Aircraft cluster security situation awareness method in autonomous operation mode

Abstract

The invention discloses an aircraft cluster security situation awareness method in an autonomous operation mode, is oriented to an operation scene of a high-altitude civil aircraft, and belongs to the field of aviation operation security. The method comprises the following steps: firstly, in order to meet the requirement of risk management refinement between aircrafts in an autonomous operation mode, comprehensively considering communication navigation monitoring factors influencing track uncertainty, introducing errors caused by CNS performance into a traditional collision model, optimizing a traditional aircraft edge connection judgment basis, and constructing an aircraft cluster situation network; secondly, designing a cluster risk evaluation index based on the constructed situation network in combination with aircraft cluster space structure representation and traffic flow parameter information, and further dividing a cluster risk mode to form an air traffic risk situation network; and finally, on the basis of a cluster structure in the risk situation network, discussing time sequence distribution, survival characteristics and spatial operation preferences of each risk mode aircraft cluster, and mining spatio-temporal evolution characteristics of the risk situation network. According to the method, the conflict risk between the aircrafts can be finely judged, the spatio-temporal evolution characteristics of the aircraft cluster can be mined, and theoretical support is provided for an autonomous operation mode.

Description

Technical Field

[0001] The present invention relates to the field of civil aviation operation safety, and in particular to a method for sensing the safety situation of an aircraft cluster in an autonomous operation mode. Background Art

[0002] In order to cope with the future operation mode based on four-dimensional trajectory and free route airspace, and adapt to the future aviation environment with high-density flight operations, there is an urgent need for a model that reflects potential flight conflicts under the autonomous operation mode. New technologies and new means are urgently needed to enhance the current air traffic situation awareness capabilities and improve the overall airspace operation safety level.

[0003] At present, the research on air traffic safety situation awareness is mainly divided into two types from the perspective of evaluating the severity of specific conflicts and measuring the stability of the macro system: risk analysis based on traffic situation measurement and risk analysis based on complex network theory. In terms of risk analysis based on traffic situation measurement, the research often measures specific conflict events to characterize the safety level of air traffic situation. However, only describing specific conflict events will make air traffic situation awareness lack globality. In recent years, graph theory and complex network theory have been introduced into the research on air traffic risk situation, providing new research ideas and methods for air traffic operation situation. In terms of risk analysis based on complex network theory, the commonality of such research is that aircraft in the airspace are regarded as network nodes, and conflicts or proximity between aircraft are regarded as edges. The specific difference lies in the edge rules between aircraft and the selected network evaluation indicators. However, this type of research, first of all, is based on the established distance for the formation of aircraft groups. The distance setting is often subjective and does not meet the requirements of fine determination of edges between aircraft in autonomous operation mode; secondly, the evolutionary characteristics of the risk situation network are analyzed, and the survival evolution law of the risk situation network is explored from the time dimension. There is a gap in the research on exploring the complex mechanisms of aircraft flow operation in airspace from a spatial perspective.

[0004] Therefore, it is necessary to provide a safety situation awareness method for aircraft clusters in autonomous operation mode to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a method for safety situation awareness of aircraft clusters under autonomous operation mode, propose a risk situation network model suitable for use in autonomous operation mode, refine the judgment of conflicts between aircraft, and then combine the survival characteristics and the risk situation evolution characteristic analysis method of spatial operation hotspot distribution to explore the evolution mechanism of aircraft clusters from a spatiotemporal perspective, so as to realize the safety situation awareness of aircraft clusters under autonomous operation conditions.

[0006] To achieve the above object, the present invention provides a method for sensing the safety situation of an aircraft cluster in an autonomous operation mode, comprising the following steps:

[0007] Step 1: Analyze the impact mechanism of CNS performance on track uncertainty, introduce the error caused by CNS performance into the traditional collision model, and build a situation network model suitable for use in autonomous operation mode;

[0008] Step 2: Combine the spatial structure representation of aircraft clusters and traffic flow parameter information to design cluster risk assessment indicators, divide cluster risk patterns, and form an air traffic risk situation network;

[0009] Step 3: Considering the significant spatiotemporal characteristics of aircraft cluster behavior, survival characteristic analysis and spatial hotspot analysis methods are used to explore the spatiotemporal evolution characteristics of the risk situation network.

[0010] Step 1 specifically includes:

[0011] Step 1.1: Analyze the operating characteristics of the autonomous operation mode.

[0012] Step 1.2: Based on the known aircraft positioning error caused by RNP, derive the aircraft positioning error caused by RCP and RSP.

[0013] Step 1.3: Based on the error distribution caused by CNS performance and the traditional collision model, a collision probability calculation model suitable for autonomous operation is derived. Aircraft pairs with collision probabilities greater than the safety target level are identified as conflicting aircraft pairs. Edges are established based on the conflict relationship to construct a situation network.

[0014] In step 1.1, the autonomous operation mode has the following characteristics: the flight pilot autonomously selects the optimal trajectory based on the actual use of the airspace. The risks between aircraft are no longer determined based on the traditional control-dominated mode and the fixed distance between aircraft. Instead, a detailed risk assessment is conducted based on the performance of the aircraft and the actual operating conditions, thereby maximizing the growth of airspace traffic while ensuring safety.

[0015] Step 2 specifically includes:

[0016] Step 2.1: Design cluster risk assessment indicators based on the spatial structure representation of aircraft clusters and traffic flow parameter information.

[0017] Step 2.2: Calculate the risk vector of each cluster and divide the risk pattern using K-means clustering algorithm.

[0018] Step 3 specifically includes:

[0019] Step 3.1: Analyze the survival characteristics of aircraft clusters in each risk mode from the time dimension.

[0020] Step 3.2: Understand the operational preferences of aircraft clusters in each risk mode from a spatial dimension.

[0021] The beneficial effects of the present invention are that the present invention can realize the refined judgment of the conflict risk between aircraft, and mine the spatiotemporal evolution characteristics of aircraft clusters, providing theoretical support for the autonomous operation mode. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 : Flowchart of a safety situation awareness method for aircraft swarms in autonomous operation mode.

[0023] Figure 2 : Schematic diagram of a collision model for a safety situation awareness method for aircraft swarms in autonomous operation mode.

[0024] Figure 3 : Schematic diagram of the risk situation network of a safety situation awareness method for aircraft swarms in autonomous operation mode.

[0025] Figure 4 :Schematic diagram of kernel density analysis of a safety situation awareness method for aircraft clusters in autonomous operation mode. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described below with reference to the accompanying drawings.

[0027] The present invention provides a method for sensing the safety situation of an aircraft cluster in an autonomous operation mode, such as Figure 1 As shown: The specific steps are:

[0028] Step 1: Analyze the impact mechanism of CNS performance on track uncertainty, introduce the error caused by CNS performance into the traditional collision model, and build a situation network model suitable for use in autonomous operation mode;

[0029] Step 1.1: Analyze the mechanism by which CNS performance affects track uncertainty. Accuracy has a direct impact on navigation error. Any RNPa means that the flight path must be within an mile of the planned path 95% of the time. And because the aircraft positioning error caused by RNP satisfies e N : Right now:

[0030]

[0031] Solved Knowable The value of is only related to a, that is, the RNP type.

[0032] The main difference between different types of RCP is the transmission time. NThe definition and solution of RCPa can be considered that any RCPa means that the flight path will be around the planned path v in 95% of the time. x an mile, where v x represents the lateral component of velocity, and a represents the transmission time. Based on the type of RCP selected and the 95% error range, the following formula is used:

[0033]

[0034] Solved Knowable Only and v x Related to a, the aircraft's airspeed and RCP type.

[0035] The difference between different types of RSP is mainly in the reaction time. S The analysis can refer to RCP and RNP, that is, any RSPa means that the flight path must be on both sides of the centerline of the established route 95% of the time. x bn mile, where v x represents the lateral component of the aircraft speed, and b represents the reaction time corresponding to RSPa. Similarly, the following formula also exists:

[0036]

[0037] Solved Knowable Only and v x Related to b, the aircraft's airspeed and RSP type;

[0038] Step 1.2: Derivation of collision probability calculation model under autonomous operation conditions Aircraft positioning error due to CNS performance e CNS , e CNS : in Assuming that the aircraft position deviation random variable grows along the x-axis, y-axis, and z-axis in the coordinate system and the errors are independent of each other, we have Indicates that the variable has an expectation of 0 and a variance of σ 2 Normal distribution. Under the performance conditions of RCPn1, RNPn2, and RSPn3:

[0039]

[0040] The collision probability is equivalent to the integral of the probability density function of the track error of aircraft B within the fusion protection zone of aircraft A. The collision model is as follows: Figure 2 As shown, the calculation formula is as follows:

[0041]

[0042] Where P conflict is the collision probability of the two aircraft; f is the probability density function of the track error transferred to the random aircraft B; Z area For integrated protection areas; x R ,y R ,z R ,x S ,y S ,z S are the positions of reference aircraft A and random aircraft B respectively.

[0043] Since the aircraft position deviation random variable grows along the x-axis, y-axis, and z-axis in the coordinate system and the errors are independent of each other, the probability density function of the normal random variable (x, y) along the horizontal axis x and the vertical axis y of the coordinate system and the calculation process of the conflict probability are simplified as follows:

[0044]

[0045] Where: f xy (x, y) is the probability density function of the two-dimensional random variable (x, y) along the horizontal axis x and the vertical axis y of the coordinate system; P xy is the instantaneous collision probability along the horizontal axis x and the vertical axis y of the coordinate system; u x 、u y are the relative positions of the two aircraft along the horizontal axis x and the vertical axis y of the coordinate system; are the variances of the two aircraft along the x-axis and y-axis respectively.

[0046] Similarly, the probability density function of the normal random variable z along the coordinate axis z and the calculation process of the conflict probability are simplified as follows:

[0047]

[0048] Where: f z (z) is the probability density function of the random variable z along the vertical axis z of the coordinate system; P z is the instantaneous collision probability along the vertical axis z of the coordinate system; u z is the relative position of the two aircraft along the vertical axis z of the coordinate system; are the variances of the two aircraft along the z-axis;

[0049] The final conflict probability: P conflict =P xy ×P z .

[0050] Step 2: Combine the spatial structure representation of aircraft clusters and traffic flow parameter information to design cluster risk assessment indicators, divide cluster risk patterns, and form an air traffic risk situation network;

[0051] Step 2.1: The indicators are designed as follows.

[0052] Conflict scale: the number of aircraft in the cluster, denoted by N.

[0053] Number of potential conflicts: the number of edges in the cluster network, denoted by E.

[0054] Conflict density: The proportion of potential conflicting pairs among all possible conflicting pairs, denoted as

[0055] Average conflict duration: Conflict duration refers to the time required for an aircraft with a conflicting relationship to fly to a non-conflicting relationship and cancel the connection while maintaining the current flight status. The average conflict duration refers to the average time required for all aircraft with conflicting relationships in the aircraft cluster to fly to a non-conflicting relationship, that is:

[0056]

[0057] in represents the average conflict duration; (v i ,v j ) represents the edge relationship between nodes i and j, 1 and 0 represent the existence of edge connection respectively; T c,ij represents the conflict duration between aircraft i and j; E is the number of network edges.

[0058] Average convergence risk intensity: First, we further calculate the spatial closeness rate V between aircraft ij (t) is used to reflect the relative motion state between the two aircraft.

[0059]

[0060] Where V ij (t) is greater than 0, it means that aircraft i and j are in a divergent state at time t; otherwise, they are in a converging state; d ij (t) represents the distance between aircraft i and j at time t.

[0061] Based on the spatial proximity rate, the concentration risk intensity R is proposed. ij The calculation method is as follows:

[0062]

[0063] Since V ij (t) There are two cases, positive and negative. Taking into account the properties of the exponential function, we ensure that the risk intensity of aircraft with a greater degree of spatial proximity is greater than that of aircraft with a smaller degree of spatial proximity when the spatial closing rate is the same. We introduce the α and β spatial closing rate adjustment coefficients to convert V ij(t) is reasonably scaled to the positive range. So the average convergence intensity

[0064]

[0065] Step 2.2: Calculate the risk vector of each cluster, and divide the risk pattern by combining the K-means clustering algorithm to form an air traffic risk situation network, such as Figure 3 shown.

[0066] Step 3: Use survival characteristic analysis and spatial hotspot analysis methods to explore the spatiotemporal evolution characteristics of the risk situation network;

[0067] Step 3.1: Analyze the changes in the number of aircraft clusters of different risk levels within 24 hours at intervals of 1 hour to illustrate the temporal evolution characteristics of aircraft clusters with different risk patterns; count the duration samples of each risk pattern cluster within the corresponding time, and fit the duration sample distribution with the distribution function; count the multi-day aviation cluster survival cycle sample data to estimate the average life cycle of each pattern; use the Kaplan-Meier method to calculate the survival rate curve of each pattern; calculate the survival time quartiles of each risk pattern; use the Nelson-Aalen method to obtain the cumulative risk curve of each risk pattern.

[0068] Step 3.2: Discretize the airspace into a 300m×300m grid and combine it with the kernel density statistical method. The principle is as follows Figure 4 As shown. The search radius is set to 3000m, and the Gaussian kernel is selected as the kernel function. The trajectory set of each risk mode aircraft cluster in a specific time period is collected, and the spatial operation hotspot distribution of each risk mode aviation cluster is statistically analyzed; the sector operation hotspots are defined and analyzed in detail in combination with the distribution of key waypoints in the study area; the distribution of operation hotspots in three-dimensional space is further analyzed, and the study area is stratified according to specific altitudes. The trajectory set of each risk mode aircraft cluster in each altitude segment is statistically analyzed, and the spatial operation hotspot distribution of each altitude segment is observed.

[0069] Among them, the kernel density statistics is specifically introduced: The kernel density method is a common spatial geographic analysis method that can present the spatial distribution of point element agglomeration hotspots. The kernel density estimation places a smooth, symmetrical kernel around each data point, and then adds the values ​​of all kernels in the entire study area to obtain a density estimate for each location and generate a smooth surface. This process can transform discrete objects into continuous fields and realize the visualization of unit density. The kernel density statistical method focuses on the calculation of kernel density values, and its calculation formula is:

[0070]

[0071] Among them, irepresents the kernel density of point i, r is the bandwidth, d ij is the distance between research object i and research object j, p is the number of research object j within the bandwidth r, and K is the kernel function; the kernel function K is a function defined on two input data points, which returns the similarity or inner product of two points.

Claims

1. A method for sensing the safety situation of an aircraft cluster in an autonomous operation mode, characterized in that: The following steps are involved: Step 1: Analyze the impact mechanism of CNS performance on track uncertainty, introduce the error caused by CNS performance into the traditional collision model, and build a situation network model suitable for use in autonomous operation mode; Step 2: Combine the spatial structure representation of aircraft clusters and traffic flow parameter information to design cluster risk assessment indicators, divide cluster risk patterns, and form an air traffic risk situation network; Step 3: Considering the significant spatiotemporal characteristics of aircraft cluster behavior, survival characteristic analysis and spatial hotspot analysis methods are used to explore the spatiotemporal evolution characteristics of the risk situation network.

2. The method for sensing the safety situation of an aircraft cluster in an autonomous operation mode according to claim 1, characterized in that: Situation network refers to a method based on graph theory, which regards active aircraft as nodes and connects aircraft with conflicting relationships with edges to generate a situation network.

3. The method for sensing the safety situation of an aircraft cluster in an autonomous operation mode according to claim 1, characterized in that: Step 1 specifically includes: Step 1.1: Analyze the operating characteristics of the autonomous operation mode; Step 1.2: Derived the aircraft positioning error caused by RCP and RSP by combining the known aircraft positioning error caused by RNP; Step 1.3: Based on the error distribution caused by CNS performance and the traditional collision model, a collision probability calculation model suitable for autonomous operation is derived. Aircraft pairs with collision probabilities greater than the safety target level are identified as conflicting aircraft pairs. Edges are established based on the conflict relationship to construct a situation network.

4. The method for sensing the safety situation of an aircraft cluster in an autonomous operation mode according to claim 3, characterized in that: In step 1.1, the autonomous operation mode has the following characteristics: the pilots will autonomously select the optimal trajectory based on the actual use of the airspace. The risk between aircraft will no longer be determined based on the traditional control-dominated mode and the fixed distance between aircraft. Instead, a detailed risk assessment will be conducted based on the performance and actual operating conditions of the aircraft, thereby maximizing the growth of airspace traffic while ensuring safety. In step 1.2, the aircraft positioning error caused by RCP and RSP is derived as follows: Accuracy has a direct impact on navigation error. Any RNPa means that the flight path must be within an mile of the planned path 95% of the time. The aircraft positioning error caused by RNP satisfies e N : Right now: Solved Knowable The value of is only related to a, that is, the RNP type; The main difference between different types of RCP is the transmission time; N Definition and solution of RCPb, assuming that any RCPb means that the flight path will be around the planned path 95% of the time x bn mile range, where v x represents the lateral component of velocity, b represents the transmission time; depending on the type of RCP selected and the 95% error range, the following formula is used: Solved Knowable Only and v x Related to b, the aircraft's airspeed and RCP type; Different types of RSP differ in their reaction time. S The analysis refers to RCP and RNP, that is, any RSPc means that the flight path must be on both sides of the centerline of the established route 95% of the time. x c′n mile, where v x represents the lateral component of the aircraft velocity, and c′ represents the reaction time corresponding to RSPc; similarly, the following formula also exists: Solved Knowable Only and v x is related to c′, the aircraft's airspeed and RSP type; In step 1.3, the collision probability calculation model under autonomous operation conditions is as follows: Aircraft positioning error due to CNS performance e CNS , e CNS : in Assuming that the aircraft position deviation random variable grows along the x-axis, y-axis, and z-axis in the coordinate system and the errors are independent of each other, then under the performance conditions of RCPn1, RNPn2, and RSPn3: Since the aircraft position deviation random variable grows along the x-axis, y-axis, and z-axis in the coordinate system and the errors are independent of each other, the probability density function of the normal random variable (x, y) along the horizontal axis x and the vertical axis y of the coordinate system and the calculation process of the conflict probability are simplified as follows: Where: f xy (x, y) is the probability density function of the two-dimensional random variable (x, y) along the horizontal axis x and the vertical axis y of the coordinate system; P xy is the instantaneous collision probability along the horizontal axis x and the vertical axis y of the coordinate system; u x 、u y are the relative positions of the two aircraft along the horizontal axis x and the vertical axis y of the coordinate system; are the variances of the two aircraft along the x-axis and y-axis respectively; The probability density function of the normal random variable z along the coordinate axis z-axis and the conflict probability calculation process are simplified as follows: Where: f z (z) is the probability density function of the random variable z along the vertical axis z of the coordinate system; P z is the instantaneous collision probability along the vertical axis z of the coordinate system; u z is the relative position of the two aircraft along the vertical axis z of the coordinate system; are the variances of the two aircraft along the z-axis; The final conflict probability: P conflict =P xy ×P z .

5. The method for sensing the safety situation of an aircraft cluster in an autonomous operation mode according to claim 1, characterized in that: Step 2 specifically includes: Step 2.1: Design cluster risk assessment indicators based on the spatial structure representation of aircraft clusters and traffic flow parameter information; Step 2.2: Calculate the risk vector of each cluster and divide the risk pattern using K-means clustering algorithm.

6. The method for sensing the safety situation of an aircraft cluster in an autonomous operation mode according to claim 5, characterized in that: In step 2.1, cluster risk assessment indicators include: Conflict scale: the number of aircraft in the cluster, denoted by N; Number of potential conflicts: the number of edges in the cluster network, denoted by E; Conflict density: The proportion of potential conflicting pairs among all possible conflicting pairs, denoted as Average conflict duration: Conflict duration refers to the time required for an aircraft with a conflicting relationship to fly to a non-conflicting relationship and cancel the connection while maintaining the current flight status, that is: in represents the average conflict duration; (v i ,v j ) represents the edge relationship between nodes i and j, 1 and 0 represent the existence of edge connection respectively; T c,ij represents the conflict duration between aircraft i and j; Average convergence risk intensity: First, we further calculate the spatial closeness rate V between aircraft ij (t) to reflect the relative motion between the two aircraft; Where V ij (t) is greater than 0, it means that aircraft i and j are in a divergent state at time t; otherwise, they are in a converging state; d ij (t) represents the distance between aircraft i and j at time t; Based on the spatial proximity rate, the concentration risk intensity R is proposed. ij The calculation method is as follows: The introduction of α and β spatial approach rate adjustment coefficients is to convert V ij (t) is reasonably scaled to the positive range, so the average convergence intensity 7. The method for sensing the safety situation of an aircraft cluster in an autonomous operation mode according to claim 1, characterized in that: Step 3 specifically includes: Step 3.1: Analyze the survival characteristics of aircraft clusters in each risk mode from the time dimension; Step 3.2: Understand the operational preferences of aircraft clusters in different risk modes from a spatial perspective.

8. The method for sensing the safety situation of an aircraft cluster in an autonomous operation mode according to claim 7, characterized in that: In step 3.1, the survival characteristics analysis includes: Analyze the changes in the number of aircraft clusters of different risk levels within 24 hours a day, and explain the temporal evolution characteristics of aircraft clusters with different risk patterns; count the duration samples of each risk pattern cluster within the corresponding time, and fit the duration sample distribution with the distribution function; count the multi-day aviation cluster survival cycle sample data, and estimate the average life cycle of each pattern; use the Kaplan-Meier method to calculate the survival rate curve of each pattern; calculate the survival time quartiles of each risk pattern; use the Nelson-Aalen method to obtain the cumulative risk curve of each risk pattern; In step 3.2, understanding the space operation preferences specifically includes: The airspace is discretized into 300m×300m grids. Combined with the kernel density statistical method, the search radius is set to 3000m, and the Gaussian kernel is selected as the kernel function. The trajectory sets of each risk mode aircraft cluster in a specific time period are collected, and the spatial operation hotspot distribution of each risk mode aviation cluster is statistically analyzed; the sector operation hotspots are defined and analyzed in detail based on the distribution of key waypoints in the study area; the distribution of operation hotspots in three-dimensional space is further analyzed, the study area is stratified according to specific altitudes, the trajectory sets of each risk mode aircraft cluster in each altitude segment are statistically collected, and the spatial operation hotspot distribution of each altitude segment is observed.

9. The method for sensing the safety situation of an aircraft cluster in an autonomous operation mode according to claim 8, characterized in that: Specific introduction to kernel density statistics: The kernel density method is used to present the spatial distribution of point element agglomeration hotspots, with the focus on the calculation of kernel density values, and its calculation formula is: Among them, i represents the kernel density of point i, r is the bandwidth, d ij is the distance between research object i and research object j, p is the number of research object j within the bandwidth r, and K is the kernel function; the kernel function K is a function defined on two input data points, which returns the similarity or inner product of two points.

Images