Fault diagnosis method, system and equipment for undercarriage damping system and medium

Abstract

The invention discloses an undercarriage damping system fault diagnosis method, system and device and a medium, and relates to the technical field of fault data processing, and the method specifically comprises the steps: building a kinetic model, collecting aircraft fuselage data and aircraft wheel data based on the kinetic model, preprocessing the aircraft fuselage data and the aircraft wheel data, and obtaining a fault diagnosis result. A fuselage vertical acceleration signal and an airplane wheel vertical acceleration signal are obtained, and more accurate fuselage data can be collected by constructing a dynamic model; establishing an undercarriage damping system fault response database based on the collected fuselage data in combination with an undercarriage dynamical model and flight measured data; the method comprises the steps of constructing an initial fault model, inputting data of a fault response database into the initial fault model to train the initial fault model, optimizing a fault training model based on the fault response database, obtaining a more accurate fault monitoring result, and fully utilizing data advantages. And prediction and trend judgment of the undercarriage damping system can be carried out more accurately, efficiently and in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of fault data processing, and particularly relates to a fault diagnosis method, system, device and medium for a landing gear shock absorption system. Background Art

[0002] The health status of the landing gear shock absorption system directly determines the safety of aircraft landing. As a key component of the landing gear shock absorption system, the shock absorber plays a crucial role during takeoff and landing of the aircraft. Currently, oil-gas shock absorbers are widely used, and the oil and gas states and the internal structure of the shock absorber determine the performance (such as stiffness, damping, etc.) of the shock absorber. Most of the monitoring and maintenance of the aircraft landing gear shock absorption system mainly rely on the manual inspection and maintenance methods according to the collection and maintenance manuals at regular intervals. Although the oil and gas states of the shock absorber can be monitored by installing a barometer, the health status of the internal structure that also affects the performance of the shock absorber cannot be monitored. Currently, the health management of the landing gear shock absorption system mainly adopts the manual method for regular inspection and maintenance according to the maintenance manual, but this detection method has problems such as high labor cost, dependence on experience, and inability to achieve prediction and trend judgment for the landing gear shock absorption system. Summary of the Invention

[0003] The technical problem to be solved by the present invention is the inaccurate prediction and trend judgment for the landing gear shock absorption system. The purpose is to provide a fault diagnosis method, system, device and medium for a landing gear shock absorption system. By constructing a dynamic model, based on the relationship of "shock absorber state - shock absorber performance - aircraft landing response" and aircraft airworthiness requirements, more accurate airframe data is collected. Based on the collected airframe data, combined with the landing gear dynamic model and flight measured data, a fault response database for the landing gear shock absorption system is established, an initial fault model is constructed, and the fault training model is optimized based on the fault response database to obtain a more accurate fault monitoring result, which can make full use of the data advantage, and perform the prediction and trend judgment for the landing gear shock absorption system more accurately, efficiently and in real time, so as to achieve fault prediction and health management.

[0004] The present invention is realized by the following technical solutions: In the first aspect of the present invention, a fault diagnosis method for a landing gear shock absorption system is provided, including the following specific steps: Construct a dynamic model, collect aircraft airframe data and wheel data based on the dynamic model, preprocess the aircraft airframe data and wheel data to obtain the airframe vertical acceleration signal and the wheel vertical acceleration signal; Construct a fault response database based on the airframe vertical acceleration signal and the wheel vertical acceleration signal; Construct an initial fault model, input the data of the fault response database into the initial fault model to train the initial fault model, and obtain a trained fault diagnosis model; Fault diagnosis is carried out based on the trained fault diagnosis model, and the diagnosis result is output, and the diagnosis result is also used to reversely optimize the fault response database.

[0005] Furthermore, the specific steps for constructing the dynamic model include: Obtain the stiffness and damping parameters of the oleo-pneumatic shock absorber of the detected target aircraft; Obtain the tire force based on the stiffness and damping parameters of the oleo-pneumatic shock absorber; Obtain the oil damping force of the oleo-pneumatic shock absorber according to the forward and reverse stroke speeds of the shock absorber and the oil data; Obtain the air spring force of the oleo-pneumatic shock absorber according to the air pressure data; Obtain the aircraft gravity, and combine the tire force, the oil damping force of the oleo-pneumatic shock absorber and the air spring force of the oleo-pneumatic shock absorber to obtain the landing impact force and construct the mechanical relationship; Construct a dynamic model according to the aircraft mechanical relationship.

[0006] Furthermore, obtaining the fuselage vertical acceleration signal and the wheel vertical acceleration signal further includes: Fuse the fuselage vertical acceleration signal and the wheel vertical acceleration signal to obtain fused data; The fused data is array data, and the array data includes: time array data, fuselage acceleration array data and wheel acceleration array data.

[0007] Furthermore, the specific steps for constructing the initial fault model include: Construct an input layer, a feature extraction layer, a feature integration layer and an output layer, where: The input layer is used to receive the fuselage acceleration and the wheel acceleration; The feature extraction layer includes multiple convolutional layers and pooling layers, and is used to extract features from the input data; The feature integration layer includes multiple fully connected layers and is used to integrate the features obtained from the feature extraction layer; The output layer is used to output fault data.

[0008] Furthermore, when the feature extraction layer extracts features from the input data, the processing process of multiple convolutional layers and pooling layers includes: The first convolution converts the input data of the first convolutional layer from None×6×64 to None×3×64; The second convolution converts the input data of the second convolutional layer from None×3×64 to None×3×64; The first pooling converts the input data of the first pooling layer from None×6×64 to None×64; Third convolution, converting the input data of the third convolutional layer from None×6×64 to None×64; Second pooling, converting the input data of the second pooling layer from None×64 to None×3.

[0009] Furthermore, during the processing of each convolutional layer and pooling layer, it also includes: Calculating the mean and standard deviation of each mini-batch of data; Performing bias adjustment and scaling on the data according to the mean and standard deviation, and normalizing the activation values to a set interval.

[0010] The second aspect of the present invention provides a landing gear shock absorption system fault diagnosis system, including: A data acquisition unit for acquiring aircraft fuselage data and wheel data; A fault response database unit for preprocessing the aircraft fuselage data and wheel data to obtain a fuselage vertical acceleration signal and a wheel vertical acceleration signal, and constructing a fault response database based on the fuselage vertical acceleration signal and the wheel vertical acceleration signal; A fault diagnosis unit for inputting the data of the fault response database into an initial fault model to train the initial fault model, and obtaining a trained fault diagnosis model; A diagnosis result unit for performing fault diagnosis based on the trained fault diagnosis model and outputting the diagnosis result to reversely optimize the fault response database.

[0011] Furthermore, the data acquisition unit includes a sensor, a collector, a storage unit and a power supply unit that are interconnected; The sensor includes a fuselage sensor and a wheel sensor. The fuselage sensor is installed on the cockpit floor, and the wheel sensor is installed on the aircraft hub and fork; Both the fuselage sensor and the wheel sensor use vibration sensors.

[0012] The third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements a landing gear shock absorption system fault diagnosis method.

[0013] The fourth aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements a landing gear shock absorption system fault diagnosis method.

[0014] Compared with the prior art, the present invention has the following advantages and beneficial effects: By constructing a kinetic model, based on the relationship of "shock absorber state - shock absorber performance - aircraft landing response" and aircraft airworthiness requirements, more accurate airframe data is collected. Based on the collected airframe data, a fault response database for the landing gear shock absorption system is established by combining the landing gear dynamics model and flight measured data. An initial fault model is constructed, and the fault training model is optimized based on the fault response database to obtain more accurate fault monitoring results, which can make full use of data advantages to predict and trend judge the landing gear shock absorption system more accurately, efficiently and in real time; By adopting a sensor network based on acceleration sensors with strong robustness and installing it at convenient positions such as the airframe and wheels, fault diagnosis of the landing gear shock absorption system is realized through a data-driven method; The designed on-line fault diagnosis system for the landing gear can realize fault diagnosis of the landing gear without affecting the landing gear structure, greatly saving labor costs. Brief Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts. In the drawings: Figure 1 is the fault diagnosis process of the landing gear shock absorption system in the embodiment of the present invention; Figure 2 is the fault diagnosis model in the embodiment of the present invention; Figure 3 is the vertical acceleration of the airframe during the landing gear simulation process in the embodiment of the present invention; Figure 4 is the vertical acceleration of the wheels during the landing gear simulation process in the embodiment of the present invention; Figure 5 is the fault diagnosis result of the trained fault diagnosis model in the embodiment of the present invention; Figure 6 is the vertical acceleration of the airframe during the actual flight of the aircraft during landing in the embodiment of the present invention; Figure 7 is the vertical acceleration of the wheels during the actual flight of the aircraft during landing in the embodiment of the present invention. Detailed Embodiments

[0016] To make the purpose, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the embodiments and the drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0017] Prognostics and Health Management (PHM) of fault prediction and health management is widely applied in various fields, mainly in the field of aeroengines. Especially in the Aircraft Logistics Information System (ALIS) of the F-35 Joint Strike Fighter, it covers several functions such as aircraft system status monitoring, health assessment, fault prediction, and maintenance planning. Through Condition-Based Maintenance (CBM), the PHM system records and analyzes the health data of electronic systems to achieve the health management of the system. With the progress of technology, the PHM system will be more intelligent and automated, capable of predicting faults more accurately, reducing unplanned downtime, and improving the reliability and efficiency of equipment. The existing Prognostics and Health Management (PHM) of fault prediction and health management has the following problems: Currently, the fault diagnosis of the landing gear mainly relies on maintenance manuals and manual experience, with high labor costs, inability to monitor faults not included in the manuals, and the accuracy of fault diagnosis depending on experience; the pressure gauge can only monitor the air pressure of the oil-gas shock absorber and cannot monitor the internal structure that also affects the performance of the shock absorber. Therefore, based on the above technical problems, this embodiment further improves the existing pre-diagnosis and health management means to obtain the specific implementation content of this implementation: As a possible implementation, as Figure 1 shown, this embodiment provides a method for fault diagnosis of a landing gear shock absorber system, specifically including the following steps: Construct a dynamic model, collect aircraft fuselage data and wheel data based on the dynamic model, preprocess the aircraft fuselage data and wheel data to obtain the fuselage vertical acceleration signal and the wheel vertical acceleration signal; Construct a fault response database based on the fuselage vertical acceleration signal and the wheel vertical acceleration signal; Construct an initial fault model, input the data of the fault response database into the initial fault model to train the initial fault model, and obtain a trained fault diagnosis model; Perform fault diagnosis based on the trained fault diagnosis model and output the diagnosis result, and the diagnosis result is also used to reverse-optimize the fault response database.

[0018] In some possible implementation manners, the specific steps for constructing the dynamic model include: Set virtual acceleration sensors at the fuselage and wheels respectively to collect the vertical accelerations of the fuselage and wheels during the landing gear simulation process. The sampling frequency of the virtual acceleration sensors is 1000 Hz. The calculation results are as Figure 3 and Figure 4 shown: According to the model of the oil-gas shock absorber of the target aircraft, obtain the stiffness and damping parameters for detecting the oil-gas shock absorber of the target aircraft, where: Based on the air pressure data, the air spring force of the oil-gas shock absorber is obtained, specifically including: The formula for the air spring force of the oil-gas shock absorber is: ; where P $p_0$ is the initial pressure of the air chamber, V $V_0$ is the initial volume of the air chamber, $A$ is the piston displacement area, $p_{atm}$ is the atmospheric pressure, μ $s$ is the compression stroke of the shock absorber, γ $n$ is the polytropic coefficient; Based on the forward and reverse stroke speeds of the shock absorber and the oil data, the oil damping force of the oil-gas shock absorber is obtained, specifically including: The formula for the oil damping force of the oil-gas shock absorber is: ; where $\rho$ is the oil density, A h $A_{net}$ is the net cross-sectional area of the oil chamber, C d $C_d$ is the oil contraction coefficient, A d $A_{main}$ is the cross-sectional area of the main oil hole, $v$ is the forward and reverse stroke speed of the shock absorber; Based on the stiffness and damping parameters of the oil-gas shock absorber, the tire force is obtained, specifically including: The formula for the tire force is: ; where k s $k_t$ is the tire spring stiffness, c δ $c_t$ is the tire spring damping coefficient, δ $x_t$ is the tire compression, $\dot{x}_t$ is the tire compression speed; The aircraft gravity is obtained, combined with the tire force, the oil damping force of the oil-gas shock absorber, and the air spring force of the oil-gas shock absorber, to obtain the landing impact force, and a mechanical relationship is constructed, specifically including: Since when the aircraft lands, the nose landing gear is mainly subjected to the following forces: gravity G $G$, F l landing impact force F v $F_{li}$, F h tire force $F_t$, oil damping force

[0019] $F_d$, air spring force $F_a$, therefore, a dynamic model is constructed according to the aircraft mechanical relationship, specifically including:

[0019] In some possible implementation manners, obtaining the fuselage vertical acceleration signal and the wheel vertical acceleration signal further includes: Fusing the fuselage vertical acceleration signal and the wheel vertical acceleration signal to obtain fusion data; The fused data becomes array data, which includes: time array data, fuselage acceleration array data, and wheel acceleration array data.

[0020] In some possible implementation manners, such as Figure 2 As shown, the specific steps for constructing the initial fault model include: Construct an input layer, a feature extraction layer, a feature integration layer, and an output layer, where: The input layer is used to receive the fuselage acceleration and the wheel acceleration; The feature extraction layer includes multiple convolutional layers and pooling layers, and is used to extract features from the input data; The feature integration layer includes multiple fully connected layers, and is used to integrate the features obtained from the feature extraction layer; The output layer is used to output the fault data.

[0021] In some possible implementation manners, the initial fault model mainly includes a convolutional layer, a flattening layer, a connection layer, and a pooling layer; Wherein: The convolutional layer slides over the input data through a series of learnable convolutional kernels to capture local feature patterns; The pooling layer downsamples the output of the convolutional layer. By selecting the maximum value in each small region, it reduces the spatial dimension of the data (i.e., dimensionality reduction), while retaining important feature information; Each neuron in the fully connected layer is connected to all neurons in the previous layer, and is responsible for integrating and learning the non-linear relationships between features, providing a basis for the final diagnostic decision; At the output end of the fully connected layer, a softmax activation function is introduced. The softmax function can convert the original output of the network into a probability distribution form, and each output value represents the prediction probability of the corresponding category, so that the model can give the confidence of each possible diagnostic result; In order to improve the training efficiency and performance of the model, the fault diagnosis model incorporates batch normalization (Batch Normalization) and dropout techniques at two key positions. Among them, batch normalization effectively normalizes the activation values into a fixed interval by calculating the mean and standard deviation on each small batch of data and performing corresponding bias adjustment and scale scaling on the data.

[0022] In some possible implementation manners, when the feature extraction layer extracts features from the input data, the processing process of multiple convolutional layers and pooling layers includes: The first convolution converts the input data of the first convolutional layer from None×6×64 to None×3×64; Second convolution, converting the input data of the second convolutional layer from None×3×64 to None×3×64; First pooling, converting the input data of the first pooling layer from None×6×64 to None×64; Third convolution, converting the input data of the third convolutional layer from None×6×64 to None×64; Second pooling, converting the input data of the second pooling layer from None×64 to None×3.

[0023] In some possible implementation manners, when processing each convolutional layer and pooling layer, it further includes: Calculating the mean and standard deviation of each mini-batch of data; Performing bias adjustment and scaling on the data according to the mean and standard deviation, and normalizing the activation values to a set interval. In some possible implementation manners, the fault response database mainly includes the vertical landing speeds of the aircraft (0.5, 1.0, 1.5, 2.0, and 2.5 m / s), 3 types of fault forms of the initial gas pressure P0, initial gas volume V0, and oil hole area As of the shock absorber, and each fault form includes 5 parameter cases (normal 1.0, 0.8, 0.9, 1.0, normal parameters), etc. The vertical vibration accelerations of the fuselage and wheels under such states are included.

[0024] In some possible implementation manners, the fault diagnosis model is trained using the vertical vibration acceleration signals of the fuselage and wheels in the fault response database. By training the initial fault model, the diagnostic accuracy rates of different faults are obtained. The trained fault diagnosis result (initial gas pressure fault) is as Figure 5 shown In some possible implementation manners, during the actual flight of the aircraft, the vibration acceleration signals collected by the sensors during the aircraft landing process are as Figure 6 and Figure 7 shown.

[0025] In some possible implementation manners, the vibration data collected by the sensors is input into the fault diagnosis model, and the fault diagnosis result is an initial gas pressure fault of the landing gear shock absorber, and the fault level is a 20% reduction in air pressure.

[0026] As a possible implementation manner, as Figure 1 shown, this embodiment provides a fault diagnosis system for a landing gear shock absorber system, including: A data acquisition unit for acquiring aircraft fuselage data and wheel data; A fault response database unit is used to preprocess the aircraft fuselage data and the landing gear data to obtain the fuselage vertical acceleration signal and the landing gear vertical acceleration signal, and construct a fault response database based on the fuselage vertical acceleration signal and the landing gear vertical acceleration signal; A fault diagnosis unit is used to input the data of the fault response database into an initial fault model to train the initial fault model, and obtain a trained fault diagnosis model; A diagnosis result unit is used to perform fault diagnosis based on the trained fault diagnosis model and output the diagnosis result to reversely optimize the fault response database.

[0027] In some possible implementation manners, a data acquisition unit includes a sensor, a collector, a storage unit, and a power supply unit that are interconnected; the sensor includes a fuselage sensor and a landing gear sensor. The fuselage sensor is installed on the cockpit floor, as close as possible to the floor center line; the landing gear sensor is installed on the aircraft hub or fork, as close as possible to the center line of the landing gear strut. Both the fuselage sensor and the landing gear sensor adopt vibration sensors. Due to the composition of the hardware system in this implementation and the use of vibration sensors, the robustness is higher, and it is easier to communicate with airworthiness certification. Moreover, the sensor system only collects the vertical accelerations of the landing gear and the fuselage, and the structure is simple.

[0028] As a possible implementation manner, this embodiment provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements a fault diagnosis method for a landing gear shock absorption system.

[0029] As a possible implementation manner, this embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements a fault diagnosis method for a landing gear shock absorption system.

[0030] The specific implementation manners described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific implementation manners of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for diagnosing faults in a landing gear shock absorption system, characterized in that, It includes the following specific steps: Construct a dynamic model, collect aircraft fuselage data and wheel data based on the dynamic model, preprocess the aircraft fuselage data and wheel data to obtain the fuselage vertical acceleration signal and the wheel vertical acceleration signal; Construct a fault response database based on the fuselage vertical acceleration signal and the wheel vertical acceleration signal; Construct an initial fault model, input the data of the fault response database into the initial fault model to train the initial fault model, and obtain a trained fault diagnosis model; Perform fault diagnosis based on the trained fault diagnosis model and output the diagnosis result, and the diagnosis result is also used to reverse-optimize the fault response database.

2. The method for diagnosing faults in a landing gear shock absorption system according to claim 1, characterized in that, The specific steps of constructing the dynamic model include: Obtain the stiffness and damping parameters of the oleo-pneumatic shock absorber of the detected target aircraft; Obtain the tire force according to the stiffness and damping parameters of the oleo-pneumatic shock absorber; Obtain the oil damping force of the oleo-pneumatic shock absorber according to the forward and reverse stroke speeds of the shock absorber and the oil data; Obtain the air spring force of the oleo-pneumatic shock absorber according to the air pressure data; Obtain the aircraft gravity, combine the tire force, the oil damping force of the oleo-pneumatic shock absorber and the air spring force of the oleo-pneumatic shock absorber to obtain the landing impact force, and construct a mechanical relationship; Construct a dynamic model according to the aircraft mechanical relationship.

3. The method for diagnosing the failure of the landing gear shock absorption system according to claim 1, characterized in that, The obtaining of the fuselage vertical acceleration signal and the wheel vertical acceleration signal further includes: Fuse the fuselage vertical acceleration signal and the wheel vertical acceleration signal to obtain fused data; The fused data is array data, and the array data includes: time array data, fuselage acceleration array data and wheel acceleration array data.

4. The method for diagnosing faults in the landing gear shock absorption system according to claim 1, characterized in that, The specific steps of constructing the initial fault model include: Construct an input layer, a feature extraction layer, a feature integration layer and an output layer, where: The input layer is used to receive the fuselage acceleration and the wheel acceleration; The feature extraction layer includes multiple convolutional layers and pooling layers, and is used to extract features from the input data; The feature integration layer includes multiple fully connected layers and is used to integrate the features obtained from the feature extraction layer; The output layer is used to output fault data.

5. The method for diagnosing the failure of the landing gear shock absorption system according to claim 4, characterized in that, When the feature extraction layer extracts features from the input data, the processing processes of multiple convolutional layers and pooling layers include: The first convolution converts the input data of the first convolutional layer from None×6×64 to None×3×64; The second convolution converts the input data of the second convolutional layer from None×3×64 to None×3×64; The first pooling converts the input data of the first pooling layer from None×6×64 to None×64; The third convolution converts the input data of the third convolutional layer from None×6×64 to None×64; The second pooling converts the input data of the second pooling layer from None×64 to None×3.

6. The method for diagnosing a failure of a landing gear shock absorption system according to claim 5, characterized in that, When processing each convolutional layer and pooling layer, it further includes: Calculate the mean and standard deviation of each mini-batch of data; Perform deviation adjustment and scale transformation on the data according to the mean and standard deviation, and normalize the activation value to a set interval.

7. A fault diagnosis system for a landing gear shock absorption system, applied to the method according to any one of claims 1-6, characterized in that, It includes: A data acquisition unit for collecting aircraft fuselage data and wheel data; A fault response database unit is used to preprocess the aircraft fuselage data and the landing gear data to obtain the fuselage vertical acceleration signal and the landing gear vertical acceleration signal, and construct a fault response database based on the fuselage vertical acceleration signal and the landing gear vertical acceleration signal; A fault diagnosis unit is used to input the data of the fault response database into an initial fault model to train the initial fault model, and obtain a trained fault diagnosis model; A diagnosis result unit is used to perform fault diagnosis based on the trained fault diagnosis model and output a diagnosis result to reversely optimize the fault response database.

8. The landing gear shock absorption system fault diagnosis system according to claim 7, characterized in that, The data acquisition unit includes a sensor, a collector, a storage unit and a power supply unit which are connected to each other; The sensor includes a fuselage sensor and a landing gear sensor. The fuselage sensor is installed on the cockpit floor, and the landing gear sensor is installed on the aircraft hub and the fork; Both the fuselage sensor and the landing gear sensor adopt vibration sensors.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the landing gear shock absorption system fault diagnosis method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the landing gear shock absorption system fault diagnosis method according to any one of claims 1 to 6.

Images