Figure BDA0003631939250000041
wherein, trThe time when the system is interfered by the outside and the performance is reduced to the lowest value; t is tdThe time when the system is interfered by the outside world and the performance begins to decline; MORP (t)d) Is the system performance at the moment the performance starts to decline, MORP (t)r) Representing the lowest value of the system performance under external interference;
robustness (Robustness, R): the system still has the capability of well maintaining the stability under the condition of being interfered by an external event;
R＝min{MORP(t)}(td<t<tns)
wherein MORP (t) represents a discrete function over time, tdIndicating the moment when the system starts to decline in performance due to disturbance, tnsThe time when the system recovers to a new stable stage is shown, R represents the maximum value of the system performance in the time period, and the maximum influence on the system under the disturbance of an external event can be measured;
recovery Time (RCT): time period t from the moment of lowest system performance to the time when the system is restored to a new steady stater<t<tnsIndicating the length of the system performance recovery time;
RCT＝tns-tr
recovery Rate (RAPI)RP): in a recovery phase tr<t<tnsRepresenting the speed at which the system recovers from minimum performance to performance in a new stable phase;
Figure BDA0003631939250000051
wherein MORP (t) represents a time-varying discrete function of system performance in the event of a disturbance, trFor the moment when the system is disturbed by the outside world and the performance is reduced to the minimum, tnsIndicating the moment at which the system recovers to a new stable phase, MORP (t)ns) Is the value at which the performance returns to the plateau, MORP (t)r) Representing the lowest value of the system performance under external interference;
recovery Ability (RA): is represented by t ≧ tnsThe system reaches a new stable stage;
Figure BDA0003631939250000052
wherein MORP (t) represents a discrete function of system performance over time in the event of a single disturbance, MORP (t)ns) Is the value at which the performance returns to the plateau, MORP (t)r) Represents the lowest value of the system performance under external interference, MORP (t)d) Is the system performance at the time the performance begins to decline;
loss of Performance (Loss of Performance, LOP): represents the total amount of performance degradation over the course of the occurrence of the confounding event;
Figure BDA0003631939250000061
average Performance Loss (Time average Performance Loss, TAPL): represents the overall performance loss of the system in the damage phase and the recovery phase;
Figure BDA0003631939250000062
wherein MORP (t) represents a discrete function of system performance over time at a single interference event; t is tdRepresenting the moment when system performance begins to decline; t is tnsThe moment when the system is restored to the new stable stage; MORPweek(t) represents a function of the change in system performance under normal conditions.
Further, the step 1 further includes establishing a non-negative comprehensive toughness index to measure the toughness of different systems under the same destructive event and the toughness of the same system under the different destructive events, and evaluating the operation toughness of the airport flight by obtaining the performance change of the airport flight in a time sequence, wherein the non-negative comprehensive toughness index (NGR):
Figure BDA0003631939250000063
further, in the step 2, constructing an airport flight operation toughness prediction model architecture based on the LSTM includes performing dimensionality reduction on data by using a PCA technology, performing deep learning on an airport performance MORP value by using message data by using the LSTM model, predicting a result through an FC layer, and processing the predicted result through a Sigmoid activation function to obtain a final predicted value MORP'.
Further, a loss function of the prediction model for the airport flight operation toughness is constructed based on the LSTM, the Mean-square error (MSE) of the MORP' and the true value MORP is adopted,
loss＝MSE(MORP,MORP')。
furthermore, the Mean Square Error (MSE) is used as a measurement index for evaluating the difference between the predicted value and the true value, the fitting degree of the predicted value and the true value of the prediction model is judged,
Figure BDA0003631939250000071
wherein y is the MORP true value (label), y' is the predicted value obtained by the prediction model on the input data, the Mean Square Error (MSE) value range is [0, + ∞ ], and the closer to 0 indicates that the predicted value is closer to the true value, and the better the model prediction performance is.
Compared with the prior art, the method for evaluating and predicting the operation toughness of the airport flight in severe weather has the following advantages:
(1) the existing toughness research method is popularized to wider and real disaster conditions, various severe weathers are involved, not only one severe weather is involved, an aviation decision maker can take measures in advance according to the change rule and the prediction result of the airport flight operation performance under different severe events to reduce the influence caused by the severe weather, and the method has important significance for reducing flight delay, guaranteeing the safety of an airport flight operation system and the like.
(2) And providing a new airport operation performance toughness index departure rate, and providing a non-negative comprehensive toughness index by combining the actual situation of airport flight operation. The change rule of the airport flight operation performance in the process from occurrence to dissipation of different severe weather is researched, the relation between severe weather events and the airport flight operation performance is explored, and the method has important practical significance for reducing the influence of severe weather on the airport flight operation.
(3) According to meteorological factors, the toughness of airport flight operation is predicted by using a deep learning model LSTM, theoretical support is provided for personnel of airport apron control and operation management centers to command flights and coordinate decisions, and safe and efficient operation of airports is guaranteed.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate an embodiment of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
FIG. 1 is a schematic diagram of the change in toughness of a system;
FIG. 2 is a schematic diagram of a model for predicting airport flight toughness in severe weather based on LSTM;
FIG. 3 is a Sigmoid function image;
FIG. 4 is a schematic diagram of changes in airport flight performance under different severe weather conditions;
FIG. 5 is a schematic diagram comparing GR and NGR;
FIG. 6 shows the value of loss of prediction model training for airport flight operation toughness in severe weather based on LSTM.
Detailed Description
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
The present invention will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
1 airport flight operation toughness evaluation model
1.1 toughness evaluation model construction
The invention provides an index which has memorability and can comprehensively measure the actual operation performance of airport flights in severe weather, namely the Departure Rate (Departure Rate) is defined as follows:
Figure BDA0003631939250000091
in the formula:
Figure BDA0003631939250000092
is [ T ]1,T2]The total number of scheduled departure flights within the time period;
Figure BDA0003631939250000093
representing planned takeoff time at [ T1,T2]Flight number with simultaneous departure delay less than 30 minutes in a time period;
Figure BDA0003631939250000094
is represented at T1Delay before time to [ T ]1,T2]Number of flights leaving the field in the time period. The index can accurately describe the actual operation performance of the airport flight, not only considers the proportion of the on-time departure flight quantity in a certain period of time to the planned departure flight quantity, but also considers the performance of the delayed flight in the later period of time.
The method adopts the departure punctuality rate as an airport flight operation toughness model evaluation index in severe weather. The model divides the system into 5 stages before and after disturbance, as shown in fig. 1, the vertical axis is a system Performance toughness index (MORP) which changes along with time, the selection of the characterization system Performance toughness index depends on the content to be researched, the value range fluctuates between [0 and 1], wherein 0 represents that the system is in a paralyzed state, and 1 represents that the system is in an ideal state.
The different stages before and after the system is disturbed and the corresponding evaluation methods are summarized in table 1:
TABLE 1 summary of various stages of the toughness process
Figure BDA0003631939250000101
1.2 toughness measurement index
In order to describe and analyze the airport flight operation performance, the following toughness measurement indexes are selected:
1. response Time (RST): from the start time of the interference event to the systemTime t at which system performance begins to declined'<t<tdIndicating the capability of the system to resist external interference;
RST＝td-td' (1.2)
wherein, td' is the beginning of bad weather, tdThe time when the system is disturbed by the outside world, the performance begins to decline.
2. Destruction Time (DSS): time period t from initial decline of system performance to lowest performanced<t<trIndicating the length of the system performance degradation time;
DSS＝tr-td (1.3)
wherein, trThe moment when the system is subjected to external interference and the performance is reduced to the lowest value.
3. Rate of destruction (RAPI)DP): in the destruction phase td<t<trRepresenting the speed at which the system degrades from initial performance to minimum performance;
Figure BDA0003631939250000102
wherein, trThe time when the system is interfered by the outside and the performance is reduced to the lowest value; t is tdThe time when the system is interfered by the outside world and the performance begins to decline; MORP (t)d) Is the system performance at the moment when performance begins to decline, MORP (t)r) Representing the lowest value of the system performance under external interference.
4. Robustness (Robustness, R): the system still has the capability of well maintaining the stability under the condition of being interfered by an external event;
R＝min{MORP(t)}(td<t<tns) (1.5)
wherein MORP (t) represents a discrete function over time, tdIndicating the time at which the performance of the system begins to decline due to disturbances, tnsThe moment when the system is restored to a new stable stage is shown, R represents the maximum value of the system performance in the time period, and the external event disturbance can be measuredThe following maximum impact on the system.
5. Recovery Time (RCT): time period t from the moment of lowest system performance to the time when the system is restored to a new steady stater<t<tnsIndicating the length of the system performance recovery time;
RCT＝tns-tr (1.6)
6. recovery Rate (RAPI)RP): in a recovery phase tr<t<tnsRepresenting the speed at which the system recovers from minimum performance to performance in a new stable phase;
Figure BDA0003631939250000111
wherein MORP (t) represents a time-varying discrete function of system performance in the event of a disturbance, trFor the moment when the system is disturbed by the outside world and the performance is reduced to the minimum, tnsIndicating the moment at which the system recovers to a new stable phase, MORP (t)ns) Is the value at which the performance returns to the plateau, MORP (t)r) Representing the lowest value of the system performance under external interference.
7. Recovery Ability (RA): is represented by t ≧ tnsThe system reaches a new stable stage;
Figure BDA0003631939250000121
wherein MORP (t) represents a discrete function of system performance over time in the event of a single disturbance, MORP (t)ns) Is the value at which the performance returns to the plateau, MORP (t)r) Represents the lowest value of the system performance under external interference, MORP (t)d) Is the system performance at the moment performance begins to decline.
8. Loss of Performance (Loss of Performance, LOP): represents the total amount of performance degradation over the course of the occurrence of the confounding event;
Figure BDA0003631939250000122
9. average Performance Loss (Time Averaged Performance Loss, TAPL): represents the overall performance loss of the system in the damage phase and the recovery phase;
Figure BDA0003631939250000123
wherein MORP (t) represents a discrete function of system performance over time at a single interference event; t is tdRepresenting the moment when system performance begins to decline; t is tnsThe moment when the system is restored to the new stable stage; MORPweek(t) represents a function of the change in system performance under normal conditions.
In order to comprehensively compare and analyze the toughness of different systems under different interference events from the overall perspective, a General toughness index (GR) is provided for comprehensively measuring the whole toughness process before and after the system is subjected to external interference, which is defined as follows:
Figure BDA0003631939250000131
the index has the disadvantages that when the system is in the process of an external interference event, the system performance level reaches the lowest value of 0, namely the robustness R of the system is 0, and the calculated comprehensive toughness index value is 0. In this case, the overall toughness of the system is 0. The index does not consider the condition that the system performance is restored again after reaching the minimum value of 0, so if the robustness is 0 under a certain interference event, the comprehensive toughness value of the system under the event is directly considered to be 0, and obviously, the comprehensive toughness value is not reasonable enough. Furthermore, this situation can lead to an inability to analyze and compare comprehensive toughness values between different systems.
On the basis, a new comprehensive toughness index is provided as a non-negative comprehensive toughness index (NGR):
Figure BDA0003631939250000132
the indexes provide important measurement for measuring the toughness of different systems under the same destructive event and the toughness of the same system under the different destructive events, and the operation toughness of the airport flight can be evaluated by acquiring the performance change of the airport flight on the time sequence.
Providing a new airport operation performance toughness index departure rate in an airport flight operation toughness evaluation model in severe weather; secondly, improving by combining with the actual situation of airport flight operation, and providing a non-negative comprehensive toughness index; the airport flight operating performance in normal weather is used as a baseline to reduce errors in calculating performance loss. The model contrasts and researches the change rule of the system under different severe weather conditions and severe weather conditions with different intensities, and provides a method for empirical research and analysis of the operation performance of the airport flights.
2 airport flight operational toughness prediction
Because severe weather events occur frequently, the functions of traffic infrastructure and systems are affected widely destructively, even casualties are caused, and therefore, if the traffic resilience space-time distribution pattern under various severe weather events can be quantitatively observed and estimated based on the performance of a traffic system and meteorological data, the specific resilience loss severity and the overall system restoration time are displayed, and the method is more effective for disaster mitigation and emergency response capability improvement.
Based on the actual situation that the flexibility of the airport flight operation performance is predicted according to the message data, an airport performance prediction model based on LSTM under severe weather is adaptively optimized and built, as shown in FIG. 2.
And (3) solving the problem that the value of the MORP of the airport flight operation performance toughness index is between 0 and 1 by adopting a Sigmoid activation function. The Sigmoid activation function is shown in formula (2.1), and the function image is shown in fig. 3.
Figure BDA0003631939250000141
Specific architecture details of the model for predicting the toughness of the airport flight operation performance in severe weather based on the LSTM are shown in the table 2. Normalizing the values of various factors of the original message data to be between-1 and 1, performing dimensionality reduction on the data by adopting a PCA (principal component analysis) technology, performing deep learning on the MORP (airport performance) value by using the message data by utilizing an LSTM (least squares metric) model, and predicting the result through an FC (fiber channel) layer. And processing the prediction result through a Sigmoid activation function to obtain a final prediction value MORP'.
TABLE 2 airport flight operational toughness prediction model details based on LSTM
Figure BDA0003631939250000142
Figure BDA0003631939250000151
The loss function of the airport performance prediction model under the severe weather based on LSTM adopts the Mean-square error (MSE) of MORP' and the actual value MORP, as shown in a formula (2.2). With the goal of reducing the loss function value, the optimizer selects Adam,
model parameters are continuously optimized through gradient descent, and model accuracy is improved.
loss＝MSE(MORP,MORP') (2.2)
In order to verify the accuracy of the predicted value obtained by the model, the Mean Square Error (MSE) is used as a measurement index for evaluating the difference between the predicted value and the true value to judge the fitting degree of the predicted value and the true value of the model provided by the text, as shown in a formula (2.3).
Figure BDA0003631939250000152
Where y is the MORP true value (label) and y' is the predicted value obtained by the model presented herein for the input data. The Mean Square Error (MSE) value range is [0, + ∞ ], and the closer to 0, the more the predicted value and the true value are, the better the model prediction performance is.
In addition, the average Relative Error (MRE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Symmetric Mean Absolute Percentage Error (SMAPE) are also used to represent the accuracy of the proposed model for airport performance prediction, as shown in equations (2.4), (2.5), (2.6), and (2.7).
Figure BDA0003631939250000153
Figure BDA0003631939250000161
Figure BDA0003631939250000162
Figure BDA0003631939250000163
And k is the total amount of the message data. The values of MRE, RMSE, MAE and SMAPE are in the range of [ [0, + ∞ ], so that the smaller the value is, the closer the value is to 0, which shows that the MORP prediction result in the text is to the true value, and the better the prediction performance of the model is.
3 analysis of excess syndrome
By selecting meteorological message data, meteorological early warning information and flight takeoff data of Beijing capital airport in a time period of 1 month and 1 day 00 hour 00 minute (5088 hours in total) of 2021 year to 8 month and 1 day and 00 hour 00 minute (total), 7 severe weather events including 2 snowfall events (snowfall event of 18 days of 1 month and snowfall event of 25 days of 1 month), 1 sandstorm event (sandstorm event of 14 days of 3 months), 4 raining events (raining event of 1 day of 7 months, 5 days of 7 months, 11 days of 7 months and 26 days of 7 months) are screened out through operations such as data preprocessing, and are respectively represented by event 1, event 2, event 3, event 4, event 5, event 6 and event 7. The weather early warning information of the severe weather events obtained on the Beijing weather station webpage is respectively as follows: a rainstorm blue early warning is issued in 16 minutes in 7 months, 1 day and 50 minutes; 7 months, 5 days and 12 days, 59 distributes a rainstorm orange early warning; a rainstorm orange early warning is distributed at 7 months, 11 days and 20 hours; a rainstorm yellow warning was issued at 21 days 25 of 7 months and 26 days, followed by a rainstorm red warning issued at 40 minutes 23.
3.1 airport flight operational toughness assessment
The indexes are calculated by the selected severe event and toughness measurement index formula, and are shown in table 3.
TABLE 3 airport flight service toughness at severe weather event measurement indexes at various stages
Figure BDA0003631939250000171
And (4) taking the departure rate in normal weather as a baseline, and drawing the change conditions of the departure rate in snowfall, sand storm and thunderstorm events. The dashed broken line in fig. 4 represents the departure rate of the airport in normal weather, and the green line represents the departure rate of the airport in severe weather; the three virtual vertical lines respectively represent the time t when the operation performance of the airport flight begins to declinedAt the time t when the performance reaches the minimumrThe performance is restored to the stable time tns. It can be seen from the figure that different adverse weather events have different degrees of impact on airport flight operations.
For different severe weather events, the comprehensive toughness index values of the GR and the NGR when the severe weather occurs are calculated respectively, and the difference between the operation toughness of the airport flights represented by the GR and the NGR under 7 severe weather events is analyzed in a comparative manner, as shown in fig. 5. The columns in the figure represent the combined toughness index GR and the broken lines represent the non-negative combined toughness index NGR after improvement. As can be seen from the graph, in the severe weather of thunderstorm rain (event 4 to event 7), the airport flight operation performance level is reduced to the lowest value of 0 at a certain time, i.e., the robustness R is 0. Therefore, the GR values of event 4 to event 7 calculated according to the formula (1.12) are all 0, which indicates that the airport flight operation system has no process of "rebounding" under the interference of event 4 to event 7, and this is contrary to the actual situation. In addition, since the GR of events 4 to 7 is 0, the law between these severe weather events and GR cannot be explored. From the change situation of the operation performance of the whole airport flight before and after the occurrence of the event, under the interference from the event 4 to the event 7, the performance of the airport flight operation system is only 0 within a certain time period, then along with the dissipation of severe weather and the coordinated scheduling of management personnel, the system gradually recovers to the original state, namely rebounds, and the GR is directly 0, so that the fact that the NGR index comprehensively estimates the operation toughness of the airport flight is more practical than the original GR is explained.
3.2 airport flight operational toughness prediction
3.2.1 data processing
The problem of large difference of the value dimension exists between different message factors in data normalization processing, as shown in a formula (3.1), wherein x represents a certain message factor.
Figure BDA0003631939250000181
Adopting Principal Component Analysis (PCA) to carry out dimensionality reduction on the data, and setting the result of the normalized message data as X*X is calculated using the equations (3.2) and (3.3)*The covariance matrix C of (2) calculates the principal component cumulative contribution ratio and the sample data of m principal components using the formulas (3.4) and (3.5), respectively.
Figure BDA0003631939250000182
Figure BDA0003631939250000183
Wherein Cov (X, y) is a matrix X*And n is the number of factors per message data.
Figure BDA0003631939250000191
Figure BDA0003631939250000192
Where k is the number of samples.
The message data can be effectively subjected to dimensionality reduction according to the PCA technology, so that the effect of compacting the data or simplifying a model is achieved, and the information of the original data is kept to the maximum extent.
3.2.1 Experimental details
Acquiring METAR messages by an airport automatic weather station every half an hour for 1 month, 1 day and 00 in 2021 year: 7/month 31/23 from 00 to 2021: 30. the method removes 3 factors of time, message type and airport code, and selects 39 factors as the message data of every half hour as follows: wind speed (m/s), visibility, shallow, flabby, low blow, blowing, paroxysmal, thunderstorm, overcooling, downy rain, snow, rice snow, ice needles, ice particles, hail, aragonite, fog, smoke, volcanic ash, floating and sinking, sand, haze, sand rolls, squall, tornado, sandstorm, dust storm, cloud cover 1, cloud cover height 1, cloud cover 2, cloud cover height 2, cloud cover 3, cloud cover height 3, cloud cover 3, temperature and dew point temperature are 39 message data serving as factors of each sample, 10176 samples are collected in total and input as a model. On the basis of message data quantization, 3 experimental settings of time sequence step length are adopted: 6 hours, 12 hours, 24 hours. 70% of the data was used for training, 20% for testing, and 10% for validation. And (3) optimizing the model by adopting a formula (2.8), and storing the model parameters which show the optimal fitting degree (MSE minimum) on the verification set in the model training process. Specific data set partitioning is shown in table 4. The experimental parameter settings are shown in table 5. The experimental environment is shown in table 6.
TABLE 4 Experimental data partitioning
Figure BDA0003631939250000201
Table 5 experimental parameter settings
Figure BDA0003631939250000202
TABLE 6 Experimental Environment
Figure BDA0003631939250000203
The principal component eigenvalues, contribution rates and cumulative contribution rates of the 39 factors obtained by the formulas (3.1-3.5) are shown in table 7:
TABLE 7 principal component eigenvalues, contribution rates and cumulative contribution rates of the respective factors
Figure BDA0003631939250000204
Figure BDA0003631939250000211
3.2.3 Experimental results and analysis
And training, verifying and testing the airport performance prediction model based on the LSTM in severe weather by adopting the processed data. The loss function loss value of the model training process is shown in fig. 6.
As can be seen from fig. 6, the fluctuation of the loss value at 24 hours was large, the loss decrease at 12 hours was slow, and the loss decrease at 6 hours was gentle, but there was still a little fluctuation. Overall, the general trend of the loss value at 3 times is gradually reduced with the increase of the training times, and finally, the loss value is reduced to the range of the same order of magnitude. The trend of decreasing loss values indicates that the prediction model of airport performance is continuously optimized in the severe weather based on LSTM.
In order to evaluate the prediction accuracy of the prediction model of airport performance in the severe weather based on LSTM, the mean square error MSE at different time sequence step lengths is calculated, as shown in table 8.
TABLE 8 MSE results at different step sizes
Figure BDA0003631939250000212
The Mean Square Error (MSE) is a measure for reflecting the difference degree between the predicted value and the actual value, and if the predicted value is closer to the actual value, the MSE is closer to 0 if the prediction capability of the model is stronger. Generally speaking, the model trained by the time sequence step of 6 hours is better than the models trained at other times, the best effect is achieved on the training set and the testing set, and the stability is better.
The average relative error (MRE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Symmetric Mean Absolute Percentage Error (SMAPE) index results were calculated and are shown in table 9.
TABLE 9 test set metric results at different step lengths
Figure BDA0003631939250000221
Table 9 is a table that evaluates the predictive performance of the model presented herein in terms of error analysis of predicted values versus actual values. From the error results of the test set with different time-series step sizes, all error index values under the condition of 6 hours are less than 0.22, which indicates that the similarity between the predicted value and the true value is higher.
In order to reasonably assess the effectiveness of the problems and corresponding improvements proposed by the present invention, ablation experiments were conducted with the results shown in table 10. An LSTM (hidden size 128 and num layers 2) is used as a basic model (Baseline), and then relevant ablation experiments are carried out according to whether data normalization processing, data PCA (principal component analysis) dimensionality reduction and Sigmoid activation functions are used. The PCA dimension reduction technique for data includes data normalization, so that the data normalization is inevitably used when PCA is used.
TABLE 10 ablation experiment
Figure BDA0003631939250000222
As can be seen from table 10, the normalization of the data has a significant effect on the performance of the model. The model performance is improved again by adding a Sigmoid function (Baseline5) on the basis of the data normalization, and the error value calculated on the test set is reduced again. The performance expressed by the model (Our) used by the three is optimal, which proves the effectiveness of the proposed method for solving the problems of data unit and dimension, overlong and redundant information, MORP value range between 0 and 1 and the like.
The method predicts the performance condition of the airport, measures by adopting a quantitative evaluation method of various error analyses, compares and analyzes error analysis indexes under different time strategies, can evaluate the optimal performance of a prediction model under message data, and predicts the operation toughness of the airport flight according to meteorological factors, thereby lightening disasters and improving the emergency response capability.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (8)

1. A method for evaluating and predicting the operation toughness of airport flights in severe weather is characterized by comprising the following steps: comprises the following steps
Step 1: constructing an airport flight operation toughness evaluation model, and evaluating the airport flight operation toughness through different measurement indexes;
and 2, step: and constructing an airport flight operation toughness prediction model based on the LSTM, and predicting the performance of the airport in severe weather.
2. The method for assessing and predicting the operation toughness of the airport flights in severe weather according to claim 1, wherein the method comprises the following steps: in the step 1, the Departure Rate is used as an evaluation index of an airport flight operation toughness model in severe weather, and the Departure Rate (Departure Rate) is defined as follows:
Figure FDA0003631939240000011
wherein:
Figure FDA0003631939240000012
is [ T ]1,T2]Total number of scheduled departed flights in the time period;
Figure FDA0003631939240000013
representing planned takeoff time at [ T1,T2]Flight number with simultaneous departure delay less than 30 minutes in a time period;
Figure FDA0003631939240000014
is represented at T1Delayed before time to [ T ]1,T2]Number of flights leaving the field in the time period.
3. The method for assessing and predicting the operation toughness of the airport flights in severe weather according to claim 1, wherein the method comprises the following steps: in the step 1, measuring the performance of the evaluation index in different stages by adopting different toughness measurement indexes in each stage, wherein the measurement indexes comprise response time, failure rate, robustness, recovery time, recovery rate, recovery capability, performance loss and average performance loss.
4. The method for assessing and predicting the operation toughness of the airport flights in severe weather according to claim 3, wherein the method comprises the following steps: wherein, each measurement index is as follows:
response Time (Response Time, RST): from the start of the interference event to the time t at which the system performance begins to declined'＜t＜tdIndicating the capability of the system to resist external interference;
RST＝td-td'
wherein, td' is the beginning of bad weather, tdThe time when the system is interfered by the outside world and the performance begins to decline;
destruction Time (DSS): time period t from initial decline of system performance to lowest performanced＜t＜trIndicating the length of the system performance degradation time;
DSS＝tr-td
wherein, trThe time when the system is interfered by the outside and the performance is reduced to the lowest value;
rate of destruction (RAPI)DP): in the destruction phase td＜t＜trRepresenting the speed at which the system degrades from initial performance to minimum performance;
Figure FDA0003631939240000021
wherein, trThe time when the system is interfered by the outside and the performance is reduced to the lowest value; t is tdThe time when the system is interfered by the outside world and the performance begins to decline; MORP (t)d) Is the system performance at the moment when performance begins to decline, MORP (t)r) The lowest value of the system performance under the external interference is represented;
robustness (Robustness, R): the system still has the capability of well maintaining the stability under the condition of being interfered by an external event;
R＝min{MORP(t)}(td＜t＜tns)
wherein MORP (t) represents a discrete function over time, tdIndicating the time at which the performance of the system begins to decline due to disturbances, tnsThe time when the system recovers to a new stable stage is shown, R represents the maximum value of the system performance in the time period, and the maximum influence on the system under the disturbance of an external event can be measured;
recovery Time (RCT): time period t from the moment of lowest system performance to the time when the system is restored to a new steady stater＜t＜tnsIndicating the length of the system performance recovery time;
RCT＝tns-tr
recovery Rate (RAPI)RP): in a recovery phase tr＜t＜tnsRepresenting the speed at which the system recovers from minimum performance to performance in a new stable phase;
Figure FDA0003631939240000031
wherein MORP (t) represents a time-varying discrete function of system performance in the event of a disturbance, trFor the moment when the system is disturbed by the outside world and the performance is reduced to the minimum, tnsIndicating the moment at which the system recovers to a new stable phase, MORP (t)ns) Is the value at which the performance returns to the plateau, MORP (t)r) The lowest value of the system performance under the external interference is represented;
recovery Ability (RA): is represented by t ≧ tnsThe system reaches a new stable stage;
Figure FDA0003631939240000041
wherein MORP (t) represents a discrete function of system performance over time in the event of a single disturbance, MORP (t)ns) Is the value at which the performance returns to the plateau, MORP (t)r) Represents the lowest value of the system performance under external interference, MORP (t)d) Is the system performance at the time the performance begins to decline;
loss of Performance (Loss of Performance, LOP): represents the total amount of performance degradation over the course of the occurrence of the confounding event;
Figure FDA0003631939240000042
average Performance Loss (Time Averaged Performance Loss, TAPL): represents the overall performance loss of the system in the damage phase and the recovery phase;
Figure FDA0003631939240000043
wherein MORP (t) represents a discrete function of system performance over time at a single interference event; t is tdRepresents the time when the system performance begins to decline; t is tnsThe moment when the system is restored to the new stable stage; MORPweek(t) represents a function of the change in system performance under normal conditions.
5. The method for assessing and predicting the operation toughness of the airport flights in severe weather according to claim 4, wherein the method comprises the following steps: the step 1 further includes establishing a non-negative comprehensive toughness index to measure the toughness of different systems under the same destructive event and the toughness of the same system under the different destructive events, and evaluating the operation toughness of airport flights by obtaining the time series of the flight performance change of the airport, wherein the non-negative comprehensive toughness index (NGR):
Figure FDA0003631939240000051
6. the method for assessing and predicting the operation toughness of the airport flights in severe weather according to claim 1, wherein the method comprises the following steps: in the step 2, constructing the airport flight operation toughness prediction model architecture based on the LSTM comprises the steps of adopting a PCA technology to carry out dimensionality reduction processing on data, utilizing the LSTM model, carrying out deep learning on an airport performance MORP value by using message data, predicting a result through an FC layer, and processing the predicted result through a Sigmoid activation function to obtain a final predicted value MORP'.
7. The method for assessing and predicting the operation toughness of the airport flights in severe weather according to claim 6, wherein the method comprises the following steps: constructing a loss function of the airport flight operation toughness prediction model based on LSTM, adopting the Mean-square error (MSE) of MORP' and the true value MORP,
loss＝MSE(MORP,MORP')。
8. the method for assessing and predicting the operation toughness of the airport flight in severe weather according to claim 7, wherein the method comprises the following steps: the Mean Square Error (MSE) is used as a measurement index for evaluating the difference between the predicted value and the true value, the fitting degree of the predicted value and the true value of the prediction model is judged,
Figure FDA0003631939240000061
wherein y is the MORP true value (label), y' is the predicted value obtained by the prediction model on the input data, the Mean Square Error (MSE) value range is [0, + ∞ ], and the closer to 0 indicates that the predicted value is closer to the true value, and the better the model prediction performance is.
CN202210492086.9A 2022-05-07 2022-05-07 Airport flight operation toughness assessment and prediction method in severe weather Pending CN114742458A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210492086.9A CN114742458A (en) 2022-05-07 2022-05-07 Airport flight operation toughness assessment and prediction method in severe weather

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210492086.9A CN114742458A (en) 2022-05-07 2022-05-07 Airport flight operation toughness assessment and prediction method in severe weather

Publications (1)

Publication Number Publication Date
CN114742458A true CN114742458A (en) 2022-07-12

Family

ID=82285606

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210492086.9A Pending CN114742458A (en) 2022-05-07 2022-05-07 Airport flight operation toughness assessment and prediction method in severe weather

Country Status (1)

Country Link
CN (1) CN114742458A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117422320A (en) * 2023-12-19 2024-01-19 长安大学 Method for extracting influence factors of weather on flight toughness operation

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117422320A (en) * 2023-12-19 2024-01-19 长安大学 Method for extracting influence factors of weather on flight toughness operation
CN117422320B (en) * 2023-12-19 2024-03-05 长安大学 Method for extracting influence factors of weather on flight toughness operation

Similar Documents

Publication Publication Date Title
Wang et al. Modeling weather impact on ground delay programs
Li et al. Flight delay prediction from spatial and temporal perspective
Reitmann et al. Advanced quantification of weather impact on air traffic management
CN111160612A (en) Off-site flight delay analysis and prediction method based on weather influence
CN114742458A (en) Airport flight operation toughness assessment and prediction method in severe weather
CN110796315B (en) Departure flight delay prediction method based on aging information and deep learning
CN112132366A (en) Prediction system for flight clearance rate
Schultz et al. Classification of weather impacts on airport operations
Li et al. Graph signal processing techniques for analyzing aviation disruptions
Lui et al. Weather impact quantification on airport arrival on-time performance through a Bayesian statistics modeling approach
Wilkinson et al. Consequence forecasting: A rational framework for predicting the consequences of approaching storms
CN112182059B (en) High-order analysis method for flight delay characteristics
CN117674119A (en) Power grid operation risk assessment method, device, computer equipment and storage medium
Chandramouleeswaran et al. Machine learning prediction of airport delays in the US air transportation network
Soldatenko et al. Managing climate risks associated with socio-economic development of the Russian Arctic
Yang et al. Departure delay prediction and analysis based on node sequence data of ground support services for transit flights
Zhu et al. Quantifying grid resilience against extreme weather using large-scale customer power outage data
Yiu et al. Flight delay predictions and the study of its causal factors using machine learning algorithms
Salamanis et al. Evaluating the effect of time series segmentation on STARIMA-based traffic prediction model
Priyanka Prediction of airline delays using K-nearest neighbor algorithm
Dai et al. Having a Bad Day? Predicting High Delay Days in the National Airspace System
Ding et al. Economic impact of a lightning strike–induced outage of air traffic control tower: case study of Baltimore–Washington International Airport
Jiang et al. A multi-index prediction method for flight delay based on long short-term memory network model
Larsen et al. A non-parametric discrete choice model for airport acceptance rate prediction
CN111090793A (en) Airport day delay characteristic classification method based on average delay time sequence

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination