Unmanned aircraft operation risk assessment method, system and equipment

Abstract

The invention relates to the field of unmanned aerial vehicles, and provides an unmanned aerial vehicle operation risk assessment method, system and device, and the method comprises the following steps: dividing a risk environment into a plurality of samples; calculating an unmanned aerial vehicle ground collision risk value of each sample; calculating an unmanned aerial vehicle air collision risk of each sample, wherein the unmanned aerial vehicle air collision risk comprises a collision risk between the unmanned aerial vehicles and a collision risk between the unmanned aerial vehicles and a bird flock; the unmanned aerial vehicle air failure risk of each sample is calculated; and determining unmanned aerial vehicle operation risk grade distribution according to the calculated unmanned aerial vehicle ground collision risk value, unmanned aerial vehicle air collision risk and unmanned aerial vehicle air failure risk of each sample. According to the method, the unmanned aerial vehicle ground collision risk value, the unmanned aerial vehicle air collision risk and the unmanned aerial vehicle air failure risk are quantitatively calculated, and the unmanned aerial vehicle operation risk grade distribution is determined according to the data, so that a quantitative SORA model is realized, and the problem of step difference caused by discrete risk factors in a traditional model is avoided.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicles, and particularly to a method, a system, and a device for assessing the operation risks of an unmanned aircraft. Background Art

[0002] The Specific Operation Risk Assessment (SORA) developed by the Joint Authorities for Rulemaking of Unmanned Systems (JARUS) is a key method for the safety management of unmanned aerial vehicles. The traditional SORA model can provide preliminary risk assessment and decision support for the operation of unmanned aerial vehicles. However, it relies on discrete qualitative risk factors, which makes it unable to provide sufficient sensitivity and accuracy when facing minor changes in the risk environment. Therefore, the traditional model is prone to stepwise differences in risk levels during the assessment process. Even after correction, the assessment results of the risk levels still do not show significant changes and there are large deviations from the actual risk environment. This limitation not only affects the accuracy of risk assessment but also restricts its application in complex unmanned aerial vehicle operation environments.

[0003] In this case, it is particularly important to improve the traditional SORA model to enhance its applicability and reliability in dynamic and complex environments. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a method, a system, and a device for assessing the operation risks of an unmanned aircraft, aiming to solve at least one of the technical problems existing in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] In the first aspect, the present invention provides a method for assessing the operation risks of an unmanned aircraft, including the following steps:

[0007] Dividing the risk environment into multiple samples;

[0008] Calculating the ground collision risk value of the unmanned aircraft for each sample;

[0009] Calculating the in-air collision risk of the unmanned aircraft, where the in-air collision risk of the unmanned aircraft includes the risk of collision between unmanned aircraft and the risk of collision between unmanned aircraft and bird flocks;

[0010] Calculating the in-air failure risk of the unmanned aircraft for each sample;

[0011] Determine the UAV operation risk level distribution based on the UAV ground collision risk value, UAV mid-air collision risk, and UAV mid-air failure risk of each calculated sample.

[0012] In some embodiments, the UAV ground collision risk value is calculated based on the number of casualties and the degree of casualties in ground impact accidents per flight hour. The specific calculation formula is as follows:

[0013] GRV = E k ·N

[0014] where GRV is the UAV ground collision risk value, E k is the degree of casualties, and N is the number of casualties in ground impact accidents per flight hour.

[0015] In some embodiments, the calculation formula for the number of casualties in ground impact accidents per flight hour is as follows:

[0016]

[0017] where A is the area of the accident impact area, ρ is the population density of the area, P sys is the probability of ground impact accidents caused by UAV mid-air accidents per flight hour, E PF is the ground shelter protection coefficient, and R is the casualty rate;

[0018] The degree of casualties is determined by the kinetic energy corresponding to 1.2 times the maximum designed speed of the UAV. The specific formula is as follows:

[0019]

[0020] where m is the mass of the UAV, and V op is the maximum designed speed of the UAV.

[0021] In some embodiments, the UAV and UAV collision risk includes the collision risk between two UAVs on parallel headings and the collision risk between two UAVs on crossing headings;

[0022] and / or, the UAV and bird flock collision risk includes the collision risk between a UAV and a flying bird on parallel headings and the collision risk between a UAV and a flying bird on crossing headings.

[0023] In some embodiments, the calculation formula for the collision risk between two UAVs on parallel headings is as follows:

[0024]

[0025] where CN is the collision risk between two UAVs on parallel headings, P x (S x ) is the probability of longitudinal collision between two UAVs on parallel headings, Py (S y ) is the probability of lateral collision between two UAVs on a parallel flight path, P z (S z ) is the probability of vertical collision between two UAVs on a parallel flight path, R is the radius of the collision template, v x , v y , v z are the relative velocities of two UAVs on a parallel flight path in the longitudinal, lateral, and vertical directions, respectively.

[0026] In some embodiments, the probability of longitudinal collision between two UAVs on the parallel flight path is:

[0027]

[0028] where S x ' is the relative distance between two UAVs in the longitudinal direction;

[0029] The probability of lateral collision between two UAVs on the parallel flight path is:

[0030]

[0031] where f y (y1) represents the probability density function based on the lateral spacing navigation error, δ is the proportion of severe errors, S y ' is the relative distance between two UAVs in the lateral direction;

[0032] In the above formula, represents the probability density function of the general navigation error presented as an exponential distribution, represents the probability density function of the severe navigation error, In the above two formulas, a1 is the parameter of the probability density function of the general navigation error, and a2 is the parameter of the probability density function of the severe navigation error.

[0033] The probability of vertical collision between two UAVs on the parallel flight path is:

[0034]

[0035] where R is the radius of the collision template, S z ' is the relative distance between two UAVs in the vertical direction, and δ is the standard deviation of the distance between two UAVs in the vertical direction.

[0036] In some embodiments, the calculation formula for the collision risk of two UAVs on a crossing flight path is as follows:

[0037]

[0038] Among them, CN2 is the risk of collision between two drones on crossing courses, and P x '(S x ) is the probability of longitudinal collision between two drones on crossing courses, and P y '(S y ) is the probability of lateral collision between two drones on crossing courses, and P z '(S z ) is the probability of vertical collision between two drones on crossing courses. R is the radius of the collision template, and v x ’, v y ’, v z ’ are the relative velocities of two drones on crossing courses in the longitudinal, lateral, and vertical directions respectively.

[0039] In some embodiments, the probability of longitudinal collision between two drones on crossing courses is:

[0040]

[0041] Among them, S x ' is the relative distance between two drones in the longitudinal direction, and θ is the crossing angle between the headings of the two drones;

[0042] The probability of lateral collision between two drones on crossing courses is:

[0043]

[0044] Among them, θ is the crossing angle between the headings of the two drones, S y ' is the relative distance between two drones in the lateral direction, and f y (y1) represents the probability density function based on the lateral spacing navigation error, δ is the proportion of serious errors;

[0045] In the above formula, represents the probability density function of the general navigation error presenting an exponential distribution, represents the probability density function of the serious navigation error, In the above two formulas, a1 is the parameter of the probability density function of the general navigation error, and a2 is the parameter of the probability density function of the serious navigation error.

[0046] The probability of vertical collision between two drones on crossing courses is:

[0047]

[0048] Among them, R is the radius of the collision template, and δ is the standard deviation of the distance between two drones in the vertical direction.

[0049] In some embodiments, the risk of collision between the parallel-heading unmanned aerial vehicle and a bird is:

[0050] P(t) = P X (t)P Y (t)P Z (t)

[0051] The risk of collision between the cross-heading unmanned aerial vehicle and a bird is:

[0052]

[0053] where Px(t) is the probability of longitudinal collision between the unmanned aerial vehicle and the bird, Py(t) is the probability of lateral collision between the unmanned aerial vehicle and the bird, and Pz(t) is the probability of vertical collision between the unmanned aerial vehicle and the bird.

[0054] In some embodiments, the method for calculating the risk of in-air failure of the unmanned aerial vehicle is as follows:

[0055] A failure assessment model is constructed using a Bayesian probability network. The failure assessment model analyzes the failure causes from three aspects: the unmanned aerial vehicle system, the operating environment, and human factors, and calculates the failure probability.

[0056] Second, the present invention also provides an operating risk assessment system for an unmanned aerial vehicle, including:

[0057] A sample determination module; used to divide the risk environment into multiple samples;

[0058] A ground collision risk calculation module, used to calculate the ground collision risk value of the unmanned aerial vehicle for each sample;

[0059] An in-air collision risk calculation module, used to calculate the in-air collision risk of the unmanned aerial vehicle for each sample, where the in-air collision risk of the unmanned aerial vehicle includes the risk of collision between unmanned aerial vehicles and the risk of collision between the unmanned aerial vehicle and a bird flock;

[0060] An in-air failure risk calculation module, used to calculate the in-air failure risk of the unmanned aerial vehicle for each sample;

[0061] A risk level distribution determination module, used to determine the operating risk level distribution of the unmanned aerial vehicle according to the calculated ground collision risk value, in-air collision risk, and in-air failure risk of the unmanned aerial vehicle for each sample.

[0062] Third, the present invention also provides an operating risk assessment device for an unmanned aerial vehicle, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned operating risk assessment method for an unmanned aerial vehicle are implemented.

[0063] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0064] The present invention provides a method, a system and a device for risk assessment of unmanned aerial vehicle operation. The method divides the risk environment into multiple samples, and quantitatively calculates the ground collision risk value, the in-air collision risk and the in-air failure risk of each sample for the unmanned aerial vehicle. Then, according to the calculation results, the risk level distribution of the unmanned aerial vehicle operation is determined, thereby realizing an innovative quantitative SORA model. The unique feature of the quantitative SORA model is the introduction of continuous quantitative risk factors. By accurately calculating and presenting the risk distribution, this model effectively avoids the step difference problem caused by discrete risk factors in traditional models. In addition, through quantitative analysis, this model significantly improves the accuracy of risk assessment, enabling it to capture subtle risk changes when facing critical thresholds, thus ensuring that the risk level assessment is more consistent with the actual environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is the overall flowchart of a method for risk assessment of unmanned aerial vehicle operation provided in this embodiment;

[0066] Figure 2 It is the flowchart of a quantitative SORA model for assessing the risk of unmanned aerial vehicle operation provided in this embodiment;

[0067] Figure 3 It is a schematic diagram of the cross-heading of unmanned aerial vehicle A and unmanned aerial vehicle B provided in this embodiment;

[0068] Figure 4 It is the calculation result of the quantitative SORA model for evaluating the simulated risk environment provided in this embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described experimental cases are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0070] SORA (Specific Operations Risk Assessment) is a risk assessment method specifically for drone operations. Its main processes include identifying the types of drone operations, analyzing potential ground and air risks, evaluating existing mitigation measures, and determining the acceptability of residual risks. The present invention aims to make innovative improvements to the SORA risk assessment method, transforming the traditional risk assessment method based on qualitative checklist into a mode based on quantitative calculation, so as to significantly improve the accuracy, reliability and adaptability of the SORA model. By introducing quantitative analysis means, the present invention seeks to overcome the stepwise errors and decision-making support limitations caused by relying on discrete qualitative risk factors in the traditional SORA model, enabling the model to more flexibly adapt to complex and dynamically changing drone operating environments.

[0071] Before conducting quantitative risk assessment, it is necessary to collect and collate multiple important data related to drone operations to ensure the comprehensiveness and accuracy of the assessment results. Specifically, the data collected includes the following aspects:

[0072] (1) Drone flight path planning route: To establish a Reich collision model for drones, it is necessary to focus on the morphological characteristics of the drone flight route and accurately measure the intersection angle and course interval to evaluate the potential collision risk or the possibility of mutual interference.

[0073] (2) Distribution of bird flocks: To calculate the collision probability between drones and birds, it is necessary to analyze the group activity patterns of birds to obtain a rough function model of their distribution. This information is particularly crucial for evaluating the probability of drones being hit by birds during flight.

[0074] (3) Meteorological report: To fully calculate the in-air failure probability of drones and construct a Bayesian probability network, it is necessary to collect and analyze meteorological data, especially paying attention to the occurrence probabilities of extreme weather conditions, including heavy rainfall, strong winds, thunderstorms and other weather factors that may affect the normal flight of drones.

[0075] (4) Geographic environment: To calculate the ground risk value, it is necessary to detailedly evaluate the geographical characteristics of the drone operation area, including the ground population density, the shielding coefficient of ground buildings or natural obstacles, and even conduct a comprehensive analysis in combination with the casualty rate. Through the collection and analysis of these data, it can provide an important basis for evaluating the potential losses when drones crash or get out of control.

[0076] As Figure 1 and Figure 2 shown, the embodiment of the present invention provides a method for assessing the operation risk of an unmanned aerial vehicle, including the following steps:

[0077] S1. Divide the risk environment into multiple samples;

[0078] Specifically, the risk environment can be divided into grids to form multiple samples. Among them, the ground risk environment can be divided into plane grids to form multiple samples with a plane distribution, and the aerial risk environment can be divided into three-dimensional grids to form multiple samples with a three-dimensional distribution. Each sample contains the UAV operation risk assessment parameters in the corresponding environment.

[0079] S2. Calculate the UAV ground collision risk value of each sample;

[0080] Specifically, calculate the corresponding UAV ground collision risk value according to each sample corresponding to the ground risk environment. Ground risk is one of the key risks during the low-altitude flight of UAVs, and potential hazards include threats to ground personnel and facilities. In some embodiments, the UAV ground collision risk value is calculated according to the number of casualties and the degree of casualties in ground impact accidents per flight hour. The specific calculation formula is as follows:

[0081] GRV = E k ·N

[0082] Where GRV is the UAV ground collision risk value, E k is the degree of casualties, and N is the number of casualties in ground impact accidents per flight hour.

[0083] In some embodiments, the calculation formula for the number of casualties in ground impact accidents per flight hour is as follows:

[0084]

[0085] Where A is the area of the accident impact area, ρ is the population density of the area, P sys is the probability of ground impact accidents caused by UAV aerial accidents per flight hour, E PF is the ground shelter protection coefficient, and R is the casualty rate.

[0086] The derivation process of this formula is as follows:

[0087] The number of casualties N in ground impact accidents per flight hour:

[0088] N = P·I (1)

[0089] In formula (1), P is the number of people on the ground affected by the accident. The method for obtaining the personnel casualty probability I of ground impact accidents per flight hour is:

[0090] I = P acc ·R (2)

[0091] In formula (2), P acc is the occurrence probability of ground impact accidents per flight hour, and R is the casualty rate. The method for obtaining the occurrence probability P of ground impact accidents acc is as follows:

[0092]

[0093] P in formula (3) sys is the probability of a ground impact accident caused by an unmanned aerial vehicle (UAV) air accident per flight hour, and E PF is the ground shelter protection coefficient. The protection coefficient of the ground shelter varies according to the material type and thickness, and generally ranges from 10 to 30.

[0094] The number of people on the ground affected by the accident P:

[0095] P = A·ρ (4)

[0096] In formula (4), A is the area of the accident impact region (square meters), and ρ is the population density of the region (number of people per square meter).

[0097] The relevant parameters have been obtained through formulas (1)-(4). Next, the comprehensive formula for the number of casualties in a ground impact accident per flight hour is obtained using the relevant parameters:

[0098]

[0099] In some embodiments, the degree of casualty is determined by the kinetic energy corresponding to 1.2 times the maximum design speed of the UAV. The specific formula is as follows:

[0100]

[0101] Among them, m is the mass of the UAV, and V op is the maximum design speed of the UAV.

[0102] By quantitatively evaluating the ground impact accident, the present invention determines the ground risk value during the operation of the UAV.

[0103] S3. Calculate the UAV air collision risks of each sample. The UAV air collision risks include the risks of collision between UAVs and the risks of collision between UAVs and bird flocks;

[0104] Specifically, calculate the corresponding UAV air collision risks according to each sample corresponding to the air risk environment. Preferably, the risks of collision between UAVs include the risks of collision between two UAVs with parallel headings and the risks of collision between two UAVs with crossing headings; the risks of collision between UAVs and bird flocks include the risks of collision between a UAV with a parallel heading and a flying bird and the risks of collision between a UAV with a crossing heading and a flying bird.

[0105] The collision between UAVs is one of the risks that cannot be ignored in the collaborative operation of UAVs in the air. The present invention uses the Reich collision model to calculate the risks of collision between UAVs.

[0106] In some embodiments, the calculation formula for the collision risk of two drones with parallel headings is as follows:

[0107]

[0108] Where CN is the collision risk of two drones with parallel headings, and P x (S x ') is the probability of longitudinal collision between two drones on a parallel heading, and P y (S y ') is the probability of lateral collision between two drones on a parallel heading, and P z (S z ') is the probability of vertical collision between two drones on a parallel heading. R is the radius of the collision template, and v x 、v y 、v z are the relative velocities of two drones on a parallel heading in the longitudinal, lateral, and vertical directions, respectively.

[0109] The derivation process of the above formula is as follows:

[0110] Set N x as the frequency of longitudinal overlap between drones; P x as the probability of longitudinal overlap between drones; t x as the average time for drone B to longitudinally pass through the collision template of drone A; V x as the relative velocity of the two drones in the longitudinal direction. Because:

[0111]

[0112] Therefore

[0113]

[0114] Similarly, according to formulas (8)-(9), the definitions of N y , N z , P y , P z , t y , t z , V y , V z can be obtained. Thus, the total number of collisions CN of drone B entering the vicinity layer of drone A in the longitudinal, lateral, and vertical directions per hour can be calculated respectively:

[0115] CN = N x P y P z + N y P x P z + Nz P x P y (10)

[0116] The probability and frequency of the collision risk CN are derived from the collision calculation under specific interval criteria. However, the traditional Reich model mainly focuses on the collision of parallel headings and does not consider the collision risk of two UAVs in operation. The present invention adds a distance function to measure the interval between UAVs in real time and calculates the collision risk accordingly. Given equations (8) to (10), the CN expression can be updated to:

[0117]

[0118] In formula (11), the longitudinal distance S x ′ = S x +|ΔL x (t)|; Similarly, the lateral distance S y ′ = S y +|ΔL y (t)|; The vertical distance S z ′ = S z +|ΔL z (t)|. Where S x , S y , S z represent the interval criteria in the longitudinal, lateral, and vertical directions of the Reich model respectively, and ΔL x (t), ΔL y (t), ΔL z (t) represent the relative displacements of the two UAVs on the longitudinal, lateral, and vertical axis directions at time t respectively.

[0119] In some embodiments, the probability of a longitudinal collision between two UAVs on the parallel heading is:

[0120]

[0121] where S x ' is the relative distance between the two UAVs in the longitudinal direction;

[0122] The probability of a lateral collision between two UAVs on the parallel heading is:

[0123]

[0124] where f y (y1) represents the probability density function based on the lateral interval navigation error, δ is the proportion of serious errors, and S y ' is the relative distance between the two UAVs in the lateral direction;

[0125] In the above formula, It is expressed as the probability density function of the general navigation error presenting an exponential distribution, It is expressed as the probability density function of the severe navigation error, In the above two formulas, a1 is the parameter of the probability density function of the general navigation error, and a2 is the parameter of the probability density function of the severe navigation error.

[0126] The probability of a vertical collision occurring between two UAVs on the parallel flight path is:

[0127]

[0128] where R is the radius of the collision template, S z ' is the relative distance between the two UAVs in the vertical direction, and δ is the standard deviation of the distance between the two UAVs in the vertical direction.

[0129] The derivation process of the above formula is as follows:

[0130] The lateral spacing standard S of the parallel flight path y remains unchanged. Assuming that the lateral position of the UAV during flight is mainly affected by the navigation error, and this error follows a two-parameter mixed exponential distribution:

[0131]

[0132] In formula (12), f y (y1) represents the probability density function based on the navigation error of the lateral spacing; δ is the proportion of severe errors; It is expressed as the probability density function of the general navigation error presenting an exponential distribution, It is expressed as the probability density function of the severe navigation error:

[0133]

[0134] In formulas (13) and (14), a1 is the parameter of the probability density function of the general navigation error, and a2 is the parameter of the probability density function of the severe navigation error. Therefore, according to the two-parameter mixed exponential distribution formula (12), the probability of a lateral collision P y (S y ) of two UAVs on the parallel flight path is obtained:

[0135]

[0136] Next, calculate the longitudinal collision probability P x (S x'), the longitudinal collision probability refers to the possibility that when two UAVs fly along parallel headings, the longitudinal interval between them is less than the standard interval. According to the known formula (9), the longitudinal overlap frequency N between UAVs per unit time of the parameters x , the probability of longitudinal collision between the two UAVs on the parallel heading is:

[0137]

[0138] where S x ' is the relative distance between the two UAVs in the longitudinal direction;

[0139] During the flight of the UAV in the air, its vertical position is affected by meteorological conditions, errors of on-board altimetry equipment and human operation errors. Assume that the flight altitude errors of UAV A and B both conform to the normal distribution, and the actual altitude difference ΔH between them:

[0140] ΔH=(H A -H B )=(D A -D B )+(X A -X B ) (17)

[0141] In formula (17), H A and H B respectively represent the designated flight altitudes of the two UAVs, and X A and X B respectively represent the altitude errors caused by on-board altimetry equipment and meteorological factors. These two errors make the altitude difference of the UAVs obey the normal distribution, and the specific expression is:

[0142]

[0143] The position altitude error ΔH of the two UAVs obeys the normal distribution with a mean of S z , so its probability density function is:

[0144]

[0145] Since the relative positions and absolute positions of the UAVs in the three directions are independent of each other, according to the probability formula, the probability of vertical collision between the two UAVs on the parallel heading is:

[0146]

[0147] where R is the radius of the collision template, S z ' is the relative distance between the two UAVs in the vertical direction, and δ is the standard deviation of the distance between the two UAVs in the vertical direction.

[0148] In some embodiments, the calculation formula for the collision risk of two UAVs in the crossing course is as follows:

[0149]

[0150] Wherein, CN2 is the collision risk of two UAVs in the crossing course, P x '(S x ') is the probability of longitudinal collision between two UAVs in the crossing course, P y '(S y ') is the probability of lateral collision between two UAVs in the crossing course, P z '(S z ') is the probability of vertical collision between two UAVs in the crossing course, R is the radius of the collision template, v x ’, v y ’, v z ’ are the relative velocities of two UAVs in the longitudinal, lateral and vertical directions in the crossing course respectively.

[0151] The derivation process of this formula is similar to that of the above formula (11), and will not be elaborated here.

[0152] In some embodiments, the probability of longitudinal collision between two UAVs in the crossing course is:

[0153]

[0154] Wherein, S x ' is the relative distance between two UAVs in the longitudinal direction, θ is the crossing angle of the headings of the two UAVs;

[0155] The probability of lateral collision between two UAVs in the crossing course is:

[0156]

[0157] Wherein, θ is the crossing angle of the headings of the two UAVs, S y ' is the relative distance between two UAVs in the lateral direction, f y (y1) represents the probability density function based on the lateral spacing navigation error, δ is the proportion of serious errors;

[0158] In the above formula, represents the probability density function of general navigation error presenting an exponential distribution, represents the probability density function of serious navigation error, In the above two formulas, a1 is the parameter of the probability density function of general navigation error, and a2 is the parameter of the probability density function of serious navigation error.

[0159] The probability of a vertical collision between two UAVs on intersecting courses is as follows:

[0160]

[0161] where R is the radius of the collision template and δ is the standard deviation of the distance between the two UAVs in the vertical direction.

[0162] The derivation process of the above formula is as follows:

[0163] There is an intersection angle in the intersecting courses, which is different from the parallel courses, and this angle will affect the collision risk. The schematic diagram of the intersecting courses is as Figure 3 shown:

[0164] Course 1 and Course 2 are parallel courses, and the intersection angle between Course 3 and Courses 1 and 2 is θ. B is flying on Course 3, while A is operating on Course 1. From Figure 3 it can be seen that the range of variation of the lateral separation standard S y is [0, S x sinθ]. Since the positions of the two UAVs are independent of each other, the lateral collision probability of the UAV pair based on the intersection angle β and the lateral position error is:

[0165]

[0166] The longitudinal collision probability is improved according to formula (16) and the intersection angle θ to:

[0167]

[0168] According to the assumption of the vertical collision probability of parallel courses, the position errors of the two follow a normal distribution. The vertical collision probability is the possibility that the vertical height difference between the two UAVs is less than the radius of the collision template:

[0169] P′ z (S′ z ) = P(|ΔH| < R) (23)

[0170] The probability density function of the position height error of the two UAVs following a normal distribution with a mean of S z is:

[0171]

[0172] Under the assumption that the flight height difference between UAV A and B is 0, the intersection course angle mainly affects the longitudinal and lateral collision probabilities, and the vertical direction is not affected. According to formula (24), in this case, the calculation method of the vertical collision probability P′ z (S′ z ) is:

[0173]

[0174] Among them, the height difference between the two UAVs follows a normal error with a standard mean of 0 for the vertical separation standard.

[0175] Combining formulas (21), (22), and (25), and since each aspect is independent and a series of events, the collision risk of UAV pairs in the cross-course direction is obtained as:

[0176]

[0177] In some embodiments, the collision risk between the horizontal-course UAV and the bird is:

[0178] P(t) = P X (t)P Y (t)P Z (t)

[0179] The collision risk between the cross-course UAV and the bird is:

[0180]

[0181] Among them, Px(t) is the probability of longitudinal collision between the UAV and the bird, Py(t) is the probability of lateral collision between the UAV and the bird, and Pz(t) is the probability of vertical collision between the UAV and the bird.

[0182] The specific process is as follows:

[0183] The present invention uses a method based on position error to measure the probability P(t) of collision between a UAV and a bird in the air. Through the probability model of position error, the collision probabilities in the longitudinal, lateral, and vertical directions are calculated, so as to predict the collision risk between the UAV and the bird. Assuming that the flight position distribution of the bird is approximately normally distributed and its flight trajectory is mostly parallel to the UAV trajectory, the normal distribution of the longitudinal position error is deduced as follows:

[0184]

[0185] The longitudinal error of the UAV or the bird at time t is i = 1, 2. i = 1 represents the UAV, i = 2 represents the bird, and x represents the longitudinal direction. ε ix (t) is the longitudinal position error of the UAV or the bird at time t; μ ix is the average distance of the longitudinal position error of the UAV or the bird; is the variance of the longitudinal position error of the UAV or the bird. At time t, the longitudinal actual distance of the UAV or the bird from a certain reference point is:

[0186] X i (t) = d ix (t) + εix (t) (28)

[0187] In formula (28), d ix (t) is the nominal distance between the drone or bird and the reference point. If the nominal longitudinal distance is L x (t) = d 1x (t) - d 2x (t), substituting the nominal longitudinal distance L x (t) into the actual longitudinal distance formula between the drone and the bird, then at time t:

[0188] X1(t) - X2(t) = L x (t) + (ε 1x (t) - ε 2x (t)) (29)

[0189] Because the positions of both conform to the normal distribution characteristics, the actual longitudinal distance between the drone and the bird:

[0190]

[0191] The collision risk between the drone and the bird at time t:

[0192] P(t) = P X (t)P Y (t)P Z (t) (31)

[0193] where Px(t) is the probability of longitudinal collision between the drone and the bird, Py(t) is the probability of lateral collision between the drone and the bird, and Pz(t) is the probability of vertical collision between the drone and the bird.

[0194] Combining formulas (27) - (30), the longitudinal, lateral, and vertical collision probabilities between the drone and the bird at time t are derived as follows:

[0195] The probability of longitudinal collision between the drone and the bird at time t is:

[0196]

[0197] The probability of lateral collision between the drone and the bird at time t is:

[0198]

[0199] The probability of vertical collision between the drone and the bird at time t is:

[0200]

[0201] Assume that at the intersection point, the vertical collision risk between the drone and the bird at the same altitude layer is P Z(t)=1, different altitude layers are P Z (t) = 0. Combining formulas (31)-(33), the collision risk between the drone and the bird when flying on a cross-trajectory is:

[0202]

[0203] S4. Calculate the risk of drone failure in the air for each sample;

[0204] Specifically, the corresponding drone air failure risk is calculated based on each sample corresponding to the air risk environment.

[0205] In some embodiments, the method for calculating the risk of drone mid-air failure is as follows:

[0206] A failure assessment model is constructed using the Bayesian probability network. The failure assessment model analyzes the failure causes from three aspects: UAV system, operating environment and human factors, and calculates the failure probability.

[0207] Specifically, drone equipment failure is one of the main sources of aerial risks. Whether it is hardware failure or external environmental impact, failure significantly increases operational risks. This paper will use the Bayesian network model to analyze the possibility of equipment failure and evaluate its impact on the overall operational safety of drones:

[0208] Based on the relevant case library, this paper analyzes the failure causes from three aspects: UAV system, operating environment and human factors, and uses Bayesian probability network to construct a failure assessment model, aiming to show how to use data to build a Bayesian network to calculate the risk probability of UAV mid-air failure.

[0209] (1) Equipment failure: Failure of any component in the UAV system may cause global failure. The Bayesian network captures the dependencies of subsystems by integrating sensor data (such as acceleration, angular velocity, etc.) and quantifies the probability of component failure through maximum likelihood estimation.

[0210] (2) Bayesian network modeling is based on directed acyclic graph (DAG), which represents the dependency and independence relationship between variables. By learning existing data, the model can infer the potential failure mode and its probability of the system. The constructed model covers factors such as equipment failure, environmental impact and human operation, and uses conditional probability distribution to characterize the dependency relationship between variables and their parent nodes. The probability distribution and joint distribution of each variable are shown in formula (36):

[0211]

[0212] In formula (36): P AF is the probability of failure event; X is the node; i is the node number; n is the total number of nodes; δ(X i ) is the node except for node Xi each node other than

[0213] (3) Analyze the impact of abnormal sensor data on the failure risk using a Bayesian network, and infer the underlying events to the top-level failure types. This model can calculate the failure probability and provide risk predictions for different operation scenarios.

[0214] S5. Determine the distribution of the UAV operation risk level according to the calculated UAV ground collision risk value, UAV mid-air collision risk, and UAV mid-air failure risk of each sample.

[0215] Specifically, by calculating the UAV ground collision risk value, UAV mid-air collision risk, and UAV mid-air failure risk for multiple samples, a series of risk value data is obtained, and then the K-means clustering algorithm is used to form the UAV operation risk level distribution. The core idea of this algorithm is to divide the risk value data into several clusters (for example, 10 clusters), and each cluster represents a risk level. Through clustering, all risk values are finally assigned to different risk levels, forming a risk level distribution.

[0216] Specifically, the UAV operation ground risk level distribution can be determined according to the UAV ground collision risk value of each sample, and the UAV operation mid-air risk level distribution can be determined according to the UAV mid-air collision risk and UAV mid-air failure risk of each sample. The risk level distribution includes multiple levels of risk levels and the risk probability or risk frequency corresponding to each risk level.

[0217] Verify the characteristics of the quantitative SORA model by comparing the differences between the quantitative SORA evaluation model in the present invention and the traditional SORA evaluation model.

[0218] The experimental environment of the present invention is selected as a three-dimensional space of 120×120×120m 3 with a grid division of 24×24×24m 3 Taking the bottom grid layer "Ground" where the coordinates (0, 0, 0), (125, 0, 0), and (0, 125, 0) are located and assigning the initial physical attributes of "ground protection coefficient = 10" and "local ground population density" to the ground grid layer, and other grids are "Air", respectively simulating the ground environment and mid-air environment under the UAV flight mission.

[0219] The elliptical proximity layer of UAV A is set according to the UAV size: the semi-major axis is 1m, the semi-minor axis is 0.5m, and the semi-height axis is 0.5m.

[0220] In the aerial environment, calculate the collision probability between the UAV and other UAVs and birds, as well as the probability of system failure risk, estimate the probability of UAV failure and falling based on the parallel relationship, and draw the initial aerial environment risk map. In the ground environment, calculate the falling probability of UAV A in the "Ground" grid, determine the initial ground risk in combination with flight mission parameters, and draw the risk assessment map.

[0221] Through risk simulation, obtain the risk data of the UAV under different conditions, then use traditional and quantitative SORA models for evaluation, and conduct hierarchical clustering analysis in the quantitative SORA model to refine the risk assessment results and master its levels and distributions.

[0222] A. Traditional SORA assesses the risk environment of the UAV in this simulation scenario:

[0223] 1. Determine the Ground Risk Category (GRC): According to the UAV weight of 0.248 kg, under the condition of a population density of 2 people per square meter and good ground protection measures (Ground Cover Protection = 10), it can be known from the table lookup and the original SORA text that GRC = 5.

[0224] 2. Determine the Aerial Risk Category (ARC): The UAV flight altitude is 200 meters, and the maximum flight speed of the UAV is 16 m / s. It can be known from the table lookup and the original SORA text that this operation belongs to the ARC-c level.

[0225] 3. Based on the calculated GRC = 5 and ARC-c, referring to Table 1, the final SAIL level is IV. The result evaluated by the traditional SORA framework method shows that: "The risk of this operation is relatively high."

[0226] Table 1 Comprehensive Risk Level SAIL Evaluation Table of Traditional SORA Model

[0227]

[0228] B. Use the quantitative SORA model framework to evaluate the risk situation in this simulation scenario to obtain specific data.

[0229] This embodiment uses the K-means algorithm: select the parameter K = 10

[0230] 1. Initialize the cluster centers: Randomly select k initial cluster centers {c1, c2,..., c k}.

[0231] 2. Sample assignment: For each sample point x i , according to the Euclidean distance formula (36), assign it to the cluster center closest to it:

[0232]

[0233] where d(x i , c j ) represents the distance between the sample x i and the cluster center c j .

[0234] 3. Update the cluster center: Calculate the new center of each cluster and set it to the average of all samples within the current cluster:

[0235]

[0236] where S j is the set of all samples assigned to cluster j.

[0237] 4. Iteration: Repeat steps 2 - 3 to calculate formulas (37) - (38) until the cluster centers no longer change or the convergence condition is reached.

[0238] After the K - means algorithm, the air risk probability data is divided into 10 clusters. By randomly selecting 6 samples as the cluster centers, according to the distance d(x i , c j ) between each sample and the cluster center, the samples are assigned to the nearest cluster, and the cluster centers are iteratively updated until the clustering result converges.

[0239] Such as Figure 4The figure shows the evaluation calculation results of the quantitative SORA model for the simulated risk environment. It can be seen from the figure that the risk values of each sample point were calculated in the experiment. Each risk value represents the potential risks in the air and on the ground under the original risk environment conditions. In this way, a series of risk value data were obtained, and there may be certain fluctuations or discreteness in these data. To better classify these discrete risk values, the K-means clustering algorithm was used in data processing in this experiment. The core idea of this algorithm is to divide the risk value data into several clusters (for example, 10 clusters), and each cluster represents a risk level. Through clustering, all the risk values were finally assigned to different risk levels, forming a risk level distribution. Each level represents a certain degree of risk, usually arranged in ascending order from low to high. For example, 1 represents the lowest risk level, and 10 represents the highest risk level. In this way, the originally complex risk values were simplified into more intuitive risk levels, which is convenient for analysis and management. The traditional SORA model mainly analyzes discrete qualitative risk factors. When the risk factors are close to the critical threshold, even if the actual risk changes slightly, the evaluation results may show significant stepwise differences, resulting in a decrease in the accuracy of risk level assessment. In addition, due to its characteristics of dealing with discrete risk factors, the traditional SORA model has a complex process and takes a long time. In contrast, the quantitative SORA model calculates the risk distribution by using continuous quantitative risk factors, significantly reducing the stepwise differences caused by the risk factors approaching the threshold, while simplifying the evaluation process and reducing the time cost. The continuity and quantification characteristics of this model enable it to capture risk changes more accurately, providing higher accuracy and efficiency for risk assessment.

[0240] Through the comparison of the experimental results, the quantitative SORA model shows higher accuracy, real-time performance and repeatability in risk environment assessment, and is especially suitable for complex scenarios that require objective and dynamic assessment.

[0241] To sum up, the present invention has the following beneficial effects compared with the prior art:

[0242] The unmanned aircraft operation risk assessment method provided by the present invention divides the risk environment into multiple samples, quantitatively calculates the ground collision risk value of the unmanned aircraft, the in-air collision risk of the unmanned aircraft, and the in-air failure risk of the unmanned aircraft for each sample, and determines the distribution of the unmanned aircraft operation risk level according to the calculated ground collision risk value of the unmanned aircraft, the in-air collision risk of the unmanned aircraft, and the in-air failure risk of each sample, realizing a quantitative SORA model. The quantitative SORA model innovatively introduces continuous quantitative risk factors, and through the precise calculation and display of the risk distribution, effectively avoids the step difference problem caused by discrete risk factors in the traditional model. The model significantly improves the accuracy of risk assessment through quantitative analysis, enabling the model to capture subtle risk changes when facing the critical threshold, ensuring that the risk level assessment is more consistent with the actual environment.

[0243] The innovation of the quantitative SORA model lies not only in its application of continuous risk factors and improvement of calculation accuracy, but also in its systematic advantages in the field of risk assessment. The model provides a quantitative assessment path for regulatory agencies and unmanned aircraft operators for review and risk mitigation, and can more scientifically quantify the impact of various corrective measures on risks. In the future, with the continuous optimization and application of the model, quantitative SORA is expected to significantly improve the risk guarantee ability of unmanned aircraft missions, promote the development of the unmanned aircraft industry towards a higher standard of standardization, and help regulatory agencies and operators more systematically and precisely manage the risk factors in unmanned aircraft operations.

[0244] Based on the same inventive concept, an embodiment of the present invention further provides one, including:

[0245] A sample determination module; used to divide the risk environment into multiple samples;

[0246] A ground collision risk calculation module, used to calculate the ground collision risk value of the unmanned aircraft for each sample;

[0247] An in-air collision risk calculation module, used to calculate the in-air collision risk of the unmanned aircraft for each sample, and the in-air collision risk of the unmanned aircraft includes the risk of collision between unmanned aircraft and the risk of collision between unmanned aircraft and bird flocks;

[0248] An in-air failure risk calculation module, used to calculate the in-air failure risk of the unmanned aircraft for each sample;

[0249] A risk level distribution determination module, used to determine the distribution of the unmanned aircraft operation risk level according to the calculated ground collision risk value of the unmanned aircraft, the in-air collision risk of the unmanned aircraft, and the in-air failure risk of each sample.

[0250] It should be noted that for the technical details not described in detail in the embodiments of this unmanned aircraft operation risk assessment system, reference can be made to the unmanned aircraft operation risk assessment method provided in any embodiment of the present invention as described above, and details will not be elaborated here.

[0251] Based on the same inventive concept, an embodiment of the present invention further provides an unmanned aircraft operation risk assessment device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the unmanned aircraft operation risk assessment method as described above are implemented.

[0252] In some embodiments, the processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor is generally used to control the overall operation of the electronic device. In this embodiment, the processor is used to run the program code stored in the memory or process data, such as the program code of the vehicle collision warning method.

[0253] The memory includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory may be an internal storage unit of the electronic device, such as the hard disk or memory of the electronic device. In other embodiments, the memory may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device. Of course, the memory may also include both the internal storage unit and the external storage device of the electronic device. In this embodiment, the memory is generally used to store the operation methods and various application software installed on the electronic device, such as the program code of the vehicle collision warning method. In addition, the memory may also be used to temporarily store various data that have been output or will be output.

[0254] Based on the same inventive concept, the present invention also provides a readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the steps of the XXX method as described above are implemented.

[0255] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories and optical memories, etc.) that contain computer-usable program code.

[0256] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0257] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0258] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0259] It can be understood that the above embodiments are merely exemplary embodiments adopted to illustrate the principles of the present invention, and the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered within the protection scope of the present invention.

Claims

1. A method for assessing the risk of unmanned aircraft operations, characterized in that: The steps include: Divide the risk environment into multiple samples; Calculate the drone ground collision risk value for each sample; Calculate the drone mid-air collision risk of each sample, where the drone mid-air collision risk includes the drone-drone collision risk and the drone-bird flock collision risk; Calculate the risk of drone mid-air failure for each sample; The distribution of drone operation risk levels is determined based on the calculated drone ground collision risk value, drone mid-air collision risk, and drone mid-air failure risk of each sample.

2. The unmanned aircraft operation risk assessment method according to claim 1, characterized in that: The UAV ground collision risk value is calculated based on the number of casualties and the degree of casualties in ground collision accidents per flight hour. The specific calculation formula is as follows: GRV=E k ·N Among them, GRV is the UAV ground collision risk value, E k is the degree of casualties, and N is the number of casualties in ground impact accidents per flight hour.

3. The unmanned aircraft operation risk assessment method according to claim 2, characterized in that: The calculation formula for the number of casualties in ground impact accidents per flight hour is as follows: Where A is the area affected by the accident, ρ is the population density of the area, and P sys is the probability of ground collision accidents caused by UAV air accidents per flight hour, E PF is the ground shelter protection factor, R is the casualty rate; The degree of casualties is determined by the kinetic energy corresponding to 1.2 times the maximum design speed of the drone. The specific formula is as follows: Where m is the mass of the drone, V op The maximum design speed of the drone.

4. The unmanned aircraft operation risk assessment method according to claim 1, characterized in that: The UAV-UAV collision risk includes the collision risk of two UAVs in parallel directions and the collision risk of two UAVs in intersecting directions; And / or, the risk of collision between the drone and the flock of birds includes the risk of collision between the drone and the birds in parallel headings and the risk of collision between the drone and the birds in cross headings.

5. The unmanned aircraft operation risk assessment method according to claim 4, characterized in that: The calculation formula for the collision risk of two UAVs in parallel directions is as follows: Among them, CN is the collision risk of two UAVs in parallel directions, P x (S x ') is the probability of longitudinal collision between two UAVs in parallel directions, P y (S y ') is the probability of a side collision between two UAVs in parallel directions, P z (S z ') is the probability of vertical collision between two UAVs in parallel directions, R is the collision template radius, v x 、v y 、v z They are the relative speeds of the two UAVs in the longitudinal, lateral and vertical directions on the parallel routes.

6. The unmanned aircraft operation risk assessment method according to claim 5, characterized in that: The probability of a longitudinal collision between two UAVs in parallel directions is: Among them, S x ' is the relative distance between the two drones in the longitudinal direction; The probability of a sideways collision between two UAVs in the parallel direction is: Among them, f y (y1) represents the probability density function of the navigation error based on lateral separation, δ is the proportion of serious errors, S y ' is the relative distance between the two UAVs in the lateral direction; In the above formula, Expressed as a general navigation error probability density function exhibiting an exponential distribution, It is expressed as the probability density function of severe navigation error, In the above two formulas, a1 is the probability density function parameter of general navigation error, and a2 is the probability density function parameter of severe navigation error. The probability of vertical collision between two UAVs in parallel directions is: Among them, R is the collision template radius, S z ' is the relative distance between the two UAVs in the vertical direction, and δ is the standard deviation of the distance between the two UAVs in the vertical direction.

7. The unmanned aircraft operation risk assessment method according to claim 4, characterized in that: The calculation formula for the collision risk of two UAVs with intersecting headings is as follows: Among them, CN2 is the risk of collision between two UAVs with intersecting headings, and P x '(S x ') is the probability of longitudinal collision between two UAVs in the crossing direction, P y '(S y ') is the probability of a lateral collision between two UAVs on a crossing heading, P z '(S z ') is the probability of vertical collision between two UAVs in the cross-direction, R is the collision template radius, v x '、v y '、v z ' are the relative speeds of the two UAVs in the longitudinal, lateral and vertical directions in the cross heading.

8. The unmanned aircraft operation risk assessment method according to claim 7, characterized in that: The probability of a longitudinal collision between two UAVs in a cross-direction is: Among them, S x ' is the relative distance between the two UAVs in the longitudinal direction, θ is the intersection angle between the headings of the two UAVs; The probability of a side collision between two UAVs on a cross-direction is: Among them, θ is the intersection angle between the headings of the two UAVs, S y ' is the relative distance between the two UAVs in the lateral direction, f y (y1) represents the probability density function of the navigation error based on lateral separation, δ is the proportion of serious errors; In the above formula, Expressed as a general navigation error probability density function exhibiting an exponential distribution, It is expressed as the probability density function of severe navigation error, In the above two formulas, a1 is the probability density function parameter of general navigation error, and a2 is the probability density function parameter of severe navigation error. The probability of vertical collision between two UAVs in intersecting directions is: Among them, R is the collision template radius, and δ is the standard deviation of the vertical distance between the two drones.

9. The unmanned aircraft operation risk assessment method according to claim 4, characterized in that: The risk of collision between a parallel-heading drone and a flying bird is: P(t)=P X (t)P Y (t)P Z (t) The risk of collision between the cross-heading drone and the bird is: Among them, Px(t) is the probability of a longitudinal collision between a drone and a bird, Py(t) is the probability of a lateral collision between a drone and a bird, and Pz(t) is the probability of a vertical collision between a drone and a bird.

10. The unmanned aircraft operation risk assessment method according to claim 1, characterized in that: The method for calculating the risk of drone mid-air failure is as follows: A failure assessment model is constructed using the Bayesian probability network. The failure assessment model analyzes the failure causes from three aspects: UAV system, operating environment and human factors, and calculates the failure probability.

11. An unmanned aircraft operation risk assessment system, characterized in that: include: Sample determination module; Used to divide the risk environment into multiple samples; The ground collision risk calculation module is used to calculate the ground collision risk value of each sample drone; An air collision risk calculation module is used to calculate the air collision risk of drones of each sample, wherein the air collision risk of drones includes the collision risk between drones and the collision risk between drones and bird flocks; The air failure risk calculation module is used to calculate the air failure risk of each sample drone; The risk level distribution determination module is used to determine the drone operation risk level distribution based on the calculated drone ground collision risk value, drone mid-air collision risk, and drone mid-air failure risk of each sample.

12. An unmanned aircraft operation risk assessment device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the unmanned aircraft operation risk assessment method as described in claims 1-10 when executing the computer program.

Images