An unmanned aerial vehicle aerodynamic deceleration control index sensitivity analysis method and system

Abstract

The application discloses a kind of unmanned aerial vehicle aerodynamic deceleration control index sensitivity analysis method and system, belong to unmanned aerial vehicle control technical field.For solving the problem of comprehensive analysis to the sensitivity of unmanned aerial vehicle.This application includes constructing a kind of unmanned aerial vehicle aerodynamic deceleration control prediction model hidden layer, using the three-layer architecture of dynamics perception layer, aerodynamic characteristic layer and control decision layer;The output index of model is constructed;Loss function is constructed as the loss function of basic loss function, speed loss term function, flight characteristic loss term function and stability loss term function composite loss function;N groups of unmanned aerial vehicle aerodynamic deceleration control prediction model input feature parameter sample points are set, input into the well-trained unmanned aerial vehicle aerodynamic deceleration control prediction model, obtain corresponding N groups of output results, record input-output corresponding relationship matrix;Unmanned aerial vehicle aerodynamic deceleration control index sensitivity analysis method is constructed.The application realizes the in-depth understanding of parameter influence mechanism.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) control technology, specifically relating to a method and system for sensitivity analysis of aerodynamic deceleration control indicators for UAVs. Background Technology

[0002] Aerodynamic deceleration control of unmanned aerial vehicles (UAVs) involves multiple control indices and parameters, which exhibit complex nonlinear relationships and coupling effects. In actual control processes, different indices have varying degrees of influence on the control effect, and this influence dynamically changes with flight conditions. Therefore, accurately identifying and analyzing the sensitivity characteristics of each control index is crucial for optimizing control strategies, improving control accuracy, and ensuring flight safety. Sensitivity analysis can clarify the importance of each index at different flight stages, providing a scientific basis for optimizing control parameter adjustments.

[0003] Existing methods for sensitivity analysis of UAV aerodynamic deceleration control parameters suffer from several limitations. First, most methods only consider the first-order linear correlation between parameters, neglecting nonlinear coupling effects and higher-order interactions, making it difficult to reflect the complex characteristics of real-world systems. Second, existing methods often treat the entire deceleration process as a whole for static analysis, failing to reflect the dynamic characteristics of parameter sensitivity changes with flight conditions, thus lacking practical guidance. Furthermore, existing methods lack a comprehensive evaluation mechanism for parameter importance, making it difficult to balance the multi-dimensional impact of parameters on deceleration performance, stability, and energy characteristics.

[0004] These limitations lead to a series of problems in practical applications. When designing control strategies, excessive focus may be placed on certain superficially important parameters while neglecting crucial actual parameters, resulting in suboptimal control performance. During different flight phases, due to a lack of accurate understanding of the dynamic changes in parameter sensitivity, adjustments to control parameters may be untimely and inaccurate, affecting the precision and reliability of deceleration control. Furthermore, the inability to comprehensively assess the combined impact of parameters may cause the control strategy to sacrifice other important performance indicators in pursuit of optimizing a particular performance metric, thus affecting the overall performance of the system.

[0005] Furthermore, traditional sensitivity analysis methods often employ simple statistical correlation analysis or local disturbance analysis, which are significantly inadequate when dealing with high-dimensional nonlinear systems. They fail to effectively identify interactions between parameters and struggle to accurately quantify the actual impact of parameters on system performance. These limitations make it difficult to establish accurate parameter importance rankings in engineering practice, impacting the efficiency of control system optimization design and parameter tuning, while also increasing the costs of system development and testing. Summary of the Invention

[0006] The problem this invention aims to solve is to conduct a comprehensive analysis of the sensitivity of unmanned aerial vehicles (UAVs), and proposes a method and system for sensitivity analysis of UAV aerodynamic deceleration control indicators.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A sensitivity analysis method for aerodynamic deceleration control indicators of unmanned aerial vehicles (UAVs) includes the following steps:

[0009] S1. Collect and normalize the input feature parameters of the UAV aerodynamic deceleration control prediction model, including flight state parameters, control input parameters and environmental parameters;

[0010] S2. Construct a hidden layer for a UAV aerodynamic deceleration control prediction model, adopting a three-layer architecture of dynamic perception layer, aerodynamic characteristics layer and control decision layer;

[0011] S3. Output index of the hidden layer construction model of the UAV aerodynamic deceleration control prediction model obtained in step S2;

[0012] S4. Construct a loss function for a prediction model of aerodynamic deceleration control for unmanned aerial vehicles (UAVs) that is a composite loss function consisting of a basic loss function, a velocity loss term function, a flight characteristic loss term function, and a stability loss term function.

[0013] S5. Based on the input feature parameter data of the UAV aerodynamic deceleration control prediction model obtained in step S1 and the output index data of the UAV aerodynamic deceleration control prediction model obtained in step S3, construct a training set and a test set, set parameters for the UAV aerodynamic deceleration control prediction model constructed in steps S2-S4, train the UAV aerodynamic deceleration control prediction model using the obtained training set, and test it using the test set to obtain a trained UAV aerodynamic deceleration control prediction model.

[0014] S6. After normalizing the sample points of the input feature parameters of the N sets of UAV aerodynamic deceleration control prediction models, input them into the UAV aerodynamic deceleration control prediction model trained in step S5 to obtain the corresponding N sets of output results. Then perform inverse normalization to obtain the output parameters with dimensions, and record them as the input-output correspondence matrix.

[0015] S7. Based on the input-output correspondence matrix obtained in step S6, divide the speed stages and construct a sensitivity analysis method for UAV aerodynamic deceleration control indicators, including calculating the sensitivity index, constructing the importance index of key parameters, and identifying the sensitivity of UAV aerodynamic deceleration control indicators.

[0016] Furthermore, the specific input feature parameters of the UAV aerodynamic deceleration control prediction model in step S1 are as follows:

[0017] Flight status parameters include airspeed X1, angle of attack X2, sideslip angle X3, altitude X4, Mach number X5, dynamic pressure X6, roll angle X7, and pitch angle X8;

[0018] Control input parameters include aileron deflection angle X9 and rudder deflection angle X. 10 Elevator deflection angle X 11 Throttle opening X 12 Air brake deflection angle X 13 and the deployment area of ​​the deceleration parachute X 14 ;

[0019] Environmental parameters include air density X 15 Temperature X 16 Wind speed X 17 Wind direction X 18 Atmospheric pressure X 19 and humidity X 20 .

[0020] Furthermore, the specific implementation method of step S2 includes the following steps:

[0021] S2.1. Based on considering airspeed and altitude in flight state parameters, and also taking into account the influence of environmental parameters on dynamic characteristics, a dynamic sensing layer is constructed, expressed as:

[0022]

[0023] in, Output for the dynamic sensing layer. For activation function, Let be the weight coefficient of the j-th input feature parameter in the UAV aerodynamic deceleration control prediction model. Let j be the j-th input feature parameter of the UAV aerodynamic deceleration control prediction model. For dynamic characteristic coefficients, As a high-impact factor, The characteristic time scale factor characterizes the characteristic response speed during the deceleration process of the UAV. The characteristic height scale factor characterizes the sensitivity of the UAV to changes in altitude; For the bias term of the dynamic sensing layer, j=1~20;

[0024] Dynamic characteristic coefficient The expression reflecting the dynamic characteristics of the UAV during deceleration is:

[0025]

[0026] in, For the quality of drones, This is the drag coefficient;

[0027] High Impact Factor The expression reflecting the relationship between altitude and atmospheric pressure is:

[0028]

[0029] in, For reference height, Standard atmospheric pressure;

[0030] S2.2. Based on dynamic characteristics, the influence of angle of attack and dynamic pressure on aerodynamic forces is considered, and the Mach number effect is also taken into account to construct an aerodynamic characteristic layer, the expression of which is:

[0031]

[0032] in, Output for aerodynamic characteristics layer; This is the bias term for the aerodynamic characteristic layer. This is the lift characteristic coefficient; As a factor influencing compressibility effect, The characteristic angle of attack characterizes the system's response to attitude changes during deceleration. Characteristic dynamic pressure, which represents the adaptability of an unmanned aerial vehicle (UAV) to changes in aerodynamics;

[0033] Lift characteristic coefficient Aerodynamic characteristics are reflected by the ratio of dynamic pressure, the deceleration parachute deployment area, and gravity, expressed as:

[0034]

[0035] Where g is the acceleration due to gravity;

[0036] Compressibility effect factor Considering the effects of air density and Mach number on compressibility, the expression is:

[0037] ;

[0038] S2.3. Construct a control decision layer. By establishing a direct information channel from the dynamic perception layer to the control decision layer, a rapid mapping from the initial dynamic state to the control decision is achieved. The expression is:

[0039]

[0040] in, To control the output of the decision-making level; The residual connectivity coefficient is determined by expert experience. To control the efficiency coefficient; As a stability influencing factor, To control the bias terms of the decision-making level; Characteristic force, representing the impact of changes in control force on the UAV; The characteristic rudder deflection angle characterizes the impact of rudder surface deflection on the UAV;

[0041] Control efficiency coefficient By considering the relationship between dynamic pressure and characteristic length to reflect control efficiency, the product of dynamic pressure and deceleration parachute deployment area characterizes the aerodynamic control force, expressed as:

[0042]

[0043] in, For characteristic length, The average aerodynamic chord length;

[0044] Stability Influencing Factors A stability evaluation index is constructed using Mach number, expressed as follows:

[0045] .

[0046] Furthermore, step S3 is specifically implemented by weighting and combining the outputs of the three hidden layers constructed in step S2, and introducing a nonlinear activation function and a feature scaling factor for the corresponding index to achieve accurate prediction of the flight state prediction index and the control effect evaluation index. The nonlinear activation function introduced for the flight state prediction index is the hyperbolic tangent function, and the nonlinear activation function introduced for the control effect evaluation index is the sigmoid function, thus obtaining the flight state prediction index and the control effect evaluation index. The flight state prediction index includes predicted airspeed Y1, predicted angle of attack Y2, predicted altitude Y3, predicted Mach number Y4, and predicted dynamic pressure Y5. The control effect evaluation index includes deceleration rate Y6, aerodynamic drag coefficient Y7, lift-to-drag ratio Y8, energy loss rate Y9, and stability index Y1. 10 Establish the mapping relationship between the model's hidden layer and the output metrics.

[0047] Furthermore, the specific implementation method of step S4 includes the following steps:

[0048] S4.1. Design the basic loss function, using the mean squared error form, and calculate all output indices Y1~Y2. 10 The mean square error between the predicted and measured values ​​is the basic loss. ;

[0049] S4.2. Design the velocity loss term function, quantify the deceleration effect by evaluating the difference between predicted and actual airspeed and the corresponding state transition efficiency, while also considering the effects of air density and parachute deployment area, resulting in the following expression:

[0050]

[0051] in, For the weighting coefficient of the velocity loss term; This is the weighting coefficient for state transition efficiency; , These represent the kinetic energy levels of the predicted state and the current state, respectively. , These respectively characterize the work done on the predicted resistance and the actual resistance;

[0052] S4.3. Design the flight characteristic loss term function. By considering the influence of Mach number on aerodynamic characteristics and evaluating the accuracy of lift, drag, and moment predictions, a comprehensive evaluation index for flight quality is constructed by introducing characteristic length and aerodynamic parameters. The expression for the flight characteristic loss term function is as follows:

[0053]

[0054] in, The Mach number effect coefficient, This is the torque weighting coefficient. , These represent the predicted aerodynamic intensity level and the actual aerodynamic intensity level, respectively. , These respectively characterize the predicted torque level and the actual torque level;

[0055] S4.4. Design a stability loss term function. By evaluating three aspects—height variation, static stability, and dynamic stability—a complete stability evaluation system is constructed. The expression for the stability loss term function is:

[0056]

[0057] in, As a high-impact coefficient, For stability weighting coefficients, and These represent the mechanical energy level of the predicted state and the mechanical energy level of the current state, respectively. and These represent the predicted stability moment level and the actual stability moment level, respectively. This is the static stability coefficient; Characteristic energy factor; The characteristic torque factor;

[0058] S4.5. The loss function of the UAV aerodynamic deceleration control prediction model is a composite loss function including a velocity loss term, a flight characteristic loss term, and a stability loss term, expressed as:

[0059]

[0060] in, , , , They are respectively , , , The weighting coefficients.

[0061] Furthermore, the input feature parameter data of the UAV aerodynamic deceleration control prediction model after normalization in step S5 and the output index data of the UAV aerodynamic deceleration control prediction model obtained in step S3 are divided into training set and test set in a ratio of 8:2.

[0062] Furthermore, in step S6, the input-output correspondence matrix is ​​recorded as R[N×(20+10)].

[0063] Furthermore, the specific implementation method of step S7 includes the following steps:

[0064] S7.1. Speed ​​phase division;

[0065] by For reference speed, the flight process is divided into three stages:

[0066] Speed ​​on the high-speed section: The corresponding flight speed range is k=1;

[0067] Speed ​​in the medium speed range: The corresponding flight speed range is k=2;

[0068] Speed ​​in the low-speed range: The corresponding flight speed range is k=3;

[0069] S7.2. Constructing a sensitivity index;

[0070] A first-order sensitivity index is set to measure the degree of direct influence between parameters by calculating the squared correlation between a single input parameter and an output parameter. The first-order sensitivity index for the k-th flight speed segment is... The expression is:

[0071]

[0072] in, This is the data set corresponding to the j-th input parameter; Let be the data set corresponding to the p-th output parameter, and corr be the correlation coefficient;

[0073] Second-order sensitivity indices are set to consider the interactions between parameters. The coupling effect of parameters is assessed through combined correlations. Second-order sensitivity indices include single-parameter correlations and parameter combination correlations. The second-order sensitivity index for the k-th flight speed segment is... The expression is:

[0074]

[0075] in, The data set corresponding to the h-th input parameter coupled with the j-th input parameter; The square of the correlation of the parameter combination;

[0076] A dynamic sensitivity index is set up to reflect the dynamic changes in parameter sensitivity by calculating the absolute value of the correlation at different speed stages. The dynamic sensitivity index at stage k is... The expression is:

[0077] ;

[0078] Set the dynamic sensitivity variability for each flight speed range. , used to characterize the dynamic changes in the influence of parameters, is expressed as:

[0079]

[0080] in, , , These are the dynamic sensitivity indicators corresponding to the high-speed, medium-speed, and low-speed segments, respectively.

[0081] S7.3. Perform a comprehensive analysis of flight speed segment characteristics based on the sensitivity index constructed in step S7.2;

[0082] A comprehensive parameter importance assessment system was constructed by weighting and combining the deceleration performance index, stability impact index, and energy characteristic impact index. Within each flight speed segment k, the comprehensive index was calculated. The expression is:

[0083]

[0084] in, The deceleration performance impact index for the k-th flight speed segment. This represents the stability impact index for the k-th flight speed segment. The energy characteristic influence index for the k-th flight speed segment;

[0085] The deceleration performance impact index for the k-th flight speed segment is constructed based on the first-order sensitivity of predicted airspeed and deceleration rate, and its expression is:

[0086] ;

[0087] The stability impact index for the k-th flight speed segment is constructed based on the first-order sensitivity of the predicted angle of attack and the stability index, and its expression is:

[0088] ;

[0089] The energy characteristic influence index for the k-th flight speed segment is constructed based on the first-order and second-order sensitivities of the drag coefficient and energy loss rate, and its expression is:

[0090] ;

[0091] S7.4. Introduce adjustment coefficients to balance the strength and stability of the influence of parameters, and construct an importance index for key parameters. Used for quantitative assessment of parameter importance;

[0092]

[0093] in, It serves as an importance index for key parameters; The adjustment coefficients are determined by expert experience;

[0094] The importance index of key parameters is classified and defined. , These are the first and second thresholds for the importance index of key parameters, determined by expert experience.

[0095] when This indicates that each flight speed range has high sensitivity and significant impact, and the corresponding indicators can be used as the core parameters for control. Real-time and precise control of the corresponding indicators is required.

[0096] when This indicates high sensitivity in a specific flight speed range. For the corresponding indicators, a segmented control strategy needs to be developed, and the control priority should be adjusted according to the flight phase.

[0097] when When the sensitivity is low across all flight speed ranges, the corresponding index is used as an auxiliary reference parameter.

[0098] A sensitivity analysis system for aerodynamic deceleration control indicators of unmanned aerial vehicles (UAVs) includes a processor, a memory, and a computer program stored in the memory and run on the processor. When the computer program runs, it implements the steps of the sensitivity analysis method for aerodynamic deceleration control indicators of UAVs as described above.

[0099] The beneficial effects of this invention are:

[0100] This invention discloses a sensitivity analysis method for aerodynamic deceleration control indicators of unmanned aerial vehicles (UAVs). This method establishes a comprehensive and dynamic sensitivity analysis approach for UAV aerodynamic deceleration control indicators. It accurately identifies and quantifies the nonlinear coupling relationships between parameters, reflects the dynamic changes in parameter sensitivity with flight conditions, and establishes a scientific comprehensive evaluation system for parameter importance. Through multi-level sensitivity analysis, it achieves a deeper understanding of the parameter influence mechanisms.

[0101] This invention discloses a sensitivity analysis method for aerodynamic deceleration control parameters of unmanned aerial vehicles (UAVs). By establishing a multi-level sensitivity index system, it achieves a comprehensive assessment of the impact of parameters, including first-order direct effects, second-order coupling effects, and dynamic change characteristics. Employing a segmented dynamic analysis strategy, it accurately captures the variation patterns of parameter sensitivity across different speed ranges. By constructing a comprehensive evaluation index, it achieves a scientific quantification of parameter importance, providing a reliable basis for optimizing control strategies. Attached Figure Description

[0102] Figure 1 This is a flowchart of a sensitivity analysis method for aerodynamic deceleration control indicators of an unmanned aerial vehicle (UAV) according to the present invention;

[0103] Figure 2 This is a comparison chart of KPI calculation results under different operating conditions of the present invention. Detailed Implementation

[0104] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0105] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0106] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 -Appendix Figure 2 Detailed explanation is as follows:

[0107] Example 1:

[0108] A sensitivity analysis method for aerodynamic deceleration control indicators of unmanned aerial vehicles (UAVs) includes the following steps:

[0109] S1. Collect and normalize the input feature parameters of the UAV aerodynamic deceleration control prediction model, including flight state parameters, control input parameters and environmental parameters;

[0110] Furthermore, the specific input feature parameters of the UAV aerodynamic deceleration control prediction model in step S1 are as follows:

[0111] Flight status parameters include airspeed X1, angle of attack X2, sideslip angle X3, altitude X4, Mach number X5, dynamic pressure X6, roll angle X7, and pitch angle X8;

[0112] Control input parameters include aileron deflection angle X9 and rudder deflection angle X. 10 Elevator deflection angle X 11 Throttle opening X 12 Air brake deflection angle X 13 and the deployment area of ​​the deceleration parachute X 14 ;

[0113] Environmental parameters include air density X 15 Temperature X 16 Wind speed X 17 Wind direction X 18 Atmospheric pressure X 19 and humidity X 20 .

[0114] S2. Construct a hidden layer for a UAV aerodynamic deceleration control prediction model, adopting a three-layer architecture of dynamic perception layer, aerodynamic characteristics layer and control decision layer;

[0115] Furthermore, the specific implementation method of step S2 includes the following steps:

[0116] S2.1. Based on considering airspeed and altitude in flight state parameters, and also taking into account the influence of environmental parameters on dynamic characteristics, a dynamic sensing layer is constructed, expressed as:

[0117]

[0118] in, Output for the dynamic sensing layer. For activation function, Let be the weight coefficient of the j-th input feature parameter in the UAV aerodynamic deceleration control prediction model. Let j be the j-th input feature parameter of the UAV aerodynamic deceleration control prediction model. For dynamic characteristic coefficients, As a high-impact factor, The characteristic time scale factor characterizes the characteristic response speed during the deceleration process of the UAV. The characteristic height scale factor characterizes the sensitivity of the UAV to changes in altitude; For the bias term of the dynamic sensing layer, j=1~20;

[0119] Dynamic characteristic coefficient The expression reflecting the dynamic characteristics of the UAV during deceleration is:

[0120]

[0121] in, For the quality of drones, This is the drag coefficient;

[0122] High Impact Factor The expression reflecting the relationship between altitude and atmospheric pressure is:

[0123]

[0124] in, For reference height, Standard atmospheric pressure;

[0125] S2.2. Based on dynamic characteristics, the influence of angle of attack and dynamic pressure on aerodynamic forces is considered, and the Mach number effect is also taken into account to construct an aerodynamic characteristic layer, the expression of which is:

[0126]

[0127] in, Output for aerodynamic characteristics layer; This is the bias term for the aerodynamic characteristic layer. This is the lift characteristic coefficient; As a factor influencing compressibility effect, The characteristic angle of attack characterizes the system's response to attitude changes during deceleration. Characteristic dynamic pressure, which represents the adaptability of an unmanned aerial vehicle (UAV) to changes in aerodynamics;

[0128] Lift characteristic coefficient Aerodynamic characteristics are reflected by the ratio of dynamic pressure, the deceleration parachute deployment area, and gravity, expressed as:

[0129]

[0130] Where g is the acceleration due to gravity;

[0131] Compressibility effect factor Considering the effects of air density and Mach number on compressibility, the expression is:

[0132] ;

[0133] S2.3. Construct a control decision layer. By establishing a direct information channel from the dynamic perception layer to the control decision layer, a rapid mapping from the initial dynamic state to the control decision is achieved. The expression is:

[0134]

[0135] in, To control the output of the decision-making level; The residual connectivity coefficient is determined by expert experience. To control the efficiency coefficient; As a stability influencing factor, To control the bias terms of the decision-making level; Characteristic force, representing the impact of changes in control force on the UAV; The characteristic rudder deflection angle characterizes the impact of rudder surface deflection on the UAV;

[0136] Control efficiency coefficient By considering the relationship between dynamic pressure and characteristic length to reflect control efficiency, the product of dynamic pressure and deceleration parachute deployment area characterizes the aerodynamic control force, expressed as:

[0137]

[0138] in, For characteristic length, The average aerodynamic chord length;

[0139] Stability Influencing Factors A stability evaluation index is constructed using Mach number, expressed as follows:

[0140] .

[0141] S3. Output index of the hidden layer construction model of the UAV aerodynamic deceleration control prediction model obtained in step S2;

[0142] Furthermore, step S3 is specifically implemented by weighting and combining the outputs of the three hidden layers constructed in step S2, and introducing a nonlinear activation function and a feature scaling factor for the corresponding index to achieve accurate prediction of the flight state prediction index and the control effect evaluation index. The nonlinear activation function introduced for the flight state prediction index is the hyperbolic tangent function, and the nonlinear activation function introduced for the control effect evaluation index is the sigmoid function, thus obtaining the flight state prediction index and the control effect evaluation index. The flight state prediction index includes predicted airspeed Y1, predicted angle of attack Y2, predicted altitude Y3, predicted Mach number Y4, and predicted dynamic pressure Y5. The control effect evaluation index includes deceleration rate Y6, aerodynamic drag coefficient Y7, lift-to-drag ratio Y8, energy loss rate Y9, and stability index Y1. 10 Establish the mapping relationship between the model's hidden layer and the output metrics.

[0143] S4. Construct a loss function for a prediction model of aerodynamic deceleration control for unmanned aerial vehicles (UAVs) that is a composite loss function consisting of a basic loss function, a velocity loss term function, a flight characteristic loss term function, and a stability loss term function.

[0144] Furthermore, the specific implementation method of step S4 includes the following steps:

[0145] S4.1. Design the basic loss function, using the mean squared error form, and calculate all output indices Y1~Y2. 10 The mean square error between the predicted and measured values ​​is the basic loss. ;

[0146] S4.2. Design the velocity loss term function, quantify the deceleration effect by evaluating the difference between predicted and actual airspeed and the corresponding state transition efficiency, while also considering the effects of air density and parachute deployment area, resulting in the following expression:

[0147]

[0148] in, For the weighting coefficient of the velocity loss term; This is the weighting coefficient for state transition efficiency; , These represent the kinetic energy levels of the predicted state and the current state, respectively. , These respectively characterize the work done on the predicted resistance and the actual resistance;

[0149] The weighting coefficient of the velocity loss term reflects the variation of the deceleration effect with flight conditions, while the air density-predicted velocity square term characterizes the kinetic energy change. Considering air density, parachute deployment area, and mass factors, the expression is:

[0150]

[0151] The state transition efficiency weighting coefficient evaluates the energy conversion efficiency during deceleration. Considering the inertial characteristics of the UAV, the characteristics of the deceleration medium, the current motion state, and the characteristics of the deceleration device, the expression for the state transition efficiency weighting coefficient is as follows:

[0152] ;

[0153] S4.3. Design the flight characteristic loss term function. By considering the influence of Mach number on aerodynamic characteristics and evaluating the accuracy of lift, drag, and moment predictions, a comprehensive evaluation index for flight quality is constructed by introducing characteristic length and aerodynamic parameters. The expression for the flight characteristic loss term function is as follows:

[0154]

[0155] in, The Mach number effect coefficient, This is the torque weighting coefficient. , These represent the predicted aerodynamic intensity level and the actual aerodynamic intensity level, respectively. , These respectively characterize the predicted torque level and the actual torque level;

[0156] The Mach number effect coefficient reflects the changes in aerodynamic characteristics during flight speed variations through the Mach number, directly reflecting the intensity of the Mach number effect. Its expression is:

[0157] ;

[0158] The torque weighting coefficient takes into account the effects of environmental characteristics, deceleration chute area, and characteristic length. The characteristic length assesses the impact of torque balance, and the predicted square velocity reflects the dynamic pressure effect. The expression is as follows:

[0159] ;

[0160] S4.4. Design a stability loss term function. By evaluating three aspects—height variation, static stability, and dynamic stability—a complete stability evaluation system is constructed. The expression for the stability loss term function is:

[0161]

[0162] in, As a high-impact coefficient, For stability weighting coefficients, and These represent the mechanical energy level of the predicted state and the mechanical energy level of the current state, respectively. and These represent the predicted stability moment level and the actual stability moment level, respectively. This is the static stability coefficient; Characteristic energy factor; The characteristic torque factor;

[0163] The altitude influence coefficient characterizes the influence of altitude, and its expression is:

[0164]

[0165] The stability weighting coefficient assesses the impact on stability, while the predicted velocity square reflects the dynamic pressure effect. It also considers the influence of environmental characteristics, deceleration chute area, mean aerodynamic chord length, and characteristic length. The expression is:

[0166] ;

[0167] S4.5. The loss function of the UAV aerodynamic deceleration control prediction model is a composite loss function including a velocity loss term, a flight characteristic loss term, and a stability loss term, expressed as:

[0168]

[0169] in, , , , They are respectively , , , The weighting coefficients.

[0170] S5. Based on the input feature parameter data of the UAV aerodynamic deceleration control prediction model obtained in step S1 and the output index data of the UAV aerodynamic deceleration control prediction model obtained in step S3, construct a training set and a test set, set parameters for the UAV aerodynamic deceleration control prediction model constructed in steps S2-S4, train the UAV aerodynamic deceleration control prediction model using the obtained training set, and test it using the test set to obtain a trained UAV aerodynamic deceleration control prediction model.

[0171] Furthermore, in step S5, the input feature parameter data of the UAV aerodynamic deceleration control prediction model after normalization is divided into training set and test set in an 8:2 ratio.

[0172] Furthermore, the output is then denormalized to restore the actual physical quantity.

[0173] S6. After normalizing the sample points of the input feature parameters of the N sets of UAV aerodynamic deceleration control prediction models, input them into the UAV aerodynamic deceleration control prediction model trained in step S5 to obtain the corresponding N sets of output results. Then perform inverse normalization to obtain the output parameters with dimensions, and record them as the input-output correspondence matrix.

[0174] Furthermore, the N sets of data are mainly obtained from the UAV monitoring and management platform, specifically including flight status parameters, control input parameters, and environmental parameters, ranging from X1 to X. 20 The input parameters. It should be noted that when the drone monitoring and management platform lacks the above data (X1~X...) 20 When conducting research, it is necessary to consider combining multiple sources, including commercial flight data from the Civil Aviation Administration of China, test data from drone manufacturers, drone design specifications, and laboratory test data. Substituting these N sets of input samples into the prediction model yields N sets of output results, which are then inversely normalized to reconstruct the actual physical quantities. Next, the aforementioned N sets of data are used for analysis.

[0175] Furthermore, in step S6, the input-output correspondence matrix is ​​recorded as R[N×(20+10)].

[0176] S7. Based on the input-output correspondence matrix obtained in step S6, divide the speed stages and construct a sensitivity analysis method for UAV aerodynamic deceleration control indicators, including calculating the sensitivity index, constructing the importance index of key parameters, and identifying the sensitivity of UAV aerodynamic deceleration control indicators.

[0177] During the deceleration process of the UAV, the system exhibits significant nonlinear characteristics and parameter coupling. To accurately assess the impact of each parameter, a global sensitivity analysis is necessary. Compared to local sensitivity analysis, global analysis considers the influence of parameters across the entire range of variation, as well as the interactions between parameters. By performing multiple sampling calculations on the prediction model and combining this with statistical correlation analysis, the importance of parameters can be comprehensively assessed, providing a basis for subsequent parameter selection and control optimization.

[0178] Furthermore, the specific implementation method of step S7 includes the following steps:

[0179] S7.1. Speed ​​phase division;

[0180] by For reference speed, the flight process is divided into three stages:

[0181] Speed ​​on the high-speed section: The corresponding flight speed range is k=1;

[0182] Speed ​​in the medium speed range: The corresponding flight speed range is k=2;

[0183] Speed ​​in the low-speed range: The corresponding flight speed range is k=3;

[0184] S7.2. Constructing a sensitivity index;

[0185] A first-order sensitivity index is set to measure the degree of direct influence between parameters by calculating the squared correlation between a single input parameter and an output parameter. The first-order sensitivity index for the k-th flight speed segment is... The expression is:

[0186]

[0187] in, This is the data set corresponding to the j-th input parameter; Let be the data set corresponding to the p-th output parameter, and corr be the correlation coefficient;

[0188] Furthermore, the strength of the influence of a single input parameter on the output can be determined. By squaring the correlation coefficient, the effects of positive and negative correlations can be eliminated, intuitively reflecting the correlation strength. First-order sensitivity indices are primarily used in UAV aerodynamic deceleration control to identify the direct impact of individual parameters. By evaluating the degree of influence of parameters on the system output, the importance ranking of deceleration device design parameters and flight control parameters can be clarified.

[0189] Second-order sensitivity indices are set to consider the interactions between parameters. The coupling effect of parameters is assessed through combined correlations. Second-order sensitivity indices include single-parameter correlations and parameter combination correlations. The second-order sensitivity index for the k-th flight speed segment is... The expression is:

[0190]

[0191] in, The data set corresponding to the h-th input parameter coupled with the j-th input parameter; The square of the correlation of the parameter combination;

[0192] Furthermore, the role of second-order sensitivity indices is mainly reflected in the evaluation of coupling effects between parameters. By analyzing the interactions between parameters, we can gain a deeper understanding of the complex characteristics of the UAV deceleration process, providing a basis for formulating multi-parameter collaborative control strategies and improving the efficiency and reliability of deceleration control.

[0193] A dynamic sensitivity index is set up to reflect the dynamic changes in parameter sensitivity by calculating the absolute value of the correlation at different speed stages. The dynamic sensitivity index at stage k is... The expression is:

[0194] ;

[0195] Furthermore, dynamic sensitivity indicators can capture the changing patterns of parameter influence during deceleration, providing a basis for optimizing control strategies at different flight stages.

[0196] Set the dynamic sensitivity variability for each flight speed range. , used to characterize the dynamic changes in the influence of parameters, is expressed as:

[0197]

[0198] in, , , These are the dynamic sensitivity indicators corresponding to the high-speed, medium-speed, and low-speed segments, respectively.

[0199] Furthermore, the dynamic sensitivity variability quantifies the degree of fluctuation in parameter sensitivity by calculating the standard deviation of the dynamic sensitivity index at different speed ranges. This index reflects the stability of the parameter's influence throughout the deceleration process and helps identify the changes in the importance of key control parameters at different speed ranges.

[0200] The comprehensive application of this multi-level sensitivity analysis method can improve system performance in terms of design optimization, performance enhancement, and safety assurance. Through scientific sensitivity analysis, it can comprehensively guide the design improvement of deceleration systems, optimize control strategies, and provide guidance for the engineering implementation of UAV deceleration control systems.

[0201] S7.3. Perform a comprehensive analysis of flight speed segment characteristics based on the sensitivity index constructed in step S7.2;

[0202] A comprehensive parameter importance assessment system was constructed by weighting and combining the deceleration performance index, stability impact index, and energy characteristic impact index. Within each flight speed segment k, the comprehensive index was calculated. The expression is:

[0203]

[0204] in, The deceleration performance impact index for the k-th flight speed segment. This represents the stability impact index for the k-th flight speed segment. The energy characteristic influence index for the k-th flight speed segment;

[0205] A comprehensive index, employing a weighted combination of deceleration performance index, stability impact index, and energy characteristic impact index, constructs a robust parameter importance assessment system. This index considers the combined impact of parameters on deceleration effectiveness, flight stability, and energy management, providing a reliable basis for parameter selection.

[0206] The deceleration performance impact index for the k-th flight speed segment is constructed based on the first-order sensitivity of predicted airspeed and deceleration rate, and its expression is:

[0207] ;

[0208] The stability impact index for the k-th flight speed segment is constructed based on the first-order sensitivity of the predicted angle of attack and the stability index, and its expression is:

[0209] ;

[0210] The energy characteristic influence index for the k-th flight speed segment is constructed based on the first-order and second-order sensitivities of the drag coefficient and energy loss rate, and its expression is:

[0211] ;

[0212] S7.4. Introduce adjustment coefficients to balance the strength and stability of the influence of parameters, and construct an importance index for key parameters. Used for quantitative assessment of parameter importance;

[0213]

[0214] in, It serves as an importance index for key parameters; The adjustment coefficient, determined by expert experience, indicates the importance of the indicator; a larger value signifies greater importance. The key parameter identification indicator establishes a quantitative assessment method for parameter importance by comprehensively considering the parameter's maximum impact and sensitivity variation characteristics. This indicator introduces an adjustment coefficient to balance the intensity and stability of the parameter's influence, providing a scientific basis for determining the core parameters in control strategies.

[0215] The importance index of key parameters is classified and defined. , These are the first and second thresholds for the importance index of key parameters, determined by expert experience.

[0216] when This indicates that each flight speed range has high sensitivity and significant impact, and the corresponding indicators can be used as the core parameters for control. Real-time and precise control of the corresponding indicators is required.

[0217] when This indicates high sensitivity in a specific flight speed range. For the corresponding indicators, a segmented control strategy needs to be developed, and the control priority should be adjusted according to the flight phase.

[0218] when When the sensitivity is low across all flight speed ranges, the corresponding index is used as an auxiliary reference parameter. Figure 2 This is a comparison chart of KPI calculation results under different operating conditions in this embodiment.

[0219] Example 2:

[0220] A sensitivity analysis system for aerodynamic deceleration control indicators of a drone includes a processor, a memory, and a computer program stored in the memory and run on the processor. When the computer program runs, it implements the steps of a sensitivity analysis method for aerodynamic deceleration control indicators of a drone as described in Example 1.

[0221] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0222] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A sensitivity analysis method for aerodynamic deceleration control indicators of unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S1. Collect and normalize the input feature parameters of the UAV aerodynamic deceleration control prediction model, including flight state parameters, control input parameters and environmental parameters; S2. Construct a hidden layer for a UAV aerodynamic deceleration control prediction model, adopting a three-layer architecture of dynamic perception layer, aerodynamic characteristics layer and control decision layer; S3. Output index of the hidden layer construction model of the UAV aerodynamic deceleration control prediction model obtained in step S2; S4. Construct a loss function for a prediction model of aerodynamic deceleration control for unmanned aerial vehicles (UAVs) that is a composite loss function consisting of a basic loss function, a velocity loss term function, a flight characteristic loss term function, and a stability loss term function. S5. Based on the input feature parameter data of the UAV aerodynamic deceleration control prediction model obtained in step S1 and the output index data of the UAV aerodynamic deceleration control prediction model obtained in step S3, construct a training set and a test set, set parameters for the UAV aerodynamic deceleration control prediction model constructed in steps S2-S4, train the UAV aerodynamic deceleration control prediction model using the obtained training set, and test it using the test set to obtain a trained UAV aerodynamic deceleration control prediction model. S6. After normalizing the sample points of the input feature parameters of the N sets of UAV aerodynamic deceleration control prediction models, input them into the UAV aerodynamic deceleration control prediction model trained in step S5 to obtain the corresponding N sets of output results. Then perform inverse normalization to obtain the output parameters with dimensions, and record them as the input-output correspondence matrix. S7. Based on the input-output correspondence matrix obtained in step S6, divide the speed stage and construct a sensitivity analysis method for UAV aerodynamic deceleration control indicators, including calculating the sensitivity index, constructing the importance index of key parameters, and identifying the sensitivity of UAV aerodynamic deceleration control indicators. A first-order sensitivity index is set to measure the degree of direct influence between parameters by calculating the squared correlation between a single input parameter and an output parameter. The first-order sensitivity index for the k-th flight speed segment is... ; The second-order sensitivity index considers the interaction between parameters and evaluates the coupling effect of parameters through combined correlation. The second-order sensitivity index includes the correlation of single parameters and the correlation of parameter combinations, thus obtaining the second-order sensitivity index for the k-th flight speed segment. ; The dynamic sensitivity index is set up by calculating the absolute value of the correlation at different speed stages to reflect the dynamic change characteristics of parameter sensitivity, thus obtaining the dynamic sensitivity index for the k-th flight speed segment. .

2. The method for sensitivity analysis of aerodynamic deceleration control indicators for unmanned aerial vehicles according to claim 1, characterized in that, The specific input feature parameters of the UAV aerodynamic deceleration control prediction model in step S1 are as follows: Flight status parameters include airspeed X1, angle of attack X2, sideslip angle X3, altitude X4, Mach number X5, dynamic pressure X6, roll angle X7, and pitch angle X8; Control input parameters include aileron deflection angle X9 and rudder deflection angle X. 10 Elevator deflection angle X 11 Throttle opening X 12 Air brake deflection angle X 13 and the deployment area of ​​the deceleration parachute X 14 ; Environmental parameters include air density X 15 Temperature X 16 Wind speed X 17 Wind direction X 18 Atmospheric pressure X 19 and humidity X 20 .

3. The method for sensitivity analysis of aerodynamic deceleration control indicators for unmanned aerial vehicles according to claim 2, characterized in that, The specific implementation method of step S2 includes the following steps: S2.

1. Based on considering airspeed and altitude in flight state parameters, and also taking into account the influence of environmental parameters on dynamic characteristics, a dynamic sensing layer is constructed, expressed as: ； in, Output for the dynamic sensing layer. For activation function, Let be the weight coefficient of the j-th input feature parameter in the UAV aerodynamic deceleration control prediction model. Let j be the j-th input feature parameter of the UAV aerodynamic deceleration control prediction model. For dynamic characteristic coefficients, As a high-impact factor, The characteristic time scale factor characterizes the characteristic response speed during the deceleration process of the UAV. The characteristic height scale factor characterizes the sensitivity of the UAV to changes in altitude; For the bias term of the dynamic sensing layer, j=1~20; S2.

2. Based on dynamic characteristics, the influence of angle of attack and dynamic pressure on aerodynamic forces is considered, and the Mach number effect is also taken into account to construct an aerodynamic characteristic layer, the expression of which is: ； in, Output for aerodynamic characteristics layer; This is the bias term for the aerodynamic characteristic layer. This is the lift characteristic coefficient; As a factor influencing compressibility effect, The characteristic angle of attack characterizes the system's response to attitude changes during deceleration; Characteristic dynamic pressure, which represents the adaptability of an unmanned aerial vehicle (UAV) to changes in aerodynamics; S2.

3. Construct a control decision layer. By establishing a direct information channel from the dynamic perception layer to the control decision layer, a rapid mapping from the initial dynamic state to the control decision is achieved. The expression is: ； in, To control the output of the decision-making level; The residual connectivity coefficient is determined by expert experience. To control the efficiency coefficient; As a stability influencing factor, To control the bias terms of the decision-making level; Characteristic force, representing the impact of changes in control force on the UAV; The characteristic rudder deflection angle characterizes the impact of rudder surface deflection on the UAV.

4. The method for sensitivity analysis of aerodynamic deceleration control indicators for unmanned aerial vehicles according to claim 3, characterized in that, The specific implementation method of step S3 is to weight and combine the outputs of the three hidden layers constructed in step S2, and introduce a nonlinear activation function and the feature scale factor of the corresponding index to achieve accurate prediction of flight state prediction index and control effect evaluation index. The nonlinear activation function introduced for the flight state prediction index is the hyperbolic tangent function, and the nonlinear activation function introduced for the control effect evaluation index is the sigmoid function, thus obtaining the flight state prediction index and control effect evaluation index. The flight state prediction index includes predicted airspeed Y1, predicted angle of attack Y2, predicted altitude Y3, predicted Mach number Y4, and predicted dynamic pressure Y5. The control effect evaluation index includes deceleration rate Y6, aerodynamic drag coefficient Y7, lift-to-drag ratio Y8, energy loss rate Y9, and stability index Y1. 10 Establish the mapping relationship between the hidden layer of the model and the output index.

5. The method for sensitivity analysis of aerodynamic deceleration control indicators for unmanned aerial vehicles according to claim 4, characterized in that, The specific implementation method of step S4 includes the following steps: S4.

1. Design the basic loss function, using the mean squared error form, and calculate all output indices Y1~Y2. 10 The mean square error between the predicted and measured values ​​is the basic loss. ; S4.

2. Design the velocity loss term function, quantify the deceleration effect by evaluating the difference between predicted and actual airspeed and the corresponding state transition efficiency, while also considering the effects of air density and parachute deployment area, resulting in the following expression: ； in, For the weighting coefficient of the velocity loss term; This is the weighting coefficient for state transition efficiency; , These respectively characterize the kinetic energy levels of the predicted state and the current state; , These respectively characterize the work done on the predicted resistance and the actual resistance; S4.

3. Design the flight characteristic loss term function. By considering the influence of Mach number on aerodynamic characteristics and evaluating the accuracy of lift, drag, and moment predictions, a comprehensive evaluation index for flight quality is constructed by introducing characteristic length and aerodynamic parameters. The expression for the flight characteristic loss term function is as follows: ； in, The Mach number effect coefficient, This is the torque weighting coefficient. , These represent the predicted aerodynamic intensity level and the actual aerodynamic intensity level, respectively. , These respectively characterize the predicted torque level and the actual torque level; S4.

4. Design a stability loss term function. By evaluating three aspects—height variation, static stability, and dynamic stability—a complete stability evaluation system is constructed. The expression for the stability loss term function is: ； in, As a high-impact coefficient, For stability weighting coefficients, and These represent the mechanical energy level of the predicted state and the mechanical energy level of the current state, respectively. and These represent the predicted stability moment level and the actual stability moment level, respectively. This is the static stability coefficient; Characteristic energy factor; The characteristic torque factor; S4.

5. The loss function of the UAV aerodynamic deceleration control prediction model is a composite loss function including a velocity loss term, a flight characteristic loss term, and a stability loss term, expressed as: ； in, , , , They are respectively , , , The weighting coefficients.

6. The method for sensitivity analysis of aerodynamic deceleration control indicators for unmanned aerial vehicles according to claim 5, characterized in that, The input feature parameter data of the UAV aerodynamic deceleration control prediction model after normalization in step S5 and the output index data of the UAV aerodynamic deceleration control prediction model obtained in step S3 are divided into training set and test set in a ratio of 8:

2.

7. The method for sensitivity analysis of aerodynamic deceleration control indicators for unmanned aerial vehicles according to claim 6, characterized in that, In step S6, the input-output correspondence matrix is ​​recorded as R[N×(20+10)].

8. The method for sensitivity analysis of aerodynamic deceleration control indicators for unmanned aerial vehicles according to claim 7, characterized in that, The specific implementation method of step S7 includes the following steps: S7.

1. Speed ​​phase division; by For reference speed, the flight process is divided into three stages: Speed ​​on the high-speed section: The corresponding flight speed range is k=1; Speed ​​in the medium speed range: The corresponding flight speed range is k=2; Speed ​​in the low-speed range: The corresponding flight speed range is k=3; S7.

2. Constructing a sensitivity index; Set the dynamic sensitivity variability for each flight speed range. , used to characterize the dynamic changes in the influence of parameters, is expressed as: ； in, , , These are the dynamic sensitivity indicators corresponding to the high-speed, medium-speed, and low-speed segments, respectively. S7.

3. Perform a comprehensive analysis of flight speed segment characteristics based on the sensitivity index constructed in step S7.2; A comprehensive parameter importance assessment system was constructed by weighting and combining the deceleration performance index, stability impact index, and energy characteristic impact index. Within each flight speed segment k, the comprehensive index was calculated. ;, S7.

4. Introduce adjustment coefficients to balance the strength and stability of the influence of parameters, and construct an importance index for key parameters. Used for quantitative assessment of parameter importance; ； in, It serves as an importance index for key parameters; The adjustment coefficients are determined by expert experience; The importance index of key parameters is classified and defined. , These are the first and second thresholds for the importance index of key parameters, determined by expert experience. when This indicates that each flight speed range has high sensitivity and significant impact, and the corresponding indicators can be used as the core parameters for control. Real-time and precise control of the corresponding indicators is required. when This indicates high sensitivity in a specific flight speed range. For the corresponding indicators, a segmented control strategy needs to be developed, and the control priority should be adjusted according to the flight phase. when When the sensitivity is low across all flight speed ranges, the corresponding index is used as an auxiliary reference parameter.

9. A sensitivity analysis system for aerodynamic deceleration control indicators of unmanned aerial vehicles (UAVs), characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed, it implements the steps of the sensitivity analysis method for aerodynamic deceleration control index of a UAV as described in any one of claims 1-8.

Images