Airworthiness Evaluation Method, Device and Medium for Handling Qualities of Large Fixed-Wing Unmanned Aerial Vehicles

Abstract

The present invention discloses an airworthiness evaluation method, device and medium for the handling quality of large fixed-wing unmanned aerial vehicles, relating to the technical fields of civil unmanned aerial vehicle airworthiness certification and intelligent flight data processing. The method includes: obtaining the airworthiness requirements for the handling quality of the large fixed-wing unmanned aerial vehicle system; constructing an evaluation index system; constructing an automatic flight profile recognition model; based on the evaluation index system, constructing an airworthiness evaluation model for the handling quality by using the fuzzy comprehensive evaluation method; obtaining the airworthiness verification flight data of the large fixed-wing unmanned aerial vehicle to be evaluated for the handling quality airworthiness, segmenting the flight data by profile using the flight profile recognition model to obtain flight parameters, calculating the Euclidean distance between the flight parameters and the design values, taking it as the input of the airworthiness evaluation model for the handling quality, and obtaining the airworthiness evaluation result of the handling quality of the unmanned aerial vehicle. The method provided by the present invention solves the problem of the lack of a reliable systematic solution for the airworthiness evaluation of the handling quality of the large fixed-wing unmanned aerial vehicle system.

Description

Technical Field

[0001] The present invention relates to the field of civil unmanned aerial vehicle airworthiness certification and flight data intelligent processing technology, and more specifically, to an airworthiness evaluation method, device and medium for the control quality of a large fixed-wing unmanned aerial vehicle. Background Art

[0002] Large fixed-wing UAV systems have the characteristics of high altitude, high speed, long flight time, and large mission load. They are the "pinnacle" of the general aviation industry system and one of the main models for developing the low-altitude economy. The control quality of large fixed-wing UAV systems is the characteristics of the accuracy and difficulty of the UAV to complete the predetermined flight mission. It is an inherent attribute of the UAV that is directly related to mission completion and flight safety through design. It needs to pass airworthiness certification to show that the UAV system conforms to the design and the control quality meets the airworthiness requirements.

[0003] Aircraft handling quality airworthiness evaluation is an important part of type certification. The existing handling quality airworthiness evaluation mainly comes from the handling quality rating method based on the subjective evaluation of pilots and the flight quality evaluation method based on flight parameters and aircraft system status data developed by the Federal Aviation Administration (FAA) for transport aircraft. There are still the following problems when these methods are directly applied to the field of large fixed-wing UAV systems:

[0004] (1) Large fixed-wing UAV systems are controlled remotely by operators through ground control stations, and the operators mainly monitor the status of the aircraft. Manual command and control of the flight is only carried out when necessary or in an emergency. Evaluating the control quality of UAVs based on the subjective feelings of operators after performing flight missions is not only highly subjective, but also fails to reflect the operating characteristics of large fixed-wing UAV systems, which are mainly separated from humans and machines and mainly fly automatically.

[0005] (2) The flight quality assessment method based on flight parameters and aircraft system status information mainly uses QAR data to identify aircraft status and monitor operational risks, identifies abnormal or over-limit events by setting thresholds, and evaluates unsafe pilot operation behaviors. This is a continuous airworthiness evaluation centered on the mission process, which does not carefully reflect the flight quality influencing factors and their measurements in each section of aircraft operation, and cannot accurately evaluate the flight quality during the airworthiness certification stage.

[0006] (3) Large fixed-wing UAV systems lack a systematic airworthiness evaluation solution that starts from airworthiness requirements and establishes a control quality evaluation index system and its measurement method based on the mission flight profile. It is unable to clearly show the flight quality of the UAV under various flight profiles and cannot provide important basis for design improvements and authority review decisions.

[0007] In summary, in view of the unique operating characteristics of large fixed-wing UAV systems, developing an airworthiness evaluation method suitable for the control quality of large fixed-wing UAV systems is an urgent issue to be solved to ensure the safe operation of civil UAV systems and ensure the high-quality development of the low-altitude economy. Summary of the invention

[0008] In order to solve the above technical problems, the present invention provides a method, device and medium for evaluating the airworthiness of the control quality of a large fixed-wing UAV, so as to overcome the deficiency of lack of a reliable systematic solution for evaluating the control quality of a large fixed-wing UAV system.

[0009] In a first aspect, the present invention provides a method for evaluating the airworthiness of a large fixed-wing unmanned aerial vehicle's control quality, the method comprising:

[0010] Obtain the control quality and airworthiness requirements of large fixed-wing UAV systems;

[0011] Based on the airworthiness requirements of the control quality of the large fixed-wing UAV system, an evaluation index system is constructed;

[0012] Construct a flight profile automatic recognition model;

[0013] Based on the evaluation index system, a fuzzy comprehensive evaluation method is used to construct a control quality airworthiness evaluation model;

[0014] The airworthiness verification flight data of a large fixed-wing UAV to be evaluated for controllable quality is obtained, and the airworthiness verification flight data of the large fixed-wing UAV to be evaluated for controllable quality is segmented by profile using the flight profile recognition model to obtain the corresponding flight parameters in the evaluation index system, and the Euclidean distance between the flight parameters and the design values ​​is calculated as the input of the controllable quality airworthiness evaluation model to obtain the controllable quality airworthiness evaluation result of the UAV.

[0015] Furthermore, based on the control quality and airworthiness requirements of the large fixed-wing UAV system, an evaluation index system is constructed, including:

[0016] Construct a large fixed-wing UAV flight profile and divide the UAV operation into six flight phases: take-off, climb, level flight, hover, descent, and landing;

[0017] An evaluation index system is established based on the performance parameters corresponding to the six flight phases and the operator's workload of controlling the UAV as the top-level indicators.

[0018] Further, the evaluation index system includes primary indicators, secondary indicators, and tertiary indicators; among them, the primary indicators include takeoff performance, climb performance, level flight performance, hover performance, glide performance, landing performance, and workload; the secondary indicators include speed, flight path, acceleration, attitude angle, attitude angle rate, signal delay time, and environmental adaptability; the tertiary indicators include longitudinal axis speed, lateral axis speed, vertical axis speed, climb rate, glide rate, longitudinal axis direction, altitude, vertical axis direction, longitudinal axis acceleration, lateral axis acceleration, vertical axis acceleration, pitch angle, sideslip angle, roll angle, longitudinal axis attitude angle rate, lateral axis attitude angle rate, vertical axis attitude angle rate, C2 link delay time, system delay time, total delay time, range, and environment.

[0019] Further, construct an automatic flight profile recognition model, including:

[0020] Construct a flight parameter feature pattern X that can represent six profiles of takeoff, climb, level flight, loiter, glide, and landing, and establish a label y for the flight profile corresponding to the flight parameter pattern;

[0021] Establish a data set {(X i , y k ) | i = 1, 2, …, n, k = 1, 2…6}; where X i represents the i-th flight parameter feature pattern, y k represents the label of X i , and k is used to indicate the type of flight profile, and n is the number of samples in the data set;

[0022] Initialize the root node of the decision tree, which contains the entire data set;

[0023] Divide the data set into profiles at each node and calculate the Gini impurity G according to formula (1):

[0024]

[0025] In the formula, p k is the proportion of the k-th type of profile samples in the data set;

[0026] Select the feature and threshold that minimize the weighted average Gini impurity for the current node division, and the weighted average Gini impurity is calculated as shown in formula (2):

[0027]

[0028] In the formula, n1, n2, G0, G 左 , G 右 respectively represent the number of samples in the left branch of the decision tree, the number of samples in the right branch, the Gini impurity of the parent node of the current node, the Gini impurity of the left branch, and the Gini impurity of the right branch;

[0029] Recursively process each child node until the stopping condition is met, adjust the minimum sample threshold of the node, and implement a pruning strategy to optimize the constructed decision tree model;

[0030] Use all samples of the dataset as the training set to train the decision tree model, and use the trained decision tree model as the flight profile automatic recognition model.

[0031] Furthermore, based on the evaluation index system, a fuzzy comprehensive evaluation method is used to construct a handling quality airworthiness evaluation model, including:

[0032] Use the analytic hierarchy process to determine the subjective weight of the secondary indicators; use the analytic hierarchy process and entropy weight method to comprehensively determine the combined weight of the tertiary indicators, where the weight calculated by the analytic hierarchy process accounts for 40% of the combined weight;

[0033] Construct the factor set and sub-factor set of the secondary and tertiary indicators in the handling quality airworthiness evaluation index system according to the secondary fuzzy comprehensive evaluation; set the comment set as {excellent, good, medium, poor, very poor} and the corresponding numerical evaluation criteria according to the Euclidean distance between the tertiary indicator data of the expected secondary indicators and the actual operation data;

[0034] Use the trapezoidal membership function to calculate the membership degree of each indicator to each comment set and generate a judgment matrix;

[0035] Determine the first-level fuzzy comprehensive evaluation set according to the single-factor evaluation and combined weight of each factor of the tertiary indicators; among them, the first-level fuzzy comprehensive evaluation set is shown in formula (3):

[0036] B i = A i ·R i = [b i1 , b i2 ,..., b im , i = 1, 2,..., s (3)

[0037] In the formula, B i represents the first-level index evaluation vector, R i represents the single-factor evaluation matrix, A i represents the secondary weight, b im represents the eigenvalue of the first-level index evaluation vector, i represents the serial number of any sub-factor set of the comment set, and s represents the number of sub-factor sets of the comment set;

[0038] Determine the second-level fuzzy comprehensive evaluation set through the following formula:

[0039] B = A·R = [b1, b2,..., b m (4)

[0040] In the formula, B represents the weight matrix of secondary indicators, A represents the weight matrix of secondary indicators, R represents the evaluation matrix of secondary indicators, and b m represents the eigenvalue of the evaluation vector of secondary indicators.

[0041] Furthermore, the airworthiness evaluation model of handling quality responds to the Euclidean distance between the flight parameters of the input indicators and the design values, substitutes the Euclidean distance between the flight parameters of the indicators and the design values into each trapezoidal membership function to obtain the evaluation value of the indicators, normalizes the evaluation value, and constructs a secondary evaluation matrix based on formula (4). According to the secondary evaluation matrix and the numerical evaluation criteria, the airworthiness evaluation result of the handling quality of the UAV is obtained.

[0042] Furthermore, the trapezoidal membership functions include trapezoidal membership functions with evaluation grades of excellent, good, medium, poor, and extremely poor.

[0043] Furthermore, the trapezoidal membership functions with evaluation grades of excellent, good, medium, poor, and extremely poor are respectively expressed as:

[0044]

[0045]

[0046] In the formula, x represents the Euclidean distance between the flight parameters of the indicators and the design values; A 优 (x), A 良 (x), A 中 (x), A 差 (x), and A 极差 (x) respectively represent the trapezoidal membership degrees with evaluation grades of excellent, good, medium, poor, and extremely poor.

[0047] In a second aspect, the present invention provides an airworthiness evaluation device for the handling quality of a large fixed-wing UAV. The device includes:

[0048] A data acquisition unit configured to acquire the airworthiness requirements for the handling quality of the large fixed-wing UAV system;

[0049] A system construction unit configured to construct an evaluation index system based on the airworthiness requirements for the handling quality of the large fixed-wing UAV system;

[0050] A first model construction unit configured to construct an automatic flight profile recognition model;

[0051] A second model construction unit configured to construct an airworthiness evaluation model for handling quality by using the fuzzy comprehensive evaluation method based on the evaluation index system;

[0052] The airworthiness evaluation unit is configured to obtain the airworthiness verification flight data of a large fixed-wing unmanned aerial vehicle (UAV) to be evaluated for handling quality airworthiness, segment the airworthiness verification flight data of the large fixed-wing UAV to be evaluated for handling quality airworthiness by profile using the flight profile recognition model, obtain the corresponding flight parameters in the evaluation index system, calculate the Euclidean distance between the flight parameters and the design values, use it as the input of the handling quality airworthiness evaluation model, and obtain the handling quality airworthiness evaluation result of the UAV.

[0053] In a third aspect, the present invention provides a readable storage medium storing one or more programs, which can be executed by one or more processors to implement the method as described above.

[0054] The present invention has at least the following beneficial effects:

[0055] The present invention realizes a systematic airworthiness evaluation solution that starts from airworthiness requirements, establishes a handling quality evaluation index system and its measurement method based on the mission flight profile, and provides a feasible idea for the airworthiness compliance verification of the handling quality of large fixed-wing UAVs; the subjective and objective integrated handling quality airworthiness evaluation method can more accurately evaluate the handling quality of UAVs during the airworthiness certification stage; it clearly shows the handling quality of UAVs under each flight profile, providing an important basis for design improvement and the certification decision of the regulatory authority. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 FIG. shows a flowchart of an airworthiness evaluation method for the handling quality of a large fixed-wing UAV according to an embodiment of the present invention.

[0057] Figure 2 FIG. shows a schematic diagram of an airworthiness evaluation framework for the handling quality of a large fixed-wing UAV system according to an embodiment of the present invention.

[0058] Figure 3 FIG. shows a flight profile recognition flowchart of a flight profile automatic recognition model according to an embodiment of the present invention.

[0059] Figure 4 FIG. shows a structural diagram of an airworthiness evaluation device for the handling quality of a large fixed-wing UAV according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention will be further described in detail below in conjunction with the accompanying drawings and specific examples, but it is not a limitation to the present invention. For the various steps described herein, if there is no necessity for a front-back relationship between them, the order in which they are described as examples herein should not be regarded as a limitation. Those skilled in the art should know that they can be adjusted in order as long as the logic between them is not destroyed and the entire process cannot be realized.

[0061] An embodiment of the present invention provides a method for evaluating the airworthiness of the handling quality of a large fixed-wing unmanned aerial vehicle. By semi-supervised optimizing the image enhancement module based on a pre-trained target detection module, directly optimizing the target detection features, and improving the utilization efficiency of unlabeled data. At the same time, a residual image enhancement foggy day target detection model is designed based on this semi-supervised optimization framework to further improve the foggy day detection performance. In addition, to solve the problem of feature differences between synthetic and real foggy day images, the present invention provides an unsupervised domain adaptation data augmentation method, which augments the training data by transferring the statistical information of unlabeled real foggy day images during the training process. Finally, the foggy day target detection technology of the present invention is optimized and trained using the proposed new foggy day target detection dataset, which is a semi-supervised dataset containing more than 120,000 high-quality labeled images and approximately 30,000 unlabeled images, providing a large-scale, semi-supervised reliable data resource for the field of foggy day target detection. The present invention significantly improves the generalization ability and detection accuracy of the model in real foggy day scenarios.

[0062] Specifically, as Figure 1 shown, it is a flowchart of the method for evaluating the airworthiness of the handling quality of a large fixed-wing unmanned aerial vehicle. The method for evaluating the airworthiness of the handling quality of a large fixed-wing unmanned aerial vehicle includes steps S10 to S50, which are introduced in detail as follows.

[0063] S10: Obtain the airworthiness requirements for the handling quality of the large fixed-wing unmanned aerial vehicle system.

[0064] Exemplarily, the top-level airworthiness requirements for the handling qualities of large unmanned aircraft are mainly Clause 92.703 of CCAR. The specific requirements are present in Chapter B of the special airworthiness conditions for large unmanned aircraft facing various operating characteristics or in the flight and performance-related content of the general requirements. 《GB / T 43504-2023》, 《GB / T38996-2020》, 《GB / T38911-2020》 and 《GB / T 42862-2023》 provide the flight test requirements for the flight performance of medium and large fixed-wing unmanned aircraft and the general requirement specifications for the flight control systems of unmanned aircraft. The comprehensive application of these standards provides a method for compliance verification of the airworthiness evaluation of handling qualities. Establish an airworthiness standard system for handling qualities, that is, CCAR92.703 as the top-level standard, the special airworthiness certification conditions for each released large unmanned aircraft model as the secondary standard, and GB as the tertiary standard, and extract the airworthiness requirements for handling qualities from top to bottom.

[0065] S20: Based on the airworthiness requirements for the handling qualities of large fixed-wing unmanned aircraft systems, construct an evaluation index system.

[0066] In some embodiments, the evaluation index system is constructed by the following method:

[0067] Construct a flight profile for large fixed-wing unmanned aircraft, and divide the operation of the unmanned aircraft into six flight phases: takeoff, climb, level flight, hover, glide, and landing. The handling qualities are defined as takeoff, climb, level flight, hover, glide, landing, and endurance performance. An airworthiness evaluation index system for handling qualities is established with these seven performances as the top-level indicators respectively. Evaluate the handling qualities of each phase and comprehensively obtain the handling qualities of the unmanned aircraft system. Considering the operating characteristics of the unmanned aircraft system mainly based on automatic flight and manual command and control in necessary or emergency situations, the workload of the operator in controlling the unmanned aircraft is also used as an airworthiness evaluation index for handling qualities. After the operator performs the flight of the manual command and control task, the NASA-TLX scale can be filled out to obtain the workload level. Based on this airworthiness evaluation framework, the airworthiness evaluation of the handling qualities of each profile and the overall handling qualities in the manual control and automatic flight modes is realized.

[0068] Exemplarily, such as Figure 2As shown in the figure, it is a schematic diagram of the airworthiness evaluation framework for the handling quality of a large fixed-wing UAV system. The evaluation index system includes first-level indicators, second-level indicators, and third-level indicators. Among them, the first-level indicators include takeoff performance, climb performance, level flight performance, hover performance, glide performance, landing performance, and workload. The second-level indicators include speed, flight path, acceleration, attitude angle, attitude angle rate, signal delay time, and environmental adaptability. The third-level indicators include longitudinal axis speed, lateral axis speed, vertical axis speed, climb rate, glide rate, longitudinal axis direction, altitude, vertical axis direction, longitudinal axis acceleration, lateral axis acceleration, vertical axis acceleration, pitch angle, sideslip angle, roll angle, longitudinal axis attitude angle rate, lateral axis attitude angle rate, vertical axis attitude angle rate, C2 link delay time, system delay time, total delay time, range, and environment.

[0069] Specifically, each first-level indicator corresponds to all second-level indicators. Taking takeoff performance as an example, the corresponding second-level indicators represent the speed, flight path, acceleration, attitude angle, attitude angle rate, signal delay time, and environmental adaptability during the takeoff phase. The corresponding second-level indicators for the other first-level indicators are similar and will not be described in detail here. One second-level indicator corresponds to specific third-level indicators. Specifically, the speed in the second-level indicators corresponds to the longitudinal axis speed, lateral axis speed, vertical axis speed, climb rate, and glide rate; the flight path corresponds to the longitudinal axis direction, altitude, and vertical axis direction; the acceleration corresponds to the longitudinal axis acceleration, lateral axis acceleration, and vertical axis acceleration; the attitude angle corresponds to the pitch angle, sideslip angle, and roll angle; the attitude angle rate corresponds to the longitudinal axis attitude angle rate, lateral axis attitude angle rate, and vertical axis attitude angle rate; the signal delay time corresponds to the C2 link delay time, system delay time, and total delay time; and the environmental adaptability corresponds to the range and environment.

[0070] S30: Construct an automatic flight profile recognition model.

[0071] In some embodiments, step S30 is implemented by the following steps:

[0072] S31. Construct a flight parameter feature pattern X that can represent six profiles of takeoff, climb, level flight, loiter, glide, and landing, and establish a label y for the flight profile corresponding to the flight parameter pattern. Given a data set {(X i , y k ) | i = 1, 2, …, n, k = 1, 2…6}, initialize the root node of the decision tree, which contains the entire data set. The data set is partitioned by profile at each node and the Gini impurity G is calculated according to formula (1). The p in formula (1) k is the proportion of the kth type of profile samples in the data set.

[0073]

[0074] S32. Select the feature and threshold that minimize the weighted average Gini impurity for the current node division. The weighted average Gini impurity is calculated as shown in formula (2), where n1, n2, G0, G 左 , G 右 respectively represent the number of samples in the left branch of the decision tree, the number of samples in the right branch, the Gini impurity of the parent node of the current node, the Gini impurity of the left branch, and the Gini impurity of the right branch.

[0075]

[0076] S33. Recursively process each child node until the stopping condition is met. Finally, adjust the minimum sample threshold of the node and implement the pruning strategy to optimize the constructed decision tree model. Use all samples of the dataset as the training set to train the decision tree model, and use the trained model to predict the input feature vector X, that is, obtain the flight profile corresponding to the feature vector.

[0077] Exemplarily, as Figure 3 shown, it is the flight profile recognition flowchart of the flight profile automatic recognition model. First, determine whether the UAV is in the climbing stage. If the height change rate ≥ 1.00017, it is determined to be in the climbing stage. If the height change rate < 1.00017, further determine whether it is in the descending stage. If the height change rate < -0.099977, it is determined to be in the descending stage. If the height change rate ≥ -0.099977, further determine whether it is in the takeoff stage or the level flight stage. If the normalized height < 0.0672824, it is determined to be in the takeoff stage. If the normalized height ≥ 0.0672824, it is determined to be in the level flight stage.

[0078] S40: Based on the evaluation index system, use the fuzzy comprehensive evaluation method to construct a handling quality airworthiness evaluation model.

[0079] In some embodiments, step S40 is implemented by the following steps:

[0080] S41. Select the UAV type certification test flight crew and the scientific research test flight crew as expert members to score the handling quality airworthiness of the tertiary indicators under each secondary indicator, and use the Analytic Hierarchy Process (AHP) to determine the subjective weights of the secondary airworthiness evaluation indicators; then use AHP and the Entropy Weight Method (EWM) to comprehensively determine the combined weights of the tertiary indicators, where the weight calculated by AHP accounts for 40% of the final weight.

[0081] S42. Construct a flight control quality airworthiness evaluation model based on fuzzy comprehensive evaluation: Construct the factor sets and sub-factor sets of the secondary and tertiary indicators in the flight control quality airworthiness evaluation index system according to the secondary fuzzy comprehensive evaluation; Set the comment set as {excellent, good, medium, poor, extremely poor} and the corresponding numerical evaluation criteria according to the Euclidean distance between the tertiary index data of the expected secondary indicators and the actual operation data, combined with expert opinions; Use the trapezoidal membership function to calculate the membership degree of each indicator to each comment set, and generate a judgment matrix. The first-level fuzzy comprehensive evaluation is determined by the single-factor evaluation of each factor of the tertiary index and the multi-factor comprehensive evaluation result of this level of index. The single-factor evaluation matrix is R i . The weight of each index is the combined weight determined by AHP+EWM. Set the first-level weight as A 1×s , and the second-level weight is Obtain the first-level fuzzy comprehensive evaluation set, as shown in Equation (3).

[0082] B i =A i ·R i =[b i1 ,b i2 ,...,b im ,i = 1,2,...,s (3)

[0083] In the formula, B i represents the first-level index evaluation vector, R i represents the single-factor evaluation matrix, A i represents the second-level weight, b im represents the eigenvalue of the first-level index evaluation vector, i represents the serial number of any sub-factor set of the comment set, and s represents the number of sub-factor sets of the comment set.

[0084] The second-level fuzzy comprehensive evaluation set is as shown in Equation (4).

[0085] B = A·R = [b1,b2,...,b m (4)

[0086] In the formula, B represents the second-level index weight matrix, A represents the second-level index weight matrix, R represents the second-level index evaluation matrix, and b m represents the eigenvalue of the second-level index evaluation vector.

[0087] The second-level fuzzy comprehensive evaluation set determines the membership degree of the second-level index to each judgment set. Combine the seven vectors obtained to form a fuzzy judgment matrix, solve the weight of each index for the flight control quality airworthiness evaluation, and perform matrix synthesis to obtain the overall evaluation result of the UAV flight control quality airworthiness.

[0088] S50: Obtain the airworthiness verification flight data of large fixed-wing UAVs for the evaluation of handling quality airworthiness. Use the flight profile recognition model to segment the airworthiness verification flight data of large fixed-wing UAVs for the evaluation of handling quality airworthiness by profile, obtain the corresponding flight parameters in the evaluation index system, calculate the Euclidean distance between the flight parameters and the design values, and use it as the input of the handling quality airworthiness evaluation model to obtain the handling quality airworthiness evaluation result of the UAV.

[0089] In this embodiment, the purpose of step S50 is to implement the airworthiness evaluation of each flight profile or the overall handling quality of large fixed-wing UAVs. Obtain the airworthiness verification flight data of large fixed-wing UAVs for the evaluation of handling quality airworthiness, use the flight profile recognition model constructed in step S30 to segment the flight data by profile, and obtain the flight parameters corresponding to all three-level indicators except the secondary indicators of environmental adaptability in the airworthiness evaluation index system in step S20. Calculate the Euclidean distance between the flight parameters and the design values and use it as the input of step S40. According to the secondary comprehensive evaluation method of step S40, the airworthiness of the handling quality of each profile of the UAV can be evaluated; the fuzzy comprehensive evaluation of the 7 primary indicators can be used to obtain the handling quality airworthiness evaluation result of the UAV.

[0090] The feasibility and progressiveness of the present invention will be further illustrated below with specific examples.

[0091] In a specific example, the airworthiness evaluation method for the handling quality of the large fixed-wing UAV is implemented by the following steps to evaluate the airworthiness of the handling quality of the large fixed-wing UAV:

[0092] Step 1: Extract the airworthiness requirements for the handling quality of the large fixed-wing UAV system according to step S10 mentioned above.

[0093] Step 2: Construct the airworthiness evaluation framework for the handling quality of the large fixed-wing UAV according to step S20 mentioned above.

[0094] Step 3: Construct an automatic flight profile recognition model for large fixed-wing UAVs based on decision trees, which is implemented by the following steps:

[0095] Step 31: Flight parameter acquisition: After the airworthiness verification flight of the large fixed-wing UAV, use the data of the UAV speed, altitude, attitude, etc. at each moment recorded by the ground station and transmitted back through the downlink of the communication link to construct a flight data set. Normalize the data sets in each dimension, select the altitude change rate and altitude as the feature patterns X of the decision tree, and manually label the flight profile labels y corresponding to the feature patterns k , {(X i , y k ) | i = 1, 2,..., n, k = 1, 2...6} as the training set for training the decision tree model.

[0096] Step 32, Model Training and Prediction: The decision tree model is constructed and trained through the ClassificationTree.fit() function in MATLAB, and the predict() function is used to implement the input feature pattern to obtain the predicted profile.

[0097] Step 33, Continuous Optimization of the Model: The minimum sample data number setting of the ClassificationTree.fit() function and the cvloss() function and prune() function are used for pruning to generate an optimized decision tree model, and the samples with correct predictions are added to the training set.

[0098] Step 4, Taking the airworthiness evaluation of the handling quality of a certain type of UAV during the manual landing phase as an example, the airworthiness evaluation is implemented through the following steps:

[0099] Step 41, Collect flight parameters, and use the constructed profile recognition decision tree model to extract the flight parameters during the landing phase. Select the UAV certification test flight crew and scientific research test flight crew as expert members, conduct airworthiness evaluation scoring on the tertiary indicators under each secondary indicator of the primary indicator "landing performance", fill in the NASA-TLX scale, and calculate the workload level. Use AHP to determine the subjective weight of the secondary indicators; then use AHP+EWM to comprehensively determine the combined weight of the tertiary indicators, where the weight calculated by AHP accounts for 40% of the total weight.

[0100] Step 42, Construct an airworthiness evaluation model for landing handling quality based on fuzzy comprehensive evaluation: Construct a fuzzy evaluation factor set U and sub-factor set U i , and the evaluation set V = {excellent, good, medium, poor, extremely poor}. Calculate the Euclidean distance between the flight parameters of the landing profile and the design values, and construct the evaluation criteria for the evaluation set in combination with expert opinions. The results are shown in Table 1. According to Table 1, establish the trapezoidal membership function formulas as shown in Formulas (5) to (9). In Formulas (5) to (9), x represents the Euclidean distance between the flight parameters of the index and the design values; A 优 (x), A 良 (x), A 中 (x), A 差 (x) and A 极差 (x) represent the trapezoidal membership degrees of the evaluation grades of excellent, good, medium, poor, and extremely poor respectively.

[0101] Table 1 Evaluation Criteria for the Evaluation Set

[0102] Evaluation level Evaluation criteria Excellent [0，0.2] Good (0.2，0.4] Medium (0.4，0.6] Poor (0.6，0.8] Very poor (0.8，+∞)

[0103] Trapezoidal membership function for the evaluation grade of "excellent":

[0104]

[0105] Trapezoidal membership function with an evaluation level of "good":

[0106]

[0107] Trapezoidal membership function with an evaluation level of "medium":

[0108]

[0109] Trapezoidal membership function with an evaluation level of "poor":

[0110]

[0111] Trapezoidal membership function with an evaluation level of "extremely poor":

[0112]

[0113] Step 43, Landing handling quality evaluation result: Substitute the Euclidean distance between the flight parameters of the corresponding index and the design value into each trapezoidal membership function to obtain the evaluation value of each index. Normalize the evaluation value and construct a secondary evaluation matrix

[0114] B = [0.22251, 0.635023, 0.039867, 0, 0.1026]

[0115] According to the principle of maximum membership degree, 0.635023 is the largest value, corresponding to the second position among "excellent, good, medium, poor, extremely poor". The airworthiness evaluation level of the landing quality of this type of UAV is "good".

[0116] The embodiment of the present invention also provides an airworthiness evaluation device for the handling quality of a large fixed-wing UAV, as Figure 4 shown. This device includes:

[0117] A data acquisition unit 401, configured to acquire the airworthiness requirements for the handling quality of the large fixed-wing UAV system;

[0118] A system construction unit 402, configured to construct an evaluation index system based on the airworthiness requirements for the handling quality of the large fixed-wing UAV system;

[0119] A first model construction unit 403, configured to construct an automatic flight profile recognition model;

[0120] A second model construction unit 404, configured to construct an airworthiness evaluation model for the handling quality by using the fuzzy comprehensive evaluation method based on the evaluation index system;

[0121] The airworthiness evaluation unit 405 is configured to obtain the airworthiness verification flight data of a large fixed-wing unmanned aerial vehicle (UAV) to be evaluated for handling quality airworthiness, segment the airworthiness verification flight data of the large fixed-wing UAV to be evaluated for handling quality airworthiness by profile using the flight profile recognition model, obtain the corresponding flight parameters in the evaluation index system, calculate the Euclidean distance between the flight parameters and the design values, use it as the input of the handling quality airworthiness evaluation model, and obtain the handling quality airworthiness evaluation result of the UAV.

[0122] In some embodiments, the system construction unit is further configured to:

[0123] Construct a flight profile of a large fixed-wing UAV, and divide the operation of the UAV into 6 flight phases: takeoff, climb, level flight, hover, glide, and landing;

[0124] Establish an evaluation index system based on the performance parameters corresponding to the 6 flight phases and the workload of the operator in controlling the UAV as the top-level indicators.

[0125] In some embodiments, the evaluation index system includes first-level indicators, second-level indicators, and third-level indicators; among them, the first-level indicators include takeoff performance, climb performance, level flight performance, hover performance, glide performance, landing performance, and workload; the second-level indicators include speed, flight path, acceleration, attitude angle, attitude angle rate, signal delay time, and environmental adaptability; the third-level indicators include longitudinal axis speed, lateral axis speed, vertical axis speed, climb rate, glide rate, longitudinal axis direction, altitude, vertical axis direction, longitudinal axis acceleration, lateral axis acceleration, vertical axis acceleration, pitch angle, sideslip angle, roll angle, longitudinal axis attitude angle rate, lateral axis attitude angle rate, vertical axis attitude angle rate, C2 link delay time, system delay time, total delay time, range, and environment.

[0126] In some embodiments, the first model construction unit is further configured to:

[0127] Construct a flight parameter feature pattern X that can represent the six profiles of takeoff, climb, level flight, hover, glide, and landing, and establish a label y for the flight profile corresponding to the flight parameter pattern;

[0128] Establish a data set {(X i , y k ) | i = 1, 2,..., n, k = 1, 2... 6}; where X i represents the i-th flight parameter feature pattern, y k represents the label of X i , k is used to indicate the type of flight profile, and n is the number of samples in the data set;

[0129] Initialize the root node of the decision tree, which contains the entire data set;

[0130] Divide the dataset into profiles at each node and calculate the Gini impurity G according to formula (1):

[0131]

[0132] In the formula, p k is the proportion of the k-th type of profile samples in the dataset;

[0133] Select the feature and threshold that minimize the weighted average Gini impurity for the current node division. The weighted average Gini impurity is calculated as shown in formula (2):

[0134]

[0135] In the formula, n1, n2, G0, G 左 , G 右 respectively represent the number of samples in the left branch of the decision tree, the number of samples in the right branch, the Gini impurity of the parent node of the current node, the Gini impurity of the left branch, and the Gini impurity of the right branch;

[0136] Recursively process each child node until the stopping condition is met, and adjust the minimum sample threshold of the node and implement the pruning strategy to optimize the constructed decision tree model;

[0137] Use all samples of the dataset as the training set to train the decision tree model, and use the trained decision tree model as the flight profile automatic recognition model.

[0138] In some embodiments, the second model construction unit is further configured to:

[0139] Determine the subjective weight of the secondary indicators using the analytic hierarchy process; comprehensively determine the combined weight of the tertiary indicators using the analytic hierarchy process and the entropy weight method, where the weight calculated by the analytic hierarchy process accounts for 40% of the combined weight;

[0140] Construct the factor set and sub-factor set of the secondary and tertiary indicators in the airworthiness evaluation index system of handling quality according to the secondary fuzzy comprehensive evaluation; set the comment set as {excellent, good, medium, poor, extremely poor} and the corresponding numerical evaluation criteria according to the Euclidean distance between the tertiary indicator data of the expected secondary indicators and the actual operation data;

[0141] Calculate the membership degree of each indicator to each comment set using the trapezoidal membership function and generate the judgment matrix;

[0142] Determine the first-level fuzzy comprehensive evaluation set according to the single-factor evaluation and combined weight of each factor of the tertiary indicators; among them, the first-level fuzzy comprehensive evaluation set is as shown in formula (3):

[0143] B i = Ai ·R i = [b i1 , b i2 ,..., b im , i = 1, 2,..., s (3)

[0144] In the formula, B i represents the first-level index evaluation vector, R i represents the single-factor evaluation matrix, A i represents the second-level weight, b im represents the eigenvalue of the first-level index evaluation vector, i represents the serial number of any sub-factor set of the comment set, and s represents the number of sub-factor sets of the comment set;

[0145] The second-level fuzzy comprehensive evaluation set is determined by the following formula:

[0146] B = A·R = [b1, b2,..., b m (4)

[0147] In the formula, B represents the second-level index weight matrix, A represents the second-level index weight matrix, R represents the second-level index evaluation matrix, and b m represents the eigenvalue of the second-level index evaluation vector.

[0148] In some embodiments, the handling quality airworthiness evaluation model responds to the Euclidean distance between the flight parameters of the input index and the design value, substitutes the Euclidean distance between the flight parameters of the index and the design value into each trapezoidal membership function to obtain the evaluation value of the index, normalizes the evaluation value, constructs a second-level evaluation matrix based on formula (4), and obtains the handling quality airworthiness evaluation result of the UAV according to the second-level evaluation matrix and the numerical evaluation standard.

[0149] In some embodiments, the trapezoidal membership function includes trapezoidal membership functions with evaluation grades of excellent, good, medium, poor, and extremely poor.

[0150] In some embodiments, the trapezoidal membership functions with evaluation grades of excellent, good, medium, poor, and extremely poor are respectively expressed as:

[0151]

[0152]

[0153] In the formula, x represents the Euclidean distance between the flight parameters of the index and the design value; A 优 (x), A 良 (x), A 中 (x), A 差 (x) and A 极差(x) represents the trapezoidal membership degrees of excellent, good, medium, poor, and extremely poor evaluation grades respectively.

[0154] It should be noted that the structures of the airworthiness evaluation devices for the handling qualities of large fixed-wing unmanned aircraft described in this embodiment belong to the same technical concept as the previously described airworthiness evaluation method for the handling qualities of large fixed-wing unmanned aircraft, and achieve the same beneficial effects through the same principle, which will not be elaborated here.

[0155] The embodiment of the present invention also provides a readable storage medium, which stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method described in any of the above embodiments.

[0156] In addition, although exemplary embodiments have been described herein, the scope includes any and all embodiments based on the present invention having equivalent elements, modifications, omissions, combinations (e.g., solutions that cross various embodiments), adaptations, or changes. The elements in the claims will be broadly interpreted based on the language used in the claims and are not limited to the examples described in this specification or during the implementation of this application, and the examples will be interpreted as non-exclusive. Therefore, this specification and examples are intended to be considered only as examples, and the true scope and spirit are indicated by the full scope of the following claims and their equivalents.

[0157] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of them) can be used in combination with each other. For example, those of ordinary skill in the art can use other embodiments when reading the above description. Additionally, in the above specific implementation manners, various features can be grouped together to simplify the present invention. This should not be construed as an intention that the features of an invention not claimed are necessary for any claim. On the contrary, the subject matter of the present invention can be less than all the features of a particular embodiment of the invention. Thus, the following claims are incorporated herein as examples or embodiments into the specific implementation manners, where each claim independently serves as a separate embodiment, and considering these embodiments, they can be combined with each other in various combinations or permutations. The scope of the present invention should be determined with reference to the appended claims and the full scope of the equivalent forms empowered by these claims.

Claims

1. An airworthiness evaluation method for the handling qualities of a large fixed-wing unmanned aerial vehicle, characterized in that, The method includes: Obtaining the airworthiness requirements for the handling qualities of large fixed-wing UAV systems; Based on the airworthiness requirements for the handling qualities of large fixed-wing UAV systems, constructing an evaluation index system; Constructing an automatic flight profile recognition model; Based on the evaluation index system, using the fuzzy comprehensive evaluation method to construct a handling quality airworthiness evaluation model; Obtaining the airworthiness verification flight data of the large fixed-wing UAV to be evaluated for handling quality airworthiness, segmenting the airworthiness verification flight data of the large fixed-wing UAV to be evaluated for handling quality airworthiness by profile using the flight profile recognition model, obtaining the corresponding flight parameters in the evaluation index system, calculating the Euclidean distance between the flight parameters and the design values, using it as the input of the handling quality airworthiness evaluation model, and obtaining the handling quality airworthiness evaluation result of the UAV.

2. The airworthiness evaluation method for the handling quality of large fixed-wing unmanned aerial vehicles according to claim 1, characterized in that, Based on the airworthiness requirements for the handling qualities of large fixed-wing UAV systems, constructing an evaluation index system, including: Constructing the flight profile of a large fixed-wing UAV, and dividing the operation of the UAV into 6 flight phases: takeoff, climb, level flight, hover, glide, and landing; Establishing an evaluation index system with the performance parameters corresponding to the 6 flight phases and the workload of the operator controlling the UAV as the top-level indicators.

3. The airworthiness evaluation method for the handling quality of large fixed-wing unmanned aircraft according to claim 2, wherein The evaluation index system includes first-level indicators, second-level indicators, and third-level indicators; among them, the first-level indicators include takeoff performance, climb performance, level flight performance, hover performance, glide performance, landing performance, and workload; the second-level indicators include speed, route, acceleration, attitude angle, attitude angle rate, signal delay time, and environmental adaptability; the third-level indicators include longitudinal axis speed, transverse axis speed, vertical axis speed, climb rate, glide rate, longitudinal axis direction, altitude, vertical axis direction, longitudinal axis acceleration, transverse axis acceleration, vertical axis acceleration, pitch angle, sideslip angle, roll angle, longitudinal axis attitude angle rate, transverse axis attitude angle rate, vertical axis attitude angle rate, C2 link delay time, system delay time, total delay time, range, and environment.

4. The airworthiness evaluation method for the handling quality of large fixed-wing unmanned aerial vehicles according to claim 1, characterized in that, Constructing an automatic flight profile recognition model, including: Constructing a flight parameter feature pattern X that can represent the six profiles of takeoff, climb, level flight, hover, glide, and landing, and establishing a label y for the flight profile corresponding to the flight parameter pattern; Establish a data set \(\{(X i ,y k )|i = 1, 2, \ldots, n, k = 1, 2 \ldots 6\}\); where, \(X i represents the \(i\)-th flight parameter feature pattern, \(y k represents the label of \(X i , \(k\) is used to indicate the type of flight profile, and \(n\) is the number of samples in the data set; Initializing the root node of the decision tree, including the entire data set; Dividing the data set by profile at each node and calculating the Gini impurity G according to formula (1); where p k is the proportion of the k-th type of profile samples in the dataset; Select the feature and threshold that minimize the weighted average Gini impurity for the current node division, and the weighted average Gini impurity is calculated as shown in formula (2): where n1, n2, G0, G 左 , G 右 represent the number of samples in the left branch of the decision tree, the number of samples in the right branch of the decision tree, the Gini impurity of the parent node of the current node, the Gini impurity of the left branch, and the Gini impurity of the right branch, respectively; Recursively processing each child node until the stopping condition is met, adjusting the minimum sample threshold of the node and implementing a pruning strategy to optimize the constructed decision tree model; Using all the samples of the data set as the training set to train the decision tree model, and using the trained decision tree model as the automatic flight profile recognition model.

5. The airworthiness evaluation method for the handling quality of large fixed-wing unmanned aerial vehicles according to claim 3, characterized in that Based on the evaluation index system, using the fuzzy comprehensive evaluation method to construct a handling quality airworthiness evaluation model, including: Using the analytic hierarchy process to determine the subjective weights of the second-level indicators; using the analytic hierarchy process and the entropy weight method to comprehensively determine the combined weights of the third-level indicators, where the weights calculated by the analytic hierarchy process account for 40% of the combined weights; Construct the factor sets and sub-factor sets of the secondary and tertiary indicators in the airworthiness evaluation index system for handling quality according to the secondary fuzzy comprehensive evaluation; set the comment set as {excellent, good, medium, poor, extremely poor} and the corresponding numerical evaluation criteria according to the Euclidean distance between the tertiary indicator data of the expected secondary indicator and the actual operation data; Calculate the membership degree of each indicator to each comment set using the trapezoidal membership function and generate a judgment matrix; Determine the first-level fuzzy comprehensive evaluation set according to the single-factor evaluation and combined weight of each factor of the tertiary indicator; among them, the first-level fuzzy comprehensive evaluation set is shown in formula (3): B i = A i ·R i = [b i1 , b i2 ,..., b im , i = 1, 2,..., s (3) Where, B i represents the evaluation vector of the first-level index, R i represents the single-factor evaluation matrix, A i represents the second-level weight, b im represents the eigenvalue of the evaluation vector of the first-level index, i represents the serial number of any sub-factor set of the evaluation set, and s represents the number of sub-factor sets of the evaluation set; Determine the second-level fuzzy comprehensive evaluation set through the following formula: B = A·R = [b1, b2,..., b m (4) Where B represents the weight matrix of secondary indicators, A represents the weight matrix of secondary indicators, R represents the evaluation matrix of secondary indicators, and b m represents the eigenvalue of the evaluation vector of secondary indicators.

6. The airworthiness evaluation method for the handling quality of large fixed-wing unmanned aerial vehicles according to claim 5, wherein The airworthiness evaluation model for handling quality responds to the Euclidean distance between the flight parameters of the input indicators and the design values, substitutes the Euclidean distance between the flight parameters of the indicators and the design values into each trapezoidal membership function to obtain the evaluation value of the indicators, normalizes the evaluation value and constructs a secondary evaluation matrix based on formula (4), and obtains the airworthiness evaluation result of the handling quality of the UAV according to the secondary evaluation matrix and the numerical evaluation criteria.

7. The airworthiness evaluation method for the handling quality of large fixed-wing unmanned aerial vehicles according to claim 5, characterized in that The trapezoidal membership function includes trapezoidal membership functions for the evaluation grades of excellent, good, medium, poor, and extremely poor.

8. The airworthiness evaluation method for the handling quality of large fixed-wing unmanned aerial vehicles according to claim 7, characterized in that, The trapezoidal membership functions for the evaluation grades of excellent, good, medium, poor, and extremely poor are respectively expressed as: Where x represents the Euclidean distance between the flight parameters of the index and the design value; A 优 (x), A 良 (x), A 中 (x), A 差 (x) and A 极差 (x) respectively represent the trapezoidal membership degrees of the evaluation grades of excellent, good, medium, poor, and extremely poor.

9. An airworthiness evaluation device for the handling quality of a large fixed-wing unmanned aerial vehicle, characterized in that The device includes: A data acquisition unit configured to acquire the airworthiness requirements for the handling quality of a large fixed-wing UAV system; A system construction unit configured to construct an evaluation index system based on the airworthiness requirements for the handling quality of the large fixed-wing UAV system; A first model construction unit configured to construct an automatic flight profile recognition model; A second model construction unit configured to construct an airworthiness evaluation model for handling quality using the fuzzy comprehensive evaluation method based on the evaluation index system; An airworthiness evaluation unit configured to acquire the airworthiness verification flight data of a large fixed-wing UAV to be evaluated for handling quality airworthiness, segment the airworthiness verification flight data of the large fixed-wing UAV to be evaluated for handling quality by profile using the flight profile recognition model to obtain the corresponding flight parameters in the evaluation index system, calculate the Euclidean distance between the flight parameters and the design values, and use it as the input of the airworthiness evaluation model for handling quality to obtain the airworthiness evaluation result of the handling quality of the UAV.

10. A non-transitory computer-readable storage medium storing instructions, which when executed by a processor, execute the method according to any one of claims 1 to 8.

Images