Digital twinning-based undercarriage retracting and releasing mechanism wear prediction method

Abstract

The invention discloses an undercarriage retracting and releasing mechanism wear prediction method based on digital twinning, and the method comprises the steps: firstly building an undercarriage three-dimensional model containing geometric dimensions and material characteristics, building a multi-body dynamics simulation model, and simulating the stress and wear process of a pin shaft under the actual working condition; and adopting an Archard wear calculation model considering key parameters to calculate the wear of the pin shaft in the simulation model in real time. On the basis, uncertainty analysis of parameters in Archard is carried out, the parameters are considered to obey normal distribution, and a prediction model under multivariate uncertainty is established by setting reasonable mean values and standard deviations. And finally, based on the digital twin model, by using a particle filtering method, initializing a particle swarm, changing a particle weight by virtue of a state and an observation equation, re-sampling to approach a real state, iteratively updating a curve, calibrating model parameters and a state, improving prediction precision, improving prediction accuracy and reliability, discovering a potential wear problem in advance, and providing a basis for a maintenance plan.

Description

Technical Field

[0001] The present invention relates to the technical field of wear reliability prediction of landing gear pin shafts, and in particular to an application of digital twin technology in the aviation industry and a wear prediction and analysis method. Background Art

[0002] With the maturity of reliability analysis methods and design theories of aircraft fixed structures, accidents caused by failures of aircraft fixed structures have become less and less. In contrast, the failure problem of motion mechanisms has become increasingly prominent. As a motion mechanism with a long service life, the landing gear retraction mechanism has achieved fruitful results in reliability research based on mechanism kinematics and dynamics in the initial design stage. However, from the perspective of continued airworthiness, considering the decisive influence of the landing gear on the safety of aircraft take-off and landing, it may be more important to study its reliability based on damage accumulation under time-varying conditions. The pins of the landing gear retraction mechanism are prone to wear due to the complex loads and stresses borne by the aircraft during frequent take-offs and landings, which seriously threatens the normal operation of the landing gear and the flight safety of the aircraft.

[0003] However, the current detection and prediction methods for landing gear pin wear have many shortcomings. Traditional manual inspection and measurement methods, such as measuring dimensions with calipers and micrometers and visually inspecting wear marks, can preliminarily determine the wear condition, but the detection accuracy is limited, and the wear process cannot be monitored in real time, and it is even more difficult to predict the wear trend. As the number of landing gear retraction and extension times increases, wear will inevitably occur at the key bearings between components, resulting in a reduction in the pin radius, which may further affect the reliability of the retraction and extension mechanism. Studying damage accumulation can help predict and evaluate the impact of this wear on landing gear performance. With the development of technology, sensor technology has been used to obtain landing gear operating parameters to infer the wear status of the pin. However, it can only provide local and indirect information, and lacks a comprehensive and accurate grasp of the pin wear process.

[0004] Digital twin technology has attracted much attention in the industrial field and has been applied to aspects such as health monitoring of aerospace aircraft structures. However, in the field of landing gear pin wear prediction, its application is still in its infancy and a complete method and system have not yet been formed.

[0005] The limitations of existing technologies in predicting landing gear pin wear have put the aerospace industry at great risk in terms of landing gear maintenance and safety assurance. If pin wear cannot be accurately predicted, it will be difficult to formulate a reasonable maintenance plan, which may lead to excessive maintenance and increased costs, or insufficient maintenance and cause safety accidents. Therefore, there is an urgent need for a more accurate and effective landing gear pin wear prediction method that can overcome the shortcomings of existing technologies, comprehensively and accurately predict pin wear, and provide strong support for landing gear maintenance and aircraft safety flight. Summary of the invention

[0006] In order to overcome the limitations of the prior art in predicting landing gear pin wear, the present invention aims to provide a landing gear retraction and extension mechanism wear prediction method based on digital twinning. By combining the wear prediction method with digital twinning technology, accurate prediction of landing gear pin wear can be achieved, providing a reliable basis for landing gear maintenance and safe flight of aircraft.

[0007] The technical solution of the present invention is as follows: Step 1: construct a three-dimensional model of the landing gear retracting and extending mechanism and a corresponding multi-body dynamics model, simulate the retracting and extending frequency, load distribution and motion state of each part of the retracting and extending mechanism, and obtain simulation data; Step 2, embedding the Archard wear calculation model that considers key factors including normal pressure, pin rotation angle, friction coefficient, and Brinell hardness, calculating the pin wear volume increment V in the multi-body dynamics simulation model in real time, and accumulating the total wear volume to update the three-dimensional model; Step 3, carry out uncertainty analysis of each key parameter in the Archard wear calculation model, set the key parameters as random variables that obey the normal distribution, and determine the value range based on the physical meaning and actual monitoring data; integrate within the interval to achieve a preset confidence level, and then build a probability distribution model of the parameters; build a multivariate random variable matrix based on the probability distribution model of the parameters, and generate different parameter combinations through a large number of random samplings, and substitute them into the Archard wear calculation model in step 2 to calculate the wear value; finally, use probability and statistical methods to analyze the wear value, obtain uncertainty distribution data on the wear amount, build an accurate wear prediction model, and obtain a wear prediction curve; Step 4: Build a digital twin model, initialize the particle swarm based on the particle filtering method, and change the particle weights based on the state and observation equations. On this basis, resampling operations are used to gradually make the particle state approach the real state. At the same time, the wear prediction curve is continuously iterated and updated, and the model parameters and states are calibrated and optimized to improve the prediction accuracy.

[0008] Preferably, in step 1, in order to drive the model to measure the contact force, add an actuator drive force , where is the friction force, is the air resistance, is the actuator efficiency.

[0009] Preferably, in the three-dimensional modeling software, a rigid body model of each component is created according to the geometric dimensions and material property parameters of the landing gear pin, and the connection and interaction between the components are simulated by constraint pairs and force elements; the pin is modeled as a cylindrical rigid body and connected to the surrounding connectors by adding contacts; in the dynamics simulation software, a multi-body dynamics model is established by considering the connection relationship of each component, motion constraints and actual load distribution; the boundary conditions of the multi-body dynamics model are set according to the actual installation configuration of the landing gear, the connection stiffness characteristics and the buffer performance parameters; the load distribution is simulated according to the characteristics of different load conditions of the aircraft.

[0010] Preferably, in step 1, based on the retraction and extension frequency and time of the landing gear, aircraft weight distribution data, landing gear structure and mechanical principles, the range of axial force, radial force and torque exerted on the pin during takeoff and landing, as well as the position area of ​​the key pin that is more affected during the retraction and extension process of the landing gear are estimated.

[0011] Preferably, in step 2, the normal pressure P is determined by simulating and calculating the contact force between the pin and adjacent components using a multi-body dynamics model, the relative slip distance L is calculated based on the kinematic analysis of the landing gear, the Brinell hardness H of the material adopts the collected parameter value, and the wear factor K adopts the reference value.

[0012] Preferably, in step 2, the Archard wear calculation model calculates the wear volume increment based on mechanical and material properties: V=KPL / H (1), where the normal pressure P and the relative sliding distance L are obtained by simulating the multi-body dynamics model, H is the Brinell hardness of the material, and K is the wear factor; a single wear process analysis is performed: according to the movement characteristics of the landing gear, the relative rotation angle of the pin is determined and the number of wear times n, calculate the relative slip distance: L = nπrφ / 180 (2), calculate the wear volume increment of a single pin based on the geometric relationship: V = [π(r+△r) 2 -πr 2 ]b=2πrb△r (3), Combining equations (1) to (3) to obtain the pin radius wear: △r=nKPφ / (360bH) (4), where r is the pin radius and b is the pin height.

[0013] Preferably, in step 2, the wear factor of the Archard wear model is dynamically calibrated according to the pin material pairing characteristics, grease type and operating temperature change curve; during the wear calculation process, the wear parameters are updated at preset time intervals to improve the adaptability of the wear calculation to the dynamic changes of actual operating conditions.

[0014] Preferably, step 4 includes: Step 4.1, taking the multi-body dynamics model, the wear calculation model and the wear prediction model as the core, and combining the metal particle monitoring data in the pin lubricating oil, a digital twin model is constructed; In step 4.2, the particle swarm is initialized based on historical experience and initial state estimation through the particle filtering algorithm to characterize the probability distribution of the pin wear state; based on the dynamics and wear model, a state equation is established to predict the change of particle state; at the same time, the observation equation is constructed using real-time monitoring data to update the particle weight; in addition, the distribution of the particle swarm is optimized through resampling technology, and the wear curve, calibration model parameters and state estimation values ​​are iteratively updated to continuously improve the accuracy of wear prediction and the overall reliability of the model.

[0015] Preferably, in step 4.2, first, the particle swarm is initialized according to the early wear monitoring data of the pin shaft; in the prediction stage, the particle state transfer is promoted by relying on the state equation and various input information obtained in real time, so as to obtain the predicted distribution of the system state; in the update stage, the observation equation is constructed using the real-time monitoring data, the particle weight is updated, the distribution of the particle swarm is optimized through the resampling technology, and the wear curve, calibration model parameters and state estimation values ​​are iteratively updated.

[0016] As a preference, considering the number of retraction and extension times n, the normal pressure P, the pin rotation angle φ, the wear factor K, the pin height b and the sampled error vector v within each iteration time interval, the state transition formula representing the pin wear state is: Wherein, subscripts k and k-1 are the k and k-1 moments, subscripts 1, 2, and 3 are the 1st, 2nd, and 3rd state variables, and ɡ(·) is the state transfer function, which is obtained by the calculation formula of the pin wear volume increment.

[0017] Preferably, step 4.2 specifically includes: step 4.21, initializing the particle swarm: at the initial moment, determine the value range of the state vector X0 according to prior knowledge or initial monitoring data, randomly sample N particles from the range, and assign an initial weight to each particle: , Where i represents the particle number; Step 4.22, state prediction: For each particle, update the particle state according to the state transfer formula: , Where f(·) is the state transfer function, u is the control input, and v k (i) The error vector obtained by sampling is assigned an initial value using a normal distribution; Step 4.23, weight update: Compare the calculated results with the actual monitored data, and redistribute the weights of the multiple groups of particles initially randomly obtained: , Where p is the value of n k The wear radius X after the updated state equation under the wear number k The likelihood value obtained by comparing with the observed value; Step 4.24, normalize the weights so that their sum is 1: , the final system state estimation is obtained by weighted averaging, expressed as: ,in is the estimated value of the system state at time k, is the state vector of the ith particle at time k, which contains the state variable information of the pin wear, Brinell hardness and radius.

[0018] Beneficial Effects

[0019] 1. The present invention establishes a wear prediction model for the landing gear retraction mechanism, and combines actual monitoring data with a reasonable wear prediction algorithm to reflect the operating status of the pin in real time while considering the uncertainty of multiple variables, thereby achieving a more accurate simulation of the wear process of the pin under actual conditions and an accurate prediction and evaluation of the wear conditions of the aircraft landing gear retraction mechanism and its main pins.

[0020] 2. The present invention also relies on the feedback and correction of actual operation data, and combines the idea of ​​particle filtering to update and correct the specific wear model state in real time, thereby improving the reliability of the prediction, thereby reducing the errors caused by model simplification or the current size being inconsistent with the theoretical model after wear in traditional methods, thereby more accurately reflecting the consistency between the theoretical model and the actual data.

[0021] 3. This method can accurately predict the wear trend of the landing gear pin within a certain number of future operations. Through the combined analysis of historical data and real-time data, it provides airlines with detailed wear trend curves and forecast reports, and enables them to formulate reasonable maintenance plans in advance based on these forecast information, reduce maintenance costs, and reduce the incidence of aircraft delays or safety accidents caused by sudden wear failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flow chart is established for a digital twin prediction model of an embodiment of the present invention; Figure 2 A landing gear multi-body dynamics model (including a key pin) of an embodiment of the present invention; Figure 3 A particle filter correction curve according to an embodiment of the present invention; Figure 4 The simulation result of the wear model of one embodiment of the present invention - the change of element force over time; Figure 5The simulation results of the wear model of one embodiment of the present invention are as follows: the variation of angular velocity, acceleration and velocity over time; Figure 6 A wear prediction result of an embodiment of the present invention - a schematic diagram of wear amount distribution corresponding to the number of wear times; Figure 7 The wear prediction result of one embodiment of the present invention is the predicted value of the wear amount per unit time and its 95% confidence interval; Figure 8 This is the wear prediction result of an embodiment of the present invention—the predicted value of the cumulative wear amount and its 95% confidence interval. DETAILED DESCRIPTION

[0023] The following will be combined with some drawings in the actual operation of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] The present invention mainly focuses on predicting the wear of the landing gear pin shaft by using digital twin technology, and provides technical and method support for the maintenance and reliability assurance of the landing gear. The present invention discloses a wear prediction method for the landing gear retraction and extension mechanism based on digital twin. First, a three-dimensional model of the landing gear including geometric dimensions and material properties is established, and a multi-body dynamics simulation model is constructed in combination with actual operating data such as the retraction and extension frequency and load distribution of the landing gear, and boundary conditions and initial parameters are set to simulate the force and wear process of the pin shaft under actual working conditions. The Archard wear calculation model considering key factors such as normal pressure, pin shaft rotation angle, friction coefficient, and Brinell hardness is used to calculate the pin shaft wear in real time in the multi-body dynamics simulation model. On this basis, the uncertainty analysis of each parameter in the Archard model is carried out. In view of the uncertainty of each parameter in the actual working condition due to the influence of many factors such as material microstructure differences, manufacturing process fluctuations, and operating environment changes, it presents certain probability distribution characteristics. Each parameter is considered to obey the normal distribution, and a prediction model under multivariate uncertainty is established by setting reasonable means and standard deviations for these parameters. Finally, by reading the information data such as the content, composition and particle size distribution of metal particles in the lubricating oil detected by relevant institutions, the actual monitoring data is obtained for comparison and verification. Based on the digital twin model, the particle swarm is initialized by using the particle filtering method, and the particle weights are changed by using the state and observation equations. The resampling approximates the true state, the curve is iteratively updated, and the model parameters and states are calibrated to improve the prediction accuracy. The prediction accuracy and reliability are improved, potential wear problems are discovered in advance, a basis is provided for maintenance plans, and the risk of failure is reduced. At the same time, a reference is provided for the design improvement of the landing gear to extend its service life.

[0025] like Figures 1 to 8 As shown, the present invention is applicable to the aircraft landing gear retracting and extending mechanism and its substructures. Taking the landing gear multi-body dynamics model established by the present invention as an example:

[0026] Step 1: Construct a three-dimensional model and a multi-body dynamics model of the landing gear retraction and extension mechanism, simulate the landing gear retraction and extension frequency, load distribution, and the motion state of each part of the retraction and extension mechanism, and obtain simulation data.

[0027] Specifically, by consulting the aircraft's flight manual, landing gear design manual, material manual and related technical documents, and combining with the aviation operator's observation database, key information and parameters are collected, covering the diameter, length, and three-dimensional shape characteristics of the landing gear pin, as well as its installation position in the landing gear, connection method, and coordination with other components, as well as the landing gear retraction and extension frequency, retraction and extension time and other operating data. The three-dimensional modeling software is used to build a three-dimensional model of the aircraft landing gear retraction and extension mechanism, and then the multi-body dynamics software is used to establish a multi-body dynamics model based on the actual structure and the pin connection method. During modeling, rigid body models of each component are created in the software, and the connection and interaction of components are simulated through constraint pairs and force elements. The pin is modeled as a cylindrical rigid body and contacts are added with surrounding connectors to accurately simulate its degrees of freedom of movement and force conditions. The model contains accurate parameters of key components (such as the elastic modulus, Poisson's ratio, Brinell hardness and yield strength of the pin material), which can accurately reflect the mechanical effects and connection conditions.

[0028] In this embodiment, driving force is added for simulation. Only when the model is able to move can the contact force be measured. Based on the actuator driving force calculation model, the actuator driving force is calculated. Perform the calculation: Among them, the theoretical driving force of the actuator ; By considering the total mass of the landing gear and related components , Gravitational acceleration , friction coefficient Determining Friction ,Right now , Refer to the manual or literature theoretical values ​​and combine them with test and flight data corrections to determine the air density based on flight conditions and atmospheric models. , obtain the air resistance coefficient by consulting the wind tunnel test data , the windward area is determined by geometric measurement and analysis , determine the retraction and extension speed according to the design specifications and operation requirements , calculate the air resistance ; The efficiency of the actuator is ; When determining the driving force, the driving force actually applied to the actuator for: .

[0029] Based on the above data, combined with the landing gear retraction and extension frequency, aircraft weight distribution, etc., the pin force range and key pin positions are estimated, and simulation data is obtained to provide support for subsequent research.

[0030] Step 2: Embed the Archard wear calculation model that considers the key factors of normal pressure, pin rotation angle, friction coefficient, and Brinell hardness into the multi-body dynamics model. Through dynamic simulation, the normal pressure, contact force, relative rotation angle, relative sliding distance and other parameters of the pin movement are accurately obtained, the wear volume increment is calculated in real time and the total wear volume is accumulated, the wear size change of the pin is accurately calculated, the wear process is accurately quantified and simulated, and a wear prediction model that can accurately describe the wear process is established. The landing gear multi-body dynamics model and the key pin model are as follows: Figure 3 As shown in the figure. A suitable wear prediction algorithm is used to consider the wear calculation model of key factors such as the size of the pin material, Brinell hardness, friction coefficient, angular velocity of the retraction and extension movement, and load size. Subsequently, real-time wear simulation is performed to obtain multiple sets of contact force, relative rotation angle and other parameters.

[0031] Specifically: Based on the multi-body dynamics model, the wear volume increment is calculated based on the Archard model for mechanical and material properties: V=KPL / H (1) where the normal pressure P is determined by simulating the contact force between the pin and adjacent components; the relative slip distance L of each movement of the pin is calculated based on the kinematic analysis of the landing gear; the Brinell hardness H of the material adopts the collected parameter value, and the wear factor K is determined by referring to existing literature.

[0032] Single wear process analysis: Determine the relative rotation angle of the pin according to the movement characteristics of the landing gear The relative slip distance L is calculated by combining formula (2) with the number of wear times n: L = nπrφ / 180 (2) The wear volume increment of a single pin is calculated based on the geometric relationship by formula (3), where b is the pin height: V = [π(r+△r) 2 -πr 2 ]b=2πrb△r (3)Combining equations (1) to (3), we can obtain the wear amount of the pin radius (Equation 4):△r=nKPφ / (360bH) (4)Equations (1) and (3) calculate the wear volume increment from two perspectives, and the formula for the wear amount of the pin radius is obtained by combining equations (1)-(3), which is convenient for model updating. In the model, according to the pin size, material parameters, and the force and motion parameters calculated by the multi-body dynamics model, the wear amount of each movement is calculated in real time and the total wear amount is accumulated. The simulation model is updated according to the pin radius wear amount and the total wear amount. For example, the pin radius wear amount ∆r is obtained, that is, ∆r is reduced from the original three-dimensional model to achieve the update.

[0033] The wear factor of the Archard wear model is dynamically calibrated based on the pin material pairing characteristics, grease type, and operating temperature change curve. During the wear calculation process, the wear parameters are updated at preset time intervals to improve the adaptability of the wear calculation to the dynamic changes of actual operating conditions.

[0034] Step 3: Explore the sources of uncertainty in the Archard model parameters, which may come from material properties (such as composition fluctuations, processing differences, and microstructural changes), load conditions (fluctuating due to flight mission diversity and environmental factors), and friction coefficients (due to dynamic changes in lubrication conditions, surface roughness evolution, and temperature effects).

[0035] In order to build an accurate wear prediction model, the wear prediction model is connected to Matlab, and each parameter is deeply analyzed. For example, the Brinell hardness of the pin is affected by material composition and heat treatment; the friction coefficient is affected by lubrication, surface roughness and other factors. The above key parameters are regarded as random variables that obey the normal distribution. The value range of each parameter is determined according to the physical meaning and monitoring data, and the probability distribution model of the wear calculation parameters is constructed by integrating the interval to reach the preset confidence level. Specifically, the angular velocity and load of the retraction and release are diverse and uncertain, which are clarified by statistical analysis of simulation data. An initial parameter distribution model is established based on multiple groups of observation data, and multiple groups of variables are substituted into the wear model calculation using a random number generation algorithm. The interval estimation is optimized using small sample test data, and the wear results are obtained according to the variable group for each retraction and release simulation. After multiple simulations, the predicted distribution of the wear radius is obtained, and the multi-body dynamics model is updated accordingly, the value range of the relevant parameters is adjusted, and the value and calculation steps are repeated. After multiple cycles, the total predicted distribution and confidence interval of the wear radius changing with the number of retraction and release simulations are obtained. In this embodiment, the simulation results of the wear model are as follows: Figures 4 to 5 As shown in Figure 2, the wear prediction results and confidence intervals of the wear simulation model are shown in Figure 2. Figures 6 to 8 As shown in the figure, these results are obtained based on the constructed multi-body dynamics model, Archard wear calculation model, uncertainty analysis and particle filtering algorithm. Figure 4 The element force of two contact points (CONTACT_1 and CONTACT_2) in the X and Y directions changes with time (the horizontal axis represents time, and the vertical axis represents the magnitude of the force). In the multi-body dynamics model, the interaction between components is simulated by kinematic pairs, drives, and force elements. The actuator drive force affects the motion state of the components, thereby generating contact forces. Figure 5The angular velocity, acceleration and speed of the component (PART_34_CM) change with time (the horizontal axis represents time, and the vertical axis represents the magnitude of the angular velocity, acceleration and speed). Specifically, CONTACT_1 has tensile or compressive force in the X direction, and the overall trend in the Y direction is relatively stable, but there are peaks at certain moments; CONTACT_2 is similar to CONTACT_1 in the X direction, but the value range is larger and the fluctuation is more severe, and it also shows a certain volatility in the Y direction. The angular velocity, acceleration and speed of the component (PART_34_CM) all show periodic fluctuations over time, with the maximum angular velocity close to 70 degrees / second, the maximum acceleration close to 5 meters / second², and the speed value is small. Figure 6 The scatter plot shows the wear amount distribution corresponding to each wear number (the horizontal axis represents the wear number, and the vertical axis represents the wear amount per unit number). The color depth represents the density of the data points. The darker the color, the more data points there are in the area. It can be seen that the wear amount per unit number fluctuates around 0.12, and as the number of wear increases, the range of wear amount changes gradually decreases. Figure 7 The predicted value of wear per unit number and its 95% confidence interval are shown (the horizontal axis represents the number of wears, and the vertical axis represents the wear per unit number). It can be seen that the predicted value is higher when the number of wears is lower, and tends to be stable when the number of wears is higher, and the confidence interval is wider when the number of wears is lower, and gradually narrows when the number of wears is higher. Figure 8 The predicted value of the cumulative wear volume and its 95% confidence interval are shown (the horizontal axis represents the number of wear times, and the vertical axis represents the cumulative wear volume). It can be seen that the cumulative wear volume increases linearly with the increase in the number of wear times, and the confidence interval is narrower when the number of wear times is lower, and gradually widens when the number of wear times is higher.

[0036] Based on this probability distribution model, a multivariate matrix containing multiple random variables is generated. Different parameter combinations are generated through a large number of random samplings and substituted into the Archard wear model to calculate the wear value. These results are analyzed using probability statistics techniques to obtain uncertainty distribution information about the wear amount.

[0037] In this embodiment, considering that there is a certain cognitive uncertainty in the key parameters, it is necessary to determine their actual observation uncertainty (b, P, φ, etc.) when they are not worn before the initial calculation. Assuming that these parameters follow a certain distribution (such as a normal distribution), a number of observations can be randomly selected from the distribution to establish a new multivariate matrix. This new matrix is ​​used to recalculate multiple sets of wear and their cumulative values.

[0038] After the first iteration, according to the current distribution of each parameter and the information obtained from the previous cycle optimization, a random number generation algorithm is used to sample again to form a new matrix. The sampling process uses a random number table sampling method based on normal distribution to ensure the randomness of the sampling and the coverage of parameter combinations. Then this new matrix is ​​input into the Archard wear model to update the wear amount estimation, and this process is repeated until the stop condition is met, as shown in formula (5).

[0039] (5).

[0040] Finally, according to the preset ratio, the maximum wear amount and the minimum wear amount at the two ends after sorting are eliminated, and only the middle part is retained as the valid data set. Based on the remaining valid data points, the uncertainty curve of the wear amount and its cumulative value changing with time is drawn.

[0041] Step 4: Based on Bayesian estimation and Monte Carlo sampling strategy, particle filtering is used to realize the update prediction of the wear curve by the digital twin model. Specifically, with the multi-body dynamics model and wear prediction model as the core, the digital twin model is constructed in combination with the metal particle monitoring data in the pin lubricant. A large number of particles are used in this model to characterize the probability distribution of the system state, which includes not only the wear amount of the pin, but also the set of parameters related to it. The state equation is constructed through the wear calculation model to describe the dynamic evolution of particles with the number of wear times.

[0042] Firstly, the particle swarm is initialized based on historical experience and initial state estimation through the particle filtering algorithm to characterize the state vector of the pin wear state.

[0043] For the wear amount X k According to Archard wear formula, considering the factors such as the number of retraction and extension n, normal pressure P, pin rotation angle φ, wear factor K, pin height b, wear number n, and sampled error vector v within each iteration time interval, its state transfer relationship can be expressed as: Subscripts k, k-1 represent the k, k-1 moments, and subscripts 1, 2, 3 represent the 1st, 2nd, and 3rd state variables, i.e., X k , X k-1 represents the wear amount at adjacent time k and k-1, X k-1,2 , X k-1,3 represents the second and third state variables of the pin wear state at the k-1th moment. Function g(·) is the state transfer function, which updates the wear amount estimate in the Archard wear model, i.e., the above formula (3).

[0044] Based on the dynamics and wear model, the state equation is established to predict the change of particle state; at the same time, the observation equation is constructed using real-time monitoring data to update the particle weight. In addition, the distribution of the particle group is optimized through resampling technology, and the wear curve, calibration model parameters and state estimation values ​​are iteratively updated to continuously improve the accuracy of wear prediction and the overall reliability of the model.

[0045] In this embodiment, an initialization operation is first performed. Based on the accumulated experience data (pin early wear monitoring data), a particle group is established in the parameter space, and an initial weight is assigned to each particle, thereby constructing a probabilistic description of the initial state of the system.

[0046] At the initial moment, the value range of the state vector Xo is determined based on prior knowledge or a small amount of initial monitoring data, N particles are randomly sampled from this range, and an initial weight is assigned to each particle: For each particle, according to the state equation: Where f(·) is the state transfer function, u is the control input, and v k (i) is the error vector obtained by sampling, assigned with normal distribution.

[0047] The new state of each particle is calculated according to the above update formulas for wear, Brinell hardness and radius. For example, for the i-th particle, according to its previous state and current input vector, combined with the state transfer function and the sampled process noise, the new state of each particle is calculated. .

[0048] Compare the calculated results with the actual monitoring data to redistribute the weights of the multiple groups of particles initially randomly obtained: Where p is the likelihood value of the wear radius Xk compared with the observed value after the state equation is updated under Nk wear times.

[0049] The particle weights are then normalized so that their sum is 1: The final system state estimate can be obtained by weighted averaging, expressed as: ,in is the estimated value of the system state at time k, is the state vector of the ith particle at time k, which contains the estimated values ​​of state variable information such as pin wear, Brinell hardness, and radius. For example, during the entire wear process, if it is assumed that the pin height does not change with wear (in fact, there may be very small changes due to wear, which is simplified here), then its estimated value remains unchanged at the initial value at each moment, that is: ,in, is the estimated value of the pin height at time k, is the initial value of the pin height. If we consider a more complex situation, assuming that the change in pin height is related to wear or other factors, the change relationship is ( is a coefficient related to the pin material and wear characteristics), the estimated value is calculated as: ,in is the number of particles in The pin height status value at the moment, is the normalized particle weight, and the same is true for the other components.

[0050] After completing the above particle filter prediction, repeat step 3 to correct the total prediction distribution curve of the wear prediction radius changing with the number of retraction and release simulations. This iterative update mechanism enables the digital twin model to be optimized over time and with more data accumulation, improving the accuracy and reliability of the model. Finally, we obtain the total prediction distribution and confidence interval of the wear prediction radius changing with the number of retraction and release simulations after iterative update. The particle filter correction curve is shown in Figure 3 shown.

[0051] In the prediction stage, the particle state transfer is promoted by relying on the state equation and various input information obtained in real time, such as complex and changeable working conditions and operating parameters, to accurately describe the particle state evolution and obtain the predicted distribution of the system state.

[0052] In the update phase, the likelihood value of each particle is calculated based on the constructed observation equation and the latest monitoring data. The new weight setting is proportional to the likelihood value and the prior weight, so as to achieve dynamic update of particle weights. Subsequently, the resampling process is carried out to copy or discard particles in a targeted manner according to the updated weights to ensure that the number of particle swarms remains constant, effectively prevent the occurrence of particle degradation, and ensure that the particle swarm can reflect the true distribution of the system state.

[0053] Through multiple iterations, the particle set gradually converges to the high likelihood region, presenting the true probability distribution of the wear state. Finally, based on the parameter expectations contained in the particle swarm, the predicted value of the wear curve is updated to improve the accuracy of wear prediction and the overall reliability of the digital twin model.

[0054] Step 5: In the actual application process, continuously collect the operation data of the aircraft landing gear pin and the information related to the actual wear. Compare and analyze the wear results predicted by the model with the actual wear data, and continuously optimize the parameters of the model according to the comparison differences. By continuously repeating this process, the prediction accuracy and reliability of the model are continuously improved to better adapt to different flight conditions and landing gear usage conditions, and provide more accurate support for the wear prediction and maintenance decision of the aircraft landing gear pin.

[0055] This application embeds the Archard wear model into multi-body dynamics simulation, and combines it with parameter uncertainty analysis (such as normal distribution random variables and Monte Carlo sampling) to solve the prediction deviation problem caused by simplified models in traditional methods. The particle filter algorithm is introduced to dynamically correct model parameters and wear curves through the interaction of state equations and real-time monitoring data, overcoming the limitation that static models are difficult to adapt to dynamic changes in working conditions. Through the digital twin framework, multi-source data (simulation data, monitoring data, material parameters) are integrated to achieve full-process closed-loop optimization of wear prediction. Compared with traditional methods, this solution significantly improves the accuracy and reliability of wear prediction (such as confidence interval analysis) by quantifying parameter uncertainty and dynamically calibrating models. It can identify potential wear risks in advance, optimize maintenance plans, and reduce costs and safety risks caused by excessive or insufficient maintenance.

[0056] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, 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 wear prediction method for landing gear retraction and extension mechanism based on digital twin, characterized in that: The following steps are involved: Step 1: construct a three-dimensional model of the landing gear retracting and extending mechanism and a corresponding multi-body dynamics model, simulate the retracting and extending frequency, load distribution and motion state of each part of the retracting and extending mechanism, and obtain simulation data; Step 2, embedding the Archard wear calculation model that considers key factors including normal pressure, pin rotation angle, friction coefficient, and Brinell hardness, calculating the pin wear volume increment V in the multi-body dynamics simulation model in real time, and accumulating the total wear volume to update the three-dimensional model; Step 3, carry out uncertainty analysis of each key parameter in the Archard wear calculation model, set the key parameters as random variables that obey the normal distribution, and determine the value range based on the physical meaning and actual monitoring data; integrate within the interval to achieve a preset confidence level, and then build a probability distribution model of the parameters; build a multivariate random variable matrix based on the probability distribution model of the parameters, and generate different parameter combinations through a large number of random samplings, and substitute them into the Archard wear calculation model in step 2 to calculate the wear value; finally, use probability and statistical methods to analyze the wear value, obtain uncertainty distribution data on the wear amount, build an accurate wear prediction model, and obtain a wear prediction curve; Step 4: Build a digital twin model, initialize the particle swarm based on the particle filtering method, and change the particle weights based on the state and observation equations. On this basis, resampling operations are used to gradually make the particle state approach the real state. At the same time, the wear prediction curve is continuously iterated and updated, and the model parameters and states are calibrated and optimized to improve the prediction accuracy.

2. The method according to claim 1, characterized in that In step 1, to drive the model to measure the contact force, add the actuator drive force , where is the friction force, is the air resistance, is the actuator efficiency.

3. The method according to claim 1, characterized in that In the three-dimensional modeling software, a rigid body model of each component is created according to the geometric dimensions and material property parameters of the landing gear pin shaft, and the connection and interaction between the components are simulated through constraint pairs and force elements; the pin shaft is modeled as a cylindrical rigid body and connected to the surrounding connecting parts by adding contacts; in the dynamics simulation software, a multi-body dynamics model is established by considering the connection relationship of each component, motion constraints and actual working load distribution; the boundary conditions of the multi-body dynamics model are set according to the actual installation configuration of the landing gear, the connection stiffness characteristics and the buffer performance parameters; The load distribution is simulated based on the characteristics of the aircraft such as different load conditions.

4. The method according to claim 3, characterized in that In step 1, based on the landing gear retraction and extension frequency, retraction and extension time, aircraft weight distribution data, landing gear structure and mechanical principles, estimate the range of axial force, radial force and torque on the pin during takeoff and landing, as well as the location area of ​​the key pin that is more affected during the landing gear retraction and extension process.

5. The method according to claim 1, characterized in that In step 2, the Archard wear calculation model calculates the wear volume increment based on mechanical and material properties: V=KPL / H (1) Wherein, the normal pressure P and the relative sliding distance L are obtained by simulating the multi-body dynamics model, H is the Brinell hardness of the material, and K is the wear factor; Conduct a single wear process analysis: Determine the relative rotation angle of the pin according to the movement characteristics of the landing gear And the number of wear times n, calculate the relative slip distance: L = nπrφ / 180 (2) Calculate the wear volume increment of a single pin based on geometric relationships: V=[π(r+△r) 2 -pr 2 ]b=2πrb△r (3) Combining equations (1) to (3) to obtain the pin radius wear: △r=nKPφ / (360bH) (4) Where r is the pin radius and b is the pin height.

6. The method according to claim 5, characterized in that In step 2, the wear factor of the Archard wear model is dynamically calibrated according to the pin material pairing characteristics, grease type and operating temperature change curve; during the wear calculation process, the wear parameters are updated at preset time intervals to improve the adaptability of the wear calculation to the dynamic changes of actual operating conditions.

7. The method according to any one of claims 1 to 6, characterized in that: Step 4 includes: Step 4.1, taking the multi-body dynamics model, the wear calculation model and the wear prediction model as the core, and combining the metal particle monitoring data in the pin lubricating oil, a digital twin model is constructed; In step 4.2, the particle swarm is initialized based on historical experience and initial state estimation through the particle filtering algorithm to characterize the state vector of the pin wear state; based on the dynamics and wear model, the state equation is established to predict the change of particle state; at the same time, the observation equation is constructed using real-time monitoring data to update the particle weight; in addition, the distribution of the particle swarm is optimized through resampling technology, and the wear curve, calibration model parameters and state estimation values ​​are iteratively updated to continuously improve the accuracy of wear prediction and the overall reliability of the model.

8. The method according to claim 7, characterized in that In step 4.2, first, the particle swarm is initialized according to the early wear monitoring data of the pin shaft; in the prediction stage, the particle state transfer is promoted based on the state equation and the input information obtained in real time, so as to obtain the predicted distribution of the system state; in the update stage, the observation equation is constructed using real-time monitoring data, the particle weight is updated, the distribution of the particle swarm is optimized through resampling technology, and the wear curve, calibration model parameters and state estimation values ​​are iteratively updated.

9. The method according to claim 8, characterized in that Considering the number of retraction and extension times n, normal pressure P, pin rotation angle φ, wear factor K, pin height b and sampled error vector v within each iteration time interval, the state transition formula representing the pin wear state is: ， Wherein, subscripts k and k-1 are the k and k-1 moments, subscripts 1, 2, and 3 are the 1st, 2nd, and 3rd state variables, and ɡ(·) is the state transfer function, which is obtained by the calculation formula of the pin wear volume increment.

10. The method according to claim 8, characterized in that Step 4.2 specifically includes: Step 4.21, initialize the particle swarm: At the initial moment, determine the value range of the state vector X0 based on prior knowledge or initial monitoring data, randomly sample N particles from this range, and assign an initial weight to each particle: ， Where i represents the particle number; Step 4.22, state prediction: For each particle, update the particle state according to the state transfer formula: ， Where f(·) is the state transfer function, u is the control input, and v k (i) The error vector obtained by sampling is assigned an initial value using a normal distribution; Step 4.23, weight update: Compare the calculated results with the actual monitored data, and redistribute the weights of the multiple groups of particles initially randomly obtained: ， Where p is the value of n k The wear radius X after the updated state equation under the wear number k The likelihood value obtained by comparing with the observed value; Step 4.24, normalize the weights so that their sum is 1: ， The final system state estimate is obtained by weighted averaging and is expressed as: ， in for The estimated value of the system state at time For the Particles in The state vector at time t contains the state variable information of pin wear, Brinell hardness and radius.

Images