A method and system for evaluating simulation flight data of a drone

Abstract

The present disclosure provides a UAV simulation flight data evaluation method and system, comprising: acquiring airflow field data in a simulation environment, and spatially layering and dividing the airflow field data to construct a layered airflow structure. The attitude change amount of the UAV at multiple sampling time points is collected, and the attitude change amount is associated with the layered airflow structure to generate an attitude disturbance response sequence. Based on the attitude disturbance response sequence, the attitude change amount of the UAV under different airflow levels is constrained to construct a flight constraint mapping sequence. According to the flight constraint mapping sequence, the UAV flight trajectory is coupled with the layered airflow structure for coupled analysis to extract the trajectory airflow coupling characteristics of the UAV flight trajectory in different airflow levels. The attitude instability section information is determined through the trajectory airflow coupling characteristics, and the UAV is evaluated for the risk of rolling over according to the attitude instability section information to output the simulation flight data evaluation result, thereby realizing accurate evaluation of flight risk.

Description

Technical Field

[0001] This disclosure pertains to the field of unmanned aerial vehicle (UAV) technology, and more specifically, relates to a method and system for evaluating UAV simulation flight data. Background Technology

[0002] In canyon-type reservoirs, the swirling airflow along the reservoir banks typically forms a vertical multi-layered airflow reversal structure, allowing UAVs to represent structural changes in a simulation environment.

[0003] Currently, in traditional technical solutions, the executability constraint mapping corresponding to the structural changes expressed by UAVs in the simulation environment is only based on a single flight constraint boundary, which makes it difficult to reflect the asymmetric disturbance of airflow on attitude control. Consequently, the flight data evaluation results are difficult to reflect the rollover risk of UAVs in real flight. Summary of the Invention

[0004] In canyon-type reservoirs, the swirling airflow along the reservoir banks forms a vertical multi-layered airflow reversal structure. Although the structural changes can be expressed in the UAV simulation environment, in traditional technical solutions, the executability constraint mapping is based only on a single flight constraint boundary, which cannot reflect the asymmetric disturbance of airflow on attitude control, resulting in the evaluation results failing to reflect the rollover risk in real flight.

[0005] The purpose of this invention is to overcome the above-mentioned defects and to propose a method and system for evaluating unmanned aerial vehicle (UAV) simulated flight data.

[0006] The present invention adopts the following technical solution.

[0007] The first aspect of this invention discloses a method for acquiring airflow field data in a simulation environment and spatially dividing the airflow field data to construct a layered airflow structure;

[0008] The attitude change of the UAV at multiple sampling times is collected, and the attitude change is correlated with the layered airflow structure to generate an attitude disturbance response sequence.

[0009] Based on the attitude disturbance response sequence, the attitude change of the UAV under different airflow levels is constrained to construct a flight constraint mapping sequence.

[0010] The UAV flight trajectory is coupled with the layered airflow structure based on the flight constraint mapping sequence to extract the trajectory-airflow coupling characteristics of the UAV flight trajectory in different airflow levels;

[0011] The attitude instability section information is determined by the trajectory airflow coupling characteristics, and the rollover risk of the UAV is assessed based on the attitude instability section information to output the simulation flight data assessment results.

[0012] Furthermore, the step of acquiring airflow field data in the simulation environment and spatially dividing the airflow field data to construct a layered airflow structure includes:

[0013] The airflow field data is time-aligned and then discretely layered according to a preset initial layer thickness to obtain multiple airflow levels.

[0014] Obtain the airflow direction and average airflow intensity corresponding to each airflow level, and calculate the airflow direction change characteristics and airflow intensity change characteristics between adjacent airflow levels based on the airflow direction and average airflow intensity;

[0015] When the change characteristics of airflow direction or airflow intensity exceed a preset change threshold, the boundary area between adjacent airflow levels is marked as an airflow abrupt change area.

[0016] The airflow level is obtained by correcting the initial layer thickness.

[0017] Furthermore, the step of acquiring airflow field data in the simulation environment and spatially dividing the airflow field data to construct a layered airflow structure further includes:

[0018] Establish local vertical windows for adjacent airflow levels, and statistically analyze the distribution density of the airflow abrupt change region within the local vertical windows to determine the swirling trend characteristics of the corresponding airflow level based on the distribution density.

[0019] The swirling trend characteristics are written into the corresponding airflow levels, and each airflow level is encoded in order of height to construct the layered airflow structure.

[0020] Furthermore, the method involves collecting the attitude changes of the UAV at multiple sampling times and correlating these attitude changes with the stratified airflow structure to generate an attitude disturbance response sequence, including:

[0021] The flight attitude and altitude of the UAV are measured at a uniform sampling time, and the flight attitude corresponding to each sampling time is written into the corresponding airflow level according to the flight altitude to determine the level dwell section.

[0022] Based on the flight attitude of the UAV at adjacent sampling times, the instantaneous rate of change of attitude for different attitude channels is calculated, and the instantaneous rate of change of attitude is smoothed according to a preset sliding time window to obtain the disturbance intensity for each attitude channel.

[0023] The disturbance intensity is statistically analyzed within the hierarchical residence section, and the attitude disturbance response of the corresponding airflow level of the hierarchical residence section to each attitude channel is determined in combination with the hierarchical airflow structure, so as to generate the attitude disturbance response sequence based on the attitude disturbance response.

[0024] Furthermore, the constraint on the attitude change of the UAV under different airflow levels based on the attitude perturbation response sequence to construct a flight constraint mapping sequence includes:

[0025] The attitude control boundary of the UAV is obtained, and the attitude control boundary of the UAV is matched with the attitude disturbance response at each time in the attitude disturbance response sequence to determine the UAV control instability section.

[0026] The attitude change of the UAV within the aforementioned UAV control instability zone is statistically analyzed, and the instability trend of the reversal region is determined based on the attitude change.

[0027] The instability trend of the inverted region is transformed into UAV flight constraints, and a mapping relationship between the UAV flight constraints and the corresponding airflow level is established to obtain the flight constraint mapping sequence.

[0028] Furthermore, the step of coupling the UAV flight trajectory with the layered airflow structure based on the flight constraint mapping sequence to extract the trajectory-airflow coupling features of the UAV flight trajectory at different airflow levels includes:

[0029] The flight trajectory of the UAV is time-aligned with the layered airflow structure to statistically analyze trajectory-level dwell segments and trajectory cross-level events, and these are then converted into layer-level dwell characteristics and cross-level frequency characteristics.

[0030] When the UAV flight trajectory is inconsistent with the stratified airflow structure, the attitude difference result is determined, and the attitude difference result is associated and matched with the stratified airflow structure and flight constraint mapping sequence to generate the trajectory airflow coupling feature by combining the stratified dwell characteristics and cross-layer frequency characteristics.

[0031] Furthermore, the step of determining attitude instability section information through the trajectory airflow coupling characteristics, and assessing the rollover risk of the UAV based on the attitude instability section information to output simulation flight data evaluation results includes:

[0032] The trajectory airflow coupling features are sliced ​​according to a unified time window, and combined with the flight constraint mapping sequence to construct a risk basis representation aligned with the time window.

[0033] The risk baseline characterization of adjacent time windows is continuously accumulated to determine the rollover risk value of the time window, and high-risk sections are determined based on the rollover risk value of the time window;

[0034] When the high-risk section meets the preset instability triggering conditions, the high-risk section is determined to be an instability triggering section, and the source of airflow action corresponding to the instability triggering section is obtained, and the rollover risk level of each time window corresponding to the instability triggering section is divided.

[0035] The high-risk section is the section where the rollover risk value exceeds a preset risk threshold for multiple consecutive time windows; the simulated flight data evaluation result is composed of the rollover risk level, the instability triggering section, the source of airflow, and the corresponding airflow level.

[0036] The second aspect of this invention discloses a UAV simulation flight data evaluation system for implementing the UAV simulation flight data evaluation method described in any one of the first aspects, the system comprising:

[0037] The airflow structure layering module is used to acquire airflow field data in the simulation environment and spatially divide the airflow field data to construct a layered airflow structure.

[0038] The attitude disturbance correlation module is used to collect the attitude change of the UAV at multiple sampling times and correlate the attitude change with the layered airflow structure to generate an attitude disturbance response sequence.

[0039] The flight constraint mapping module is used to constrain the attitude change of the UAV under different airflow levels based on the attitude disturbance response sequence, so as to construct the flight constraint mapping sequence.

[0040] The feature coupling analysis module is used to perform coupling analysis between the UAV flight trajectory and the layered airflow structure according to the flight constraint mapping sequence, so as to extract the trajectory airflow coupling features of the UAV flight trajectory in different airflow levels;

[0041] The flight data evaluation module is used to determine the attitude instability section information through the trajectory airflow coupling characteristics, and to perform a rollover risk assessment on the UAV based on the attitude instability section information, so as to output the simulation flight data evaluation results.

[0042] A third aspect of the present invention discloses a terminal, including a processor and a storage medium;

[0043] The storage medium is used to store instructions;

[0044] The processor is configured to operate according to the instructions to perform the steps of the method described in the first aspect.

[0045] A fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0046] The beneficial effects of the present invention are as follows: Compared with the prior art, the present invention has the following advantages:

[0047] (1) This invention spatially divides the airflow field data in the simulation environment into several airflow levels according to the height direction, and extracts the airflow direction change characteristics, airflow intensity change characteristics, and swirling trend characteristics in each level unit to form a set of vertically layered airflow structures. Then, the attitude change of the UAV at each sampling time is associated with the airflow level, and the response of different levels of airflow to attitude roll change, pitch change, and yaw change is extracted to construct a set of attitude response abrupt change segments. Based on the set of attitude response abrupt change segments, the boundary and stability constraints of the UAV attitude control capability are introduced, and the attitude change amplitude and controllable range under the action of different airflow levels are compared and analyzed. The attitude change segments that exceed the stable control range are marked to form a flight executability constraint mapping sequence under the action of asymmetric airflow, and the control instability trend in the cross-layer airflow reversal region is enhanced. This can effectively reflect the asymmetric disturbance of airflow on attitude control, so that the final flight data evaluation results can reflect the rollover risk of the UAV in the real flight process.

[0048] (2) Based on the flight executability constraint mapping sequence, this invention couples the positional changes of the UAV flight trajectory in vertical space with the stratified airflow structure, extracting the residence time, cross-layer frequency, and attitude change continuity characteristics of the trajectory in different airflow levels. Furthermore, by performing consistency judgment on the coupling relationship between the trajectory and the airflow structure, a vertical airflow structure coupling evaluation result is generated to reflect the degree of matching between flight behavior and airflow distribution. Finally, combining the attitude instability section information under asymmetric airflow, the rollover risk at each time period during flight is graded and evaluated, outputting a simulated flight data evaluation result including the rollover risk level, instability triggering section, and airflow source, further realizing an accurate assessment of the actual flight risk of UAVs in the strong swirling airflow area of ​​the canyon reservoir. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating a method for evaluating simulated flight data of unmanned aerial vehicles (UAVs) provided by the present invention.

[0050] Figure 2 This is a schematic diagram of the structure of a drone simulation flight data evaluation system provided by the present invention. Detailed Implementation

[0051] The present application will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, and should not be construed as limiting the scope of protection of the present application.

[0052] like Figure 1 As shown, in one embodiment, a method for evaluating drone simulation flight data includes the following steps:

[0053] Step S110: Obtain airflow field data in the simulation environment and spatially divide the airflow field data to construct a layered airflow structure.

[0054] In some embodiments, the UAV simulation flight data evaluation method provided by the present invention includes the following steps in step S110:

[0055] Step S111: Time alignment processing is performed on the airflow field data, and the airflow field data is discretized and layered according to the preset initial layer thickness to obtain multiple airflow layers.

[0056] The airflow level is obtained by correcting the initial layer thickness.

[0057] Step S112: Obtain the airflow direction and average airflow intensity corresponding to each airflow level, and calculate the airflow direction change characteristics and airflow intensity change characteristics between adjacent airflow levels based on the airflow direction and average airflow intensity.

[0058] Step S113: When the change characteristics of airflow direction or airflow intensity exceed the preset change threshold, the boundary area between the corresponding adjacent airflow levels is marked as the airflow abrupt change area.

[0059] The threshold for directional change can be set to 80-85 degrees to identify significant wind direction deflection; the threshold for intensity change can be set to 0.65-0.70 to identify areas of sudden aerodynamic increase.

[0060] In some embodiments, the UAV simulation flight data evaluation method provided by the present invention further includes the following steps in step S110:

[0061] Step S114: Establish local vertical windows for adjacent airflow levels, and statistically analyze the distribution density of abrupt airflow change regions within the local vertical windows to determine the swirling trend characteristics of the corresponding airflow levels based on the distribution density.

[0062] Step S115: Write the swirling trend characteristics into the corresponding airflow level, and encode each airflow level in order of height to construct a layered airflow structure.

[0063] In a specific embodiment, the present invention provides a method for evaluating unmanned aerial vehicle (UAV) simulation flight data, comprising steps 1 to 5:

[0064] Step 1: Construct the expression sequence of vertical stratified airflow structure of the canyon reservoir.

[0065] The airflow field data in the simulation environment is spatially layered, dividing the airflow into several airflow level units according to the height direction. The airflow direction change characteristics, airflow intensity change characteristics, and swirling trend characteristics of each level unit are extracted to form a sequence of layered airflow structure representations. Finally, the airflow reversal relationships between different levels are encoded to form a set of vertically layered airflow structures, including the following sub-steps:

[0066] Sub-step 1.1: Preprocessing of airflow field data.

[0067] Specifically, firstly, the airflow field data in the simulation environment is uniformly aligned at all times. This airflow field data includes at least the three-dimensional position, altitude, airflow velocity, and horizontal angle of arrival of each spatial sampling point at each sampling time. Then, using the lowest and highest flight altitudes in the normal flight airspace of the canyon reservoir as the vertical stratification range, discrete stratification is performed along the altitude direction. Unified thickness division is not used during stratification because the backflow zone near the reservoir surface, the bank-side lifting zone, and the upper-middle swirling zone often have significantly different vertical thicknesses. Therefore, a method of base layer thickness plus local subdivision correction is adopted. That is, a base layer thickness is first set, and then the altitude intervals with large airflow gradients are further subdivided. The base layer thickness is 2-8 meters. When the average wind speed change rate of adjacent altitude intervals exceeds 0.3 meters per second per meter, or the horizontal wind direction change rate exceeds 8 degrees per meter, the interval is further subdivided into sub-layer units of 1-3 meters in thickness.

[0068] After completing the above stratification, the sampling points within each stratum unit are aggregated, and the average airflow intensity, dominant direction, and local dispersion of that stratum at the current sampling time are calculated, thereby forming the basic airflow statistics within the stratum. Simultaneously, a stratum boundary height sequence is output, which is used to construct the stratum directional variation characteristics.

[0069] Sub-step 1.2, hierarchical mutation marker.

[0070] Specifically, after completing the hierarchical division, it is necessary to further transform the airflow characteristics of each layer into the extent of variation between adjacent layers. During this transformation, based on the dominant direction and average airflow intensity of each layer, the degree of directional and intensity variation between adjacent layers is calculated layer by layer along the height direction. In this example, simple difference cannot be used alone, because the strong swirling airflow in canyon-type reservoirs often exhibits abrupt changes in direction but not necessarily a synchronous change in intensity, or it may exhibit a sharp increase in intensity but a slow deflection in direction. Therefore, it is necessary to construct directional variation characteristics and intensity variation characteristics separately.

[0071] In this embodiment, the dominant direction is obtained by weighted circular statistical analysis of the horizontal wind direction angles of all sampling points within the layer, with the weights positively correlated with the airflow velocity at the sampling points. A greater directional change characteristic indicates a higher likelihood of attitude control disturbance switching zones forming between that layer and adjacent layers; a greater intensity change characteristic indicates a higher likelihood of aerodynamic surge zones forming at the interface of that layer. Subsequently, vertical abrupt change zones are marked based on the joint thresholds of the directional and intensity change characteristics. The directional abrupt change threshold is set between 30 and 90 degrees, and the intensity abrupt change threshold is a dimensionless value between 0.25 and 1.

[0072] The expression for the directional change characteristic is as follows:

[0073]

[0074] In the formula, Indicates the first Layer and First The directional variation characteristics between layers are dimensionless. Indicates the first The dominant horizontal wind direction angle at each level, in radians; Indicates the first The dominant horizontal wind direction angle at each level, in radians; and These represent the average airflow intensity of two adjacent layers, in meters per second. This represents the normalized velocity constant for directional change, expressed in meters per second, with a range of 0.5-3. This formula couples the degree of angular deflection with the intensity of the airflow involved in the deflection. If the directional reversal is significant and the wind speeds at the upper and lower layers are high, this value will rise rapidly, providing a more realistic characterization of the intensity of interlayer lateral disturbance switching.

[0075] The characteristic expression for intensity variation between adjacent layers is:

[0076]

[0077] In the formula, Indicates the first Layer and First The intensity variation characteristics between layers are dimensionless. and These represent the average airflow intensity of two adjacent layers, in meters per second. This represents the intensity change stabilization constant, expressed in meters per second, with a value ranging from 0.1 to 0.5. This formula uses the inter-layer average wind speed as a reference to normalize the absolute intensity difference, making inter-layer abrupt changes comparable under strong and weak wind conditions.

[0078] Sub-step 1.3 transforms the directional change characteristics and intensity change characteristics into cyclonic trend characteristics.

[0079] Specifically, taking each layer as the center, several adjacent layers above and below it form a local vertical window. Within this window, the cumulative degree of directional change, the concentration of intensity change, and the distribution density of abrupt change regions are statistically analyzed. If a layer exhibits continuous directional deflection, enhanced intensity shear, and multiple abrupt change regions simultaneously, it indicates that the region containing that layer is not a typical transition layer but a cyclotron-dominant layer. The number of layers in the window should be 3-7; too few layers will lose cyclotron continuity, while too many layers will dilute local cyclotron flow.

[0080] To avoid misclassifying an entire stratum as a swirling-dominant zone due to a momentary spike, a temporal continuity check of the swirling trend characteristics is necessary. That is, only when the same stratum maintains a high swirling trend over 3-10 consecutive sampling times is it included in the set of vertically dominant swirling zones. For example, within a local vertical window, the dominant effects of directional deflection, intensity shear amplification, and abrupt change suppression or enhancement are unified to identify strata with genuine swirling characteristics, rather than ordinary transition zones with only slight wind direction changes.

[0081] Sub-step 1.4 outputs a representation of the layered airflow structure.

[0082] Specifically, after obtaining the swirling trend characteristics, the entire vertical airflow structure needs to be transformed into a standardized representation for subsequent attitude perturbation analysis. During this transformation, four core elements are recorded for each layer: layer height range, layer average airflow intensity, layer dominant direction, and layer swirling trend level. Simultaneously, relationships are encoded regarding whether there are opposing dominant airflows between adjacent layers, whether they are at abrupt transition boundaries, and whether they are located within a swirling dominant region.

[0083] In this embodiment, the airflow reversal relationship does not simply refer to two layers with directions differing by 180 degrees. Rather, it means that the dominant directions of adjacent layers are opposite, the directional change characteristics are high, and at least one layer is in the swirling dominant segment. Only when these conditions are met is the inter-layer relationship marked as a valid reversal relationship. Subsequently, all layer expression items are concatenated in height order to obtain the stratified airflow structure expression sequence. Finally, all valid reversal relationships are associated and encapsulated with the corresponding layer entries to obtain the vertical stratified airflow structure set.

[0084] In addition, the swirling trend level is divided into four levels: low, lower, higher, and high. The threshold for determining the effective reversal relationship should be calibrated in conjunction with historical flight tests. The threshold for directional change characteristics is 0.3-1.5, and the threshold for swirling trend is 0.5-3.

[0085] Step S120: Collect the attitude change of the UAV at multiple sampling times, and correlate the attitude change with the stratified airflow structure to generate an attitude disturbance response sequence.

[0086] In some embodiments, the UAV simulation flight data evaluation method provided by the present invention includes the following steps in step S120:

[0087] Step S121: Measure the flight attitude and altitude of the UAV at a uniform sampling time, and write the flight attitude corresponding to each sampling time into the corresponding airflow level according to the flight altitude to determine the level residence section.

[0088] Step S122: Based on the flight attitude of the UAV at adjacent sampling times, calculate the instantaneous change rate of attitude for different attitude channels, and smooth the instantaneous change rate of attitude according to a preset sliding time window to obtain the disturbance intensity for each attitude channel.

[0089] Step S123: Statistically measure the disturbance intensity within the hierarchical residence section, and determine the attitude disturbance response of each attitude channel to the corresponding airflow level of the hierarchical residence section in combination with the hierarchical airflow structure, so as to generate an attitude disturbance response sequence based on the attitude disturbance response.

[0090] In a specific embodiment, the present invention provides a method for evaluating UAV simulated flight data. Step 2 involves constructing an attitude disturbance response sequence under the influence of stratified airflow. Based on the stratified airflow structure expression sequence obtained in Step 1, the attitude change of the UAV at each sampling time is correlated with the airflow level, and the responses of different airflow levels to attitude roll, pitch, and yaw changes are extracted to form an attitude disturbance response sequence. Simultaneously, the airflow reversal regions between adjacent levels are marked to construct a set of attitude response abrupt change segments, including the following sub-steps:

[0091] Sub-step 2.1: Construct the set of attitude hierarchy association results and the sequence of hierarchy dwelling segments.

[0092] Specifically, firstly, using a unified sampling time sequence as a benchmark, the UAV attitude measurement sequence and altitude sequence are time-aligned. The attitude measurement sequence includes roll, pitch, and yaw angles, while the altitude sequence determines which airflow level unit the UAV falls into at the current sampling time. Then, the level boundary altitude information generated in step 1 is used to compare the UAV altitude at each sampling time with the upper and lower boundaries of the corresponding level, completing the point-to-point assignment of time to level. If the UAV crosses a level boundary between two adjacent sampling times, this time period is marked as a cross-level transition segment; if multiple consecutive sampling times fall within the same level, they are merged to form a level dwelling segment.

[0093] In this embodiment, to avoid frequent hierarchical jumps caused by minor altitude fluctuations, boundary buffer zones with a thickness of 0.5-2 meters are set on both sides of the hierarchical boundary. When the UAV altitude fluctuates only within the buffer zone, it retains the hierarchical affiliation from the previous sampling moment. Only when the altitude crosses the buffer zone for 2-5 consecutive sampling moments is it determined to have entered a new hierarchical level. This effectively avoids pseudo-cross-level events caused by boundary jitter. Finally, the set of attitude hierarchical association results and the sequence of hierarchical dwell segments are output.

[0094] Sub-step 2.2: Extract the attitude change at each sampling time from the original attitude measurement values.

[0095] Specifically, firstly, based on the differences in roll, pitch, and yaw angles between adjacent sampling times, the instantaneous rate of change for each attitude channel is calculated. Then, a sliding time window is used to locally smooth the rate of change of attitude, suppressing high-frequency, minor fluctuations caused by normal fine-tuning of the control system, thereby highlighting the attitude changes truly caused by airflow disturbances. The sliding time window length is 0.3-1 second, corresponding to 3-10 sampling points. If the UAV is large or has strong inertia, it needs to be appropriately extended to 1.5 seconds.

[0096] Special attention needs to be paid to the angle rotation issue when calculating yaw angle. If the yaw angle is directly subtracted from the 360° to 0° boundary, false large changes may occur. Therefore, the yaw angle difference should first be constrained to the radian range corresponding to -180° to 180° before being included in the rate of change calculation.

[0097] In this embodiment, the expression for the roll change rate is:

[0098]

[0099] In the formula, Indicates the sampling time The weighted value of the rate of change of roll, in radians per second; Indicates the sampling time The roll angle, in radians; Represents the roll angle at the previous sampling time, in radians; This represents the time interval between adjacent sampling moments, in seconds, with a value range of 0.05-0.2 seconds. This represents the roll change suppression constant, expressed in radians, with a value range of 0.03-0.2. While maintaining the dimensional accuracy of the roll change rate, this formula preserves excessively small changes and softly suppresses instantaneous spikes, making the results more closely approximate the effective attitude changes caused by airflow disturbances.

[0100] The expression for the intensity of attitude change is:

[0101]

[0102] In the formula, Indicates the sampling time The intensity of attitude change, expressed in radians per second; This represents the weighted value of the roll rate of change, in radians per second. This represents the weighted value of the pitch rate of change, in radians per second. The calculation method is the same as that of the roll rate of change, except that the roll angle is replaced by the pitch angle. This represents the weighted value of the yaw rate of change, expressed in radians per second. Its calculation method is the same as the roll rate of change, except that the roll angle is replaced with the yaw angle after cycle-changing processing. This formula unifies the changes in the three attitude channels into a comparable disturbance intensity quantity, used for subsequent identification of the combined effect of stratified airflow on attitude.

[0103] Sub-step 2.3: Extract the response of airflow levels to the three types of attitudes.

[0104] Specifically, taking the sub-level residence section as the basic analysis unit, within each residence section, the average intensity, peak intensity, and cumulative change of roll change, pitch change, and yaw change are statistically analyzed; then, combined with the average airflow intensity, direction change characteristics, and swirling trend level of the corresponding level, the response of that level to the three attitudes is extracted.

[0105] In this embodiment, roll changes better reflect lateral airflow and swirling thrust, pitch changes better reflect updrafts and changes in angle of attack, and yaw changes more readily reflect the traction effect of horizontal wind direction switching on the aircraft's heading. Therefore, the responses to these three types of effects need to be modeled separately and cannot be directly replaced by a single index. The response analysis window length needs to be set to the full length of the corresponding dwell segment; if a dwell segment is too long, it should be segmented and statistically analyzed in units of 1-3 seconds.

[0106] Subsequently, the response results of the same level in multiple dwelling sections are summarized to form the roll action response sequence, pitch action response sequence and yaw action response sequence of that level; then the three are spliced ​​together in chronological order to obtain the attitude disturbance response sequence.

[0107] In this embodiment, the expression for the roll action response value of the dwell section is:

[0108]

[0109] In the formula, Indicates the first The roll response value of each level of residence section, in radians per second; This indicates the duration of the stay, in seconds; Indicates the time within the stationing section. The weighted value of the rate of change of roll, in radians per second; In step 1, the first The cyclical trend characteristics corresponding to each level are dimensionless. This represents the intensity variation characteristics between adjacent layers, and is dimensionless. If there are adjacent layers both above and below this layer, the larger value or a weighted average is taken. This formula first extracts the basic intensity of the roll disturbance using a time-averaged method, and then amplifies it using the cyclotron trend and interlayer intensity shear, thereby distinguishing between ordinary attitude oscillations and continuous roll disturbances caused by cyclotron layers.

[0110] Sub-step 2.4: Determine the abrupt change segment in attitude response.

[0111] Specifically, key areas of airflow reversal between adjacent layers are marked to identify truly risky attitude response abrupt change segments. First, the effective reversal relationships from step 1 are used to locate all flight time periods spanning airflow reversal layers. Then, near these time periods, the attitude disturbance response sequence is checked to see if at least one channel of roll, pitch, or yaw increases rapidly, and this increase is not a single-point spike, but rather maintains an upward trend or high-level fluctuation over several consecutive sampling moments. If the above conditions are met, the segment is marked as an attitude response abrupt change segment.

[0112] To avoid misjudging normal turns, climbs, and other active maneuvers as abrupt events, maneuver rejection constraints need to be added. For example, when there are continuous and stable active turn or climb commands in the flight control instructions, and the direction of attitude change is consistent with the direction of the control command, the priority of abrupt event judgment in that segment is reduced. The threshold for abrupt event duration is generally set at 0.2-1 seconds, and the threshold for attitude change intensity is set at 1.5-3 times the average value of normal stable flight.

[0113] In this embodiment, the expression for the mutation determination value of the candidate segment is:

[0114]

[0115] In the formula, Indicates the first The mutation determination value of each candidate segment, in radians per second; Indicates the first The duration of each candidate segment, in seconds; Indicates the first Time within each candidate segment The intensity of attitude change, expressed in radians per second; Indicates the first The enhancement coefficient for each candidate segment is located near an effective inversion relation. It is dimensionless, and is zero if located between ordinary layers and 0.3-2 if located between inversion layers. Indicates the first The number of sampling points exceeding the basic threshold within each candidate segment, with a value that is a positive integer; This represents the normalization constant for the abrupt change density, with a value ranging from 3 to 20. This formula simultaneously considers the average disturbance intensity of the candidate segment, whether it is located between inversion layers, and the concentration of high-disturbance points. Only segments that simultaneously satisfy these three characteristics are more likely to be genuine attitude abrupt changes caused by stratified airflow reversal, rather than occasional jitter.

[0116] Step S130: Based on the attitude disturbance response sequence, constrain the attitude change of the UAV under different airflow levels to construct a flight constraint mapping sequence.

[0117] In some embodiments, the UAV simulation flight data evaluation method provided by the present invention includes the following steps in step S130:

[0118] Step S131: Obtain the UAV attitude control boundary and match the UAV attitude control boundary with the attitude disturbance response at each time point in the attitude disturbance response sequence to determine the UAV control instability section.

[0119] Step S132: Calculate the attitude change of the UAV within the unstable control zone, and determine the instability trend of the reversal region based on the attitude change.

[0120] Step S133: Transform the instability trend of the reverse region into UAV flight constraints, and establish a mapping relationship between UAV flight constraints and corresponding airflow levels to obtain a flight constraint mapping sequence.

[0121] In a specific embodiment, the present invention provides a method for evaluating UAV simulated flight data. Step 3 involves constructing a flight executability constraint mapping sequence under asymmetric airflow. Based on the attitude disturbance response sequence output in step 2, the boundary and stability constraints of the UAV attitude control capability are introduced. The attitude change amplitude and controllable range under different airflow levels are compared and analyzed. Attitude change segments exceeding the stable control range are marked, forming a flight executability constraint mapping sequence under asymmetric airflow. The control instability trend in the cross-layer airflow reversal region is further enhanced. Step 3 finally takes the attitude disturbance response sequence as input and outputs the flight executability constraint mapping sequence, including the following sub-steps:

[0122] Sub-step 3.1, attitude control capability boundary matching.

[0123] Specifically, firstly, the attitude disturbance response sequence output in step 2 is expanded chronologically, and the roll response, pitch response, yaw response, and corresponding level number at each sampling moment are read point by point. Then, the UAV's own attitude control capability boundary is introduced. This boundary is not a single fixed angle limit, but should include at least three categories: the first is the upper limit of the allowable rate of attitude change; the second is the upper limit of the allowable attitude recovery time; and the third is the reduction value of the control margin under the current flight weight, current speed, and current altitude conditions. These three types of parameters can be provided by the actual flight control calibration results, flight test envelope, or the manufacturer's control manual.

[0124] In this embodiment, the control load ratios for the roll, pitch, and yaw channels are calculated at each sampling time to indicate how much control capability is occupied by the current attitude disturbance. If the ratio is close to 1, it indicates that the control capability is nearing saturation; if it consistently exceeds 1, it indicates that the attitude disturbance has exceeded the stable control range. Since the lateral recirculation airflow in the strong swirling airflow region of the canyon reservoir typically has a stronger impact on the roll channel than the pitch channel, the roll channel should be allowed a higher weight during boundary matching. During weight calibration, the roll channel weight is set to 1.2-2, the pitch channel weight to 0.8-1.5, and the yaw channel weight to 0.6-1.2.

[0125] The expression for the control load ratio is:

[0126]

[0127] In the formula, Indicates the sampling time The control load ratio is dimensionless. Indicates the sampling time The roll response, measured in radians per second; Indicates the sampling time The pitch response, in radians per second; Indicates the sampling time Yaw response, in radians per second This indicates the upper limit of the allowable rate of attitude change for the roll channel, in radians per second, with a value range of 0.5-3. This indicates the upper limit of the allowable rate of attitude change for the pitch channel, in radians per second, with a value range of 0.4-2.5. This indicates the upper limit of the allowable rate of attitude change in the yaw channel, in radians per second, with a value range of 0.3-2. , , These represent the weighting coefficients of the three channels, all of which are dimensionless constants.

[0128] Sub-step 3.2: Construct the attitude stability margin sequence and the candidate unstable segment sequence.

[0129] Specifically, after obtaining the control load ratio, it is necessary to further calculate how much stability control margin remains at each sampling time. In this example, a simple numerical subtraction of the load ratio from 1 is not performed because the impact of control load on flight stability has a time-cumulative effect. That is, a short-term over-limit may not lead to immediate instability, but sustained high load will cause the aircraft to gradually lose its attitude recovery capability. Therefore, a time-cumulative penalty term is introduced when calculating the stability margin.

[0130] First, the control load ratio is accumulated over time using a sliding window with a length of 0.5 to 2 seconds. If the proportion of high load points within the window is too large, the stability margin at the current moment is further reduced, thereby distinguishing between single-point strong disturbances and persistent strong disturbances. Subsequently, continuous time periods with stability margins lower than the stability margin threshold are extracted to form a candidate instability segment sequence. This stability margin threshold is set to 0.1-0.4; if the model is lighter and has weaker disturbance rejection capability, a higher threshold needs to be used.

[0131] In this embodiment, the expression for attitude stability margin is:

[0132]

[0133] In the formula, Indicates the sampling time The attitude stability margin is dimensionless. Indicates the sampling time The control load ratio is dimensionless. This indicates the length of the sliding time window, in seconds, with a value range of 0.5-2. Indicates the time within the time window The control load ratio is dimensionless. This formula first uses the control load at the current moment to obtain the instantaneous margin, and then subtracts it from the cumulative load within the time window, thus reflecting the erosive effect of continuous disturbances on stable control capabilities.

[0134] Sub-step 3.3: Construct a reversal region to reinforce the unstable trend sequence.

[0135] Specifically, because the dominant airflow directions in the upper and lower layers of the inversion region within the canyon reservoir are opposite and the swirling trend is relatively high, the actual risk of instability is higher even if the instantaneous attitude change amplitude is similar to that of ordinary disturbance regions. Therefore, it is necessary to strengthen the representation of candidate instability segments in the inversion region based on the stability margin.

[0136] First, the valid reversal relationship formed in step 1 is invoked and compared segment by segment with the candidate unstable segments to determine whether each candidate segment is located near the reversal interlayer. Then, the number of cross-layers, the intensity of inter-layer directional change, and the swirling trend level within the candidate segment are statistically analyzed, and a reversal enhancement coefficient is constructed accordingly. A larger enhancement coefficient indicates that the attitude disturbance in that segment is more likely caused by asymmetric airflow switching rather than ordinary flight adjustments. The threshold for the number of cross-layers is set to 1-3 times, the intensity of directional change refers to the directional change characteristics in step 1, and the swirling trend level can be mapped to four levels (low, lower, higher, high) with enhancement values ​​ranging from 0.2 to 2.

[0137] In this embodiment, the expression for the inversion region strengthening instability trend value of the candidate instability segment is:

[0138]

[0139] In the formula, Indicates the first The reversal region of each candidate instability segment reinforces the instability trend value, which is dimensionless; Indicates the first The number of sampling points within each candidate instability zone ranges from 3 to 100. Indicates the first [unit] within this segment The attitude stability margin at each sampling point is dimensionless. This represents the stabilization constant, which is dimensionless and ranges from 0.05 to 0.2. Indicates the first The inter-layer directional variation enhancement coefficient corresponding to each candidate segment is dimensionless and ranges from 0 to 3. Indicates the first The cyclonic trend enhancement coefficient corresponding to each candidate segment is dimensionless and ranges from 0 to 2.

[0140] Sub-step 3.4 outputs the flight executability constraint mapping sequence.

[0141] Specifically, the aforementioned results are uniformly converted into a flight executability constraint mapping sequence for step 4. That is, for each sampling time or each candidate segment, the following are included: current level number, roll load state, pitch load state, yaw load state, stability margin level, whether it is in the inversion region, instability trend level, and executability judgment result.

[0142] In this embodiment, the executability determination results can be divided into three categories: the first category is stable executability, corresponding to a control load ratio significantly lower than the upper limit and sufficient stability margin; the second category is critically executable, corresponding to a control load close to the upper limit or, although not exceeding the limit, located in the strong reversal region; the third category is unstable executability, corresponding to a stability margin that is consistently too low and has a significant tendency to instability. To make the mapping results applicable to subsequent trajectory coupling analysis, each result segment, along with its corresponding time start and end range and altitude start and end range, needs to be output together to form a spatiotemporally continuous executability constraint expression.

[0143] Step S140: Based on the flight constraint mapping sequence, the UAV flight trajectory is coupled with the stratified airflow structure to extract the trajectory-airflow coupling characteristics of the UAV flight trajectory in different airflow levels.

[0144] In some embodiments, the UAV simulation flight data evaluation method provided by the present invention includes the following steps in step S140:

[0145] Step S141: Perform time alignment processing on the UAV flight trajectory and the stratified airflow structure to statistically analyze trajectory stratified dwell segments and trajectory cross-stratification events, and transform the trajectory stratified dwell segments and trajectory cross-stratification events into stratified dwell features and cross-stratification frequency features.

[0146] Step S142: When the UAV flight trajectory is inconsistent with the stratified airflow structure, determine the attitude difference result, and associate and match the attitude difference result with the stratified airflow structure and flight constraint mapping sequence to generate trajectory airflow coupling features by combining stratified dwell features and cross-layer frequency features.

[0147] Generally, the continuous attitude change trend can be aligned with the corresponding airflow level and the executable constraint state on a time-by-time basis. Within the level dwelling section, it can be determined whether the attitude change is consistent with the swirling trend, and corrections can be made in combination with the cross-layer frequency characteristics. For example, if the roll continues to increase when entering a high swirling layer and crossing layers frequently, it is determined to be a high coupling disturbance characteristic.

[0148] In a specific embodiment, the present invention provides a method for evaluating UAV simulation flight data. Step 4 involves performing a consistency evaluation of the coupling between the vertical airflow structure and the flight trajectory. Based on the flight executability constraint mapping sequence output in step 3, the positional change of the UAV flight trajectory in vertical space is coupled with the layered airflow structure for analysis. The dwell time, cross-layer frequency, and attitude change continuity characteristics of the trajectory in different airflow levels are extracted. By performing consistency judgment on the coupling relationship between the trajectory and the airflow structure, a vertical airflow structure coupling evaluation result is generated to reflect the degree of matching between flight behavior and airflow distribution. This includes the following sub-steps:

[0149] Sub-step 4.1: Construct the trajectory hierarchical dwell statistics sequence and the trajectory cross-layer event sequence.

[0150] Specifically, firstly, the flight trajectory is aligned with the vertically stratified airflow structure on a time-by-time basis. During the alignment process, the altitude, vertical velocity, and corresponding executability constraint strength and stability level of the UAV at each sampling time are read. Based on the stratification boundary altitude sequence given in step 1, each trajectory point is mapped to a unique airflow stratum. Subsequently, trajectory points continuously within the same stratum are merged into segments to form trajectory stratification dwell segments. At the same time, situations where the stratum to which two adjacent sampling times belong changes are identified as cross-stratum events.

[0151] In the strong swirling airflow zone of the canyon reservoir, simply counting the number of layers entered is insufficient; it is also necessary to distinguish between stable layer crossings and repeated layer crossings. Therefore, when constructing the cross-layer event sequence, it is necessary to record the start time, end time, number of layers crossed, average vertical velocity during the crossing process, and whether it crosses inversion layers for each cross-layer event. In this example, if the height change between adjacent sampling times does not exceed 0.3-1 meter, and the layer switching occurs only once, it is considered a normal boundary crossing; if it crosses the same layer multiple times within 1-3 seconds, it is marked as a repeated layer crossing event. Finally, the output trajectory layer dwell time statistics sequence is directly used to extract the dwell time features of each layer, and the trajectory cross-layer event sequence is directly used to extract the cross-layer frequency features and inversion zone coupling features.

[0152] Sub-step 4.2: Extract hierarchical dwell features and cross-layer frequency features.

[0153] Specifically, based on the trajectory-level dwell statistics sequence, two core features are extracted. The first feature is the level dwell feature, which describes whether the dwell distribution of the UAV in different levels is stable. The second feature is the cross-level frequency feature, which describes whether the UAV frequently switches levels in a short period of time.

[0154] When considering tiered dwell time characteristics, it's insufficient to focus solely on absolute dwell time, as the total duration varies across different flight missions. In this example, the dwell time for each tier is normalized to a dwell time percentage, which is then adjusted in conjunction with the executability constraint strength from step 3. If the UAV dwells for an extended period at tiers with lower executability constraint strength and higher stability margins, it indicates a good match between the trajectory and the airflow structure; conversely, if it dwells for an extended period at tiers with high constraint strengths, it suggests a strong dependence of the trajectory on unfavorable airflow layers.

[0155] For cross-layer frequency characteristics, the number of cross-layer events per unit time needs to be calculated using a sliding time window, with higher weights given to cross-layer events traversing inverted layers. The sliding time window length is 3-10 seconds; if the UAV is performing a low-speed inspection mission, it can be appropriately extended to 15 seconds. The weight of ordinary cross-layer events is 1, while the weight of cross-layer events traversing effective inverted layers is 1.5-3. Finally, the output layer dwell characteristic sequence and cross-layer frequency characteristic sequence are used to construct a coupling metric between trajectory and airflow structure.

[0156] Sub-step 4.3 determines whether the attitude changes in the flight trajectory are consistent with the vertical airflow structure.

[0157] Specifically, after obtaining the statistical results of dwell time and cross-layer movement, it is also necessary to determine whether the attitude changes in the flight trajectory are consistent with the vertical airflow structure. Here, continuity does not mean that the attitude changes should be as small as possible, but rather that there should be a reasonable correspondence between the attitude changes and the layer switching and the crossing of the reversal zone.

[0158] In this embodiment, when the UAV enters a high gyroscopic trend layer or traverses an effective reversal layer, if the intensity of attitude change shows a continuous change that matches the characteristics of that layer, for example, the roll response gradually increases, the yaw response shifts synchronously, and then gradually decreases after leaving the layer, it indicates that there is a real coupling relationship between the trajectory and the airflow structure. Conversely, if the layer has been switched, but the attitude change has not been adjusted accordingly, or the attitude change suddenly increases without corresponding layer change support, it indicates that there is an inconsistency between the trajectory and the airflow structure.

[0159] Therefore, continuous differences in roll, pitch, and yaw responses need to be calculated for each level of dwell and its preceding and following transition sections, and the continuous differences are then smoothed over time, with the smoothing window length ranging from 0.2 to 1 second. Subsequently, the obtained continuous difference results are jointly matched with the swirling trend characteristics from step 1 and the executability constraint strength from step 3 to generate attitude change continuity characteristics. Finally, the level dwell characteristics, cross-level frequency characteristics, and attitude change continuity characteristics are fused to form a trajectory-airflow coupling characteristic sequence.

[0160] Sub-step 4.4 outputs the evaluation results of vertical airflow structure coupling.

[0161] Specifically, the aforementioned features are uniformly converted into vertical airflow structure coupling evaluation results that can be called in step 5. For each evaluation time window or each level dwell segment, the results include the rationality of level dwell, the rationality of cross-level frequency, the attitude continuity level, the degree of adaptation of the reversal zone, and the overall coupling consistency level.

[0162] In this embodiment, the overall coupling consistency should reflect the combined effect of three directions: first, whether it preferentially stays in layers with low constraint strength and relatively controllable rotational perturbations; second, whether it avoids frequent crossings of effective reversal layers; and third, whether attitude changes maintain a continuous correspondence with layer switching. If all three conditions are met simultaneously, it is judged as high consistency; if only one of them is met, it is judged as medium consistency; if it stays in layers with high constraint strength for a long time, crosses layers frequently, and loses the continuous correspondence of attitude changes, it is judged as low consistency.

[0163] To facilitate subsequent steps in implementing rollover risk classification, the overall coupling consistency level, along with the corresponding time segment, altitude segment, and dominant risk source, must be output together to ultimately form a vertical airflow structure coupling assessment result.

[0164] Step S150: Determine the attitude instability section information through trajectory airflow coupling characteristics, and conduct a rollover risk assessment of the UAV based on the attitude instability section information, so as to output the simulation flight data assessment results.

[0165] In some embodiments, the UAV simulation flight data evaluation method provided by the present invention includes the following steps in step S150:

[0166] Step S151: The trajectory airflow coupling features are sliced ​​according to a unified time window, and combined with the flight constraint mapping sequence to construct a risk basis representation aligned with the time window.

[0167] Step S152: Continuously accumulate the risk baseline characteristics of adjacent time windows to determine the rollover risk value of the time window, and determine the high-risk section based on the rollover risk value of the time window.

[0168] Step S153: When the preset instability triggering conditions are met in the high-risk section, the high-risk section is determined to be the instability triggering section, and the source of airflow action corresponding to the instability triggering section is obtained, and the rollover risk level of each time window corresponding to the instability triggering section is divided.

[0169] Among them, the instability triggering condition refers to the judgment rule that the following three types of constraints are met simultaneously in the candidate high-risk section: (1) the trajectory crosses the effective reversal layer; (2) the attitude disturbance response dominated by roll continues to increase; (3) the overall coupling consistency continues to decrease. When setting, it should be based on a continuous time window. When the above three conditions are met for 2-8 seconds and the rollover risk value of the time window is continuously higher than the risk threshold, it is judged as an instability triggering section. Among them, the roll dominance should be significantly higher than the pitch and yaw channels, and the attitude change should show a continuous trend rather than an instantaneous peak. The high-risk section is the section where the rollover risk value of the time window exceeds the preset risk threshold for multiple consecutive time windows; the simulation flight data evaluation result is composed of the rollover risk level, the instability triggering section, the source of airflow, and the corresponding airflow level.

[0170] In a specific embodiment, the present invention provides a method for evaluating simulated flight data of unmanned aerial vehicles (UAVs). Step 5 involves generating simulated flight data evaluation results driven by rollover risk. Based on the vertical airflow structure coupling evaluation results output in step 4, and combined with the attitude instability section information under asymmetric airflow, the rollover risk at each time period during flight is graded and evaluated. The simulated flight data evaluation results, including the rollover risk level, instability triggering section, and airflow source, are output, thereby achieving an accurate assessment of the actual flight risk in the strong swirling airflow area of ​​the canyon reservoir. This includes the following sub-steps:

[0171] Sub-step 5.1: Construct the time window risk basic representation sequence.

[0172] Specifically, firstly, the comprehensive coupling consistency value, hierarchical dwell rationality, and reversal zone adaptation degree in step 4 are sliced ​​using a unified evaluation time window. Simultaneously, the executability constraint strength, stability margin level, and instability trend level within the corresponding time period in step 3 are retrieved to form a risk baseline representation sequence aligned to each time window. The length of these time windows is 1-5 seconds. If the UAV's flight speed is high, this is shortened to 0.5-2 seconds. The aim is to uniformly express whether the trajectory conforms to the airflow structure and whether control is close to instability under the same time reference.

[0173] Sub-step 5.2 verifies the persistence of attitude instability.

[0174] Specifically, after obtaining the baseline risk value, the persistence of attitude instability needs to be further considered. This is because the actual rollover risk in the strong swirling airflow zone of the canyon reservoir is often not triggered by a single large disturbance, but rather by the combined effects of continuous roll imbalance, inversion layer crossing, and insufficient control recovery. Therefore, it is necessary to continuously accumulate the baseline risk values ​​for adjacent time windows to construct a time window rollover risk value sequence. If multiple consecutive time windows maintain a high-risk baseline value, the segment is marked as a candidate high-risk segment. The continuous time threshold is set at 2-8 seconds; if the aircraft is smaller and has lower inertia, it is appropriately lowered to 1-3 seconds.

[0175] Sub-step 5.3 identifies the instability triggering zone that truly leads to an increase in risk.

[0176] Specifically, if a candidate segment simultaneously meets the following conditions, its preceding segment is determined to be an instability triggering segment: first, it traverses an effective reversal interlayer; second, the intensity of roll-dominated attitude change continues to increase; and third, the overall coupling consistency continues to decrease. Then, based on the cyclone trend characteristics in step 1, the reversal-enhanced instability trend in step 3, and the cross-layer frequency characteristics in step 4, the triggering segment that truly leads to the expansion of instability is located by utilizing the combined effects of high risk value, low consistency, and high cross-layer frequency. This allows the determination of the main airflow source of the instability triggering segment, which is divided into three categories: interlayer reversal-dominated type, cyclone shear-dominated type, and cross-layer disturbance superposition type.

[0177] Sub-step 5.4 outputs the simulation flight data evaluation results.

[0178] Specifically, rollover risk is classified for each time period, primarily based on rollover risk value and trigger significance, and further adjusted by considering whether the instability triggering segment is located near an inter-layer inversion zone. The risk levels are divided into four categories: low risk, medium risk, high risk, and extremely high risk. Low risk indicates a good match between the trajectory and airflow structure with no sustained instability trend; medium risk indicates the presence of unfavorable strata or mild inter-layer disturbances; high risk indicates the presence of a clear instability triggering segment; and extremely high risk indicates a continuously expanding triggering segment primarily caused by inter-layer inversion and cyclonic shearing. Finally, the output simulation flight data evaluation results include the rollover risk level, the start and end times of the instability triggering segment, the corresponding altitude range, the source of the dominant airflow, and an evaluation description.

[0179] The following describes a UAV simulation flight data evaluation system provided by the present invention. The UAV simulation flight data evaluation system described below and the UAV simulation flight data evaluation method described above can be referred to and correspond to each other.

[0180] like Figure 2 As shown in one embodiment, a UAV simulation flight data evaluation system includes an airflow structure layering module, an attitude disturbance correlation module, a flight constraint mapping module, a feature coupling analysis module, and a flight data evaluation module.

[0181] The airflow structure layering module is used to acquire airflow field data in the simulation environment and to spatially divide the airflow field data into layers to construct a layered airflow structure.

[0182] The attitude disturbance correlation module is used to collect the attitude changes of the UAV at multiple sampling times and correlate the attitude changes with the stratified airflow structure to generate an attitude disturbance response sequence.

[0183] The flight constraint mapping module is used to constrain the attitude change of UAVs under different airflow levels based on the attitude disturbance response sequence, so as to construct the flight constraint mapping sequence.

[0184] The feature coupling analysis module is used to perform coupling analysis between the UAV flight trajectory and the stratified airflow structure based on the flight constraint mapping sequence, so as to extract the trajectory airflow coupling features of the UAV flight trajectory in different airflow levels.

[0185] The flight data evaluation module is used to determine the attitude instability section information through trajectory airflow coupling characteristics, and to conduct a rollover risk assessment of the UAV based on the attitude instability section information, so as to output the simulation flight data evaluation results.

[0186] The applicant of this invention has provided a detailed description of the embodiments of the invention in conjunction with the accompanying drawings. However, those skilled in the art should understand that the above embodiments are merely preferred embodiments of the invention. The detailed description is only intended to help readers better understand the spirit of the invention and is not intended to limit the scope of protection of the invention. On the contrary, any improvements or modifications made based on the inventive spirit of the invention should fall within the scope of protection of the invention.

Claims

1. A method for evaluating simulated flight data of unmanned aerial vehicles (UAVs), characterized in that, The method includes: Acquire airflow field data in the simulation environment and spatially divide the airflow field data to construct a layered airflow structure; The attitude change of the UAV at multiple sampling times is collected, and the attitude change is correlated with the layered airflow structure to generate an attitude disturbance response sequence. Based on the attitude disturbance response sequence, the attitude change of the UAV under different airflow levels is constrained to construct a flight constraint mapping sequence. The UAV flight trajectory is coupled with the layered airflow structure based on the flight constraint mapping sequence to extract the trajectory-airflow coupling characteristics of the UAV flight trajectory in different airflow levels; The attitude instability section information is determined by the trajectory airflow coupling characteristics, and the rollover risk of the UAV is assessed based on the attitude instability section information to output the simulation flight data assessment results.

2. The method for evaluating UAV simulated flight data according to claim 1, characterized in that, The process of acquiring airflow field data in a simulation environment and spatially dividing the airflow field data to construct a layered airflow structure includes: The airflow field data is time-aligned and then discretely layered according to a preset initial layer thickness to obtain multiple airflow levels. Obtain the airflow direction and average airflow intensity corresponding to each airflow level, and calculate the airflow direction change characteristics and airflow intensity change characteristics between adjacent airflow levels based on the airflow direction and average airflow intensity; When the change characteristics of airflow direction or airflow intensity exceed a preset change threshold, the boundary area between adjacent airflow levels is marked as an airflow abrupt change area. The airflow level is obtained by correcting the initial layer thickness.

3. The method for evaluating UAV simulated flight data according to claim 2, characterized in that, The step of acquiring airflow field data in the simulation environment and spatially dividing the airflow field data to construct a layered airflow structure further includes: Establish local vertical windows for adjacent airflow levels, and statistically analyze the distribution density of the airflow abrupt change region within the local vertical windows to determine the swirling trend characteristics of the corresponding airflow level based on the distribution density. The swirling trend characteristics are written into the corresponding airflow levels, and each airflow level is encoded in order of height to construct the layered airflow structure.

4. The method for evaluating UAV simulated flight data according to claim 1, characterized in that, The method collects the attitude changes of the UAV at multiple sampling times and correlates these attitude changes with the stratified airflow structure to generate an attitude disturbance response sequence, including: The flight attitude and altitude of the UAV are measured at a uniform sampling time, and the flight attitude corresponding to each sampling time is written into the corresponding airflow level according to the flight altitude to determine the level dwell section. Based on the flight attitude of the UAV at adjacent sampling times, the instantaneous rate of change of attitude for different attitude channels is calculated, and the instantaneous rate of change of attitude is smoothed according to a preset sliding time window to obtain the disturbance intensity for each attitude channel. The disturbance intensity is statistically analyzed within the hierarchical residence section, and the attitude disturbance response of the corresponding airflow level of the hierarchical residence section to each attitude channel is determined in combination with the hierarchical airflow structure, so as to generate the attitude disturbance response sequence based on the attitude disturbance response.

5. The method for evaluating UAV simulated flight data according to claim 4, characterized in that, The constraint on the attitude change of the UAV under different airflow levels based on the attitude perturbation response sequence to construct a flight constraint mapping sequence includes: The attitude control boundary of the UAV is obtained, and the attitude control boundary of the UAV is matched with the attitude disturbance response at each time in the attitude disturbance response sequence to determine the UAV control instability section. The attitude change of the UAV within the aforementioned UAV control instability zone is statistically analyzed, and the instability trend of the reversal region is determined based on the attitude change. The instability trend of the inverted region is transformed into UAV flight constraints, and a mapping relationship between the UAV flight constraints and the corresponding airflow level is established to obtain the flight constraint mapping sequence.

6. The method for evaluating UAV simulated flight data according to claim 1, characterized in that, The step of coupling analysis between the UAV flight trajectory and the layered airflow structure based on the flight constraint mapping sequence to extract the trajectory-airflow coupling features of the UAV flight trajectory at different airflow levels includes: The flight trajectory of the UAV is time-aligned with the layered airflow structure to statistically analyze trajectory-level dwell segments and trajectory cross-level events, and these are then converted into layer-level dwell characteristics and cross-level frequency characteristics. When the UAV flight trajectory is inconsistent with the stratified airflow structure, the attitude difference result is determined, and the attitude difference result is associated and matched with the stratified airflow structure and flight constraint mapping sequence to generate the trajectory airflow coupling feature by combining the stratified dwell characteristics and cross-layer frequency characteristics.

7. The method for evaluating UAV simulated flight data according to claim 1, characterized in that, The process of determining attitude instability segment information through the trajectory airflow coupling characteristics and assessing the rollover risk of the UAV based on the attitude instability segment information to output simulation flight data assessment results includes: The trajectory airflow coupling features are sliced ​​according to a unified time window, and combined with the flight constraint mapping sequence to construct a risk basis representation aligned with the time window. The risk baseline characterization of adjacent time windows is continuously accumulated to determine the rollover risk value of the time window, and high-risk sections are determined based on the rollover risk value of the time window; When the high-risk section meets the preset instability triggering conditions, the high-risk section is determined to be an instability triggering section, and the source of airflow action corresponding to the instability triggering section is obtained, and the rollover risk level of each time window corresponding to the instability triggering section is divided. The high-risk section is the section where the rollover risk value exceeds a preset risk threshold for multiple consecutive time windows; the simulated flight data evaluation result is composed of the rollover risk level, the instability triggering section, the source of airflow, and the corresponding airflow level.

8. A UAV simulation flight data evaluation system, characterized in that, The system for implementing the UAV simulation flight data evaluation method according to any one of claims 1 to 7, the system comprising: The airflow structure layering module is used to acquire airflow field data in the simulation environment and spatially divide the airflow field data to construct a layered airflow structure. The attitude disturbance correlation module is used to collect the attitude change of the UAV at multiple sampling times and correlate the attitude change with the layered airflow structure to generate an attitude disturbance response sequence. The flight constraint mapping module is used to constrain the attitude change of the UAV under different airflow levels based on the attitude disturbance response sequence, so as to construct the flight constraint mapping sequence. The feature coupling analysis module is used to perform coupling analysis between the UAV flight trajectory and the layered airflow structure according to the flight constraint mapping sequence, so as to extract the trajectory airflow coupling features of the UAV flight trajectory in different airflow levels; The flight data evaluation module is used to determine the attitude instability section information through the trajectory airflow coupling characteristics, and to perform a rollover risk assessment on the UAV based on the attitude instability section information, so as to output the simulation flight data evaluation results.

9. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-7.

Images