CN113269336B

CN113269336B - Flight event cause and effect detection method and device, electronic equipment and readable storage medium

Info

Publication number: CN113269336B
Application number: CN202110810842.3A
Authority: CN
Inventors: 夏欢; 罗谦; 田海涛; 文涛; 陈肇欣; 薛方冉; 毛宏黎; 汤孝川; 成翔; 刘畅; 党婉丽
Original assignee: Second Research Institute of CAAC
Current assignee: Second Research Institute of CAAC
Priority date: 2021-07-19
Filing date: 2021-07-19
Publication date: 2021-10-29
Anticipated expiration: 2041-07-19
Also published as: CN113269336A

Abstract

The application provides a flight event cause and effect detection method, a flight event cause and effect detection device, electronic equipment and a readable storage medium. The method comprises the following steps: acquiring a first data set of the flight to be detected from the historical flight data set, wherein the first data set comprises time information of a plurality of links of the flight to be detected; preprocessing the first data set to obtain a second data set, wherein the second data set comprises delay time of each link in the plurality of links obtained through preprocessing; and carrying out independence detection on the delay time of the first target link and the second target link in the second data set by a preset independence detection strategy to obtain the causal relationship between the first target link and the second target link. The second data set is subjected to independence detection through a preset independence detection strategy, manual analysis is replaced, and therefore the causal relationship between two target links can be obtained quickly, and the causal detection efficiency is improved.

Description

Flight event cause and effect detection method and device, electronic equipment and readable storage medium

Technical Field

The application relates to the field of computer data processing, in particular to a flight event cause and effect detection method, a flight event cause and effect detection device, electronic equipment and a readable storage medium.

Background

Under the existing flight ground support system, the ground support system comprises a plurality of links such as passenger getting-off, cleaning and refueling. And taking the wheel gear withdrawing as the last link of the ground support system, judging whether the flight is delayed or not, wherein the final time is determined by the duration of each guarantee link from the wheel gear adding to the wheel gear withdrawing. Problems in the process of each guarantee link may cause flight delay. Currently, the reason for the delay of the historical flights is generally analyzed manually, and the efficiency is low.

Disclosure of Invention

An object of the embodiments of the present application is to provide a method, an apparatus, an electronic device, and a readable storage medium for causal detection of a flight event, which can improve the efficiency of causal analysis of the flight event.

In order to achieve the above object, embodiments of the present application are implemented as follows:

in a first aspect, an embodiment of the present application provides a method for causal detection of a flight event, where the method includes:

acquiring a first data set of a flight to be detected from a historical flight data set, wherein the first data set comprises time information of a plurality of links of the flight to be detected;

preprocessing the first data set to obtain a second data set, wherein the second data set comprises delay duration of each link in the plurality of links obtained through preprocessing;

and carrying out independence detection on the delay time of a first target link and a second target link in the second data set by a preset independence detection strategy to obtain the causal relationship between the first target link and the second target link, wherein the first target link and the second target link are any two different links in the multiple links.

In the above embodiment, the second data set is obtained by preprocessing the first data set to obtain the delay duration of each link, so as to reduce interference data, and then the second data set is subjected to independence detection by using a preset independence detection strategy to replace manual analysis, so that the causal relationship between two target links can be quickly obtained, and the efficiency of causal detection is improved.

With reference to the first aspect, in some optional embodiments, the method further comprises:

traversing, by the preset independence detection strategy, independence detection on delay duration of two target links in the second data set to obtain a causal relationship between any two links in the multiple links, where the two target links are not the first target link and the second target link at the same time in the multiple links;

and creating a causal network structure of the plurality of links based on the causal relationship of any two links in the plurality of links.

In the above embodiment, causal detection is performed on any two links of the multiple links, so that a causal network structure between the multiple links can be obtained, and calculation of a causal effect based on the causal network structure is facilitated.

in the causal network structure, a causal effect between the first target link and the second target link is determined according to a third target link and a fourth target link corresponding to the first target link and the second target link through a preset causal effect calculation strategy, wherein the third target link is a link in the causal network structure as an independent variable of the first target link and as a dependent variable of the second target link, the fourth target link is a link in the causal network structure as a dependent variable of the first target link and the second target link, and the preset causal effect calculation strategy includes any one of a front door criterion, a back door criterion, and a tool variable.

In the above embodiment, by calculating the causal effect between two target links, it is convenient for the user to check the causal strength between the two target links and optimize the links of each event of the flight.

With reference to the first aspect, in some optional embodiments, preprocessing the first data set to obtain a second data set includes:

when the time information of partial links is lost in the first data set, determining the time information of the lost partial links based on the average duration corresponding to the lost partial links in a pre-stored preset corresponding table;

adding the time information of the partial links into the first data set to obtain an updated first data set;

and determining the delay time corresponding to each link based on the plan time information and the actual time information in the time information of each link in the updated first data set.

In the above embodiment, when the data in the first data set is not complete, the missing data is supplemented, which is beneficial to improve the accuracy and reliability of causal detection.

With reference to the first aspect, in some optional embodiments, before determining a delay time corresponding to each link based on the planned time information and the actual time information in the time information of each link in the updated first data set, the method further includes:

based on other data sets of the flight to be detected when the flight to be detected is different from the current shift, the preset corresponding table corresponding to the flight to be detected is created, the other data sets comprise time information of each link of the flight to be detected in the shift corresponding to other dates, the preset corresponding table comprises average time information corresponding to the link, and the other dates are dates different from the dates of the flight to be detected.

In the above embodiment, the creation of the preset mapping table is beneficial to completing the time information of the missing link, and improves the accuracy and reliability of causal detection.

and marking links with delay time length larger than or equal to second preset time length in the causal network structure and sending prompt information.

With reference to the first aspect, in some optional embodiments, the preset independence detection policy includes a Nataf transformation policy and a preset formula, and the independent detection of the delay time of the first target link and the delay time of the second target link in the second data set through the preset independence detection policy to obtain the causal relationship between the first target link and the second target link includes:

determining a joint probability density function of random vectors in the second data set according to a Nataf conversion strategy and a preset formula, wherein the random vectors comprise delay durations corresponding to the first target link and the second target link;

obtaining a causal relationship between the first target link and the second target link according to a joint probability density function of the first target link and the second target link, wherein the preset formula is as follows:

in the pre-set formula, the formula is defined as,

is a joint probability density function of the random vectors of link X,

is a probability density function of the nth link, n is an integer greater than 1,

is composed of

A probability density function of a standard normal distribution, y denotes a random variable following the standard normal distribution,

based on a mean value of 0, a variance of 1 and a correlation coefficient of

The calculation formula of the n-dimensional standard normal distribution is as follows:

wherein,

refers to the correlation coefficient and T refers to the matrix transposition.

In a second aspect, the present application also provides a flight event cause and effect detection apparatus, comprising:

the flight monitoring system comprises a data acquisition unit, a data acquisition unit and a data processing unit, wherein the data acquisition unit is used for acquiring a first data set of a flight to be detected from a historical flight data set, and the first data set comprises time information of a plurality of links of the flight to be detected;

the preprocessing unit is used for preprocessing the first data set to obtain a second data set, and the second data set comprises the delay duration of each link in the plurality of links obtained through preprocessing;

and the cause and effect determination unit is used for performing independence detection on the delay duration of a first target link and a second target link in the second data set through a preset independence detection strategy to obtain the cause and effect relationship of the first target link and the second target link, wherein the first target link and the second target link are any two links in the links which are different.

In a third aspect, the present application further provides an electronic device, which includes a processor and a memory coupled to each other, wherein the memory stores a computer program, and when the computer program is executed by the processor, the electronic device is caused to perform the method described above.

In a fourth aspect, the present application also provides a computer-readable storage medium having stored thereon a computer program which, when run on a computer, causes the computer to perform the method described above.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.

Fig. 1 is a schematic flow chart of a flight event cause and effect detection method according to an embodiment of the present application.

Fig. 2 is a schematic diagram of a causal network structure of flight delays according to an embodiment of the present application.

Fig. 3 is a schematic view of a link model for causal analysis in flight delay according to an embodiment of the present disclosure.

Fig. 4 is a schematic flow chart of a flight event cause and effect detection apparatus according to an embodiment of the present application.

Icon: 200-flight event cause and effect detection means; 210-a data acquisition unit; 220-a pre-processing unit; 230-cause and effect determination unit.

Detailed Description

The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application. The embodiments described below and the features of the embodiments can be combined with each other without conflict.

The application provides an electronic device, which can automatically detect the reason of flight delay based on the historical flight data of delayed flights and is beneficial to improving the efficiency of flight delay analysis.

The electronic device may include a processing module and a memory module. The memory module stores a computer program that, when executed by the processing module, enables the electronic device to perform the steps of the flight event cause and effect detection method described below.

The electronic device may be, but is not limited to, a personal computer, a server, etc. The processing module, the storage module and other elements in the electronic device are directly or indirectly electrically connected with each other to realize data transmission or interaction. For example, the components may be electrically connected to each other via one or more communication buses or signal lines.

Referring to fig. 1, the present application further provides a method for causal detection of a flight event, which can be applied to the electronic device, and is executed or implemented by the electronic device, where the method includes the following steps:

step S110, acquiring a first data set of a flight to be detected from a historical flight data set, wherein the first data set comprises time information of a plurality of links of the flight to be detected;

step S120, preprocessing the first data set to obtain a second data set, wherein the second data set comprises the delay duration of each link in the plurality of links obtained through preprocessing;

step S130, performing independence detection on delay durations of a first target link and a second target link in the second data set by using a preset independence detection strategy, to obtain a causal relationship between the first target link and the second target link, where the first target link and the second target link are any two different links in the multiple links.

The individual steps of the process are explained in detail below, as follows:

in step S110, the electronic device may obtain a first data set of flights to be tested from a server or device storing historical flight data sets. The flight to be detected is the flight with the assigned airplane number at the assigned date, and can be flexibly determined according to the actual condition. The ground support process of a single flight includes, but is not limited to, wheel gear loading, oil, passenger stairs, inspection of the flight, cabin door opening, cabin cleaning, food, sewage treatment, garbage treatment, unit boarding, cabin door closing, wheel gear removing and other links, and each link can be provided with a corresponding serial number or serial number for distinguishing. The historical flight data set typically includes historical information for a plurality of flights, including a history of the relevant flights to be tested. The first data set is a set of historical record information corresponding to the number of the airplane of the flight to be tested and the number of the flight. The first data set comprises recorded time information of flights to be tested in a plurality of links. The time information includes, but is not limited to, the start time, the end time, etc. of the individual link.

In step S120, the preprocessing may include processing operations such as detecting and completing missing data, calculating a delay duration corresponding to a single link, and the like, and may be determined flexibly according to actual conditions. For example, step S120 may include:

Understandably, the electronic device may determine whether the data is missing in the first data set based on link types included in data sets of flights to be detected on other dates. When the missing data exists, the missing data can be supplemented based on the average data of the data which is not missing in other dates of the flight to be detected.

For example, in the shift of the flight to be detected on other dates, the time information of the "food navigation" link is included, and in the first data set, the time information of the "food navigation" link does not exist, so that the time information of the "food navigation" link in the first data set is determined to be absent. Then, based on the acquired time information of the flight in the "food navigation" link in other days, the average time information of the "food navigation" link is calculated (for example, the average time of the beginning of the "food navigation" link and the average time of the end of the "food navigation" link are calculated) to be used as the time information of the "food navigation" link missing from the flight to be measured.

In this embodiment, before determining a delay duration corresponding to each link based on planned time information and actual time information in each link time information in the updated first data set, the method further includes:

Understandably, the other data set of the flight to be tested is usually the shift with the same number as the airplane of the flight to be tested, different flight date and the same flight time interval. For example, the flight to be tested is a shift with the aircraft number "XXX" flying at 9:00 am on 1 st; the other data set is a set of data collected for a shift of the aircraft on a day other than 1 month and 1 day at 9:00 am for an aircraft with aircraft number "XXX". For example, the other data set includes data related to an aircraft with aircraft number "XXX" flying at 9:00 AM on days 1 month 3, 1 month 5, etc.

When a preset correspondence table corresponding to the flight to be measured is created, an average value of time information needs to be calculated based on time information of each link in other data sets. When the average value is calculated, abnormal data needs to be filtered, wherein the abnormal data is time information of the dispersion exceeding a preset threshold value in the same link. The preset threshold value can be determined according to actual conditions. For example, if it is assumed that the start time of the "aviation oil" link of the same-numbered aircraft in the shift on different dates is usually 8:00 a.m., and the start time of the "aviation oil" link in the shift on a certain date is 10:00 a.m., and exceeds a preset threshold, the time information that the start time of the "aviation oil" link is 10:00 a.m. is determined to be abnormal data, and when the average start time of the "aviation oil" link is calculated, the abnormal data is filtered, and the average start time of the "aviation oil" link is not calculated by using the abnormal data. In this way, the reliability of the calculated average time can be improved.

In this embodiment, by calculating the average time information of each link of the airplanes with the same number in other shifts, a preset correspondence table corresponding to the airplane number (or the flight to be tested) can be created.

When the first data set lacks time information of partial links, filling and completion of the link time can be carried out by using the ending time of the preorder link, the starting time of the subsequent link and the average time length between the preorder link and the link in the non-missing flight data, so that the delay time length of each link is calculated according to the filling and completion. Illustratively, the specific operations of data padding are (b denotes a link start, e denotes a link end):

starting time of aviation oil link (t)_fuelb) = gear shifting moment (t)_ongear) + average duration (t) between shift and oil-in-flight links for non-missing data_ongear - t_fuelb)；

End of the voyage oil link (t)_fuele) = time of starting of aviation and fuel supply link (t)_fuelb) + average duration of flight segment (t) for non-missing data flights_fuele - t_fuelb) ；

Passenger ladder car in-place time (t)_stairsb) = gear shifting moment (t)_ongear) + average duration of time (t) between upper wheel block of non-missing data flight and passenger elevator car in-place_ongear - t_stairsb)；

Passenger ladder vehicle leaving ending time (t)_stairse) = passenger elevator car in-place time (t)_stairsb) + average duration of passenger stairs (t) for non-missing data flights_stairse - t_stairsb)；

Starting time (t) of engineering inspection link_maintenanceb) = gear shifting moment (t)_ongear) + average duration (t) of starting of shift and engineering inspection link of non-missing data flight_maintenanceb - t_ongear)；

The end time (t) of the engineering inspection link_maintenancee) = engineering inspection link start time (t)_maintenanceb) + mean duration of flight inspection link (t) for non-missing data flights_maintenancee - t_maintenanceb)；

Moment of opening door (t)_opencabin) = departure time (t) of passenger elevator car_stairse) + mean duration between departure and departure of a passenger car from a non-missing data flight (t)_stairse - t_opencabin)；

Cabin cleaning segment start time (t)_cleanb) = moment of opening door (t)_opencabin) + average duration (t) between the start of a flight door and the start of cabin cleaning for non-missing data_cleanb - t_opencabin)；

Cabin cleaning segment end time (t)_cleane) = cabin cleaning session start time (t)_cleanb) + mean duration of cabin cleaning segment (t) for non-missing data flights_cleane - t_cleanb)；

Beginning of the food navigation link (t)_foodb) = moment of opening door (t)_opencabin) + average duration (t) between hatch of non-missing data flight and start of flight_foodb - t_opencabin)；

End of the food segment (t)_foode) = flight food link start moment (t)_foodb) + flight without missing dataAverage duration of food link (t)_foode - t_foodb)；

Beginning of the Sewage treatment Process (t)_sewageb) = moment of opening door (t)_opencabin) + average duration (t) between hatch of non-missing data flight and start of sewage treatment_sewageb - t_opencabin)；

End of the Sewage treatment Link (t)_sewagee) = start time of sewage treatment link (t)_sewageb) + average duration (t) of sewage treatment link for non-missing data flights_sewagee - t_sewageb)；

Beginning of the refuse treatment process (t)_garbageb) = moment of opening door (t)_opencabin) + average duration (t) between the opening of a door to a non-missing data flight and the start of the garbage disposal_garbageb - t_opencabin)；

End of the refuse treatment step (t)_garbagee) = refuse handling link start time (t)_garbageb) + average duration of the garbage disposal link (t) for non-missing data flights_garbagee - t_garbageb)；

Starting time of unit boarding link (t)_boardingb) = cabin cleaning segment end time (t)_cleane) + average duration (t) between the end of flight cabin cleaning procedure and the start of unit boarding_boardingb - t_cleane)；

End of unit boarding link (t)_boardinge) = aircraft boarding link start time (t)_boardingb) + average duration of unit boarding link (t) for non-missing data flights_boardinge - t_boardingb)；

Moment of closing the door (t)_closecabin) = aircraft boarding link end time (t)_boardinge) + average duration (t) between the end of the unit boarding link and the closing of the doors of the flight without missing data_closecabin - t_boardinge)；

Moment of gear shift withdrawing (t)_regear) = closing the door time (t)_closecabin) + average duration (t) between closing door and removing gear of non-missing data flight_regear - t_closecabin)。

The non-missing data refers to data which is not missing in other shifts but is missing in the flight to be detected, wherein the data is the same number as the flight to be detected.

In step S130, a preset independence detection strategy is used to perform causal detection on any two target links in the second data set to obtain a causal relationship between the two target links, where the causal relationship is whether there is a causal association between delay times of the two target links. The pre-defined independence detection strategy may include, but is not limited to, a Nataf transformation strategy and a pre-defined formula. The Nataf transformation strategy is a strategy which can effectively normalize input variables, and can solve the problem of conditional independent test by using unconditional independent test.

For example, step S130 may include:

（1）

in the preset formula, f_X(x) A joint probability density function of random vectors for a link X, X_nRefers to the sample value, f, of any guaranteed link delay duration_n(x_n) As a function of the probability density of the nth element, n being an integer greater than 1, ∅ (y)_n) Is y_nA probability density function of a standard normal distribution, y denotes a random variable following the standard normal distribution,

based on a mean of 0, a variance of 1 and a correlation coefficientIs composed of

（2）

wherein,

refers to the correlation coefficient, T refers to the matrix transposition,

by empirical estimation of the components of

And (5) realizing. Wherein F is more than or equal to 1, and F is a correlation coefficient

And a relation function of the edge probability density function. In that

In the method, i and j refer to the serial numbers of any two different links, and exp operation refers to an exponential function with a natural constant e as a base;

finger-shaped

The determinant, the way of calculation, of the matrix is well known to those skilled in the art.

In this embodiment, the method may further include:

Understandably, the processing module of the electronic device may perform the independence detection on the delay durations of any two target links in the second data set, so as to obtain a causal network structure among multiple links, as shown in fig. 2, which facilitates calculation of a causal effect based on the causal network structure. The causal network structure shown in fig. 2 includes links such as wheel guard, cabin door opening, cabin cleaning, aviation oil, passenger elevator, machine boarding, food shipping, inspection of air service, cabin door closing, sewage treatment, garbage treatment, and wheel guard removing.

In other embodiments, the causal network structure obtained based on the data sets of different flights may be different from the causal network structure shown in fig. 2.

In the causal network structure shown in fig. 2, two links connected by an arrow line are links having causal relationship, and two links connected by a bidirectional arrow line affect processing time (or delay time) of the links mutually. In the two links connected by the one-way arrow, the link indicated by the arrow is the affected link, for example, in the link of "opening the door" and the link of "cleaning the cabin," the arrow of the line indicates the link of "cleaning the cabin," which indicates that the processing duration of the link of "opening the door" affects the processing duration (or delay duration) of the link of "cleaning the cabin," and the link of "cleaning the cabin does not affect the processing duration (or delay duration) of the link of" opening the door.

Referring to fig. 3, in order to facilitate understanding of the processing flow of the preset independence detection policy, an implementation process of the preset independence detection policy will be described below by way of example, as follows:

assuming that the first target link is cabin cleaning, and X is used for indicating the delay time of the cabin cleaning; the second target link is a planned gear-removing gear, and Y is used for representing the delay time of the planned gear-removing gear; the third target link is machine set boarding, and the delay time of the machine set boarding is represented by Z; the fourth target link is the door opening, the delay time of the door opening is represented by U, and X, Y, Z, U is a random variable understandably. At this time, it is necessary to calculate the influence of the delay of the cabin cleaning link on the planned gear-withdrawal delay, that is, to detect X, Y whether a causal relationship exists, that is, whether the following assumption of formula (3) is satisfied, where formula (3) is:

（3）

in the formula (3), Pr represents a causal relationship between X, Y, and when the value of Pr is 1, the causal relationship exists; when the value of Pr is 0, no causal relationship exists. F refers to the joint distribution corresponding to the delay duration of the corresponding link. x, y and u respectively refer to the delay time of different links. For example, x refers to a cabin cleaning delay time sample value, y refers to a planned wheel gear-withdrawing delay time sample value, and u refers to an opening door delay time sample value.

To simplify the test, the above equation (3) may be changed to F_x,y,u(x,y,u)F_u(u) and F_x,u(x,u)F_y,uAnd (y, u) whether the correlation coefficient between the (y, u) is equal to 0 or not is checked, namely, whether the combined distribution of the delay time of the three links of passenger cabin cleaning, wheel gear removing planning and unit boarding is multiplied by the distribution of the delay time of the link of cabin door opening or not is checked, and whether the combined distribution of the delay time of the passenger cabin cleaning and the delay time of the link of cabin door opening is multiplied by the combined distribution of the delay time of the wheel gear removing planning and the delay time of the link of cabin door opening. Then, an edge density function of a variable (indicating delay duration) of a link is estimated through a Gaussian kernel density function, and then an empirical distribution function is obtained through Gasser-Muller estimation. The Gasser-Muller estimation is a common core estimation method, and the distribution function of the delay time of the passenger cabin cleaning core estimated by the Gasser-Muller is as follows:

（4）

in the formula (4), i refers to the number of samples included in each link, i.e., the number of flights; k refers to a kernel function for x; h refers to a smoothing parameter, and can be flexibly set according to actual conditions. Wherein,

，s_imeans the average value of the ith delay time period sample value and the (i + 1) th delay time period sample value, Y_iThe delay time of the planned gear-removing of each sample is indicated.

Based on the formula (4), the edge probability density function of the variable of each link can be obtained, and then, the independence test is carried out on the variable set by adopting Nataf transformation. Wherein each variable is the delay time of any link. Let K (x) be a Gaussian kernel function selected for the delay period of cabin cleaning and satisfy

That is to say have

Where h is a smoothing parameter, and the choice of h is obtained by Mean Integral Square Error (MISE).

The Nataf transformation strategy is a strategy capable of normalizing input variables (the input variables comprise delay time of each link), and the strategy does not need to know a joint probability density function of random input, but needs to know a correlation coefficient between a marginal probability density function of each random variable and the random variable. Additionally, based on the Nataf transform, the problem of conditional independent testing can be solved using unconditional independent testing. For the Nataf transformation, the correlation coefficient for calculating two random variables needs to be converted into the correlation coefficient in a standard normal space under the condition of a cumulative empirical distribution function. At this time, the electronic device may automatically create conditions, i.e., set random variable vectors consisting of all security link delay durations

Wherein the random variable M₁Probability density function of

And cumulative distribution function

Are known. Introducing standard normal random vector

The joint probability density function of the random vector M can be derived by using the implicit function derivation rule, which is the above formula (1).

Based on the formula (1), a joint probability density function of the first target link X and the second target link Y can be calculated. If Z is not contained in a separate set of X and Y, then the path from X to Y is blocked by Z, i.e., X and Y are independent given Z and no causal relationship exists. A causal relationship exists if Z is contained in a separate set of X and Y, i.e. X and Y are not independent given Z.

If the common neighbor link W of X and Y is not included in the split set, it can be determined that the link X, W, Y is a V-shaped structure (X → W ← Y), where the link W is a security link other than X, Y, Z, U. Based on equation (1), the directions of all edges in the network can be determined in turn, and new V-shaped structures and directed loops are avoided in the process. Therefore, the causal network structure of all links of the flight to be detected can be obtained.

In this embodiment, the method may further include: in the causal network structure, a causal effect between the first target link and the second target link is determined according to a third target link and a fourth target link corresponding to the first target link and the second target link through a preset causal effect calculation strategy, wherein the third target link is a link in the causal network structure as an independent variable of the first target link and as a dependent variable of the second target link, the fourth target link is a link in the causal network structure as a dependent variable of the first target link and the second target link, and the preset causal effect calculation strategy includes any one of a front door criterion, a back door criterion, and a tool variable.

Exemplarily, assuming that the resulting causal relationship is Y =2X, it means that the causal effect of X on Y is 2, i.e. every 1 increase in X, Y increases by 2 each time X increases by 1.

To facilitate an understanding of the causal effect calculation process, the following is illustrated in connection with fig. 3:

for example, first, assuming that a link model having a local structure diagram as shown in FIG. 3 exists in the causal network structure, when both U and Z can be observed (observable means that there is U, Z delay time and the causal relationship shown in FIG. 3 is satisfied; unobservable means that there is no assumed variable satisfying the relationship of FIG. 3), the following relationship exists:

（5）

assuming that (U, X, Z, Y) obeys a multivariate normal distribution N (0, Σ), where the covariance matrix Σ is a symmetric matrix, a matrix made up of the covariance between (U, X, Z, Y) two by two, P (Z | X) represents the conditional distribution of Z given X, there is a matrix as shown in equation (6) as follows:

（6）

in equation (6), the parameters a are as follows:

A11=1

A12=a_xu

A13=a_xu·a_zx

A14=a_yu+a_xu·a_zx·a_yz

A21=a_xu

A22=1+a² _xu

A23=a_zx(1+ a² _xu)

A24=a_zx·a_yz (1+a² _xu)+a_xu·a_yu

A31= a_xu·a_zx

A32= a_zx(1+ a² _xu)

A33= 1+a² _zx(1+ a² _xu)

A34= a_yz ·[1+ a² _zx(1+ a² _xu)]+ a_xu·a_zx·a_yu

A41= a_yu +a_xu·a_zx·a_yz

A42= a_zx·a_yz(1+ a² _xu)+ a_xu·a_yu

A43= a_yz[1+ a² _zx(1+ a² _xu)]+ a_xu·a_zx·a_yu

A44=1+a² _yz(1+ a² _zx)+ (a_yu+a_xu·a_zx·a_yz)²

in addition, in equation (6), Σ is a symmetric matrix. a refers to a correlation coefficient, and can be determined according to actual conditions.

Based on equation (6), there is E (y | do (x)) = a_zx·a_yzX, causal effect is τ_yx=a_zx·a_yz. That is, x has a causal strength of a to y_zx·a_yzAnd do (X) indicates that the condition of intervening the link X to be a fixed value and E to be y is expected.

If U is not observable and Z is observable, then to calculate from the front door criteria

First, calculate

Wherein X' is a member of the group consisting ofA fixed value. Because of the fact that

Wherein the symbol ″) represents independently,

all figures containing links from X are shown, so there are

. Then calculate

Due to the fact that

And is

，

Representing all links containing a point X, and thus there is a

. And because of

Therefore, it is

. Through the calculation, the method can obtain

。

Then, P (y-do (x)) is calculated. Because of the fact that

Therefore, it is

. And because of

Therefore, it is

Thus having

In summary, the causal effect is now identified as

Wherein a is_zxExpressing the regression coefficient, beta, of Z in the regression model of Z versus X_yz.xExpressing the regression coefficient, σ, of X in the regression model of Y versus X, Z_yz.xDenotes the partial covariance, σ, of X, Y with respect to Z_zxRepresents the covariance between Z and X.

If Z cannot be observed and U can be observed, the above similar method is adopted to obtain the following door criteria:

then the causal effect is

. The causal effect identification method based on the tool variables is characterized in that a causal graph is rewritten by adopting a structural equation, and the rewriting is as follows:

（7）

wherein a is_xu，a_zx，a_yz，a_yuAre not 0, ϵ_iIs a normal disturbance term independent of each other, then (U, T, X, Y, Z) follows a multivariate normal distribution N (0, Σ'), as shown in equation (8):

（8）

in equation (8), the parameters a are as follows:

A11=1

A12=0

A13=a_xu

A14= a_xu·a_zx

A15= a_yu+a_xu·a_zx·a_yz

A21=0

A22=1

A23= a_xt

A24= a_xt·a_zx

A25= a_xt·a_zx·a_yz

A31= a_xu

A32= a_xt

A33=1+ a² _xu+ a² _xt

A34= a_zx·σ_xx

A35= a_zx·a_yz·σ_xx+ a_xu·a_yu

A41= a_xu·a_zx

A42= a_xt·a_zx

A43= a_zx·σ_xx

A44=1+ a² _zx·σ_xx

A45= a_yz·σ_zz+ a_xu·a_zx·a_yu

A51= a_yu +a_xu·a_zx·a_yz

A52= a_xt·a_zx·a_yz

A53= a_zx·a_yz·σ_xx+ a_xu·a_yu

A54= a_yz·σ_zz+ a_xu·a_zx·a_yu

A55=1+a² _yu+ a² _yz·σ_zz+2 a_xu·a_yu·a_zx·a_yz

in the formula (8), a denotes a correlation coefficient, and σ denotes a covariance, which can be determined according to actual conditions.

Assuming that Z cannot be observed at this time, but there is a tool variable T that can be observed, the causal effect τ can be identified by the tool variable method using the observed data of (T, X, Y) at this time_yx=β_yt/β_xt. Wherein the tool variables are: if there is a set of variables T for a link such that none of the variables belong to the descendants of X nor Y, and T is related only to X and not to Y, then T is said to be a tool variable related to (X, Y). By calculating the causal effect of any two links, all causal effects between the links in the flight to be detected can be obtained, and the subsequent adjustment of corresponding strategies on the links based on the obtained causal effects is facilitated (for example, links with causal relationships are obtained, and the processing time of the preamble links is reduced).

Based on the design, compared with a manual analysis mode in the prior art, the causal effect identification efficiency is high in the embodiment, flexible subsequent distribution of flight guarantee resources (guarantee resources refer to processing time of each link) is facilitated, and causal detection and guarantee resource distribution have higher flexibility and accuracy.

As an optional implementation, the method may further include:

In this embodiment, the second preset duration generally refers to a duration with a longer delay duration, and may be flexibly determined according to actual situations, for example, the second preset duration may be 1 hour. In the embodiment, the links with longer delay time are marked in a differentiated mode and a prompt is sent, so that an administrator can find the links with longer delay time in time, a user can conveniently find the reason of the flight delay, and each link of subsequent airplane flights in the same shift is optimized, and the occurrence frequency of the flight delay is reduced.

The mode of the differential marking can be determined according to actual conditions, for example, a link with a delay time length greater than or equal to a second preset time length is used as a key link, and in the causal network structure, the line color, thickness, pattern size and the like of the key link can be different from non-key links of the flight to be detected. The prompt message may be, but is not limited to, a text message, a voice message, etc., and is not limited to this.

Referring to fig. 4, an embodiment of the application further provides a flight event cause and effect detection apparatus 200, which can be applied to the electronic device described above for executing the steps of the method. The flight event cause and effect detection device 200 includes at least one software function module which can be stored in a memory module in the form of software or Firmware (Firmware) or solidified in an Operating System (OS) of an electronic device. The processing module is used for executing executable modules stored in the storage module, such as software function modules and computer programs included in the flight event cause and effect detection apparatus 200.

The flight event cause and effect detection apparatus 200 may include a data obtaining unit 210, a preprocessing unit 220, and a cause and effect determination unit 230, and may perform the following operations:

the data acquisition unit 210 is configured to acquire a first data set of a flight to be detected from a historical flight data set, where the first data set includes time information of multiple links of the flight to be detected;

a preprocessing unit 220, configured to preprocess the first data set to obtain a second data set, where the second data set includes a delay duration of each of the plurality of links obtained through the preprocessing;

the cause and effect determining unit 230 is configured to perform independence detection on delay durations of a first target link and a second target link in the second data set through a preset independence detection strategy to obtain a cause and effect relationship between the first target link and the second target link, where the first target link and the second target link are any two links different from each other in the multiple links.

Optionally, the cause and effect determining unit 230 may be further configured to traverse, through the preset independence detection policy, independence detection on delay durations of two target links in the second data set to obtain a cause and effect relationship between any two links in the multiple links, where the two target links are not the first target link and the second target link at the same time in the multiple links; and the network structure creating unit is used for creating a causal network structure of the links based on the causal relationship of any two links in the links.

The flight event cause and effect detection apparatus 200 may further include an effect determination unit, configured to determine, in the cause and effect network structure, a cause and effect between the first target link and the second target link according to a third target link and a fourth target link corresponding to the first target link and the second target link through a preset cause and effect calculation policy, where the third target link is a link in the cause and effect network structure that is used as an independent variable of the first target link and is used as a dependent variable of the second target link, the fourth target link is a link in the cause and effect network structure that is simultaneously used as a dependent variable of the first target link and the second target link, and the preset cause and effect calculation policy includes any one of a front gate criterion, a back gate criterion, and a tool variable.

Optionally, the preprocessing unit 220 may be further configured to:

Optionally, the flight event cause and effect detection apparatus 200 may further include a correspondence table creation unit. Before the preprocessing unit 220 determines the delay duration corresponding to each link based on the planned time information and the actual time information in each link time information in the updated first data set, the correspondence table creating unit is used for creating the preset correspondence table corresponding to the flight to be detected based on other data sets of the flight to be detected when the flight to be detected is different from the current shift, the other data sets comprise the time information of each link in the shift corresponding to the flight to be detected on other dates, the preset correspondence table comprises the average time information corresponding to the link, and the other dates are dates different from the flight to be detected.

Optionally, the flight event cause and effect detection apparatus 200 may further include a prompting unit, configured to mark, in the cause and effect network structure, a link with a delay duration greater than or equal to a second preset duration and send a prompting message.

Optionally, the cause and effect determination unit 230 may be further configured to:

in the pre-set formula, the formula is defined as,

is a joint probability density function of the random vectors of link X,

is composed of

based on a mean value of 0, a variance of 1 and a correlation coefficient of

wherein,

refers to the correlation coefficient and T refers to the matrix transposition.

In this embodiment, the processing module may be an integrated circuit chip having signal processing capability. The processing module may be a general purpose processor. For example, the processor may be a Central Processing Unit (CPU), a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component, and may implement or execute the methods, steps, and logic blocks disclosed in the embodiments of the present Application.

The memory module may be, but is not limited to, a random access memory, a read only memory, a programmable read only memory, an erasable programmable read only memory, an electrically erasable programmable read only memory, and the like. In this embodiment, the storage module may be configured to store historical flight data sets, preset independence detection policies, and the like. Of course, the storage module may also be used to store a program, and the processing module executes the program after receiving the execution instruction.

It should be noted that, as will be clear to those skilled in the art, for convenience and brevity of description, the specific working process of the electronic device described above may refer to the corresponding process of each step in the foregoing method, and will not be described in detail herein.

The embodiment of the application also provides a computer readable storage medium. The computer-readable storage medium has stored therein a computer program which, when run on a computer, causes the computer to perform the method of causal detection of a flight event as described in the above embodiments.

From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by hardware, or by software plus a necessary general hardware platform, and based on such understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a usb disk, a removable hard disk, etc.), and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the method described in the embodiments of the present application.

In summary, the present application provides a method, an apparatus, an electronic device and a readable storage medium for causal detection of a flight event. The method comprises the following steps: acquiring a first data set of the flight to be detected from the historical flight data set, wherein the first data set comprises time information of a plurality of links of the flight to be detected; preprocessing the first data set to obtain a second data set, wherein the second data set comprises delay time of each link in the plurality of links obtained through preprocessing; and carrying out independence detection on the delay time of a first target link and a second target link in a second data set by a preset independence detection strategy to obtain the causal relationship between the first target link and the second target link, wherein the first target link and the second target link are any two links different from each other in the multiple links. According to the scheme, the second data set is obtained by preprocessing the first data set, so that the delay time of each link is obtained, interference data are reduced, independence detection is carried out on the second data set through a preset independence detection strategy, manual analysis is replaced, and therefore the causal relationship between two target links can be quickly obtained, and the causal detection efficiency is improved.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus, system, and method may be implemented in other ways. The apparatus, system, and method embodiments described above are illustrative only, as the flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions. In addition, functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for causal detection of a flight event, the method comprising:

performing independence detection on delay duration of a first target link and a second target link in the second data set through a preset independence detection strategy to obtain a causal relationship between the first target link and the second target link, wherein the first target link and the second target link are any two links different from each other in the plurality of links, the preset independence detection strategy comprises a Nataf transformation strategy and a preset formula, and the step of performing independence detection on the delay duration of the first target link and the second target link in the second data set through the preset independence detection strategy to obtain the causal relationship between the first target link and the second target link comprises:

determining a joint probability density function of random vectors in the second data set according to the Nataf transformation strategy and the preset formula, wherein the random vectors comprise delay durations corresponding to the first target link and the second target link;

in the pre-set formula, the formula is defined as,

is a joint probability density function of the random vectors of link X,

the value of the delay time of the nth link is given,

is composed of

refers to the y of the nth link,

based on a mean value of 0, a variance of 1 and a correlation coefficient of

wherein,

refers to the correlation coefficient and T refers to the matrix transposition.

2. The method of claim 1, further comprising:

3. The method of claim 2, further comprising:

4. The method of claim 1, wherein preprocessing the first data set to obtain a second data set comprises:

5. The method of claim 4, wherein before determining the delay period corresponding to each link based on the planned time information and the actual time information in each link time information in the updated first data set, the method further comprises:

6. The method of claim 2, further comprising:

7. A flight event cause and effect detection apparatus, the apparatus comprising:

a cause and effect determining unit, configured to perform independence detection on delay durations of a first target link and a second target link in the second data set through a preset independence detection policy to obtain a cause and effect relationship between the first target link and the second target link, where the first target link and the second target link are any two different links among the multiple links, the preset independence detection policy includes a Nataf transformation policy and a preset formula, and the cause and effect determining unit is further configured to:

in the pre-set formula, the formula is defined as,

is a joint probability density function of the random vectors of link X,

the value of the delay time of the nth link is given,

is composed of

refers to the y of the nth link,

based on a mean value of 0, a variance of 1 and a correlation coefficient of

wherein,

refers to the correlation coefficient and T refers to the matrix transposition.

8. An electronic device, characterized in that the electronic device comprises a processor and a memory coupled to each other, the memory storing a computer program which, when executed by the processor, causes the electronic device to perform the method according to any of claims 1-6.

9. A computer-readable storage medium, in which a computer program is stored which, when run on a computer, causes the computer to carry out the method according to any one of claims 1-6.