Performance prediction method for unmanned aerial vehicle climbing process, electronic device, and storage medium

Abstract

The performance prediction method of the unmanned aerial vehicle climbing process, the electronic equipment and the storage medium belong to the unmanned aerial vehicle control technical field. In order to improve the performance prediction precision of the unmanned aerial vehicle climbing process, the input parameters of the performance prediction of the unmanned aerial vehicle climbing process are collected and normalized, including power system input parameters, motion state input parameters and environmental input parameters; a main body layer of a performance prediction model of the unmanned aerial vehicle climbing process is constructed, a three-layer architecture of a power characteristic layer, a state response layer and an interlayer interaction layer is adopted; an output index of the model is constructed, and a correlation between the main body layer of the model and the output index is established; a loss function of the performance prediction model of the unmanned aerial vehicle climbing process is constructed, including a basic loss, a power dynamic mapping loss, a state migration loss and a response feature loss, and the overall optimization of the model is carried out through a comprehensive mapping loss; the performance prediction model of the unmanned aerial vehicle climbing process is set, trained, verified and tested.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) control technology, specifically relating to a performance prediction method, electronic equipment, and storage medium for the UAV climbing process. Background Technology

[0002] With the rapid development of drone technology, its application areas are constantly expanding, ranging from agricultural plant protection to logistics delivery and many other fields. In these application scenarios, climb performance is one of the key flight performance indicators for drones. The climb process involves complex dynamic characteristics and environmental factors, making accurate prediction of climb performance crucial for ensuring flight safety and improving operational efficiency.

[0003] Traditional methods for predicting the climb performance of unmanned aerial vehicles (UAVs) mainly rely on simplified mathematical models and empirical formulas. These methods often overlook the coupling effects of multiple important factors. For example, there are complex nonlinear relationships between factors such as motor performance, aerodynamic characteristics, and environmental conditions, which traditional methods struggle to accurately describe. Furthermore, uncertainties in actual flight (such as sudden airflow and temperature changes) can significantly affect climb performance, and these factors are difficult to adequately consider in traditional prediction methods.

[0004] While deep learning methods have made some progress in UAV performance prediction in recent years, most existing deep learning models treat the prediction process as a "black box," focusing only on the correspondence between input and output, lacking description and explanation of the physical mechanisms involved in the UAV's ascent. This lack of interpretability not only makes it difficult to verify the physical rationality of the prediction results but may also produce predictions that violate physical laws. Furthermore, because the model structure cannot be specifically optimized and improved, it is difficult to diagnose and trace faults when encountering abnormal situations, and the model's generalization ability is also limited, making it difficult to handle unseen operating conditions.

[0005] Accurate prediction of multiple key performance indicators during the climb of unmanned aerial vehicles (UAVs) has significant practical application value. These indicators involve various aspects of UAVs, including maneuverability, navigation accuracy, energy management, flight safety, and attitude stability. The prediction results can be directly used to optimize flight control strategies, improve flight efficiency, ensure flight safety, and provide important reference for UAV design improvements.

[0006] Current research suffers from several problems, including oversimplified prediction models, poor environmental adaptability, insufficient prediction accuracy, lack of analysis of multi-system coupling effects, and low computational efficiency. In particular, the lack of model interpretability makes it difficult to effectively integrate physical mechanisms with prediction models, thus limiting further optimization and improvement of prediction methods. Summary of the Invention

[0007] The problem to be solved by this invention is to improve the performance prediction accuracy of UAVs during the climb process, and to propose a performance prediction method, electronic equipment and storage medium for UAVs during the climb process.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] A method for predicting the performance of a drone's climb process includes the following steps:

[0010] S1. Collect and normalize the input parameters for performance prediction during the UAV's climb process, including power system input parameters, motion state input parameters, and environmental input parameters;

[0011] S2. Construct the main layer of a performance prediction model for the UAV climbing process, adopting a three-layer architecture of dynamic characteristics layer, state response layer and inter-layer interaction layer;

[0012] S3. Based on the performance prediction model of the UAV climb process obtained in step S2, the main layer of the model is constructed to output the following indicators: dynamic characteristic indicator group and climb capability indicator group. A fully connected approach is used to establish the correlation between the main layer of the model and the output indicators. The indicators in the dynamic characteristic indicator group include thrust response time D1, attitude adjustment rate D2, velocity tracking error D3, power fluctuation rate D4, thrust-to-weight ratio dynamic value D5, and attitude stability D6. The indicators in the climb capability indicator group include average climb rate D7, energy efficiency D8, track holding accuracy D9, and power margin D1. 10 , lift-to-drag ratio D 11 and climb stability index D 12 ;

[0013] S4. Construct a loss function for a performance prediction model of the UAV climb process, including basic loss, power dynamic mapping loss, state transition loss and response feature loss, and optimize the model as a whole by integrating the mapping loss;

[0014] S5. Based on the input parameter data for the performance prediction of the UAV climbing process determined in step S1 and the normalized output index data obtained in step S3, construct the training set, validation set and test set;

[0015] S6. Set the parameters of the performance prediction model of the UAV climb process constructed in steps S2-S4, then train the performance prediction model of the UAV climb process using the training set obtained in step S5, validate it using the validation set, and test the validated performance prediction model of the UAV climb process using the test set.

[0016] Furthermore, the specific input parameters for performance prediction of the UAV's climb process in step S1 are as follows:

[0017] The power system input parameters include motor input voltage A1, motor input current A2, propeller speed A3, measured thrust A4, motor surface temperature A5, motor remaining capacity A6, motor power factor A7, and motor battery internal resistance A8.

[0018] Motion state input parameters include vertical velocity A9 and horizontal velocity A9. 10 Instantaneous height A 11 Pitch angle A 12 Total weight A 13 Vertical acceleration A 14 Dynamic pressure A 15 and lift coefficient A 16 ;

[0019] Environmental input parameters include atmospheric pressure A 17 Ambient temperature A 18 Relative humidity A 19 Horizontal wind speed A 20 Vertical wind speed A 21 Air density A 22 Gravitational acceleration A 23 Air viscosity A 24 and Reynolds number A 25 .

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

[0021] S2.1. The power characteristic layer is constructed by modeling four dimensions: motor input, thrust-weight characteristics, energy efficiency characteristics, and thrust output. Weighting coefficients and bias terms are added. The motor input sub-layer focuses on the basic power output capability; the thrust-weight characteristic sub-layer focuses on the power distribution effect; the energy efficiency characteristic sub-layer reflects the energy conversion efficiency; and the thrust output sub-layer reflects the final power effect.

[0022] S2.2. Constructing a state response layer describes the response characteristics of the UAV during the climb process from the velocity response sublayer, acceleration response sublayer, attitude response sublayer and energy efficiency sublayer, and introduces an LSTM structure to capture the temporal evolution law;

[0023] S2.3. The inter-layer interaction layer first extracts environmental impact features, dynamic state features, and motion features, and then performs feature fusion.

[0024] Furthermore, the specific implementation method of step S4 includes the following steps:

[0025] S4.1. Construct the basic loss;

[0026] The basic loss function is expressed as mean squared error, which is calculated by comparing the predicted and measured values ​​of all output indicators; this is the basic loss. ;

[0027] S4.2. Construct a dynamic power mapping loss;

[0028] Based on the dynamic evolution of temperature changes and thrust-weight characteristics during the UAV's climb, which involves the input voltage and current undergoing electromechanical conversion to generate thrust output, and characterized by power dynamic mapping loss, the expression is obtained as follows:

[0029]

[0030] in, For power dynamic mapping loss; This is the power deviation coefficient. The temperature response coefficient, This is the thrust-to-weight ratio coefficient;

[0031] S4.3. Construct the state transition loss;

[0032] The climbing process involves state-space transition characteristics, including the coupling of multi-dimensional forces and the dynamic changes in velocity and acceleration. Characterized by state transition loss, the expression is:

[0033]

[0034] in, For state transition loss, For cosine function, For the characteristic area, For air speed, For the characteristic time interval, For dynamic balance coefficient, This refers to the velocity characteristic coefficient;

[0035] S4.4. Construct the response feature loss;

[0036] The system response characteristics during the UAV's climb process exhibit a high-order nonlinear dynamic process, including the coupling of attitude dynamic characteristics, efficiency response characteristics, and stability characteristics. Characterized by response feature loss, the expression is:

[0037]

[0038] in, For response feature loss; For feature parameters, For the transmission coefficient; , , These are the weighting coefficients for attitude dynamic characteristics, efficiency response characteristics, and stability characteristics, respectively. For the conversion coefficient between angle and acceleration; For stability and lift-to-drag ratio conversion coefficient;

[0039] S4.5. Construct the comprehensive mapping loss;

[0040] Based on a unified description of dynamic characteristics, state characteristics, and response characteristics, a comprehensive mapping loss function is constructed. The expression is:

[0041]

[0042] in, , , , These are the weighting coefficients for the basic loss, power dynamic mapping loss, state transition loss, and response feature loss, respectively.

[0043] Furthermore, in step S5, the data is divided into a training set, a validation set, and a test set in a ratio of 8:1:1.

[0044] Furthermore, the model training strategy in step S6 is as follows:

[0045] The model training uses the backpropagation algorithm. During training, the model calculates the predicted value through forward propagation, substitutes the predicted value and the true value into the loss function to calculate the loss, and then updates the network parameters through backpropagation. To prevent overfitting, an early stopping strategy is adopted, and training is stopped when the validation set loss does not decrease for 10 consecutive epochs.

[0046] Once the model is trained, the relationship between the input and output is established, and the output is then restored to the actual physical quantity through inverse normalization.

[0047] An electronic device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the performance prediction method for the climb process of a UAV.

[0048] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned performance prediction method for the climb process of a UAV.

[0049] The beneficial effects of this invention are:

[0050] The present invention discloses a performance prediction method for the climbing process of a UAV. It constructs a prediction model that combines the advantages of deep learning with the interpretability of physical mechanisms, realizes comprehensive analysis of multi-dimensional parameters during the UAV climbing process, ensures that the model has good adaptability and robustness to dynamic environments, and guarantees the accuracy and real-time performance of multiple performance index predictions.

[0051] This invention discloses a performance prediction method for the climb process of a UAV. By comprehensively considering 25 input parameters, covering multiple dimensions such as motor performance, flight state, and environmental conditions, it ensures the comprehensiveness of the prediction. It innovatively divides the deep learning model into a three-layer architecture: "power characteristic layer - state response layer - inter-layer interaction," and constructs a loss function through three parts: power dynamic mapping loss, state transition loss, and response feature loss. This allows the model to maintain the advantages of deep learning while reflecting the physical mechanism of the UAV climb process. This method not only provides strong physical interpretability but also achieves high-precision simultaneous prediction of multiple indicators. It has good environmental adaptability and engineering practical value, providing comprehensive and reliable technical support for UAV flight control and mission planning. Attached Figure Description

[0052] Figure 1 This is a flowchart of a performance prediction method for the climbing process of a UAV as described in this invention;

[0053] Figure 2 This is a loss curve diagram of the training process of the performance prediction model for the UAV's climb process in this invention;

[0054] Figure 3 This is a comparison chart of the predicted and actual values ​​of the performance prediction model for the UAV's climb process according to the present invention. Detailed Implementation

[0055] 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.

[0056] 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.

[0057] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 -Appendix Figure 3 Detailed explanation is as follows:

[0058] Example 1:

[0059] A method for predicting the performance of a drone's climb process includes the following steps:

[0060] S1. Collect and normalize the input parameters for performance prediction during the UAV's climb process, including power system input parameters, motion state input parameters, and environmental input parameters;

[0061] Furthermore, the specific input parameters for performance prediction of the UAV's climb process in step S1 are as follows:

[0062] The power system input parameters include motor input voltage A1, motor input current A2, propeller speed A3, measured thrust A4, motor surface temperature A5, motor remaining capacity A6, motor power factor A7, and motor battery internal resistance A8.

[0063] Motion state input parameters include vertical velocity A9 and horizontal velocity A9. 10 Instantaneous height A 11 Pitch angle A 12 Total weight A 13 Vertical acceleration A 14 Dynamic pressure A 15 and lift coefficient A 16 ;

[0064] Environmental input parameters include atmospheric pressure A 17 Ambient temperature A 18 Relative humidity A 19 Horizontal wind speed A 20 Vertical wind speed A 21 Air density A 22 Gravitational acceleration A 23 Air viscosity A 24 and Reynolds number A 25 .

[0065] Meanwhile, to eliminate the dimensional differences between different physical quantities and improve the training effect and prediction accuracy of the model, all input parameters and output indicators were normalized and uniformly mapped to the [0,1] interval. Normalization not only speeds up the model convergence but also avoids numerical overflow caused by excessive differences in magnitude during numerical calculations. Furthermore, it helps improve the weight balance of different features in the model.

[0066] Specifically, for the input parameters: the motor input voltage is normalized by dividing by the rated voltage; the motor input current is normalized by dividing by the rated current; the propeller speed is the actual speed divided by the maximum design speed; the measured thrust is divided by the theoretical maximum thrust; the motor surface temperature is linearly mapped based on the highest and lowest temperature ranges statistically analyzed from historical operating data; the remaining battery capacity is itself in percentage form; the power factor is originally in the range of 0-1; and the battery internal resistance is the measured value divided by the nominal internal resistance.

[0067] Furthermore, Table 1 shows some of the input data for the performance prediction model of the UAV's climb process;

[0068] Table 1

[0069]

[0070] For flight status parameters: vertical speed and horizontal speed are divided by their respective historical maximum speeds; instantaneous altitude is normalized based on the design maximum takeoff altitude; pitch angle is linearly mapped based on the angle range statistically determined from historical flight data; total weight is the actual weight divided by the maximum takeoff weight; acceleration indicators are normalized based on the maximum acceleration statistically determined from historical data; dynamic pressure and lift coefficient are mapped based on the range determined by theoretical calculations and historical data.

[0071] Regarding environmental parameters: atmospheric pressure is normalized based on standard atmospheric pressure; temperature and humidity are linearly mapped based on the expected operating environment range; wind speed is normalized based on the maximum wind speed allowed by the design; air density, viscosity, and other parameters are normalized based on standard atmospheric conditions; and the Reynolds number is mapped based on the numerical range under typical operating conditions.

[0072] This normalization method, based on practical engineering experience and historical data statistics, ensures the scientific validity and rationality of the normalization process. This hierarchical characteristic parameter system design not only fully describes the entire process of energy input, conversion, and output, but also demonstrates clear correlations between parameters, accurately reflecting the system's operating state and performance characteristics.

[0073] S2. Construct the main layer of a performance prediction model for the UAV climbing process, adopting a three-layer architecture of dynamic characteristics layer, state response layer and inter-layer interaction layer;

[0074] The climb process of an unmanned aerial vehicle (UAV) is a complex process involving continuous interaction between power output and state response. Climb performance is primarily determined by two core factors: first, the output characteristics of the power system, including motor input, thrust-weight ratio, energy efficiency, and thrust output; and second, the state response characteristics of the aircraft, including velocity, acceleration, attitude, and energy efficiency response. These two factors influence and constrain each other; therefore, the model design adopts a three-layer architecture: "power characteristics layer - state response layer - inter-layer interaction." The power characteristics layer is responsible for characterizing the power output patterns, the state response layer describes the motion characteristics, and the inter-layer interaction design captures the coupling relationship between the two, thereby achieving accurate prediction of climb performance.

[0075] Furthermore, the specific implementation method of step S2 includes the following steps:

[0076] S2.1. The power characteristic layer is constructed by modeling four dimensions: motor input, thrust-weight characteristics, energy efficiency characteristics, and thrust output. Weighting coefficients and bias terms are added. The motor input sub-layer focuses on the basic power output capability; the thrust-weight characteristic sub-layer focuses on the power distribution effect; the energy efficiency characteristic sub-layer reflects the energy conversion efficiency; and the thrust output sub-layer reflects the final power effect.

[0077] S2.1.1. The design of the motor input sublayer uses the ReLU activation function for nonlinear mapping. This function multiplies all power system input parameters and environmental parameters (atmospheric pressure, ambient temperature, and relative humidity) by their corresponding weighting coefficients, sums the results, adds a bias term, and finally obtains the output value of the motor input sublayer through the ReLU function. ;

[0078] S2.1.2. The thrust-weight characteristic sublayer uses the ReLU activation function to focus on the relationship between thrust and weight, utilizing... Characterizing thrust-to-weight ratio, Characterizing the relationship between power consumption and battery capacity, The nonlinear effect of wind speed on thrust-weight characteristics is characterized, and the output value of the thrust-weight characteristic sublayer is obtained. The expression is:

[0079]

[0080] in, To push the bias term of the feature sublayer, The weighting coefficient is used to influence the push-to-weight ratio. The power consumption rate weighting coefficient. This represents the weighting coefficient of vertical acceleration on the thrust-weight characteristic. This represents the weighting coefficient of dynamic pressure on the thrust-weight characteristic. This is the wind field disturbance weighting coefficient. This represents the weighting coefficient of air pressure on thrust-weight characteristics. This is the weighting coefficient of gravitational acceleration on the thrust-weight characteristic;

[0081] S2.1.3. The energy efficiency characteristic sublayer uses the sigmoid activation function to focus on energy conversion efficiency, utilizing... Characterizing the effect of temperature gradient on energy conversion efficiency Characterizing the impact of aerodynamic effects on energy conversion, the output value of the energy efficiency characteristic sublayer is obtained. The expression is:

[0082]

[0083] in, This is a bias term for the energy efficiency characteristic sublayer; This is the weighting coefficient for temperature difference loss. The weighting coefficient for the impact of the motor's remaining capacity on energy efficiency. The power factor is the weighting factor for the impact of energy efficiency. The weighting coefficient for the impact of the motor battery internal resistance on energy efficiency. This is a dynamic energy consumption weighting coefficient. Aerodynamic efficiency weighting coefficient Adjust the weighting coefficients for density;

[0084] S2.1.4. The thrust output sublayer uses the tanh function to focus on the final output effect, utilizing... The coupling effect between air and gravity loads is characterized to obtain the output value of the thrust output sublayer. The expression is:

[0085]

[0086] in, For the bias terms of the inference output sublayer; The weighting coefficient for the effect of rotational speed on thrust output. This is a weighting coefficient representing the effect of measured thrust on thrust output. The weighting coefficient for the effect of vertical velocity on thrust output. The horizontal velocity is the weighting factor for the impact of thrust output. The weighting factor for the effect of altitude on thrust output This is the weighting coefficient for the effect of pitch angle on thrust output. This is the weighting factor for the effect of the lift coefficient on thrust output. This is the weighting coefficient for the air-gravity coupling effect;

[0087] S2.1.5. Constructing the output of the dynamic characteristic layer fusion The expression is:

[0088]

[0089] in, , , , These are the weighting coefficients for the motor input sublayer, thrust-weight characteristic sublayer, energy efficiency characteristic sublayer, and thrust output sublayer, respectively.

[0090] During the drone's ascent, the power system parameters, environmental parameters, and motion parameters are closely interrelated. As the drone begins to climb, the motors need to provide greater power output to overcome gravity. Simultaneously, environmental conditions, such as changes in air density with altitude, affect thrust output, while the motion state, in turn, influences power requirements. Therefore, the power characteristics layer needs to be modeled from four dimensions: motor input, thrust-weight characteristics, energy efficiency characteristics, and thrust output. Specifically: the motor input sublayer focuses on basic power output capability; the thrust-weight characteristics sublayer emphasizes power distribution effectiveness; the energy efficiency characteristics sublayer reflects energy conversion efficiency; and the thrust output sublayer reflects the final power effect. This layered design comprehensively captures the characteristic changes of the power system during the ascent process.

[0091] S2.2. Constructing a state response layer describes the response characteristics of the UAV during the climb process from the velocity response sublayer, acceleration response sublayer, attitude response sublayer and energy efficiency sublayer, and introduces an LSTM structure to capture the temporal evolution law;

[0092] S2.2.1. The speed response sublayer integrates motor operating parameters, flight state parameters, and environmental parameters, employs the ReLU activation function to capture dynamic characteristics during speed changes, and utilizes... This reflects the direct relationship between the motor input power and speed response, yielding the output value of the speed response sublayer. The expression is:

[0093]

[0094] in, For the bias term of the velocity response sublayer; The weighting coefficient for the influence of vertical velocity. The weighting coefficient for the influence of horizontal velocity. The weighting coefficient for the effect of altitude on velocity response. The weighting coefficient for the influence of horizontal wind speed on the velocity response. This represents the weighting coefficient for the influence of vertical wind speed on the velocity response. The weighting coefficient for the effect of input power on speed response. The weighting factor for the influence of thrust velocity response. The weighting coefficient for the effect of air density on velocity response;

[0095] S2.2.2. The acceleration response sublayer uses the tanh function to describe the dynamic characteristics during acceleration, comprehensively considering thrust system parameters and aerodynamic characteristics to achieve accurate modeling of acceleration performance. Considering the impact of load variation on acceleration characteristics, the acceleration capability of the UAV is reflected, and the output value of the acceleration response sublayer is obtained. The expression is:

[0096]

[0097] in, For the bias term of the acceleration response sublayer; , , , , , , The weighting coefficients for the effects of vertical acceleration, thrust-to-weight ratio, gravitational acceleration, aerodynamic pressure, propeller speed, lift coefficient, and air viscosity on acceleration characteristics are respectively.

[0098] S2.2.3. The attitude response sublayer uses the sigmoid function to characterize the attitude adjustment process, coupling flight attitude parameters with the influence of environmental factors to construct a dynamic attitude response model. The intensity of horizontal wind disturbance experienced by an aircraft during vertical motion directly affects its attitude stability; The intensity of the disturbance of updrafts or downdrafts on the horizontal motion of the aircraft is reflected, which affects the attitude control requirements, and the output value of the attitude response sublayer is obtained. The expression is:

[0099]

[0100] in, For the bias term of the attitude response sublayer; , , , , , These are the weighting coefficients for the effects of pitch angle, dynamic pressure, lateral disturbance, longitudinal disturbance, Reynolds number, and atmospheric pressure on the attitude of the UAV.

[0101] S2.2.4. The energy efficiency sublayer establishes an energy efficiency model using the ReLU function, combining motor characteristics, thrust characteristics, and environmental parameters to reflect the efficiency variation during energy conversion. It considers energy conversion efficiency under environmental disturbances and utilizes... The effectiveness of thrust output under given wind field conditions, i.e., the ability of thrust output to counteract air kinetic energy loss, is reflected in the output value of the energy efficiency sublayer. The expression is:

[0102]

[0103] in, This is the bias term for the energy efficiency sublayer; , , , , , These are the weighting coefficients for the effects of lift coefficient, kinetic energy compensation, motor remaining capacity, motor power factor, motor battery internal resistance, and relative humidity on energy efficiency.

[0104] S2.2.5. An LSTM network is used to process the output sequences of the velocity response sublayer, acceleration response sublayer, attitude response sublayer, and energy efficiency sublayer to perform deep mining of the temporal features of the state response, thereby obtaining the output value of the temporal feature embedding layer. The expression is:

[0105]

[0106] in, It is a long short-term memory neural network module;

[0107] S2.2.6. Construct the output of the state response layer fusion The expression is:

[0108]

[0109] in, The weighting coefficients are the output values ​​of the temporal feature embedding layer. Weighting coefficients for fusing features from the velocity response sublayer, acceleration response sublayer, attitude response sublayer, and energy efficiency sublayer;

[0110] During a climb, a drone undergoes several phases, including accelerated climb, constant speed climb, and deceleration adjustment. Each phase involves complex processes of velocity changes, attitude adjustments, and energy allocation. For example, during the accelerated climb phase, it is necessary to coordinate the relationship between thrust increase and attitude adjustment; during the constant speed climb phase, it is necessary to precisely balance the effects of various forces; and during the adjustment phase, the impact of environmental disturbances must be considered. Therefore, the state response layer describes these complex response characteristics through four sub-layers: velocity response, acceleration response, attitude response, and energy efficiency, and introduces an LSTM structure to capture the temporal evolution.

[0111] S2.3. The inter-layer interaction layer first extracts environmental impact features, dynamic state features, and motion features, and then performs feature fusion;

[0112] There are multiple coupling relationships during the climb: power output affects flight status, which in turn affects power demand; environmental conditions simultaneously influence both the propulsion system and flight status; and various parameters are also correlated. For example, when the ambient temperature rises, it not only reduces motor efficiency but also affects air density, thus altering the required thrust. This complex coupling relationship needs to be described through a specialized interaction mechanism.

[0113] S2.3.1. The environmental impact feature sublayer assigns corresponding weight coefficients to the environmental input parameters. The sum of the weighted environmental input parameters is processed using the ReLU function to obtain the output value E of the environmental impact feature sublayer. This design fully considers the comprehensive impact of external environmental conditions on the UAV's climb performance and can effectively characterize the system response characteristics caused by environmental changes. Environmental impact feature sublayer output. This will serve as an important input for subsequent inter-layer interaction analysis, used to assess the impact of environmental factors on the overall performance of the UAV.

[0114] S2.3.2. The dynamic state characteristic sublayer assigns corresponding weighting coefficients to the input parameters of the power system. The sum of the weighted input parameters is processed using the ReLU function to obtain the output value F of the dynamic state characteristic sublayer. This design method effectively captures the operating state characteristics of the motor and thrust system, accurately reflecting the output characteristics of the power system. (Power State Characteristic Sublayer Output) This will serve as an important basis for subsequent inter-layer interaction analysis, used to assess the impact of the power system on the overall performance of the UAV.

[0115] S2.3.3. The motion feature sublayer assigns corresponding weight coefficients to the motion state input parameters. The sum of the weighted motion state input parameters is processed using the ReLU function to obtain the output value G of the motion feature sublayer. This design method comprehensively characterizes the kinematic properties of the UAV and can effectively reflect the state changes of the aircraft during space motion. Motion feature sublayer output. This data will serve as the basis for subsequent inter-layer interaction analysis, used to assess the impact of motion state on the overall performance of the drone.

[0116] S2.3.4. Constructing the output of inter-layer interaction layer fusion The expression is:

[0117]

[0118] in, , , , , These are the weight coefficients corresponding to the fusion output of the dynamic characteristic layer, the fusion output of the state response layer, the output of the environmental influence feature sub-layer, the output of the dynamic state feature sub-layer, and the output of the motion feature sub-layer, respectively.

[0119] Furthermore, this three-layer architecture design fully considers the various complex relationships during the UAV's climb process. Through reasonable parameter combinations and feature extraction, it can accurately describe and predict climb performance. At the same time, the model has good interpretability, and the role of each layer and sublayer has a clear physical meaning.

[0120] S3. Based on the performance prediction model of the UAV climb process obtained in step S2, the main layer of the model is constructed to output the following indicators: dynamic characteristic indicator group and climb capability indicator group. A fully connected approach is used to establish the correlation between the main layer of the model and the output indicators. The indicators in the dynamic characteristic indicator group include thrust response time D1, attitude adjustment rate D2, velocity tracking error D3, power fluctuation rate D4, thrust-to-weight ratio dynamic value D5, and attitude stability D6. The indicators in the climb capability indicator group include average climb rate D7, energy efficiency D8, track holding accuracy D9, and power margin D1. 10 , lift-to-drag ratio D 11 and climb stability index D 12 ;

[0121] Furthermore, the specific implementation method of step S3 includes the following steps:

[0122] S3.1. Establish the correlation between the main body of the model and the indicators in the dynamic characteristic indicator group;

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129] in, , , , , , These are the fully connected components corresponding to thrust response time, attitude adjustment rate, velocity tracking error, power fluctuation rate, thrust-to-weight ratio dynamic value, and attitude stability, respectively.

[0130] S3.2. Establish the correlation between the main body of the model and the indicators in the climbing ability indicator group;

[0131]

[0132]

[0133]

[0134]

[0135]

[0136]

[0137] in, , , , , , These are the full connections corresponding to the average rate of climb, energy efficiency index, track-keeping accuracy, power margin, lift-to-drag ratio, and climb stability index, respectively.

[0138] Furthermore, during the UAV's climb, system performance manifests in two core aspects: dynamic characteristics and climb capability. This dictates that the design of output metrics must encompass both dimensions. It's important to note that output metrics require normalization before use, similar to the normalization of input parameters. Output metric normalization is also based on statistical analysis and theoretical calculations: thrust response time is divided by the maximum allowable response time required by the design specifications; attitude adjustment rate and velocity tracking error are mapped based on control accuracy requirements; power fluctuation rate and thrust-to-weight ratio are normalized using theoretically calculated optimal values; attitude stability, climb rate, energy efficiency, and other metrics are all normalized based on reasonable variation ranges obtained from extensive historical flight data statistics. Track holding accuracy, power margin, lift-to-drag ratio, and climb stability index are also normalized within ranges determined based on design specifications and actual operational data.

[0139] S4. Construct a loss function for a performance prediction model of the UAV climb process, including basic loss, power dynamic mapping loss, state transition loss and response feature loss, and optimize the model as a whole by integrating the mapping loss;

[0140] Furthermore, the specific implementation method of step S4 includes the following steps:

[0141] S4.1. Construct the basic loss;

[0142] The basic loss function is expressed as mean squared error, which is calculated by comparing the predicted and measured values ​​of all output indicators; this is the basic loss. ;

[0143] S4.2. Construct a dynamic power mapping loss;

[0144] Based on the dynamic evolution of temperature changes and thrust-weight characteristics during the UAV's climb, which involves the input voltage and current undergoing electromechanical conversion to generate thrust output, and characterized by power dynamic mapping loss, the expression is obtained as follows:

[0145]

[0146] in, For power dynamic mapping loss; This is the power deviation coefficient. The temperature response coefficient, This is the thrust-to-weight ratio coefficient;

[0147] S4.3. Construct the state transition loss;

[0148] The climbing process involves state-space transition characteristics, including the coupling of multi-dimensional forces and the dynamic changes in velocity and acceleration. Characterized by state transition loss, the expression is:

[0149]

[0150] in, For state transition loss, For cosine function, For the characteristic area, For air speed, For the characteristic time interval, For dynamic balance coefficient, This refers to the velocity characteristic coefficient;

[0151] S4.4. Construct the response feature loss;

[0152] The system response characteristics during the UAV's climb process exhibit a high-order nonlinear dynamic process, including the coupling of attitude dynamic characteristics, efficiency response characteristics, and stability characteristics. Characterized by response feature loss, the expression is:

[0153]

[0154] in, For response feature loss; For feature parameters, For the transmission coefficient; , , These are the weighting coefficients for attitude dynamic characteristics, efficiency response characteristics, and stability characteristics, respectively. For the conversion coefficient between angle and acceleration; For stability and lift-to-drag ratio conversion coefficient;

[0155] S4.5. Construct the comprehensive mapping loss;

[0156] Based on a unified description of dynamic characteristics, state characteristics, and response characteristics, a comprehensive mapping loss function is constructed. The expression is:

[0157]

[0158] in, , , , These are the weighting coefficients for the basic loss, power dynamic mapping loss, state transition loss, and response feature loss, respectively.

[0159] S5. Based on the input parameter data for the performance prediction of the UAV climbing process determined in step S1 and the normalized output index data obtained in step S3, construct the training set, validation set and test set;

[0160] Furthermore, in step S5, the data is divided into a training set, a validation set, and a test set in a ratio of 8:1:1.

[0161] S6. Set the parameters of the performance prediction model for the UAV climb process constructed in steps S2-S4, then train the performance prediction model for the UAV climb process using the training set obtained in step S5, validate it using the validation set, and test the validated performance prediction model for the UAV climb process using the test set.

[0162] Furthermore, the model training strategy in step S6 is as follows:

[0163] The model training uses the backpropagation algorithm. During training, the model calculates the predicted value through forward propagation, substitutes the predicted value and the true value into the loss function to calculate the loss, and then updates the network parameters through backpropagation. To prevent overfitting, an early stopping strategy is adopted, and training is stopped when the validation set loss does not decrease for 10 consecutive epochs.

[0164] Once the model is trained, the relationship between the input and output is established, and the output is then restored to the actual physical quantity through inverse normalization.

[0165] Furthermore, the data sources include commercial flight data obtained from the Civil Aviation Administration of China, test data obtained from large UAV manufacturers, design parameters obtained from UAV design specifications, experimental data obtained from UAV laboratory tests, and monitoring data obtained from UAV monitoring and management platforms. This data includes input data A1~A25 and output data D1~D12.

[0166] Furthermore, the model parameter settings include hyperparameters that need to be manually set based on expert experience, such as network structure parameters, number of neurons in each layer, and LSTM hidden layer dimension; hyperparameters that need to be manually set during model training, such as learning rate, batch size, number of training epochs, and weight coefficients in the loss function; and parameters that need to be determined by the model during training, such as weight coefficients and bias terms for each layer, and weight coefficients for each fusion layer.

[0167] Furthermore, some input data for the performance prediction model of the UAV's climb process are shown in Table 2. Figure 3 This invention demonstrates the technical effectiveness of a performance prediction method for the climb process of a UAV.

[0168] Table 2

[0169]

[0170] Example 2:

[0171] An electronic device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the performance prediction method for the climb process of a UAV as described in Embodiment 1.

[0172] The computer device of the present invention may include a processor and a memory, such as a microcontroller containing a central processing unit. The processor executes the computer program stored in the memory to implement the steps of the aforementioned performance prediction method for the climbing process of a UAV. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0173] Example 3:

[0174] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the performance prediction method for the climb process of a UAV as described in Embodiment 1.

[0175] The computer-readable storage medium of the present invention can be any form of storage medium that can be read by the processor of a computer device, including but not limited to non-volatile memory, volatile memory, ferroelectric memory, etc. The computer-readable storage medium stores a computer program. When the processor of the computer device reads and executes the computer program stored in the memory, the aforementioned steps of the UAV climbing process can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0176] 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.

[0177] 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 method for predicting the performance of a UAV during its climb process, characterized in that, Includes the following steps: S1. Collect the input parameters for performance prediction during the UAV's climb process and normalize them. The specific input parameters are as follows; The power system input parameters include motor input voltage A1, motor input current A2, propeller speed A3, measured thrust A4, motor surface temperature A5, motor remaining capacity A6, motor power factor A7, and motor battery internal resistance A8. Motion state input parameters include vertical velocity A9 and horizontal velocity A9. 10 Instantaneous height A 11 Pitch angle A 12 Total weight A 13 Vertical acceleration A 14 Dynamic pressure A 15 and lift coefficient A 16 ; Environmental input parameters include atmospheric pressure A 17 Ambient temperature A 18 Relative humidity A 19 Horizontal wind speed A 20 Vertical wind speed A 21 Air density A 22 Gravitational acceleration A 23 Air viscosity A 24 and Reynolds number A 25 ; S2. Construct the main layer of a performance prediction model for the UAV climb process, adopting a three-layer architecture consisting of a dynamic characteristic layer, a state response layer, and an inter-layer interaction layer, including the following steps: S2.

1. The power characteristic layer is constructed by modeling the motor input, thrust-weight characteristics, energy efficiency characteristics and thrust output from four dimensions, and weighting coefficients and bias terms are added. The motor input sub-layer focuses on the basic power output capability. The thrust-weight characteristic sublayer focuses on power distribution effectiveness; the energy efficiency characteristic sublayer reflects energy conversion efficiency. The thrust output sublayer reflects the final dynamic effect; S2.

2. Constructing a state response layer describes the response characteristics of the UAV during the climb process from the velocity response sublayer, acceleration response sublayer, attitude response sublayer and energy efficiency sublayer, and introduces an LSTM structure to capture the temporal evolution law; S2.

3. The inter-layer interaction layer first extracts environmental impact features, dynamic state features, and motion features, and then performs feature fusion; S3. Based on the performance prediction model of the UAV climb process obtained in step S2, the main layer of the model is constructed to output the following indicators: dynamic characteristic indicator group and climb capability indicator group. A fully connected approach is used to establish the correlation between the main layer of the model and the output indicators. The indicators in the dynamic characteristic indicator group include thrust response time D1, attitude adjustment rate D2, velocity tracking error D3, power fluctuation rate D4, thrust-to-weight ratio dynamic value D5, and attitude stability D6. The indicators in the climb capability indicator group include average climb rate D7, energy efficiency D8, track holding accuracy D9, and power margin D1. 10 , lift-to-drag ratio D 11 and climb stability index D 12 ; S4. Construct a loss function for a performance prediction model of the UAV climb process, including basic loss, power dynamic mapping loss, state transition loss and response feature loss, and optimize the model as a whole by integrating the mapping loss; S5. Based on the input parameter data for the performance prediction of the UAV climbing process determined in step S1 and the output index data obtained in step S3, construct the training set, validation set and test set; S6. Set the parameters of the performance prediction model of the UAV climb process constructed in steps S2-S4, then train the performance prediction model of the UAV climb process using the training set obtained in step S5, validate it using the validation set, and test the validated performance prediction model of the UAV climb process using the test set.

2. The performance prediction method for the climb process of a UAV according to claim 1, characterized in that, The specific implementation method of step S2 includes the following steps: S2.1.

1. The design of the motor input sublayer uses the ReLU activation function for nonlinear mapping. This function multiplies all power system input parameters and environmental input parameters (atmospheric pressure, ambient temperature, and relative humidity) by their corresponding weighting coefficients, sums the results, adds a bias term, and finally obtains the output value of the motor input sublayer through the ReLU function. ; S2.1.

2. The thrust-weight characteristic sublayer uses the ReLU activation function to focus on the relationship between thrust and weight, utilizing... Characterizing thrust-to-weight ratio, This characterizes the relationship between power consumption and battery capacity. The nonlinear effect of wind speed on thrust-weight characteristics is characterized, and the output value of the thrust-weight characteristic sublayer is obtained. The expression is: ； in, To push the bias term of the feature sublayer, The weighting coefficient is used to influence the push-to-weight ratio. The power consumption rate weighting coefficient. This represents the weighting coefficient of vertical acceleration on the thrust-weight characteristic. This represents the weighting coefficient of dynamic pressure on the thrust-weight characteristic. This is the wind field disturbance weighting coefficient. This represents the weighting coefficient of air pressure on thrust-weight characteristics. This is the weighting coefficient of gravitational acceleration on the thrust-weight characteristic; S2.1.

3. The energy efficiency characteristic sublayer uses the sigmoid activation function to focus on energy conversion efficiency, utilizing... Characterizing the effect of temperature gradient on energy conversion efficiency Characterizing the impact of aerodynamic effects on energy conversion, the output value of the energy efficiency characteristic sublayer is obtained. The expression is: ； in, This is a bias term for the energy efficiency characteristic sublayer; This is the weighting coefficient for temperature difference loss. The weighting coefficient for the impact of the motor's remaining capacity on energy efficiency. The power factor is the weighting factor for the impact of energy efficiency. The weighting coefficient for the impact of the motor battery internal resistance on energy efficiency. This is a dynamic energy consumption weighting coefficient. Aerodynamic efficiency weighting coefficient Adjust the weighting coefficients for density; S2.1.

4. The thrust output sublayer uses the tanh function to focus on the final output effect, utilizing... The coupling effect between air and gravity loads is characterized to obtain the output value of the thrust output sublayer. The expression is: ； in, For the bias terms of the inference output sublayer; The weighting coefficient for the effect of rotational speed on thrust output. This is a weighting coefficient representing the effect of measured thrust on thrust output. The weighting coefficient for the effect of vertical velocity on thrust output. The horizontal velocity is the weighting factor for the impact of thrust output. The weighting factor for the effect of altitude on thrust output This is the weighting coefficient for the effect of pitch angle on thrust output. This is the weighting factor for the effect of the lift coefficient on thrust output. This is the weighting coefficient for the air-gravity coupling effect; S2.1.

5. Construct the output of the dynamic characteristic layer fusion The expression is: ； in, , , , These are the weighting coefficients for the motor input sublayer, thrust-weight characteristic sublayer, energy efficiency characteristic sublayer, and thrust output sublayer, respectively. S2.2.

1. The speed response sublayer integrates motor operating parameters, flight state parameters, and environmental parameters, employing the ReLU activation function to capture dynamic characteristics during speed changes. This reflects the direct relationship between the motor's input power and speed response, yielding the output value of the speed response sublayer. The expression is: ； in, For the bias term of the velocity response sublayer; The weighting coefficient for the influence of vertical velocity. The weighting coefficient for the influence of horizontal velocity. The weighting coefficient for the effect of altitude on velocity response. The weighting coefficient for the influence of horizontal wind speed on the velocity response. This represents the weighting coefficient for the influence of vertical wind speed on the velocity response. The weighting coefficient for the effect of input power on speed response. The weighting factor for the influence of thrust velocity response. The weighting coefficient for the effect of air density on velocity response; S2.2.

2. The acceleration response sublayer uses the tanh function to describe the dynamic characteristics during acceleration, comprehensively considering thrust system parameters and aerodynamic characteristics to achieve accurate modeling of acceleration performance. Considering the impact of load variation on acceleration characteristics, the acceleration capability of the UAV is reflected, and the output value of the acceleration response sublayer is obtained. The expression is: ； in, For the bias term of the acceleration response sublayer; , , , , , , The weighting coefficients for the effects of vertical acceleration, thrust-to-weight ratio, gravitational acceleration, aerodynamic pressure, propeller speed, lift coefficient, and air viscosity on acceleration characteristics are respectively. S2.2.

3. The attitude response sublayer uses the sigmoid function to characterize the attitude adjustment process, coupling flight attitude parameters with the influence of environmental factors to construct a dynamic attitude response model. The intensity of horizontal wind disturbance experienced by an aircraft during vertical motion directly affects its attitude stability; The intensity of the disturbance of updrafts or downdrafts on the horizontal motion of the aircraft is reflected, which affects the attitude control requirements, and the output value of the attitude response sublayer is obtained. The expression is: ； in, For the bias term of the attitude response sublayer; , , , , , These are the weighting coefficients for the effects of pitch angle, dynamic pressure, lateral disturbance, longitudinal disturbance, Reynolds number, and atmospheric pressure on the attitude of the UAV. S2.2.

4. The energy efficiency sublayer establishes an energy efficiency model using the ReLU function, combining motor characteristics, thrust characteristics, and environmental parameters to reflect the efficiency variation during energy conversion. It considers energy conversion efficiency under environmental disturbances and utilizes... The effectiveness of thrust output under given wind field conditions, i.e., the ability of thrust output to counteract air kinetic energy loss, is reflected in the output value of the energy efficiency sublayer. The expression is: ； in, This is a bias term for the energy efficiency sublayer; , , , , , These are the weighting coefficients for the effects of lift coefficient, kinetic energy compensation, motor remaining capacity, motor power factor, motor battery internal resistance, and relative humidity on energy efficiency. S2.2.

5. An LSTM network is used to process the output sequences of the velocity response sublayer, acceleration response sublayer, attitude response sublayer, and energy efficiency sublayer to perform deep mining of the temporal features of the state response, thereby obtaining the output value of the temporal feature embedding layer. The expression is: ； in, It is a long short-term memory neural network module; S2.2.

6. Construct the output of the state response layer fusion The expression is: ； in, The weighting coefficients are the output values ​​of the temporal feature embedding layer. Weighting coefficients for fusing features from the velocity response sublayer, acceleration response sublayer, attitude response sublayer, and energy efficiency sublayer; S2.3.

1. The environmental impact feature sublayer assigns corresponding weight coefficients to the environmental input parameters, and processes the sum of the weighted environmental input parameters through the ReLU function to obtain the output value E of the environmental impact feature sublayer; S2.3.

2. The dynamic state feature sublayer assigns corresponding weight coefficients to the input parameters of the dynamic system, and processes the sum of the weighted input parameters of the dynamic system through the ReLU function to obtain the output value F of the dynamic state feature sublayer; S2.3.

3. The motion feature sublayer assigns corresponding weight coefficients to the motion state input parameters, and processes the weighted sum of the motion state input parameters through the ReLU function to obtain the output value G of the motion feature sublayer; S2.3.

4. Constructing the output of inter-layer interaction layer fusion The expression is: ； in, , , , , These are the weight coefficients corresponding to the fusion output of the dynamic characteristic layer, the fusion output of the state response layer, the output of the environmental influence feature sub-layer, the output of the dynamic state feature sub-layer, and the output of the motion feature sub-layer, respectively.

3. The performance prediction method for the climb process of a UAV according to claim 2, characterized in that, The specific implementation method of step S3 includes the following steps: S3.

1. Establish the correlation between the main body of the model and the indicators in the dynamic characteristic indicator group; ； ； ； ； ；； ； in, , , , , , These are the fully connected components corresponding to thrust response time, attitude adjustment rate, velocity tracking error, power fluctuation rate, thrust-to-weight ratio dynamic value, and attitude stability, respectively. S3.

2. Establish the correlation between the main body of the model and the indicators in the climbing ability indicator group; ； ； ； ； ； ； in, , , , , , These are the full connections corresponding to the average rate of climb, energy efficiency index, track-keeping accuracy, power margin, lift-to-drag ratio, and climb stability index, respectively.

4. The performance prediction method for the climb process of a UAV according to claim 3, characterized in that, The specific implementation method of step S4 includes the following steps: S4.

1. Construct the basic loss; The basic loss function is expressed as mean squared error, which is calculated by comparing the predicted and measured values ​​of all output indicators; this is the basic loss. ; S4.

2. Construct a dynamic power mapping loss; Based on the dynamic evolution of temperature changes and thrust-weight characteristics during the UAV's climb, which involves the input voltage and current undergoing electromechanical conversion to generate thrust output, and characterized by power dynamic mapping loss, the expression is obtained as follows: ； in, For power dynamic mapping loss; This is the power deviation coefficient. The temperature response coefficient, This is the thrust-to-weight ratio coefficient; S4.

3. Construct the state transition loss; The climbing process involves state-space transition characteristics, including the coupling of multi-dimensional forces and the dynamic changes in velocity and acceleration. Characterized by state transition loss, the expression is: ； in, For state transition loss, For cosine function, For the characteristic area, For air speed, For the characteristic time interval, For dynamic balance coefficient, This refers to the velocity characteristic coefficient; S4.

4. Construct the response feature loss; The system response characteristics during the UAV's climb process exhibit a high-order nonlinear dynamic process, including the coupling of attitude dynamic characteristics, efficiency response characteristics, and stability characteristics. Characterized by response feature loss, the expression is: ； in, For response feature loss; For characteristic parameters, For the transmission coefficient; , , These are the weighting coefficients for attitude dynamic characteristics, efficiency response characteristics, and stability characteristics, respectively. For the conversion coefficient between angle and acceleration; For stability and lift-to-drag ratio conversion coefficient; S4.

5. Construct the comprehensive mapping loss; Based on a unified description of dynamic characteristics, state characteristics, and response characteristics, a comprehensive mapping loss function is constructed. The expression is: ； in, , , , These are the weighting coefficients for the basic loss, power dynamic mapping loss, state transition loss, and response feature loss, respectively.

5. The performance prediction method for the climb process of a UAV according to claim 4, characterized in that, Step S5 divides the data into training set, validation set and test set in a ratio of 8:1:

1.

6. The performance prediction method for the climb process of a UAV according to claim 5, characterized in that, The model training strategy for step S6 is as follows: The model training uses the backpropagation algorithm. During training, the model calculates the predicted value through forward propagation, substitutes the predicted value and the true value into the loss function to calculate the loss, and then updates the network parameters through backpropagation. To prevent overfitting, an early stopping strategy is adopted, and training is stopped when the validation set loss does not decrease for 10 consecutive epochs. Once the model is trained, the relationship between the input and output is established, and the output is then restored to the actual physical quantity through inverse normalization.

7. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the performance prediction method for the climb process of a UAV as described in any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the performance prediction method for the climbing process of a UAV as described in any one of claims 1-6.

Images