Marine seat fatigue test system based on digital twinning and iterative learning control

Abstract

The invention discloses a marine seat fatigue test system based on digital twinning and iterative learning control, and the system comprises a data collection and working condition simulation module which is used for obtaining marine environment, ship motion and seat mechanical response data, and generating a composite load spectrum simulating a real working condition; the marine seat digital twin model module constructs a virtual model corresponding to a physical seat, integrates structural dynamics, a material constitutive model and a damage accumulation model, and is used for mapping the real-time state of the physical seat; the iterative learning and intelligent optimization control module is used for receiving feedback of the digital twinborn model, correcting a test load spectrum on line, and allocating computing resources by using a dynamic priority scheduling model so as to accelerate the test; and the life prediction and evaluation module is used for probabilistically predicting the residual fatigue life of the seat. Through combination of the high-fidelity digital twinborn model and iterative learning control, precise simulation of complex and random ocean loads is realized, and a test environment is closer to a real working condition.

Description

Technical Field

[0001] This invention relates to the field of marine equipment testing technology, specifically to a marine seat fatigue testing system that combines digital twin and iterative learning control, belonging to the category of software system inventions. Background Technology

[0002] Marine seats operate under complex marine conditions for extended periods, enduring complex loads with multiple degrees of freedom and varying amplitudes. Fatigue life testing is crucial for ensuring safety and reliability. Traditional physical testing methods are time-consuming, costly, and struggle to reproduce the coupled loads of random waves and ship motion in the real marine environment. Existing digital simulation tests often employ fixed load spectra, failing to consider time-varying loads and system nonlinearities, resulting in insufficient prediction accuracy and low testing efficiency.

[0003] Currently, there is a lack of systematic solutions capable of integrating operational data in real time, accelerating testing through intelligent algorithms, and accurately predicting lifespan. Therefore, there is an urgent need for a testing system that can accurately simulate real marine loads and possess autonomous learning and optimization capabilities. Summary of the Invention

[0004] The purpose of this invention is to provide a marine seat fatigue testing system based on digital twin and iterative learning control. This system constructs a high-fidelity digital twin model and combines an improved intelligent optimization algorithm and dynamic scheduling strategy to achieve accurate simulation of complex marine loads, intelligent acceleration of the testing process, and accurate prediction of seat life.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A fatigue testing system for marine seats based on digital twin and iterative learning control includes:

[0007] The data acquisition and operational condition simulation module is used to acquire marine environmental parameters of the target navigation area and generate a six-degree-of-freedom motion time-domain sequence of the ship based on the ship's hydrodynamic model and wave spectrum; it simultaneously acquires multi-axis mechanical response data at the seat mounting point; based on coordinate transformation and Newton-Euler equations, it transforms the ship's motion sequence into basic motion and inertial loads acting on the seat structure; and it superimposes additional dynamic loads generated by the parameterized model of local vibration and impact events to synthesize a time-varying composite load spectrum for driving the digital twin model module.

[0008] Marine Seat Digital Twin Model Module: Constructs a virtual model that fully corresponds to the physical seat. This model integrates structural dynamics model, material constitutive model and damage accumulation model, and is used to map the real-time state and evolution process of the physical seat in virtual space.

[0009] Iterative Learning and Intelligent Optimization Control Module: This is the core module. It receives feedback data from the digital twin model, corrects the test load spectrum online through a built-in improved intelligent optimization algorithm, and allocates computing resources using a dynamic priority scheduling model to accelerate the damage convergence process.

[0010] Life Prediction and Assessment Module: Based on the damage data output by the digital twin model, combined with the principles of capacity balance and material availability calculation, the module makes a probabilistic prediction of the remaining fatigue life of the seats and generates a test report.

[0011] The data acquisition and operational condition simulation module, which converts the ship's motion sequence into basic motion and inertial loads, specifically includes:

[0012] Based on the position vector of the seat mounting point in the ship coordinate system Calculate the ship's angular velocity With angular acceleration Additional linear acceleration caused ;

[0013] The additional linear acceleration and the acceleration along the ship's center of gravity By superimposing these values, the total motion excitation at the seat mounting point can be obtained. ;

[0014] Based on the total motion excitation and the mass properties of the seat and the occupant, the equivalent six-component force load acting on the seat mounting interface is calculated.

[0015] The core algorithm of the iterative learning and intelligent optimization control module includes:

[0016] An improved hybrid particle swarm optimization-genetic optimization algorithm (HPSO-GA) is used to optimize the application strategy of test loads to induce potential damage at the fastest speed. Its rate update formula incorporates the crossover and mutation concepts of a genetic algorithm into the standard particle swarm optimization algorithm.

[0017]

[0018] in, , The first The particle in the first Velocity and position at the next iteration; Inertial weight; , For learning factors; , It is a random number; For the individual's optimal; It is the global optimum; and These represent crossover and mutation operations, respectively. and This is an adaptive adjustment coefficient, which is related to the number of iterations and the particle aggregation degree.

[0019] Dynamic priority scheduling model: used to manage system computing resources, prioritizing load conditions that have the greatest impact on lifetime prediction. Dynamic priority of each test task The calculation formula is:

[0020]

[0021] In the formula, For this task within the time window Increased damage caused internally; Estimate the time required for the task; Initial weights for the tasks; The current system's allocation of computing resources to this task; This is a factor representing the correlation between tasks; These are weighting coefficients, and .

[0022] The correlation factor between tasks Calculated using the following formula:

[0023]

[0024] in, The total number of tasks. The similarity function represents the time series of damage increments between two tasks, calculated using Pearson correlation coefficient or cosine similarity.

[0025] A material availability-based life prediction method for the structural system: The seat structure is treated as a "production line," and micro-damage is considered as "materials to be assembled." Life is predicted by calculating the "availability" of damage under a given load spectrum. The overall damage availability of the structural system is also considered. Defined as:

[0026]

[0027] in, For the number of key components, Damage mode type (e.g., crack, creep, etc.); This represents the actual cumulative damage. The damage threshold that leads to failure; This represents the weighting coefficient for the damage mode. When When the value approaches 1, the lifespan is considered to have ended. Remaining lifespan Calculated by extrapolation:

[0028]

[0029] in, This represents the current testing duration. This represents the rate of change in the matching rate.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] (1) By combining high-fidelity digital twin model with iterative learning control, the accurate simulation of complex and random marine loads is realized, and the test environment is closer to the real working conditions.

[0032] (2) The improved HPSO-GA algorithm and dynamic priority scheduling significantly accelerated the testing process, quickly located structural weaknesses, and improved testing efficiency.

[0033] (3) An innovative lifetime prediction model based on capacity balance and damage matching rate assesses the damage synergy effect from the perspective of the whole system, and the prediction results are more accurate and reliable.

[0034] (4) The entire system is software-based, which reduces the hardware dependence and cost of traditional physical testing and has high flexibility and scalability. Attached Figure Description

[0035] Figure 1 This is a block diagram of the overall architecture of the system of the present invention.

[0036] Figure 2 This is a flowchart of the iterative learning and intelligent optimization control module.

[0037] Figure 3 This is a schematic diagram of lifetime prediction based on damage fitting rate. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] like Figure 1As shown, this system comprises four main modules. The data acquisition and operational condition simulation module collects six-degree-of-freedom motion and load data through accelerometers, force sensors, and a ship motion attitude instrument deployed at key points of the seat (such as the seat cushion, armrests, and leg joints). Simultaneously, it accesses a marine environmental database to obtain the wave spectrum of the target navigation area (e.g., using the JONSWAP spectrum, with a significant wave height of 2.5 meters and a peak period of 8.5 seconds). This module integrates this information to generate a composite load time series containing low-frequency wave loads, mid-frequency ship vibrations, and high-frequency impacts, which serves as the input boundary conditions for the digital twin model.

[0040] Specific implementation and numerical examples of the data acquisition and operating condition simulation module:

[0041] The core task of this module is to generate a high-fidelity virtual test load environment.

[0042] Data acquisition and ship motion generation:

[0043] Determine the typical sea state of the target navigation area: significant wave height Spectral peak period JONSWAP wave spectrum was used. Modeling.

[0044] Using ship hydrodynamic analysis software, the motion response amplitude operator (RAO) of the target vessel (e.g., a 50-meter-long coastal workboat) under this wave spectrum is calculated. It is assumed that the RAO of the heave motion is obtained at the encounter frequency... The maximum value is obtained at the location. .

[0045] Based on linear wave theory, a time-domain sequence of heave displacement at the ship's center of gravity is synthesized. For example, at a certain moment The heave acceleration of the ship's center of gravity was calculated. Pitch angular acceleration .

[0046] Example of composite load spectrum synthesis calculation:

[0047] Given conditions:

[0048] The position vector of the seat mounting point relative to the ship's center of gravity: (The direction of the stern is positive on the x-axis).

[0049] Ship's center of gravity acceleration: .

[0050] Ship angular velocity: (Mainly pitch angular velocity).

[0051] Ship angular acceleration: .

[0052] Calculate the total excitation acceleration at the seat mounting point:

[0053] First, calculate the acceleration term caused by rotation:

[0054]

[0055] Calculate the first term (caused by angular acceleration):

[0056]

[0057] Calculate the second term (centripetal acceleration):

[0058] First calculate

[0059]

[0060] Recalculate

[0061]

[0062] therefore,

[0063]

[0064] The final total acceleration at the seat mounting point is obtained as follows:

[0065]

[0066] Generating equivalent six-component forces: assuming the total mass of the seat and occupants. The vector of the centroid relative to the mounting point is The inertia tensor is known. According to the Newton-Euler equations, it can be obtained from... The equivalent six-component force acting on the mounting interface is calculated based on the angular motion. This process is completed automatically by the system.

[0067] Load spectrum synthesis: Perform the above calculations on the continuous time series to obtain the basic load. Simultaneously, establish the host vibration (amplitude). ,frequency ) and wave impact (amplitude) Duration probability of occurrence The parameterized model of ) is superimposed on it. The final composite load spectrum is formed. It is divided into multiple load blocks.

[0068] The digital twin model module for marine seats establishes a high-fidelity finite element model in ANSYS or a similar simulation platform. The model includes the seat frame (made of Q235 steel, defined as a nonlinear elastoplastic constitutive model), the shock absorber (defined as a viscoelastic model), and the skin (defined as a composite material model). The model is calibrated through initial static load and modal tests (e.g., applying 1000N of static pressure and measuring the first natural frequency as 25Hz). During testing, real-time strain data from physical sensors (e.g., actual strain at a measuring point is 550με, model prediction is 520με) is compared with the twin model using a Kalman filter algorithm to synchronize the data frequencies, continuously correcting model parameters to ensure mapping accuracy. The aforementioned data synchronization technology is specifically implemented through the Kalman filtering algorithm, and its functions include: ① filtering the noise data collected by physical sensors; ② comparing the filtered real-time strain and acceleration data with the predicted values ​​of the corresponding nodes in the digital twin model; ③ correcting the local stiffness, damping, or boundary condition parameters in the model in real time through state estimation, ensuring that the virtual model and the physical entity are consistent in dynamic response, thereby improving the mapping accuracy of damage prediction.

[0069] This module is built in simulation software, and material parameters (such as the elastic modulus of Q235 steel) are specified. Yield strength (Calibrated through testing)

[0070] During the test, the model received the load spectrum. The stress and strain are calculated, and the damage is updated according to Miner's linear cumulative damage rule or the Coffin-Manson formula. For example, after a weld joint experiences a load block, the cumulative damage increases from... Increase to .

[0071] Iterative learning and intelligent optimization control module (see flowchart) Figure 2 This is the core of the system. Its operational flow and specific numerical examples are as follows:

[0072] 1. Numerical example of the improved HPSO-GA algorithm:

[0073] Optimization objective: Adjust the application order and amplitude scaling factor of the three load blocks (A, B, C). This maximizes the maximum damage increment of the digital twin model within 100 iterations.

[0074] Parameter settings: Number of particles = 4 , , , Maximum iteration .

[0075] Iterative process: Assume the first At the next iteration, the position of particle 1 ,speed Individual optimal Global Optimum .

[0076] Calculate the standard PSO term:

[0077]

[0078]

[0079]

[0080] Sum of standard PSO terms:

[0081] Genetic operation: Assume an arithmetic crossover is performed with particle 2 (weights of 0.5 each), and the crossover result is... Perform Gaussian mutation on itself (standard deviation 0.05), the mutation result is as follows: .

[0082] Coefficient adaptive: The current population is relatively dispersed. , .

[0083] Final speed update:

[0084]

[0085] Location update:

[0086] The system will then generate a new load sequence for testing.

[0087] 2. Numerical example of dynamic priority scheduling model:

[0088] Assume the system has three parallel test tasks (corresponding to different load conditions or model parameters):

[0089] Task 1: Damage increment over the past 5 minutes Estimated remaining time Weight Currently using CPU resources .

[0090] Task 2: , , , .

[0091] Task 3: , , , .

[0092] set up , , Task relevance factor , , .

[0093] Calculate dynamic priority:

[0094]

[0095]

[0096]

[0097] Scheduling decision: Although the incremental damage per unit time for Task 3 is small, its computational resource utilization efficiency ( Task 3 has extremely high priority and a short estimated completion time, therefore it receives the highest priority. The system will prioritize resource allocation for Task 3, and may allocate some computing resources from Task 2 to Task 3.

[0098] 3. Life Prediction and Assessment Module

[0099] This module makes comprehensive predictions based on damage data output from the digital twin model.

[0100] A. Example of calculating damage fitting rate:

[0101] Known data:

[0102] Component 1 (critical solder joint): fatigue damage threshold Weight Plastic deformation threshold Weight .

[0103] Component 2 (Shock absorber): Fatigue damage threshold Weight Plastic deformation threshold Weight .

[0104] Calculation process:

[0105]

[0106]

[0107] B. Example of remaining lifetime prediction:

[0108] Current test duration .

[0109] By monitoring the last 50 hours The value changes, and the fitting rate of change is obtained by fitting. .

[0110] Calculate remaining lifetime:

[0111]

[0112] The system predicts the total lifespan of the seat to be approximately The system will be continuously updated. and The predicted values ​​are dynamically adjusted.

[0113] It should be noted that the damage matching rate model proposed in this invention does not replace the classical fatigue cumulative damage theory (such as the Miner linear cumulative criterion), but rather, based on classical theory, performs weighted normalization and collaborative evaluation of the cumulative damage of each component and each damage mode from a system integration perspective. The actual cumulative damage of each damage mode... The calculation is still based on the Miner criterion or the Coffin-Manson formula, while the kitting rate... This provides an indicator for judging damage progress at the system level as a whole, and is especially suitable for complex structural systems with multi-path and multi-mode damage coupling.

[0114] Example effect:

[0115] This system was used to test a certain type of marine seat. Traditional methods require 600 hours to complete the evaluation. This system, through precise load simulation and algorithm optimization, enables the digital twin model to reach the critical damage state within 240 hours. The system improved efficiency by 150%. The final predicted lifespan was 41,200 hours, with an error of less than ±3% compared to the statistical median of actual service data (approximately 40,000 hours), verifying the accuracy and efficiency of the system.

[0116] The terms "capacity balance" and "material availability rate" used in this paper are originally from the field of production management. In this invention, they are extended to the assessment of structural fatigue damage: "capacity" refers to the ability of a structure to produce damage under load; "material availability rate" refers to the degree to which each damage quantity is "matched" with its failure threshold under multiple damage modes. By calculating the overall damage availability rate Q of the structural system, the coordinated progress of damage in multiple locations and modes can be comprehensively reflected, thereby achieving a more accurate assessment of the overall fatigue life of the system.

[0117] It should be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0118] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A fatigue testing system for marine seats based on digital twin and iterative learning control, characterized in that, include: The data acquisition and operating condition simulation module is used to acquire data on marine environment, ship motion and seat mechanical response, and generate composite load spectra that simulate real operating conditions. The digital twin model module for marine seats constructs a virtual model corresponding to the physical seat, integrating structural dynamics, material constitutive model and damage accumulation model to map the real-time state of the physical seat; The iterative learning and intelligent optimization control module receives feedback from the digital twin model, corrects the test load spectrum online through the built-in improved intelligent optimization algorithm, and uses a dynamic priority scheduling model to allocate computing resources to accelerate testing. The life prediction and assessment module, based on the damage data output by the digital twin model and combined with the principles of capacity balance and material availability calculation, makes a probabilistic prediction of the remaining fatigue life of the seats.

2. The system according to claim 1, characterized in that, The improved intelligent optimization algorithm is an improved hybrid particle swarm optimization-genetic optimization algorithm (HPSO-GA), and its particle velocity update formula is: in, , The first The particle in the first Velocity and position at the next iteration; Inertial weights; , For learning factors; , It is a random number; For the individual's optimal; It is the global optimum; and These represent crossover and mutation operations, respectively. and This is an adaptive adjustment coefficient related to the number of iterations and particle aggregation degree.

3. The system according to claim 1, characterized in that, In the dynamic priority scheduling model, the first... Dynamic priority of each test task The calculation formula is: In the formula, For this task within the time window Increased damage caused internally; Estimate the time required for the task; Initial weights for the tasks; The current system's allocation of computing resources to this task; This is a factor representing the correlation between tasks; These are weighting coefficients, and .

4. The system according to claim 1, characterized in that, In the life prediction and assessment module, a material availability rate calculation method based on capacity balance is used to calculate the overall damage availability rate of the structural system. Defined as: in, For the number of key components, Damage pattern type; This represents the actual cumulative damage. The damage threshold that leads to failure; These are the weighting coefficients.

5. The system according to claim 4, characterized in that, The remaining lifespan Calculated by extrapolation using the following formula: in, This represents the current testing duration. This represents the rate of change in the matching rate.

6. The system according to any one of claims 1-5, characterized in that, The composite load spectrum generated by the data acquisition and operating condition simulation module includes the coupling of wave load, inertial load caused by the six degrees of freedom motion of the ship, and equipment vibration load.

7. The system according to any one of claims 1-5, characterized in that, The marine seat digital twin model module uses data synchronization technology to calibrate the virtual model parameters in real time using feedback data from physical sensors.

8. The system according to claim 1, characterized in that, The data acquisition and operating condition simulation module is specifically used for: The system acquires marine environmental parameters for the target navigation area and generates a six-degree-of-freedom motion time-domain sequence of the ship based on the ship's hydrodynamic model and wave spectrum. Simultaneously, it collects multi-axis mechanical response data at the seat mounting point. Based on coordinate transformation and the Newton-Euler equations, the ship motion sequence is transformed into basic motion and inertial loads acting on the seat structure. Additional dynamic loads generated by the parameterized model of local vibration and impact events are superimposed to synthesize a time-varying composite load spectrum for driving the digital twin model module.

9. The system according to claim 8, characterized in that, The process of converting the ship's motion sequence into basic motion and inertial loads specifically includes: Based on the position vector of the seat mounting point in the ship coordinate system Calculate the ship's angular velocity With angular acceleration Additional linear acceleration caused ; The additional linear acceleration and the acceleration along the ship's center of gravity By superimposing these values, the total motion excitation at the seat mounting point can be obtained. ; Based on the total motion excitation and the mass properties of the seat and the occupant, the equivalent six-component force load acting on the seat mounting interface is calculated.

Images